Interventions for preventing falls in Parkinson's disease
Abstract
Background
Most people with Parkinson’s disease (PD) experience at least one fall during the course of their disease. Several interventions designed to reduce falls have been studied. An up‐to‐date synthesis of evidence for interventions to reduce falls in people with PD will assist with informed decisions regarding fall‐prevention interventions for people with PD.
Objectives
To assess the effects of interventions designed to reduce falls in people with PD.
Search methods
CENTRAL, MEDLINE, Embase, four other databases and two trials registers were searched on 16 July 2020, together with reference checking, citation searching and contact with study authors to identify additional studies. We also conducted a top‐up search on 13 October 2021.
Selection criteria
We included randomised controlled trials (RCTs) of interventions that aimed to reduce falls in people with PD and reported the effect on falls. We excluded interventions that aimed to reduce falls due to syncope.
Data collection and analysis
We used standard Cochrane Review procedures. Primary outcomes were rate of falls and number of people who fell at least once. Secondary outcomes were the number of people sustaining one or more fall‐related fractures, quality of life, adverse events and economic outcomes. The certainty of the evidence was assessed using GRADE.
Main results
This review includes 32 studies with 3370 participants randomised. We included 25 studies of exercise interventions (2700 participants), three studies of medication interventions (242 participants), one study of fall‐prevention education (53 participants) and three studies of exercise plus education (375 participants). Overall, participants in the exercise trials and the exercise plus education trials had mild to moderate PD, while participants in the medication trials included those with more advanced disease. All studies had a high or unclear risk of bias in one or more items. Illustrative risks demonstrating the absolute impact of each intervention are presented in the summary of findings tables.
Twelve studies compared exercise (all types) with a control intervention (an intervention not thought to reduce falls, such as usual care or sham exercise) in people with mild to moderate PD. Exercise probably reduces the rate of falls by 26% (rate ratio (RaR) 0.74, 95% confidence interval (CI) 0.63 to 0.87; 1456 participants, 12 studies; moderate‐certainty evidence). Exercise probably slightly reduces the number of people experiencing one or more falls by 10% (risk ratio (RR) 0.90, 95% CI 0.80 to 1.00; 932 participants, 9 studies; moderate‐certainty evidence).
We are uncertain whether exercise makes little or no difference to the number of people experiencing one or more fall‐related fractures (RR 0.57, 95% CI 0.28 to 1.17; 989 participants, 5 studies; very low‐certainty evidence). Exercise may slightly improve health‐related quality of life immediately following the intervention (standardised mean difference (SMD) ‐0.17, 95% CI ‐0.36 to 0.01; 951 participants, 5 studies; low‐certainty evidence). We are uncertain whether exercise has an effect on adverse events or whether exercise is a cost‐effective intervention for fall prevention.
Three studies trialled a cholinesterase inhibitor (rivastigmine or donepezil). Cholinesterase inhibitors may reduce the rate of falls by 50% (RaR 0.50, 95% CI 0.44 to 0.58; 229 participants, 3 studies; low‐certainty evidence). However, we are uncertain if this medication makes little or no difference to the number of people experiencing one or more falls (RR 1.01, 95% CI 0.90 to 1.14; 230 participants, 3 studies) and to health‐related quality of life (EQ5D Thermometer mean difference (MD) 3.00, 95% CI ‐3.06 to 9.06; very low‐certainty evidence). Cholinesterase inhibitors may increase the rate of non fall‐related adverse events by 60% (RaR 1.60, 95% CI 1.28 to 2.01; 175 participants, 2 studies; low‐certainty evidence). Most adverse events were mild and transient in nature. No data was available regarding the cost‐effectiveness of medication for fall prevention.
We are uncertain of the effect of education compared to a control intervention on the number of people who fell at least once (RR 10.89, 95% CI 1.26 to 94.03; 53 participants, 1 study; very low‐certainty evidence), and no data were available for the other outcomes of interest for this comparisonWe are also uncertain (very low‐certainty evidence) whether exercise combined with education makes little or no difference to the number of falls (RaR 0.46, 95% CI 0.12 to 1.85; 320 participants, 2 studies), the number of people sustaining fall‐related fractures (RR 1.45, 95% CI 0.40 to 5.32,320 participants, 2 studies), or health‐related quality of life (PDQ39 MD 0.05, 95% CI ‐3.12 to 3.23, 305 participants, 2 studies). Exercise plus education may make little or no difference to the number of people experiencing one or more falls (RR 0.89, 95% CI 0.75 to 1.07; 352 participants, 3 studies; low‐certainty evidence). We are uncertain whether exercise combined with education has an effect on adverse events or is a cost‐effective intervention for fall prevention.
Authors' conclusions
Exercise interventions probably reduce the rate of falls, and probably slightly reduce the number of people falling in people with mild to moderate PD.
Cholinesterase inhibitors may reduce the rate of falls, but we are uncertain if they have an effect on the number of people falling. The decision to use these medications needs to be balanced against the risk of non fall‐related adverse events, though these adverse events were predominantly mild or transient in nature.
Further research in the form of large, high‐quality RCTs are required to determine the relative impact of different types of exercise and different levels of supervision on falls, and how this could be influenced by disease severity. Further work is also needed to increase the certainty of the effects of medication and further explore falls prevention education interventions both delivered alone and in combination with exercise.
PICOs
Plain language summary
Interventions for preventing falls in Parkinson's disease
Review Question
In this review we assessed the evidence on the effect of interventions designed to reduce falls in people with Parkinson’s disease (PD). The interventions included exercise, medication, fall‐prevention education and exercise plus education combined. We excluded interventions that aimed to reduce falls due to syncope (e.g. dizziness and fainting). The evidence in this review is current to 16 July 2020.
Background
In people with PD, the emergence of frequent falls is one of the most serious disease milestones. Information about effective fall‐prevention strategies will aid the implementation of fall‐prevention interventions.
Study characteristics
We included 32 randomised controlled trials with 3370 participants. Of these, 25 studies with 2700 participants were exercise trials. Three studies with 242 participants were medication trials. One study with 53 participants was an education trial. Three studies with 375 participants were exercise plus education trials. Overall, the exercise and exercise plus education studies included people with mild to moderate PD.
Key results
Twelve studies compared exercise with a control intervention not thought to reduce falls. Exercise probably reduces the number of falls by around 26%. Exercise probably slightly reduces the number of people experiencing one or more falls by around 10%. Exercise may slightly improve health‐related quality of life immediately after the exercise program. However, we are uncertain if it reduces the number of fall‐related fractures, if it has an effect on the number of adverse events or if it is a cost‐effective intervention for fall prevention.
Three studies compared a cholinesterase inhibitor (either rivastigmine or donepezil) with placebo medication (an inactive treatment) and found that this medication may reduce the rate of falls by around 50%. However, the effect of this medication on the number of people experiencing one or more falls, and on health‐related quality of life was uncertain. Cholinesterase inhibitor medication may increase the number of non fall related adverse events by around 60%. There was no information about the cost‐effectiveness of medication for fall prevention.
One study compared education alone and three studies compared exercise plus education with a control group. Exercise plus education may make little or no difference to the number of people experiencing one or more falls. However, we are uncertain of the effects of these interventions on the other fall and non‐fall outcomes.
Certainty of the evidence
All studies had high or unclear risk of bias in at least one area. This could have influenced how the studies were conducted and how the outcomes were assessed.
For the exercise interventions, the certainty of the evidence for the rate of falls and the number of people experiencing one or more falls was moderate. The certainty of the evidence was low or very low for all other outcomes.
For medication, the education and the exercise plus education interventions, the certainty of the evidence was low to very low for all outcomes.
Authors' conclusions
Summary of findings
| Exercise (all types) compared with control (e.g. usual activities) for preventing falls in people with Parkinson's disease | ||||||
| Patient or population: People with Parkinson's disease Settings: Any Intervention: Exercise of all types Comparison: Control ‐ usual care or a non‐active intervention | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Control | Exercise (all types) | |||||
| Rate of Falls (falls per person‐year) Follow‐up: range 2 weeks to 12 months | All exercise trials population | Rate ratio 0.74 (0.63 to 0.87) | 1456 | ⊕⊕⊕⊝ | Overall, exercise probably reduces the number of falls by 26% (95% CI 37% reduction to 13% reduction).
| |
| 8250 falls per 1000 people
| 6105 falls per 1000 people | |||||
| Number of people who experienced one or more falls Follow‐up: range 2 weeks to 12 months | All exercise trials population | Risk ratio 0.90 (0.80 to 1.00) | 932 | ⊕⊕⊕⊝ | Overall, exercise probably slightly reduces the number of people experiencing one or more falls by 10% (95% CI 20% reduction to no change).
| |
| 634 fallers per 1000 people
| 571 fallers per 1000 people | |||||
| Number of people sustaining one or more fall‐related fractures Follow‐up: range 20 weeks to 12 months | All exercise trials population | Risk ratio 0.57 (0.28 to 1.17) | 989 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the findings that exercise may make little or no difference in the number of people experiencing one or more fall‐related fractures. | |
| 40 people with fracture per 1000
| 23 people with fracture per 1000 | |||||
| Quality of life immediately after the intervention assessed with various measures Follow‐up: range 8 weeks to 6 months A lower score indicates better quality of life | ‐ | The mean quality of life score in the intervention groups was 0.17 standard deviations lower (0.36 lower to 0.01 higher). |
| 951 | ⊕⊕⊝⊝ | Overall, exercise may slightly improve quality of life by 2.6 points in the PDQ39 score (MD = 2.6 lower, 95% CI 5.5 lower to 0.2 higher). Of note is that the 95% CI includes the possibility of both increased and no change in quality of life.
The SMD was converted back to MD using the PDQ39 scale (0‐100), using the pooled SD from the baseline scores of the largest included trial (Chivers Seymour 2019). The MID for the PDQ39 is about 1.6 (Peto 2001).
SMD was calculated from 2 trials using the PDQ39, 1 trial using the PDQ8, 1 trial using the EQ‐5D visual analogue scale and 1 trial using the EQ‐5D index score. |
| Adverse events | Adverse events were reported inconsistently and often only for the exercise group. Three studies reported there were no adverse events related to the exercise intervention and one reported there were no falls during exercise. The remaining four studies reported minor adverse events such as muscle or joint soreness and non‐injurious falls. | Not estimable | 1242 (8 RCTs) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether exercise has an effect on adverse events. | |
| Economic outcomes | We were unable to compare ICERs due to variations in the methods used, however reported ICERs suggest that exercise may be cost‐effective in preventing falls. | Not estimable | 923 (4 RCTs) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether exercise is a cost‐effective intervention for falls prevention. | |
| *The assumed risk is the median control group risk across studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. | ||||||
| aDowngraded due to indirectness as most of the included participants had mild to moderate disease and good cognition. There was no downgrading for risk of bias as most trials had low or unclear risk of bias and the unclear risk of bias (predominantly performance bias and detection bias) unlikely to lower the confidence in the estimation of the effect. bDowngraded due to indirectness as most of the included participants had mild to moderate disease and good cognition. Downgraded by two levels due to imprecision as there was a small number of events and a wide confidence interval. There was no downgrading for risk of bias as most trials had low or unclear risk of bias and the unclear risk of bias (predominantly performance bias and detection bias) unlikely to lower the confidence in the estimation of the effect. cDowngraded by one level due to risk of bias as most trials were at high or unclear risk of bias for performance bias and detection bias as quality of life is a self‐reported measure. Downgraded by a further level due to indirectness as most of the included participants had mild to moderate disease and good cognition. dDowngraded by three levels due to incomplete data.
| ||||||
| Cholinesterase inhibitors compared with placebo medication for preventing falls in people with Parkinson's disease | ||||||
| Patient or population: People with Parkinson's disease Settings: Any Intervention: cholinesterase inhibitor medication (rivastigmine, donepezil) Comparison: placebo medication | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Placebo | Cholinesterase inhibitor | |||||
| Rate of falls (falls per person‐year) Follow‐up: range 12 weeks to 12 months | Cholinesterase inhibitor trial population | Rate ratio 0.50 (0.44 to 0.58) | 229 | ⊕⊕⊝⊝ | Overall, cholinesterase inhibitors may reduce the number of falls by 50% (95% CI 42% reduction to 56% reduction).
| |
| 28,800 falls per 1000
| 14,400 falls per 1000 | |||||
| Number of people who experienced one or more falls Follow‐up: range 12 weeks to 12 months | Cholinesterase inhibitor trial population | Risk rato 1.01 (0.90 to 1.14) | 230 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that cholinesterase inhibitors make little or no difference to the number of people experiencing one or more falls. | |
| 774 fallers per 1000
| 782 fallers per 1000 | |||||
| Number of people sustaining one or more fall‐related fractures Follow‐up: 12 weeks | Reported in one study, with no fractures in either group. | Not estimable | 23 (1 RCT) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether cholinesterase inhibitors make little or no difference to the number of people sustaining one or more fall‐related fractures. | |
| Quality of Life immediately after the intervention (EQ5D Thermometer, scale 0 to 100; and EQ5D Index Score, scale 0‐1, high score is better quality of life) Follow‐up: 8 months | The mean EQ5D thermometer score was 63 and the mean EQ5D Index Score was 0.66 in the placebo group. | In the cholinesterase inhibitor group the mean EQ5D Thermometer Score was 3 points higher (3.06 lower to 9.06 higher) and the mean EQ5D Index Score was 0.01 points lower (0.08 lower to 0.07 higher). |
| 121 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that cholinesterase inhibitors may make little or no difference to health‐related quality of life immediately after the intervention. |
| Rate of adverse events excluding falls (per person year) Follow‐up: range 12 weeks to 8 months
| Cholinesterase inhibitor trial population | Rate ratio 1.60 (1.28 to 2.01) | 175 | ⊕⊕⊝⊝ | Overall, cholinesterase inhibitors may increase the number of non fall‐related adverse events by 60% (95% CI 28% increase to 101% increase).
| |
| 1970 adverse events per 1000
| 3152 adverse events per 1000 | |||||
| Economic outcomes |
|
|
|
| No data reported for this outcome | |
| *The assumed risk is the median control group risk across studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. | ||||||
| aDowngraded by two levels for imprecision due to the relatively small sample size. There was no downgrading for risk of bias as the sensitivity analyses to remove trials at high risk of bias in any item, or high/unclear risk of bias in any domain, made little difference to the result (Table 1). bDowngraded by one level due to risk of bias as results changed when removing the two trials with a high risk of bias in any item (Henderson 2016, Chung 2010) (Table 2). Downgraded an additional two levels due to imprecision because of the relatively small sample size. There was no downgrading for inconsistency as results were essentially unchanged with removal of the comparison responsible for the high heterogeneity (Li 2015a) (Table 2). cDowngraded by two levels for imprecision due to the very small sample size. Downgraded a further one level as only one of the three studies included in the review for this comparison contributed to the outcome. dDowngraded by two levels for imprecision due to the relatively small sample size. Downgraded a further one level as only one of the three studies included in the review for this comparison contributed to the outcome. eDowngraded by two levels for imprecision due to the relatively small sample size. | ||||||
| Sensitivity analysis | Pooled impact of intervention on fall rate, Rate ratio, 95% CI |
| Exercise trials vs control | |
| Primary analysis, all trials, random effects meta‐analysis | 0.74, 0.63 to 0.87; participants = 1456; trials = 12 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.74, 0.61 to 0.90; participants = 1,245; trials = 9 |
| Sensitivity analysis 2, removing trials with unclear or high risk of bias on random sequence generation | 0.90, 0.76 to 1.05; participants = 995; trials = 7 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 0.80, 0.70 to 0.91; participants = 1299; trials = 8 |
| Sensitivity analysis 4, removing trials with unclear or high risk of bias on assessor blinding | 0.92, 0.73 to 1.16; participants = 692; trials = 2 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 0.77, 0.65 to 0.92; participants = 1260; trials = 11 |
| Sensitivity analysis 6, removing trials with less than three months falls monitoring | 0.79, 0.68 to 0.92; participants = 1268; trials = 9 |
| Sensitivity analysis 8, all exercise trials, fixed effects meta‐analysis | 0.79, 0.71 to 0.88; participants = 1456; trials = 12 |
| Primary analysis, subgrouped by exercise type Gait, balance and functional training Resistance training 3D exercise | 0.80, 0.67 to 0.95; participants = 1146; trials = 9 0.72, 0.55 to 0.94; participants = 137; trials = 2 0.41, 0.23 to 0.72; participants = 174; trials = 2 Test for subgroup differences Chi2 = 4.92, df = 2 (P = 0.09), I2 = 59.3% |
| Sensitivity analysis 10, classification of interventions that included functional strength training from resistance training to gait, balance and functional training Gait, balance and functional training Resistance training 3D exercise | 0.78, 0.68 to 0.91; participants = 1244; trials = 10 0.84, 0.28 to 2.53; participants = 38; trials = 1 0.41, 0.23 to 0.72; participants = 174; trials = 2 Test for subgroup differences Chi2 = 4.8, df = 2 (P = 0.09), I2 = 58.3% |
| Medication trials ‐ cholinesterase inhibitor vs placebo | |
| Primary analysis, all trials, fixed effects meta‐analysis | 0.50, 0.44 to 0.58; participants = 229; trials = 3 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.43, 0.32 to 0.56; participants = 81; trials = 1 |
| Sensitivity analysis 2, removing trials with unclear or high risk of bias on random sequence generation | 0.60, 0.38 to 0.96; participants = 129; trials = 1 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 0.60, 0.38 to 0.96; participants = 129; trials = 1 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 0.60, 0.38 to 0.96; participants = 129; trials = 1 |
| Sensitivity analysis 9, all cholinesterase inhibitor trials, random effects meta‐analysis | 0.50, 0.43 to 0.58; participants = 229; trials = 3 |
| Exercise plus education trials vs control | |
| Primary analysis, all trials, random effects meta‐analysis | 0.46, 0.12 to 1.85; participants = 320; trials = 2 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 1.58, 0.74 to 3.40; participants = 124; trials = 1 |
| Sensitivity analysis 4, removing trials with unclear or high risk of bias on assessor blinding | 0.24, 0.10 to 0.61; participants = 196; trials = 1 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 1.58, 0.74 to 3.40; participants = 124; trials = 1 |
| Sensitivity analysis 7, removing the comparison responsible for the high level of heterogeneity (Morris 2017) | 0.24, 0.10 to 0.61; participants = 196; trials = 1 |
| Sensitivity analysis 8, all exercise plus education trials, fixed effects meta‐analysis | 0.54, 0.33 to 0.89; participants = 320; trials = 2 |
| Sensitivity analysis | Pooled impact of intervention on risk of falling, Risk ratio, 95% CI |
| Exercise trials vs control | |
| Primary analysis, all exercise trials, random effects meta‐analysis | 0.90, 0.80 to 1.00; participants = 932; trials = 9 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.87, 0.75 to 1.02; participants = 721; trials = 6 |
| Sensitivity analysis 2, removing trials with unclear or high risk of bias on random sequence generation | 0.89, 0.76 to 1.04; participants = 516; trials = 5 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 0.91, 0.81 to 1.03; participants = 838; trials = 7 |
| Sensitivity analysis 4, removing trials with unclear or high risk of bias on assessor blinding | 0.93, 0.78 to 1.11; participants = 231; trials = 1 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 0.89, 0.79 to 1.00; participants = 736; trials = 8 |
| Sensitivity analysis 6, removing trials with less than three months falls monitoring | 0.89, 0.77 to 1.02; participants = 789; trials = 7 |
| Sensitivity analysis 8, all exercise trials, fixed effects meta‐analysis | 0.90, 0.80 to 1.00; participants = 932; trials = 9 |
| Primary analysis, subgrouped by exercise type Gait, balance and functional training Resistance training 3D exercise | 0.92, 0.81 to 1.04; participants = 622; trials = 6 0.87, 0.43 to 1.74; participants = 136; trials = 2 0.59, 0.36 to 0.95; participants = 174; trials = 2 Test for subgroup differences Chi2 = 3.14, df = 2 (P = 0.21), I2 = 36.2% |
| Sensitivity analysis 10, classification of interventions that included functional strength training from resistance training to gait, balance and functional training Gait, balance and functional training Resistance training 3D exercise | 0.93, 0.83 to 1.05; participants = 720; trials = 7 0.58, 0.30 to 1.13; participants = 38; trials = 1 0.59, 0.36 to 0.95; participants = 174; trials = 2 Test for subgroup differences Chi2 = 5.02, df = 2 (P = 0.08), I2 = 60.1% |
| Medication trials ‐ cholinesterase inhibitor vs placebo | |
| Primary analysis, all trials, fixed effects meta‐analysis | 1.01, 0.90 to 1.14; participants = 230; trials = 3 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.31, 0.12 to 0.78; participants = 81; trials = 1 |
| Sensitivity analysis 2, removing trials with unclear or high risk of bias on random sequence generation | 1.00, 0.87 to 1.15; participants = 130; trials = 1 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 1.00, 0.87 to 1.15; participants = 130; trials = 1 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 1.03, 0.92 to 1.16; participants = 149; trials = 2 |
| Sensitivity analysis 7, removing the comparison responsible for the high level of heterogeneity (Li 2015a) | 1.03, 0.92 to 1.16; participants = 149; trials = 2 |
| Sensitivity analysis 9, all cholinesterase inhibitor trials, random effects meta‐analysis | 0.95, 0.70 to 1.28; participants = 230; trials = 3 |
| Exercise plus education trials vs control | |
| Primary analysis, all trials, random effects meta‐analysis | 0.89, 0.75 to 1.07; participants = 352; trials = 3 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.84, 0.65 to 1.08; participants = 156; trials = 2 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 0.90, 0.75 to 1.08; participants = 320, trials = 2 |
| Sensitivity analysis 4, removing trials with unclear or high risk of bias on assessor blinding | 0.93, 0.73 to 1.19; participants = 228, trials = 2 |
| Sensitivity analysis 8, all exercise plus education trials, fixed effects meta‐analysis | 0.89, 0.75 to 1.07; participants = 352; trials = 3 |
| Health education compared with usual care for preventing falls in people with Parkinson's disease | ||||||
| Patient or population: People with Parkinson's disease Settings: Any Intervention: Education about falls prevention Comparison: Usual care | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Usual care | Health education | |||||
| Rate of falls (falls per person‐year) |
|
|
|
| No data reported for this outcome | |
| Number of people who experienced one or more falls Follow‐up: 12 months | All exercise trials population* | Risk ratio 10.89 (1.26 to 94.03) | 53 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that health education increases the number of people who experience one or more falls. | |
| 634 fallers per 1000 people | 6,911 per 1000 | |||||
| Number of people sustaining one or more fall‐related fractures |
|
|
|
| No data reported for this outcome | |
| Quality of life |
|
|
|
| No data reported for this outcome | |
| Adverse events |
|
|
|
| No data reported for this outcome | |
| Economic outcomes |
|
|
|
| No data reported for this outcome | |
| *The assumed risk is the median control group risk across exercise versus control studies, as there were no data to calculate the illustrative risk in the health education trial. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. | ||||||
| aDowngraded three levels due to risk of bias (single study with high risk of bias for method of ascertaining falls (recall bias) and unclear risk for allocation concealment, performance bias, detection bias, attrition bias and reporting bias). Also downgraded for imprecision due to the relatively small sample size and very wide confidence interval. | ||||||
| Exercise (all types) plus education for falls prevention compared with control (e.g. usual activities) for preventing falls in people with Parkinson's disease | ||||||
| Patient or population: People with Parkinson's disease Settings: Any Intervention: Exercise of all types plus fall‐prevention education Comparison: Control ‐ usual care or a non‐active intervention | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Control | Exercise plus education | |||||
| Rate of Falls (falls per person‐year) Follow‐up: 12 months | Exercise plus education trials population | Rate ratio 0.46 (0.12 to 1.85) | 320 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that exercise plus education makes little or no difference to the number of falls. | |
| 16,400 falls per 1000 people | 7,544 per 1000 | |||||
| Number of people who experienced one or more falls Follow‐up: range 6 months to 12 months | Exercise plus education trials population | Risk Ratio 0.89 (0.75 to 1.07) | 352 | ⊕⊕⊝⊝ | Overall, exercise plus education may make little or no difference to the number of people experiencing one or more falls (11% reduction (95% CI 25% reduction to 7% increase)).
| |
| 672 per 1000 | 598 per 1000 | |||||
| Number of people sustaining one or more fall‐related fractures Follow‐up: 12 months | Exercise plus education trials population | Risk ratio 1.45 (0.40 to 5.32) | 320 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that exercise plus education makes little or no difference to the number of people experiencing one or more fall‐related fractures. | |
| 25 per 1000 | 36 per 1000 | |||||
| Quality of life immediately after the intervention assessed with the PDQ39 (range 0 to 100) Follow‐up: 6 weeks A lower score indicates better quality of life | ‐ | The mean PDQ39 in the intervention groups was |
| 305 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that exercise plus education makes little or no difference to health‐related quality of life immediately after the intervention. |
| Adverse events | Adverse events related to the exercise intervention only were reported. One study reported there were no adverse events, while the other reported minor adverse events such as muscle soreness and a fall while exercising. | Not estimable | 343 (2 RCTs) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether exercise plus education has an effect on adverse events. | |
| Economic Outcomes | Costs per fall prevented were not calculated as there was no reduction in falls in this study | Not estimable | 133 (1 RCT) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether exercise plus education is a cost‐effective intervention for falls prevention. | |
| *The assumed risk the median control group risk across studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. | ||||||
| aDowngraded by three levels due to risk of bias as results changed when removing the study with a high risk of bias on assessor blinding (Morris 2017) and for inconsistency due to a high level of heterogeneity, with the result changed when the comparison responsible for the high heterogeneity (Morris 2017) was removed (Table 1). Downgraded for imprecision due to the wide confidence interval and small sample size and indirectness as most of the included participants had mild to moderate disease and good cognition. Additionally downgraded as the result changed when fixed effects analysis was used (Table 1). bDowngraded one level for imprecision due to the relatively small sample size and an additional level for indirectness as most of the included participants had mild to moderate disease and good cognition. There was no downgrading for risk of bias as the sensitivity analyses to remove trials at high risk of bias in any item, or high/unclear risk of bias in any domain, made little difference to the result (Table 2). cDowngraded by two levels for imprecision due to the relatively small sample size, the small number of events and the very wide confidence interval. Downgraded a further level for indirectness as most of the included participants had mild to moderate disease and good cognition. dDowngraded by one level due to risk of bias as the studies included for this outcome were at unclear risk of bias for performance bias and high risk of bias for detection bias as quality of life is a self‐reported measure. Downgraded by one level for imprecision due to the relatively small sample size and wide confidence interval. Downgraded a further level for indirectness as most of the included participants had mild to moderate disease and good cognition. eDowngraded by three levels due to incomplete data and serious risk of bias from reporting bias. fDowngraded by two levels for imprecision due to the small sample size. Downgraded a further level for indirectness as most of the included participants had mild to moderate disease and good cognition. Downgraded a further one level as only one of the three studies included in the review for this comparison contributed to the outcome. | ||||||
Background
Description of the condition
People with Parkinson’s disease (PD) fall frequently and recurrently with approximately 60% of individuals falling each year and two thirds of these people falling recurrently (Allen 2013; Bloem 2001; Latt 2009; Paul 2013; Pickering 2007). These rates are double those reported for the general older population (Sherrington 2019). In addition, falls in people with PD are associated with injury (Paul 2017; Walker 2013; Wielinski 2005), with the incidence of hip fracture reported to be two (Kalilani 2016) to four times (Walker 2013) that of older people of the same age without PD. It is not surprising that falls are associated with escalating healthcare costs (Paul 2017; Pressley 2003), and are major contributors to reduced health‐related quality of life (Rascol 2015; Soh 2011).
A large number of fall risk factors have been identified in people with PD (Canning 2014; Fasano 2017) . Consistently identified risk factors include a history of past falls (Allcock 2009; Latt 2009; Paul 2013; Pickering 2007); disease severity (Allcock 2009; Kerr 2010; Latt 2009; Paul 2013; Pickering 2007), which are fixed and not remediable. However, a number of risk factors which contribute to loss of balance and falls have the potential to be modified with exercise or pharmaceutical interventions (Allen 2011; Fasano 2017; Shen 2016; Tomlinson 2013), which may in turn reduce falls. These include: freezing of gait (i.e. an episodic inability to initiate or continue walking) (Kerr 2010; Latt 2009; Paul 2013); balance deficits, mobility impairments and lower limb muscle strength deficits (Kerr 2010; Latt 2009; Paul 2013); fear of falling (Mak 2009), and cognitive deficits (Allcock 2009; Latt 2009; Paul 2013). While falls are commonly monitored as adverse events in intervention trials (Nieuwboer 2007; van Nimwegen 2013), only recently have interventions designed primarily to reduce falls in people with PD been developed and investigated (e.g. Canning 2015a; Chivers Seymour 2019; Li 2012; Mirelman 2016; Morris 2015).
Description of the intervention
Interventions designed to reduce falls in people with PD include exercise and/or movement strategy training, pharmacological and/or surgical management, increasing knowledge about fall prevention (education), environmental modifications, assistive technology, management of urinary incontinence, fluid or nutrition therapy, psychological interventions, social environment, and any other intervention designed to reduce falls in this population. Interventions are classified as single interventions (e.g. exercise), multiple interventions (e.g. exercise plus environmental modifications) or multifactorial interventions (i.e. multiple interventions tailored to the individual's identified risk factors).
How the intervention might work
Each intervention type is designed to target specific, potentially remediable fall risk factors. Exercise interventions aim to reduce falls by targeting physical and/or cognitive risk factors, including poor balance, reduced muscle strength and freezing of gait (Canning 2014; Mirelman 2016). Cholinesterase inhibitors address the central nervous system (CNS) cholinergic neuron loss associated with PD and may reduce falls by enhancing cognitive and attentional resources (Chung 2010), and/or reducing gait variability contributing to falls (Henderson 2016). Education interventions aim to increase awareness of the risk of falls and may include behaviour modification to avoid high‐risk activities (Stack 2013), while environmental modifications focus on reducing environmental hazards, such as poor lighting, or slippery surfaces (Bhidayasiri 2015).
Why it is important to do this review
Recently, a number of large‐scale randomised controlled trials and several smaller trials specifically testing interventions designed to reduce falls in people with PD have been published. In addition, participants with PD are excluded from the Cochrane Reviews of interventions for preventing falls in older people living in the community (Hopewell 2018; Sherrington 2019). Further, while falls as an outcome is addressed in Cochrane Reviews of physiotherapy interventions for PD (Tomlinson 2013; Tomlinson 2014), these reviews do not differentiate between physiotherapy interventions primarily designed to reduce falls versus other interventions. In addition, the scope of the physiotherapy reviews is limited to physical interventions. Therefore, there is a need to systematically review the literature to identify trials of all interventions aimed at reducing falls in people with PD and summarise this evidence for people with PD, clinicians, researchers and policymakers.
Objectives
To assess the effects of interventions designed to reduce the incidence of falls in people with Parkinson's disease (PD).
Methods
Criteria for considering studies for this review
Types of studies
We included randomised controlled trials (RCTs) and quasi‐randomised trials, including cluster‐ and cross‐over trials, evaluating the effects of interventions on falls in people with PD. Eligible randomised cross‐over trials of exercise interventions had the first phase data only included in order to minimise the risk of carry‐over effects of the interventions. For eligible randomised cross‐over trials of medication interventions we included data from both phases as washout phases ensured no carry‐over effects. We did not include studies published only in abstract form.
Types of participants
We included trials of participants with idiopathic PD who had been diagnosed by the UK Parkinson’s Disease Brain Bank criteria (Hughes 1992), or by a clinical definition. No restrictions were made with regard to gender, age or disease duration. We included studies reporting an intervention carried out in a mixed sample of participants, including people with idiopathic PD, if separate data were available for participants with idiopathic PD.
Types of interventions
We included interventions where a stated primary or secondary aim was to reduce falls in people with PD. Therefore, any intervention which did not have a stated aim of preventing falls, and which reported falls as an adverse event, was not included. We did not include interventions designed to primarily address syncopal falls (e.g. falls associated with neurogenic postural hypotension) as the aetiology and intervention for syncopal falls are different from falls arising from loss of balance due to physical, cognitive and emotional risk factors associated with PD (Fasano 2012; van der Marck 2014). We included studies where a fall‐prevention intervention was compared with ‘usual care’ (i.e. no change in usual activities or treatments), a ‘placebo’ or other control intervention (i.e. an intervention not thought to have an effect on falls, such as very gentle or 'sham' exercise), or another fall‐prevention intervention.
We grouped interventions using the fall‐prevention classification taxonomy developed by the Prevention of Falls Network Europe (ProFaNE) (Lamb 2011). Interventions were classified according to intervention type: exercises, medication (drug target, i.e. withdrawal, dose reduction or increase, substitution, provision), surgery, management of urinary incontinence, fluid or nutrition therapy, psychological interventions, environment/assistive technology, social environment, interventions to increase knowledge (education), or other interventions. Interventions were also classified according to combination of intervention types: single, multiple (more than one intervention type) or multifactorial (more than one intervention type specifically targeting person‐specific fall risk factors). Full details are available in the ProFaNE Taxonomy Manual (Lamb 2011).
We used the ProFaNE taxonomy (Lamb 2011) to categorise exercise types. Exercise categories were: i) gait, balance and functional training; ii) resistance training (including muscle power training); iii) flexibility exercise; iv) 3D exercise (e.g. Tai Chi); v) general physical activity; vi) endurance exercise, and vii) other forms of exercise (including where the exercise was not described in sufficient detail to allocate a category) (Table 3).
| Exercise Category | ProFaNE exercise description | How the criteria were applied in this review* |
| Gait, balance and functional training | Gait training involves specific correction of walking technique (e.g., posture, stride length and cadence) and changes of pace, level and direction. Balance training involves the efficient transfer of bodyweight from one part of the body to another or challenges specific aspects of the balance systems (e.g. vestibular systems). Balance retraining activities range from the re‐education of basic functional movement patterns to a wide variety of dynamic activities that target more sophisticated aspects of balance. Functional training utilises functional activities as the training stimulus, and is based on the theoretical concept of task specificity. All gait, balance and functional training should be based on an assessment of the participant's abilities prior to starting the program; tailoring of the intervention to the individuals abilities; and progression of the exercise program as ability improves. | Selected as the primary exercise category when the majority of the exercise was conducted in standing and when the intervention focus and the majority of time spent was on exercise in this category. Movement strategy training and cueing are included in this category. |
| Resistance training | The term Resistance Training covers all types of weight training.i.e contracting the muscles against a resistance to overload and bring about a training effect in the muscular system. The resistance is an external force, which can be ones own body placed in an unusual relationship to gravity (e.g. prone back extension) or an external resistance (e.g. free weight). All strength/resistance training should be based on an assessment of the participant's abilities prior to starting the program; tailoring of the intervention to the individuals abilities; and progression of the exercise program as ability improves. | Selected as the primary category for interventions where additional resistance was used or where it was clear that overload was sufficient without external resistance and where the intervention focus and the majority of time spent was on exercise in this category. |
| Flexibility | Flexibility training is the planned process by which stretching exercises are practised and progressed to restore or maintain the optimal Range Of Movement (ROM) available to a joint or joints. The ranges of motion used by flexibility programs may vary from restoration/maintenance of the entire physiological range of motion, or alternatively, maintenance of range that is essential to mobility or other functions. | Selected as the primary category for interventions where flexibility training was a stated aim of the intervention and where the intervention focus and the majority of time spent was on exercise in this category. |
| 3D | 3D training involves constant movement in a controlled, fluid, repetitive way through all 3 spatial planes or dimensions (forward and back, side to side, and up and down). Tai Chi and Qi Gong incorporate specific weight transferences and require upright posture and subtle changes of head position and gaze direction. Dance involves a wide range of dynamic movement qualities, speeds and patterns. | Selected as the primary exercise category where the intervention focus and the majority of time was spent on exercise in this category (e.g., Tai Chi or dance). |
| General Physical activity | Physical activity is any bodily movement produced by skeletal muscle contraction resulting in a substantial increase in energy expenditure. Physical activity has occupational, transportation and recreational components and includes pursuits like golf, tennis and swimming. It also includes other activities and pastimes like gardening, cutting wood and carpentry. Physical activity can provide progressive health benefits and is a catalyst for improving health attitudes, health habits and lifestyle. Increasing habitual physical activity should be with specific recommendations as to duration, frequency and intensity if a physical or mental health improvement is indicated. | Selected as the primary category where the intervention focus and the majority of time was spent on exercise in this category (e.g. unstructured physical activity, including unstructured waking). |
| Endurance | Endurance training is aimed at cardiovascular conditioning and is aerobic in nature and simultaneously increases the heart rate and the return of blood to the heart. | Selected as the primary category for interventions where the intervention focus and the majority of time spent was on structured aerobic training (e.g. exercise with a target heart rate range). |
| Other | Other kind of exercises not described. | Selected as the primary category if the intervention did not meet the other categories listed and where the intervention focus and the majority of time was spent in this category. This category included interventions where the exercise was not described in sufficient detail to allocate a category. |
*Interventions were allocated primary categories using categorisation based on Sherrington 2019.
Types of outcome measures
We included studies that reported the rate or number of falls, or the number of participants experiencing at least one fall during the follow‐up. We included studies that recorded falls either prospectively or retrospectively.
Primary outcomes
-
Rate (number) of falls
-
Number of people who fell at least once (i.e. the number of fallers)
Secondary outcomes
-
Number of participants sustaining one or more fall‐related fractures
-
Quality of life
-
Rate (number) and type of adverse events (excluding falls)
-
Economic outcomes
Adverse events were only included in meta‐analyses when they were monitored using the same methods in all groups over the entire study period. We used the rate of adverse events excluding falls, as the rate of falls is presented separately in the analyses.
Timing of outcome measurement
One time point from each study was used for the primary outcomes. Where studies reported outcomes measured at multiple time periods, we used the longest time period available unless outcomes were monitored for over 12 months, in which case we used results reported at 12 months if these were available. We chose a 12‐month limit as nearly all fall studies in PD measure falls for 12 months or less. Where studies reported falls data for different time periods, we combined the data for the different time periods when possible. If this was not possible, we used the data from the time period closest to the end of the intervention period. For the quality of life outcomes, we used data from immediately after the end of the intervention, and data from follow‐up at a later time in separate analyses.
Search methods for identification of studies
Electronic searches
We performed searches up until the 16 July 2020 and conducted a top‐up search on the 13 October 2021. Studies identified in the top‐up search were added to 'Studies awaiting classification.' We searched the Cochrane Movement Disorders Group Trial Register and the Cochrane Central Register of Controlled Trials (CENTRAL, in The Cochrane Library; 2021, issue 11), MEDLINE (OvidSP from 1946), Embase (OvidSP from 1947), CINAHL (Cumulative Index to Nursing and Allied Health Literature) (EBSCO from 1982), PsycINFO (OvidSP from 1806), AMED (OvidSP from 1985), and the Physiotherapy Evidence Database (PEDro)(The University of Sydney, https://pedro.org.au/).
The full search strategy for each database can be found in Appendix 1.
Searching other resources
To identify any further published or ongoing trials, we:
-
searched trial registers: ClinicalTrials.gov (http://clinicaltrials.gov/), and the World Health Organization's International Clinical Trials Registry Platform Search Portal (http://apps.who.int/trialsearch/) (January 20, 2022) (see Appendix 1);
-
checked reference lists of relevant articles;
-
contacted trialists and researchers in the field;
-
used Science Citation Index Cited Reference Search;
-
checked studies included in the Cochrane Review of interventions for preventing falls in older people living in the community (Gillespie 2012; Hopewell 2018; Sherrington 2019) and the Cochrane Review of interventions for preventing falls in older people in care facilities and hospitals (Cameron 2018) for any trial which includes a subgroup of people with PD.
We did not apply any language restrictions.
Data collection and analysis
The intended methods for data collection and analysis for this review are published in our protocol (Canning 2015b). These are based on the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Selection of studies
Review authors CC and NL separately screened the search results (title, abstract and descriptors) to identify studies for possible inclusion. Trial register results were excluded at this stage and searched separately through the trials registries as previously described. Any study that either researcher identified for possible inclusion was progressed to full‐text screening. CC and NL then separately assessed the eligibility of studies based on full text. Where a researcher involved in selecting studies was an author of a potentially eligible study, review author AN replaced them to assess the eligibility of that study. Again, disagreements were resolved through discussion or third‐party adjudication. Study authors were contacted for additional information if necessary.
Data extraction and management
Information for the included studies' table was extracted by pairs of review authors (LA, NA and TY).
Review authors NA and GV independently extracted data using a pre‐tested data extraction form (based on the one used in Sherrington 2019). Disagreement was resolved by consensus or third‐party adjudication. Review authors were not blinded to authors or sources.
The following information was collected.
-
General information: review author's name, study ID and first author of study.
-
Study details: study design and interventions, sample size, baseline fall rates, number of dropouts, cluster randomisation.
-
Rate of falls, number of people experiencing one or more falls, number of people experiencing one or more fall‐related fractures, rate and type of adverse events, quality of life, and cost and cost‐effectiveness information related to fall outcomes. Where data were provided in graphical form, we used the software program Web‐PlotDigitizer to extract the data (WebPlotDigitizer 2020).
We collected data from full‐text journal articles. Where a study had more than one journal article published, we consulted all articles for details. Where there was insufficient information reported, we contacted the study authors, requesting additional details.
Assessment of risk of bias in included studies
Pairs of review authors (NA, SK and NL) independently assessed risk of bias using the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017) and using a pre‐tested risk of bias assessment form. Review authors were not blinded to author and source institution. Review authors did not assess their own studies. Disagreement was resolved by consensus or third‐party adjudication.
We assessed the following domains, using the criteria developed by Gillespie 2012 for judging risk of bias in fall‐prevention trials (as outlined in Table 4): random sequence generation (selection bias); allocation concealment (selection bias); blinding of participants and personnel (performance bias); blinding of outcome assessment (detection bias) for falls and the number of people who fell at least once, and for fractures separately; incomplete outcome data (attrition bias) for falls and the number of people who fell at least once separately, and selective outcome reporting bias. We assessed bias in the recall of falls due to unreliable methods of ascertainment (Hannan 2010). We also used the specific criteria for assessing attrition bias in falls trials developed by Gillespie 2012 (Appendix 2). Additionally, we assessed the trials for any other potential sources of bias.
| Domain | Criteria for judging risk of bias |
| Random sequence generation: selection bias (biased allocation to interventions) due to inadequate generation of a randomised sequence | • Judgement of ’low risk’ if the trial authors described a random component in the sequence generation, e.g. referring to a random number table; using a computer random number generator; coin tossing; shuffling cards or envelopes; throwing dice; drawing of lots; minimisation. • Judgement of ’high risk’ if the trial used a systematic nonrandom method, e.g. date of admission; odd or even date of birth; case record number; clinician judgement; participant preference; patient risk factor score or test results; availability of intervention. • Judgement of ’unclear risk’ if there is insufficient information about the sequence generation process to permit judgement of ’low risk’ or ’high risk’. |
| Allocation concealment: selection bias (biased allocation to interventions) due to inadequate concealment of allocations prior to assignment | • Judgement of ’low risk’ in studies using: ◦ cluster randomisation if allocation of all cluster units performed at the start of the study and individual participant recruitment was completed prior to assignment of the cluster, and the same participants were followed up over time or individual participants were recruited after cluster assignment, but recruitment carried out by a person unaware of group allocation and participant characteristics (e.g. fall history) or individual participants in intervention and control arms were invited by mail questionnaire with identical information. • Judgement of ’high risk’ in studies using: • Judgement of ’unclear risk’ if insufficient information to permit judgement of ’low risk’ or ’high risk’. This is usually the case if the method of concealment is not described or not described in sufficient detail to allow a definite judgement, e.g. if the use of assignment envelopes is described, but it remains unclear whether envelopes were sequentially numbered, opaque and sealed. |
| Blinding of participants and personnel: performance bias due to knowledge of the allocated interventions by participants and personnel carrying out the interventions | • Judgement of ’low risk’ if blinding of participants and personnel implementing the interventions was ensured, and unlikely that the blinding could have been broken. |
| Blinding of outcome assessment: detection bias due to knowledge of the allocated interventions by outcome assessors | • Falls, fallers: • Fractures: |
| Incomplete outcome data: attrition bias due to amount, nature or handling of incomplete outcome data | • Judgement of ’low risk’ if there are no missing outcome data, or less than 20% of outcome data are missing and losses are balanced in numbers across intervention groups with similar reasons for missing data across groups or missing data have been imputed using appropriate methods. |
| Selective reporting: reporting bias due to selective outcome reporting | • Judgement of ’low risk’ if the study protocol is available (i.e., published protocol or trial registry) and all prespecified study outcomes are reported in the prespecified way or the study protocol is unavailable, but it is clear the published report includes all expected outcomes. |
| Method of ascertaining falls: bias in the recall of falls due to unreliable methods of ascertainment | • Judgement of ’low risk’ if the study used some form of concurrent collection of data about falling, e.g. participants given postcards to fill in daily and mail back monthly, calendar to mark monthly, or more frequent, follow‐up by the researchers. |
We adapted this from Table 8.5.a 'The Cochrane Collaboration's tool for assessing risk of bias’ and Table 8.5.d 'Criteria for judging risk of bias in the 'Risk of bias’ assessment tool’ (Higgins 2017) and from Sherrington 2019.
We rated the risk of bias in each domain as high, low or unclear.
Measures of treatment effect
We reported treatment effect for rate of falls and rate of adverse events as a rate ratio (RaR) and 95% confidence interval (CI). The RaR compares the rate of events (falls or adverse events) between two groups in any given trial, where rate of events is the total number of events per unit of person time that events were monitored (e.g. falls per person year). If the RaR was reported in the included trial (e.g. incidence rate ratio or hazard ratio (HR)), we used the reported values. If both adjusted and unadjusted RaRs were reported, we used the unadjusted RaR, unless the adjustment was for clustering. If a RaR was not reported, but appropriate raw data were available, we used Excel to calculate a RaR and 95% CI. To do this, we used the reported rate of events (per person year) in each group or the reported total number of events in each group. If the rate of events in each group was not reported, where possible we calculated this as events per person year from the total number of events in that group, the length of time events were monitored and the number of participants contributing to the data. If there were no participants lost to follow‐up, or data were only available for participants completing the study, we assumed that participants' data had been collected for the maximum possible period of time.
It is possible that individual multiple fallers may have excessive influence on the rate of falls results. To investigate this possibility, we recorded procedures used by investigators to decrease this influence, such as randomisation stratified by fall history or analyses adjusted for previous falls. We also extracted baseline falling rates for each group (where available).
For the number of people who fell at least once and number of participants experiencing fall‐related fractures, we reported a risk ratio (RR) and 95% CI. The RR compares the number of people experiencing events (i.e. participants who fell once or more, or participants who experienced one or more fall‐related fractures) between groups. If the RR and 95% CI was reported (including relative risk, HR for first fall or odds ratio (OR)), we used the reported values. If both adjusted and unadjusted RRs were reported, we used the unadjusted RR, unless the adjustment was for clustering. If a RR was not reported, but data were available to calculate the relative risk and 95% CI, then this was calculated using the calculator function in RevMan 5.4. For these calculations, we used the number of participants reported contributing data to each group. If the number of participants contributing data was not known, we used the number randomised to each group.
Quality of life was reported as a continuous outcome. For these data we calculated mean differences (MD) with 95% CIs where data using one measurement were pooled, or standardised mean differences (SMD) and 95% CIs where data using different outcome measures were pooled. Where study authors reported median and interquartile range (IQR), the mean and standard deviation (SD) was estimated by review authors. For studies with smaller sample size (e.g. 40 participants), this was conducted using the technique described by Wan 2014. For larger trials (e.g. over 100 participants), this was conducted using the technique described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017).
Where comprehensive economic evaluations were incorporated in the included studies, we reported the incremental cost per fall prevented and/or per quality‐adjusted life‐year (QALY) gained by the intervention compared with the comparator group, as stated by the authors. We also extracted from studies reporting a cost analysis or cost description, the type of resource use (e.g. delivering the intervention, hospital admissions, outpatient visits) and the intervention and healthcare service costs per participant in each group.
Unit of analysis issues
We incorporated studies with more than one intervention arm compared with a control group, and therefore needed to avoid 'double‐counting' of control participants from these studies in any one meta‐analysis. To achieve this, each intervention was included in a separate comparison. For the RaRs and RRs, the standard errors (SEs) of the natural log of the between‐group difference were increased by 25% and participant numbers in the control group were allocated in proportion to the participant numbers in each intervention arm. For example, if a study had 70 participants in exercise group A, 70 in exercise group B and 70 in a control group, the SEs of the natural log of the between‐group difference in the exercise A versus control and exercise B versus control were increased by 25% and the number of control participants was shown as 35 in each comparison. For the continuous data (i.e. quality of life), the number of participants in the control group was divided equally among the comparisons and the control mean and SD were unchanged (Higgins 2017).
Data from cluster‐randomised trials were adjusted for clustering (Higgins 2017), if this had not already been done by the trial authors. If no estimate of the intra‐class correlation coefficient (ICC) was available, we used an ICC of 0.01 as reported by Smeeth 2002.
Dealing with missing data
We provided an overview of missing data from our selected studies in raw data tables. We did not use a cut‐off for missing data as an inclusion criterion. When outcome data were not reported, we contacted the study authors to request the data. We addressed the potential impact of missing data in the assessment of risk of bias.
Assessment of heterogeneity
We performed meta‐analyses when we considered study interventions to be similar enough to pool results. We assessed heterogeneity of these meta‐analyses by visual inspection of forest plots, as well as considering both the Chi2 test (with statistical significance set at P < 0.10) and the I2 statistic. We interpreted the I2 statistic according to Higgins 2017 who suggested: 0% to 40% may not be important; 30% to 60% may indicate moderate heterogeneity; 50% to 90% may indicate substantial heterogeneity; and 75% to 100% may indicate considerable heterogeneity. We performed subgroup analyses to determine whether heterogeneity was explained by study and/or participant characteristics.
Assessment of reporting biases
We minimised reporting bias by comprehensively searching multiple databases, searching for studies in languages other than English, and searching the grey literature and trial registries. We observed funnel plots for outcomes with more than 10 data points and considered reporting bias when using the GRADE approach to inform the certainty of the evidence in the summary of findings tables.
Data synthesis
We performed separate analyses to pool results of studies comparing an active fall‐prevention intervention with either ‘usual care’ or a ‘placebo’ control intervention, and studies comparing two active fall‐prevention interventions. We grouped similar intervention types together using the fall‐prevention classification taxonomy for intervention descriptors developed by ProFaNE (Lamb 2011). Furthermore, similar exercise interventions were grouped together according to ProFaNE exercise categories (Lamb 2011) (Table 3). Where meta‐analyses were appropriate (i.e. studies with comparable interventions and participant characteristics), we pooled results using fixed‐effect models, except where the review authors felt that it was unlikely that there would be a single true effect of the intervention on falls (i.e. exercise interventions and exercise plus education interventions), in which case random‐effects models were used. We considered it to be inappropriate to perform meta‐analyses where two active fall‐prevention interventions were compared. When meta‐analyses were not performed, trial‐level data are presented in forest plots and tables and narrative reviews are provided.
Where appropriate, pooled RaRs (for falls and adverse events) and pooled RRs (number of people who fell at least once and number of people sustaining one or more fall‐related fractures) were calculated using the generic inverse variance method in Review Manager software (RevMan 5.4). This involves entering the natural logarithm of the RaR or RR and its SE for each study. These values were calculated using Excel with the method developed for the Gillespie and colleagues Cochrane Review of interventions to prevent falls (Gillespie 2012).
The continuous quality of life outcomes were presented as MDs where one outcome measure was pooled, or SMDs where different outcome measures were pooled. Where SMDs were presented, the SMD was converted back to an MD in the summary of findings tables. This was done for the most commonly used outcome measure, using the SD from the baseline scores of the largest included study.
Subgroup analysis and investigation of heterogeneity
We performed subgroup analyses to determine whether intervention impacts on primary outcomes varied according to baseline level of fall risk (increased fall risk due to previous fall or specified high fall risk versus fall risk not specified), or disease severity. For exercise trials, subgroup analysis was undertaken for the type of exercise (ProFaNE exercise category) and the proportion of exercise that was supervised.
For the subgroup analyses on disease severity, we extracted and pooled subgroup data from included studies that reported results by disease severity subgroups, and pooled these data using random‐effects meta‐analyses. This was because we were unable to categorise studies based on disease severity as most studies used populations with a range of disease severity and used different definitions of disease severity.
We used the random‐effects model to pool data in all analyses testing for subgroup differences due to the high risk of false‐positive results when comparing subgroups in a fixed‐effect model (Higgins 2017). We used the test for subgroup differences available in RevMan 5.4 to determine whether there was evidence of a difference in treatment effects between subgroups.
Sensitivity analysis
We performed sensitivity analyses to explore the impact of risk of bias on pooled estimates of treatment effect for the primary outcomes. We removed studies from pooled analyses if they were assessed as having high risk of bias in any item, or as having high or unclear risk of bias in a key domain: random‐sequence generation (selection bias), allocation concealment (selection bias), blinding of outcome assessors (detection bias), and incomplete outcome data (attrition bias) (see Higgins 2017). We performed a sensitivity analyses to explore the impact of fall monitoring time by removing studies from pooled analyses that monitored falls for less than three months. Additionally, we performed sensitivity analyses on comparisons with a high heterogeneity (I2 > 50%) by removing the studies that were responsible for the high levels of heterogeneity. We explored the impact of the model of meta‐analysis chosen by performing sensitivity analyses using fixed‐effect rather than random‐effects analyses on the exercise versus control and exercise plus education versus control studies and using random effects rather than fixed effects analyses on the cholinesterase inhibitor versus placebo studies. Additionally, we considered there was some subjectivity in the classification of exercise categories, so we performed a sensitivity analysis where studies that utilised functional strength training (e.g. using body weight, weighted vests and/or ankle weights) were re‐classified from resistance exercise to gait, balance and functional training.
Summary of findings and assessment of the certainty of the evidence
Summary of findings tables were prepared for each comparison where interventions were compared with control or placebo interventions. The certainty of the evidence in these tables for all outcomes where meta‐analyses had been conducted was assessed using the GRADE approach (Schűnemann 2013), utilising GRADEpro GDT (GRADEPro GDT 2015). This approach categorises the certainty of the evidence as high, moderate, low or very low depending on the evaluation of five factors: risk of bias; inconsistency of the effect; indirectness; imprecision; and publication bias. The certainty of the evidence and effect size were then used to determine the appropriate standardised statements to describe the certainty of the evidence (Cochrane Norway 2017). Decisions regarding whether to downgrade the evidence are described in the footnotes of the summary of findings tables.
Results
Description of studies
Results of the search
A flow diagram of the study selection process is shown in Figure 1. A total of 3459 records were downloaded, with the number from each database as follows: Cochrane Movement Disorders Group Trial Register and CENTRAL (663); MEDLINE (687), Embase (1665), CINHAL (174), PsycINFO (159), AMED (40) and PEDro (71).
Study flow diagram.
a Ashburn 2019 was identified through contacting researchers in the field.
Following removal of duplicates, we screened the abstracts and titles of 2752 papers, resulting in 156 full‐text papers being considered. From these we removed 89 papers, leaving 67 reports of 36 studies. We contacted the authors of four studies (one with two reports (Hill 2015)) to request additional information regarding eligibility of the study (Hill 2015; Kurlan 2015; Sparrow 2016; Thaut 2019). We received responses from three (Hill 2015; Sparrow 2016;Thaut 2019). Three studies were excluded from the review (Hill 2015; Kurlan 2015; Sparrow 2016). Subsequently, a fourth study was excluded (Sato 2011) as the integrity of the data has been questioned (Bolland 2016) and the publication has been retracted by the journal. Information about the excluded studies is in the Characteristics of excluded studies. Consequently, there were 32 studies reported in 62 articles in the review. A flow‐diagram of the study selection process is in Figure 1.
Following the 'top‐up' search on 13 October 2021, an additional two eligible trials were identified. One trialled peroneal nerve functional electrical stimulation and the other trialled perturbation training. These have been added to the "Studies awaiting classification."
Included studies
This review includes 32 studies with 3370 participants randomised. There were 29 studies of a single intervention and three studies of multiple interventions. In the single intervention studies there were 25 studies of exercise (2700 participants randomised), three studies of cholinesterase inhibitors (242 participants randomised) and one study of education (53 participants with PD). The three studies of multiple interventions all trialled exercise plus education (375 participants randomised). Details of the studies are presented in the Characteristics of included studies.
We contacted the authors of 24 included studies for further information: 18 exercise studies, three medication studies, one education study and two exercise plus education studies. For the exercise studies, nine authors responded, and six authors provided further information that was used in the review (Ashburn 2007; Chivers Seymour 2019; Goodwin 2011; Harro 2014; Paul 2014; Thaut 2019). The remaining three authors were unable to provide the requested information (Martin 2015; Munneke 2010; Protas 2005). The authors for all three medication studies were contacted for further information, and two responded, providing information that was used in the review (Chung 2010; Henderson 2016). There was no response to our request for further information about the education study (Ward 2004). One of the two authors contacted regarding the exercise plus education studies responded with information that was used in the review (Morris 2015), but there was no response from the other author (Cattaneo 2019).
Trial design
All included studies were randomised controlled trials (RCTs), with one exercise study being cluster randomised by community hospitals and their catchment areas (Munneke 2010). The exercise studies had a total of 54 groups, with 10 exercise studies having two groups, one of which was a control group (i.e. usual care, or sham exercise) (Ashburn 2007; Canning 2015a; Chivers Seymour 2019; Gao 2014; Goodwin 2011; Martin 2015; Paul 2014; Protas 2005; Song 2018; Wong‐Yu 2015). A further 11 studies had two groups which compared two different exercise interventions (Gandolfi 2017; Gandolfi 2019; Harro 2014; Mirelman 2016; Munneke 2010; Penko 2019; Shen 2015; Smania 2010; Thaut 2019; Volpe 2014a; Volpe 2014b). There were four studies that compared three groups; two of these had two exercise groups and one control group (Li 2012; Sedaghati 2016) and two had three exercise groups (Pelosin 2017; Ricciardi 2015). All three medication studies had two groups and compared a cholinesterase inhibitor with a placebo (Chung 2010; Henderson 2016; Li 2015a). One of these studies was a randomised cross‐over trial (Chung 2010), the two others had parallel arms. The education study compared personalised health education, including education about falls prevention, with a control group (Ward 2004). Two of the exercise plus education studies compared the intervention with a control group (Morris 2017; Cattaneo 2019) while the third study had two intervention groups and one control group (Morris 2015).
Trial size
The median number of participants randomised per study in the exercise studies was 60 (interquartile range (IQR) 34 to 130), with sample size ranging from 18 (Protas 2005) to 474 (Chivers Seymour 2019). For the medication studies, the median number of participants randomised per study was 89 (IQR 56 to 109.5), with sample size ranging from 23 (Chung 2010) to 130 (Henderson 2016). There were 53 participants with PD in the education study (Ward 2004). The exercise plus education studies had a median of 133 participants randomised (IQR 83 to 172), with sample size ranging from 32 (Cattaneo 2019) to 210 (Morris 2015).
Trial setting
Of the exercise studies, 13 were conducted at a facility with full supervision (Gao 2014; Harro 2014; Li 2012; Mirelman 2016; Paul 2014; Pelosin 2017; Penko 2019; Protas 2005; Ricciardi 2015; Sedaghati 2016; Smania 2010; Volpe 2014a; Volpe 2014b); five were conducted partially at a facility and partially at home with four of these having an average of 35% (range 13% to 55%) of sessions supervised (Canning 2015a; Goodwin 2011; Shen 2015; Wong‐Yu 2015), and the proportion of supervision in the remaining study was unclear (Gandolfi 2019). Five studies were conducted entirely in the participants’ homes, and in four of these there was an average of 10% (range 5% to 18%) of sessions supervised (Ashburn 2007; Chivers Seymour 2019; Martin 2015; Song 2018). The proportion of supervision in the remaining home‐based study was unclear (Thaut 2019). One exercise study included both a group that attended a facility with full supervision, and a group that was home‐based and fully supervised in pairs via telehealth (Gandolfi 2017). There was one study where the setting of the study was unclear (Munneke 2010).
Of the three exercise plus education studies, two were conducted partially at a facility and partially at home (Cattaneo 2019; Morris 2015). One of these had 14% of the exercise supervised, and the education session delivered in a group setting (Cattaneo 2019). The remaining two studies both had 50% of the exercise sessions supervised, and the education sessions delivered individually (Morris 2015; Morris 2017), with one of these studies delivered wholly in participants’ homes (Morris 2017).
Participants
In the exercise studies, 2601 participants contributed data for the rate of falls (1456 in the exercise versus control meta‐analysis) and 1044 participants for the number of people who fell at least once (932 in the exercise versus control meta‐analysis). In the cholinesterase inhibitor versus placebo studies, 229 participants contributed data for the rate of falls outcome and 230 contributed data for the number of people who fell at least once. The study of an education intervention versus control did not report the rate of falls and included 53 participants in the number of people who fell at least once outcome. The three studies of exercise plus education versus control included 352 participants (320 participants from two RCTs for the rate of falls meta‐analysis and 352 participants from three RCTs in the number of people who fell at least once meta‐analysis). The inclusion and exclusion criteria and other participant details are presented in the Characteristics of included studies table.
The included studies described disease severity in a variety of ways, and overall, participants in the included studies had mild to moderate PD (see Characteristics of included studies), though the increased fall rates and inclusion of people with impaired cognition in the medication trials indicates these participants had more advanced disease overall than participants in the trials of other interventions.
For the exercise studies the average disease duration was 7.9 years and the average age was 68.3 years. Thirteen exercise studies specified that participants had to either have a recent history of one or more falls, or a fall risk factor to be included (Ashburn 2007; Canning 2015a; Chivers Seymour 2019; Gao 2014; Goodwin 2011; Mirelman 2016; Penko 2019; Protas 2005; Sedaghati 2016; Smania 2010; Thaut 2019; Volpe 2014a; Volpe 2014b). One study included only participants with no history of falls (Wong‐Yu 2015). For the medication studies, the average disease duration was 7.9 years and average age was 68.3 years. Two of the three cholinesterase inhibitor versus placebo studies specified that participants required a history of falls to be included (Henderson 2016; Li 2015a), with one study requiring at least one fall in the prior year (Henderson 2016), and the other requiring two or more falls or near falls each week, without freezing of gait (Chung 2010).
The single study of an education intervention did not report age, disease severity or disease duration for the PD subgroup, and did not require a history of falls for participation (Ward 2004).
Of the three studies of exercise plus education versus control, one included people with and without PD and reported data for, but not the characteristics of the PD subgroup (Cattaneo 2019). The remaining two studies included people with mild to moderately severe PD with an average age of 69 years (Morris 2015; Morris 2017). An average disease duration of 6.7 years was reported in one of these studies (Morris 2015). There was no requirement for participants in any of these studies to have a history of falls.
Most studies excluded participants with significant cognitive impairment (usually defined as a Mini‐mental State Examination score of below 24). There was one exercise study (Mirelman 2016) and one exercise plus education study (Cattaneo 2019) that included participants with mild cognitive impairment (Mini‐mental State Examination ≥ 21). Two studies only excluded people with dementia; one medication study (Henderson 2016) and the education study (Ward 2004). Another medication study (Li 2015a) recruited only people with cognitive impairment.
Interventions
In the exercise studies, exercise was compared with a control intervention (i.e. usual care or an intervention not expected to have an effect on falls, such as ‘sham’ exercise or upper limb exercise) in 12 studies (Ashburn 2007; Canning 2015a; Chivers Seymour 2019; Gao 2014; Goodwin 2011; Li 2012; Martin 2015; Paul 2014; Protas 2005; Sedaghati 2016; Song 2018; Wong‐Yu 2015), and with an alternative form of exercise in 15 studies (Gandolfi 2017; Gandolfi 2019; Harro 2014; Li 2012; Mirelman 2016; Munneke 2010; Pelosin 2017; Penko 2019; Ricciardi 2015; Sedaghati 2016; Shen 2015; Smania 2010; Thaut 2019; Volpe 2014a; Volpe 2014b). Three of these studies compared more than one exercise intervention with a control intervention (Li 2012; Pelosin 2017; Sedaghati 2016). Overall, there were 42 exercise interventions and 12 control interventions.
The exercise interventions were grouped into categories based on the ProFaNE taxonomy (Table 3). The features of the exercise interventions are presented in Table 5. Most exercise interventions (34/42, 81%) were categorised as primarily gait, balance and functional training. PD‐specific exercises such as movement strategy training and cueing were included in this category. There were three resistance training interventions (7%) (Li 2012; Paul 2014; Shen 2015). Two interventions (5%) were of 3D exercise (Tai Chi; Li 2012; Gao 2014) and one intervention (2%) utilised flexibility exercises (Smania 2010). A further two interventions (5%) were from a study that compared physiotherapy provided by therapists with specific PD training according to evidence‐based guidelines with physiotherapy provided by usual therapists, but the specific details of the interventions were not provided (Munneke 2010). The duration of the exercise interventions ranged from 6 to 26 weeks (mean (SD) 11.3 (SD 6.9) weeks).
| Study ID | Exercise description | Primary exercise category | Duration of exercise intervention (weeks) | Group/Individual | Location | % supervision* |
| Exercise trials | ||||||
|
| Functional strength, range of movement, balance and walking exercise. | Gait, balance and functional training | 6 | Individual | Home‐based | 18% |
|
| Functional strength, balance and cueing exercise (some participants attended monthly group classes). | Gait, balance and functional training | 24 | Both (most individual but some participants attended monthly exercise classes) | Both (most home‐based (but classes were held at a facility) | 13% |
|
| Functional strength and balance exercise and strategies for fall and freezing avoidance. | Gait, balance and functional training | 26 | Individual | Home‐based | 7% |
| Virtual reality balance training delivered via telehealth
| Gait, balance and functional training | 7 | Group (pairs) | Home‐based | 100% | |
| Sensory‐integration balance training | Gait, balance and functional training | 7 | Individual | Facility‐based | 100% | |
| Trunk‐specific exercise | Gait, balance and functional training | 4 | Individual | Both | Unclear ‐ 100% at facility, number of unsupervised home‐sessions prescribed unclear | |
| General exercise | Gait, balance and functional training | 4 | Individual | Both | Unclear ‐ 100% at facility, number of unsupervised home‐sessions prescribed unclear | |
|
| Tai Chi classes. | 3D (Tai Chi) | 12 | Group | Facility‐based | 100% |
|
| Functional strength and balance exercise. | Gait, balance and functional training | 10 | Both | Both | 33% |
|
| Rhythmic auditory cued overground walking. | Gait, balance and functional training | 6 | Group | Facility‐based | 100% |
|
| Treadmill‐based gait training. | Gait, balance and functional training | 6 | Individual | Facility‐based | 100% |
|
| Tai Chi classes. | 3D (Tai Chi) | 24 | Group | Facility‐based | 100% |
|
| Functional strength exercise with weighted vests and ankle weights. | Resistance training | 24 | Group | Facility‐based | 100% |
|
| Exercises to address freezing of gait and associated falls, and walking using cues. | Gait, balance and functional training | 24 | Individual | Home‐based | 5% |
|
| Treadmill training in a virtual reality environment. | Gait, balance and functional training | 6 | Individual | Facility‐based | 100% |
|
| Treadmill‐based gait training. | Gait, balance and functional training | 6 | Individual | Facility‐based | 100% |
|
| Physiotherapy provided by ParkinsonNet therapists. | Other ‐ ParkinsonNet trained therapists | 24 | Individual | Unclear | ND |
|
| Physiotherapy usual care. | Other ‐ usual therapists | 24 | Individual | Unclear | ND |
|
| Progressive lower limb muscle power training using strength training machines. | Resistance training | 12 | Group (pairs) | Facility‐based | 100% |
| High frequency treadmill training (5 times per week for 10 sessions) | Gait, balance and functional training | 2 | Individual | Facility‐based | 100% | |
| Intermediate frequency treadmill training (3 times per week for 10 sessions) | Gait, balance and functional training | 3.3 | Individual | Facility‐based | 100% | |
| Low frequency treadmill training (2 times per week for 10 sessions) | Gait, balance and functional training | 5 | Individual | Facility‐based | 100% | |
| Gait and cognitive training practised together | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% | |
| Gait and cognitive training practised separately | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% | |
|
| Gait and stepping training. | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% |
|
| Strength, balance and gait training targeting the more affected side. | Gait, balance and functional training | 12 | Unclear | Facility‐based | 100% |
|
| Strength, balance and gait training targeting the less affected side. | Gait, balance and functional training | 12 | Unclear | Facility‐based | 100% |
|
| Functional strength, balance and gait training. | Gait, balance and functional training | 12 | Unclear | Facility‐based | 100% |
|
| Progressive balance and gait training with a balance pad (i.e. foam to stand on). | Gait, balance and functional training | 10 | Unclear | Facility‐based | 100% |
|
| Progressive balance and gait training without a balance pad. | Gait, balance and functional training | 10 | Unclear | Facility‐based | 100% |
|
| Balance and gait training. | Gait, balance and functional training | 12 | Unclear | Both | 55% |
|
| Lower limb resistance training using strength training machines (facility) and functional strength training (home) | Resistance training | 12 | Unclear | Both | 55% |
|
| Balance exercises. | Gait, balance and functional training | 7 | Individual | Facility‐based | 100% |
|
| Flexibility and coordination exercises not aimed at improving balance. | Flexibility | 7 | Individual | Facility‐based | 100% |
| Stepping videogame exercise | Gait, balance and functional training | 12 | Individual | Home‐based | 8% | |
| Gait training with rhythmic auditory stimulation throughout intervention period | Gait, balance and functional training | 24 | Individual | Home‐based | Unclear | |
| Gait training with rhythmic auditory stimulation, with no training in middle 8 weeks of intervention period | Gait, balance and functional training | 16 | Individual | Home‐based | Unclear | |
|
| Balance training using external perturbations wearing a proprioceptive stabiliser. | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% |
|
| Balance training using external perturbations with a sham proprioceptive stabiliser. | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% |
|
| Hydrotherapy with perturbation‐based balance training. | Gait, balance and functional training | 8 | Unclear | Facility‐based | 100% |
|
| Land‐based therapy with perturbation‐based balance training. | Gait, balance and functional training | 8 | Unclear | Facility‐based | 100% |
|
| Strength and balance exercise, including dance and modified Wing Chun martial art. | Gait, balance and functional training | 8 | Both | Both | 40% |
| Exercise plus education trials | ||||||
| Tailored mobility and balance exercises (plus fall prevention education). | Gait, balance and functional training | 8 | Individual | Home‐based | 14% | |
|
| Functional progressive resistance training with weighted vests and resistance bands. | Resistance training | 8 | Individual | Both | 50% |
|
| Movement strategy training. | Gait, balance and functional training | 8 | Individual | Both | 50% |
|
| Functional strength, movement strategy training (plus falls prevention education). | Gait, balance and functional training | 6 | Individual | Home‐based | 50% |
* % supervision calculated according to the % of exercise sessions supervised.
ND: no useable data
In the medication studies, three trials compared a cholinesterase inhibitor with a placebo. Two of these studies trialled rivastigmine, for either eight months (Henderson 2016) or 12 months (Li 2015a). The other trialled donepezil for six weeks (Chung 2010).
The education study (Ward 2004) provided individualised education and information in the form of a 12‐month health action plan, designed to improve each participant’s physical, social and psychological well‐being, including addressing fall risk. The education was delivered in participants’ homes by an occupational therapist through one home visit and a subsequent phone call.
In all three of the exercise plus education studies, the intervention was compared with a control intervention (Cattaneo 2019; Morris 2015; Morris 2017), with one of these also comparing with an alternative form of exercise plus education (Morris 2015). Two studies used home‐based exercise that was categorised as gait, balance and functional training (Cattaneo 2019; Morris 2017). The remaining study conducted the exercise interventions at a facility and at home, with one intervention categorised as gait, balance and functional training and the other as resistance training (Morris 2015). The features of the exercise interventions are presented in Table 5. The fall‐prevention education was provided individually at the time of the weekly supervised exercise session in two studies (Morris 2015; Morris 2017). In the remaining study there was a single one‐hour group education session about fall prevention which occurred before the exercise program was prescribed (Cattaneo 2019).
Outcomes
The source and time period of the data used for the generic inverse variance analysis (falls, fractures and adverse events (adverse events for the medication studies only)) outcomes for each study is shown in Table 6 and Table 7. Raw data for these outcomes and baseline falls data, when available, are shown in Table 7 and Table 8, respectively. Raw quality of life data is in Table 9. Data from studies reporting an economic analysis related to the cost of the intervention and/or fall outcomes is in Table 10, and information related to adverse events is in Table 11.
| Study ID and comparison | Source for rate ratio: rate of falls | Source for risk ratio: number of fallers | Source for risk ratio: number with fractures | Source for risk ratio: number with adverse events |
| Exercise trials | ||||
| Gait, balance and functional training vs Control | 3* | 7 | 7 | NA |
| Gait, balance and functional training vs Control | 1 | 5 | 7 | NA |
| Gait, balance and functional training vs Control | 1a++ | NA | 7 | NA |
| Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | 3‡‡‡ | NA | NA | NA |
| Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | 3‡‡‡ | NA | NA | NA |
| 3D exercise (Tai Chi) vs Control | 3 | 7 | NA | NA |
| Gait, balance and functional training vs Control | 1a++ | 6a | 7 | NA |
| Gait, balance and functional training (cueing training) vs Gait, balance and functional training (treadmill‐based gait training) | 3 | 7 | NA | NA |
| 3D exercise (Tai Chi) vs Resistance training (functional strength) and 3D exercise (Tai Chi) vs Control | 1 | 7 | NE | NA |
| Resistance training (functional strength) vs Control | 3 | 7 | NE | NA |
| Gait, balance and functional training vs Control | 1* | 7 | NA | NA |
| Gait, balance and functional training (virtual reality treadmill training) vs Gait, balance and functional training (treadmill‐based gait training) | 1a | NA | NA | NA |
| Other exercise (ParkinsonNet therapists) vs Other exercise (standard therapists) | 3c | NA | NA | NA |
| Resistance training vs Control | 1** | 5 | 7 | NA |
| Gait, balance and functional training (treadmill training at high frequency) vs Gait balance and functional training (treadmill training at intermediate frequency) vs Gait, balance and functional training (treadmill training at low frequency) | 3‡‡‡ | NA | NA | NA |
| Gait, balance and functional training (Gait and cognitive training practised together) vs Gait, balance and functional training (Gait and cognitive training practised separately) | 3‡‡‡ | NA | NA | NA |
| Gait, balance and functional training vs Control | 3 | 7 | NA | NA |
| Gait, balance and functional training (best side therapy) vs Gait, balance and functional training (worst side therapy) vs Gait, balance and functional training (standard therapy) | 3 | NA | NA | NA |
| Gait, balance and functional training (with a balance pad) vs Gait, balance and functional training (without a balance pad) vs Control | 3 | NA | NA | NA |
| Shen 2015*** Gait, balance and functional training vs Resistance training | 1a | 7 | 7 | NA |
| Gait, balance and functional training vs Flexibility exercise | 3 | NA | NA | NA |
| Gait, balance and functional training vs Control | 1 | 7 | NA | NA |
| Gait, balance and functional training (rhythmic auditory stimulation training throughout intervention period) vs Gait, balance and functional training (rhythmic auditory stimulation training with no training in middle 8 weeks of intervention period) | NA | 7 | NA | NA |
| Gait, balance and functional training (with proprioceptive stabiliser) vs Gait, balance and functional training (without proprioceptive stabiliser) | 3 | NA | NA | NA |
| Gait, balance and functional training (hydrotherapy) vs Gait, balance and functional training (land‐based therapy) | 3 | NA | NE | NA |
| Gait, balance and functional training vs Control | 1 | 6 | NA | NA |
| Medication trials | ||||
| Donepezil vs placebo | 3 | 7 | NE | 3 |
| Rivastigmine vs placebo | 1* | 7 | NA | 3 |
| Rivastigmine vs placebo | 3 | 6 | NA | ND |
| Education trial | ||||
| Personalised education vs control (standardised printed information) | NA | 6a | NA | NA |
| Exercise plus education trials | ||||
| Gait, balance and functional training + education vs Control | NA | 4 | NA | NA |
| Resistance training (functional strength) + education vs Control and Gait, balance and functional training (movement strategy training) + education vs Control | 1 | 5 | 7 | NA |
| Resistance training (functional strength) + education vs Gait, balance and functional training (movement strategy training) + education | 3 | 7 | 7 | NA |
| Gait, balance and functional training + education vs Control | 1 | 5 | 7 | NA |
ND: no useable data; NA: not applicable (not reported as an outcome in the trial OR not applicable for adverse events for exercise and exercise plus education trials as these were not pooled); NE (no events in either group.)
*One participant with excessive number of falls removed from analysis.
**Two participants with excessive number of falls assigned a value of 10 falls.
***One participant from the balance group and 2 from the resistance group with excessive number of falls at baseline removed from the analysis.
+randomisation stratified by falls history
++adjusted for previous falls
+++Incidence rate ratio using Poisson‐Inverse Gaussian regression, with unpublished 95% confidence interval provided by trial authors.
‡0 to 6 months data used as 0 to 12 months not available
‡‡0 to 10 weeks data used for rate ratio as 0 to 20 weeks not available
‡‡‡the separate time periods of falls data were combined
Codes for source of rate ratio:
1. Incidence rate ratio reported by trial authors
2. Hazard ratio/relative hazard (multiple events) reported by trial authors
3. Incidence rate ratio calculated by review authors
a. Adjusted for confounders by trial authors
b. Adjusted for clustering by trial authors
c. Adjusted for clustering by review authors
Codes for source of risk ratio:
4. Hazard ratio/relative hazard (first fall only) reported by trial authors
5. Relative risk reported by trial authors
6. Odds ratio reported by trial authors
7. Relative risk calculated by review authors
a. Adjusted for confounders by trial authors
b. Adjusted for clustering by trial authors
c. Adjusted for clustering by review authors
| Study ID and comparison | Intervention group: falls per person year | Intervention group: number (%) of fallers | Intervention group: number (%) of people sustaining one or more fall‐related fractures | Intervention group: non‐fall‐related adverse events per person year | Intervention group: number in analysis | Control group: falls per person year | Control group: number (%) of fallers | Control group: number (%) of people sustaining one or more fall‐related fractures | Control group: non‐fall‐related adverse events per person year | Control group: number in analysis | Length of falls/adverse events monitoring |
| Exercise trials | |||||||||||
| Gait, balance and functional training vs Control
| 6.3 | 46 (73%) | 2 (3%) | NA | Rate of falls and number of fallers: 63 Number sustaining fracture: 67
| 7.9* | 49 (78%) | 6 (9%) | NA | Rate of falls: 62 Number of fallers: 63 Number sustaining fracture: 67 | 6 months |
| Gait, balance and functional training vs Control
| 8.2 | 75 (65%) | 3 (3%) | NA | 115 | 14.0 | 81 (70%) | 4 (3%) | NA | 116 | 6 months |
| Gait, balance and functional training vs Control | 0‐6 months: 6.8 6‐12 months: 5.4 | NA | 0‐6 months: 5 (2%) 6‐12 months: 7 (5%) | NA | 0‐6 months: 231 6‐12 months: 127 | 0‐6 months: 5.4 6‐12 months: 5.6 | NA | 0‐6 months: 9 (4%) 6‐12 months: 3 (2%) | NA | 0‐6 months: 230 6‐12 months: 147 | 12 months, divided into 0‐6 and 6‐12 month time periods |
| Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | 4.0/8.5 | NA | NA | NA | 36/34 | NA | NA | NA | NA | NA | 2 months‡‡ |
| Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | 6.3/2.9 | NA | NA | NA | 19/18 | NA | NA | NA | NA | NA | 2 months‡‡ |
| 3D exercise (Tai Chi) vs Control | 0.6 | 8 (22%) | NA | NA | 37 | 1.3 | 19 (49%) | NA | NA | 39 | 6 months |
| Gait, balance and functional training vs Control | 0‐10 weeks: 93.9 10‐20 weeks: 34.5 | 0‐20 weeks: 52 (85%) | 0‐20 weeks: 0 (0%) | NA | 61 | 0‐10 weeks: 168.1 10‐20 weeks: 155.4 | 0‐20 weeks: 55 (86%) | 0‐20 weeks: 1 (2%) | NA | 64 | 20 weeks |
| Gait, balance and functional training (cueing training) / Gait, balance and functional training (treadmill‐based gait training) | 0.4/1.0 | 2 (20%)/4 (40%) | NA | NA | 10/10 | NA | NA | NA | NA | NA | 6 months |
| 3D exercise (Tai Chi) / Resistance training vs Control | 1.9/4.1 | 19 (29%)/31 (48%) | 0 (0%)/0 (0%) | NA | 65/65 | 5.7 | 26 (40%) | 0 (0%) | NA | 65 | 6 months |
| Gait, balance and functional training vs Control | 166.4 (using fall rate data from week 24‐28)* | 10 (100%) | NA | NA | Fall rate: 9 Number of fallers: 10 | 140.4 (using fall rate data from week 24‐28) | 9 (100%) | NA | NA | Fall rate: 8 Number of fallers: 9 | 6 months |
| Gait, balance and functional training (virtual reality treadmill training) / Gait, balance and functional training (treadmill‐based gait training) | ND | NA | NA | NA | 66/64 | NA | NA | NA | NA | NA | 6 months |
| Other exercise (ParkinsonNet therapists) / Other exercise (standard therapists) | 1.5/1.4 | NA | NA | NA | 329/312 | NA | NA | NA | NA | NA | 24 weeks |
| Resistance training vs Control | 6.5 | 7 (37%) | 1 (5%) | NA | 19 | 11.6 | 12 (63%) | 0 (0%) | NA | 19 | 6 months |
| Gait, balance and functional training (treadmill training at high frequency) vs Gait balance and functional training (treadmill training at intermediate frequency) vs Gait, balance and functional training (treadmill training at low frequency) | Unclear, as timeframe for falls monitoring not reported | NA | NA | NA | 10/10/10 | NA | NA | NA | NA | NA | Unclear |
| Gait, balance and functional training (Gait and cognitive training practised together) vs Gait, balance and functional training (Gait and cognitive training practised separately) | 9.1/8.2 | NA | NA | NA | 10/9 | NA | NA | NA | NA | NA | 2 months‡‡ |
| Gait, balance and functional training vs Control | 23.1 | 5 (56%) | ND | NA | 9 | 37.6 | 6 (67%) | ND | NA | 9 | 2 weeks |
| Gait, balance and functional training (best side therapy) / Gait, balance and functional training (worst side therapy) / Gait, balance and functional training (standard therapy) | 11.1/7.2/4.9 | NA | NA | NA | 9/9/9 | NA | NA | NA | NA | NA | 16 weeks |
| Gait, balance and functional training (with a balance pad) / Gait, balance and functional training (without a balance pad) vs Control | 1.04/4.16 | NA | NA | NA | 15/14 | 7.8 | NA | NA | NA | 15 | 10 weeks |
| Gait, balance and functional training / Resistance training | 0.41/1.02 | 6 (27%)/13 (57%) | 1 (5%)/1 (4%) | NA | Fall rate: 21/21 Number of fallers and fractures: 22/23 | NA | NA | NA | NA | NA | 15 months |
| Gait, balance and functional training / Flexibility exercise | 15.6/49.2 | NA | NA | NA | 28/27 | NA | NA | NA | NA | NA | 1 month |
| Gait, balance and functional training vs Control | 9.4 | 16 (55%) | NA | NA | 29 | 8.6 | 17 (68%) | NA | NA | 25 | 6 months |
| Gait, balance and functional training (rhythmic auditory stimulation training throughout intervention period) vs Gait, balance and functional training (rhythmic auditory stimulation training with no training in middle 8 weeks of intervention period) | NA | 24 (96%)/22 (100%) | NA | NA | 25/22 | NA | NA | NA | NA |
|
|
| Gait, balance and functional training (with proprioceptive stabiliser) / Gait, balance and functional training (without proprioceptive stabiliser) | 11.4/18.6 | NA | NA | NA | 20/20 | NA | NA | NA | NA | NA | 4 months |
| Gait, balance and functional training (hydrotherapy) / Gait, balance and functional training (land‐based therapy) | 3.6/9.6 | NA | 0 (0%)/0 (0%) | NA | 17/17 | NA | NA | NA | NA | NA | 2 months |
| Gait, balance and functional training vs Control | 0.38 | 6 (19%) | NA | NA | 32 | 0.38 | 8 (22%) | NA | NA | 36 | 6 months |
| Medication trials | |||||||||||
| Donepezil vs placebo+ | 47.45 | 18 (95%) | 0 (0%) | 3.0 | Fall rates and number of fallers and fractures:19 Adverse events: 23 | 91.25 | 16 (84%) | 0 (0%) | 1.1 | Fall rates and number of fallers and fractures: 19 Adverse events: 23 | 12 weeks |
| Rivastigmine vs placebo | 16.8* | 56 (86%) | NA | 4.4 | Fall rate and adverse events: 64 Number of fallers and fractures:65 | 28.8 | 56 (86%) | NA | 2.8 | 65 | 8 months |
| Rivastigmine vs placebo | 1.82 | 13 (32%) | NA | ND | 41 | 4.26 | 24 (60%) | NA | ND | 40 | 12 months |
| Education trial | |||||||||||
| Personalised education vs control (standardised printed information) | NA | ND | NA | NA | 27 | NA | ND | NA | NA | 26 | 12 months |
| Education plus exercise trials | |||||||||||
| Gait, balance and functional training plus education vs Control | NA | ND | NA | NA | 15 | NA | ND | NA | NA | 17 | 6 months |
| Resistance training / Gait, balance and functional training (movement strategy training) vs Control | 2.8/6.6 | 36 (52%)/44 (66%) | 3 (4%)/3 (4%) | NA | 69/67 | 18.6 | 37 (63%) | 2 (3%) | NA | 59 | 12 months |
| Gait, balance and functional training plus education vs Control | 21.9 | 39 (61%) | 2 (3%) | NA | 64 | 14.2 | 43 (72%) | 1 (2%) | NA | 60 | 12 months |
ND: no useable data; NA: not applicable (not reported as an outcome in the trial OR not applicable for adverse events for exercise and education trials as these were not pooled).
*outlier with excessive number of falls excluded
+randomised cross‐over trial
‡‡the two separate months of falls data were combined
| Study ID and groups | Intervention group: Number of participants | Intervention group: falls per person year | Intervention group: number (%) of fallers | Control group: number of participants | Control group: falls per person year | Control group: number (%) of fallers | Randomisation stratified by fall history | Timeframe for baseline falls monitoring |
| Exercise trials | ||||||||
| Gait, balance and functional training vs Control | 70 | 60 | 70 (100%) | 72 | 61 | 72 (100%) | No | 12 months, measured retrospectively |
| Gait, balance and functional training vs Control | 115 | 2 | 90 (78%) | 116 | 2 | 90 (78%) | Yes | 12 months, measured retrospectively |
| Gait, balance and functional training vs Control | 3 months prospective: 237 12 months retrospective: 238 | 3 months prospective: 23.6 12 months retrospective: 26 | 12 months retrospective: 238 (100%) | 3 months prospective and 12 months retrospective: 236 | 3 months prospective: 12 12 months retrospective: 19 | 236 (100%) | No | 3 months, measured prospectively prior to commencing intervention and 12 months, measured retrospectively |
| Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | 38/38 | 6.9/22.1 | ND | NA | NA | NA | No | 1 month, unclear if measured prospectively or retrospectively |
| Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | 19/18 | 19.6/7.9 | ND | NA | NA | NA | No | 1 month, unclear if measured prospectively or retrospectively |
| 3D exercise (Tai Chi) vs Control | 40 | ND | ND | 40 | ND | ND | No | ND |
| Gait, balance and functional training vs Control | Rate of falls analysis: 60 Number of fallers analysis: 64 | 137.8 | 55 (86%) | Rate of falls analysis: 62 Number of fallers analysis: 66 | 156.2 | 54 (82%) | No | 10 weeks, measured prospectively prior to commencing intervention |
| Gait, balance and functional training (cueing training) / Gait, balance and functional training (treadmill‐based gait training) | 10/10 | 1/1.4 | 3 (30%)/5 (50%) | NA | NA | NA | No | 6 months, measured retrospectively |
| 3D exercise (Tai Chi) / resistance training (functional strength) vs Control | 65/65 | ND | ND | 65 | ND | ND | No | ND |
| Gait, balance and functional training vs Control | Rate of falls analysis: 11* Number of fallers analysis: 12 | 202.8 | 9 (75%) | 9 | 150.8 | 6 (67%) | No | 5 weeks, measured prospectively from the point of study entry ‐ unclear if this overlaps with the first 3 weeks of the intervention period |
| Gait, balance and functional training (virtual reality treadmill training) / Gait, balance and functional training (treadmill‐based gait training) | 66/64 | 36.5/38.5 | 66 (100%)/64 (100%) | NA | NA | NA | No | 6 months, measured retrospectively |
| Other exercise (ParkinsonNet therapists) / Other exercise (standard therapists) | 358/341 | ND | ND | NA | NA | NA | No | ND |
| Resistance training vs Control | 20 | ND | 5 (25%) | 20 | ND | 7 (35% | No | 12 months, measured retrospectively |
| Gait, balance and functional training (treadmill training at high frequency) vs Gait balance and functional training (treadmill training at intermediate frequency) vs Gait, balance and functional training (treadmill training at low frequency) | 10/10/10 | Unclear, as timeframe for falls monitoring not reported | NA | NA | NA | NA | No | Unclear |
| Gait, balance and functional training (Gait and cognitive training practised together) vs Gait, balance and functional training (Gait and cognitive training practised separately) | 10/9 | 28/7.4 | ND | NA | NA | NA | No | 30 days, measured retrospectively |
| Gait, balance and functional training vs Control | 9 | 66.4 | 5 (56%) | 9 | 66.4 | 6 (67%) | No | 2 weeks, measured prospectively |
| Gait, balance and functional training (best side therapy) / Gait, balance and functional training (worst side therapy) / Gait, balance and functional training (standard therapy) | 9/9/10 | ND | ND | NA | NA | NA | No | ND |
| Gait, balance and functional training (with a balance pad) / Gait, balance and functional training (without a balance pad) vs Control | 15/14 | 6.8/6.7 | ND | 15 | 6.2 | ND | No | 10 weeks, unclear if measured retrospectively or prospectively |
| Gait, balance and functional training / Resistance training | 22/23 | 0.57/0.76** | 9 (41%)/10 (43%) | NA | NA | NA | No | 12 months, measured retrospectively |
| Gait, balance and functional training / Flexibility exercise | 28/27 | 51.6/55.2 | ND | NA | NA | NA | No | 1 month, measured prospectively |
| Gait, balance and functional training vs Control | 31 | ND | 17 (55%) | 29 | ND | 16 (55%) | No | 6 months, measured retrospectively |
| Gait, balance and functional training (rhythmic auditory stimulation training throughout intervention period) vs Gait, balance and functional training (rhythmic auditory stimulation training with no training in middle 8 weeks of intervention period) | 25/22 | 4.5/4.2 | ND | NA | NA | NA | No | 12 months, measured retrospectively |
| Gait, balance and functional training (with proprioceptive stabiliser) / Gait, balance and functional training (without proprioceptive stabiliser) | 20/20 | ND | 16 (80%)/12 (60%) | NA | NA | NA | No | 2 months, measured prospectively |
| Gait, balance and functional training (hydrotherapy) / Gait, balance and functional training (land‐based therapy) | 17/17 | 18/12.6 | 17 (100%)/17 (100%) | NA | NA | NA | No | Rate of falls: 2 months, measured prospectively Number of fallers: 12 months, measured retrospectively |
| Gait, balance and functional training vs Control | 32 | 0 | 0 (0%) | 38 | 0 | 0 (0%) | No | 6 months, measured retrospectively |
| Medication trials | ||||||||
| Donepezil vs placebo | 19 | ND | 19 (100%) | 19 | ND | 19 (100%) | No | Unclear: participants had all fallen or nearly fallen 2 or more times per week, measured retrospectively |
| Rivastigmine vs placebo | 65 | 5.0 | 65 (100%) | 65 | 5.5 | 65 (100%) | No | 12 months, measured retrospectively |
| Rivastigmine vs placebo | 41 | 3.6 | 22 (54%) | 40 | 3.8 | 23 (58%) | No | Unclear |
| Education trial | ||||||||
| Personalised education vs control (standardised printed information) | 27 | ND | ND | 26 | ND | ND | No | ND |
| Exercise plus education trials | ||||||||
| Gait, balance and functional training plus education vs Control | 15 | ND | ND | 17 | ND | ND | No | ND |
| Resistance training (functional strength) / Gait, balance and functional training (movement strategy training) vs Control | 70/69 | ND | 38 (54%)/40 (58%) | 71 | ND | 38 (54%) | No | 12 months, measured retrospectively |
| Gait, balance and functional training plus education vs Control | 67 | ND | 38 (57%) | 66 | ND | 35 (53%) | No | 12 months, measured retrospectively |
ND: no useable data; NA: not applicable
*One participant with excessive number of falls removed from analysis
**One participant from the balance group and 2 from the resistance group with excessive number of falls removed from the analysis
| Study ID and comparison | Intervention group baseline n, mean (SD) | Control group baseline n, mean (SD) | Intervention group post timeframe, n, mean (SD) | Control group post timeframe, n, mean (SD) | Intervention group follow‐up timeframe, n, mean (SD) | Control group follow‐up timeframe, n, mean (SD) |
| Exercise trials | ||||||
| Parkinson’s Disease Questionnaire 39 (PDQ39) and 8 (PDQ8) (range 0‐100)* | ||||||
| Gait, balance and functional training vs Control |
115, 28 (13.9) |
116, 30.7 (15.4) | 26 weeks, 104, 29.7 (14.8) | 26 weeks, 115, 32.5 (15.9) | NA | NA |
| Gait, balance and functional training vs Control | 126, 27.4 (14.3) | 153, 28.7 (15.9) | 6 months, 126, 28.3 (15.0) | 6 months, 153, 29.5 (16.5) | 12 months, 77, 29.1 (15.4) | 12 months, 100, 31.7 (15.5) |
| Gandolfi 2017 (PDQ8) Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | 36, 30.7 (15.5)/ 34, 30.5 (16.0) | NA | 7 weeks, 36, 24.1 (14.8)/ 34, 24.2 (15.9) | NA | 11 weeks, 36, 25.8 (14.9)/ 34, 23.9 (13.2)
| NA |
| Gandolfi 2019 (PDQ8) Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | 19, 25.5 (11.8)/ 18, 18.7 (10.8) | NA | 4 weeks, 19, 21.5 (10.0)/ 18, 15.3 (8.6) | NA | 8 weeks, 19, 23.0 (12.6)/ 18, 21.0 (8.8) | NA |
| Gait, balance and functional training (cueing training) / Gait, balance and functional training (treadmill‐based gait training) |
11, 31.1 (14.8)/ 11, 40.1 (17.5 | NA | 6 weeks, 10, 27.5 (17.9)/ 10, 27.4 (10.0) | NA | 3 months, 10, 25.4 (15.0)/ 9, 30.0 (12.9) | NA |
| Li 2012 (PDQ8) 3D exercise (Tai Chi) / Resistance training vs Control |
65, 25.1 (16.8)/ 65, 25.3 (14.7) |
65, 25.2 (16.3) | 6 months, 65, 15.5 (11.4)/ 65, 21.4 (12.7) | 6 months, 65, 25.1 (15.6) | NA | NA |
| Gait, balance and functional training (with proprioceptive stabiliser) / Gait, balance and functional training (without proprioceptive stabiliser) |
20, 62.7 (19.5)/ 20, 61.4 (38.9) | NA | 2 months, 20, 44.0 (22.3)/ 20, 58.5 (37.9) | NA | 4 months, 20, 53.7 (22.3)/ 20, 61.0 (35.1) | NA |
| Gait, balance and functional training (hydrotherapy) / Gait, balance and functional training (land‐based therapy) |
17, 60.3 (19.9)/ 17, 64.4 (28.6) | NA | 2 months 17, 41.9 (20.9)/ 17, 56.4 (26.8) | NA | NA | NA |
| EQ5D Thermometer (0‐100) | ||||||
| Gait, balance and functional training vs Control |
70, 63.1 (17.1) |
71, 64.6 (14.5) | 8 weeks, 67, 61.3 (19.8) | 8 weeks, 66, 61.7 (14.5) | 6 months, 65, 63.0 (18.7) | 6 months, 64, 56.6 (16.9) |
| EQ5D Index score (range 0‐1) | ||||||
| Gait, balance and functional training vs Control |
61, 0.7 (0.1) |
63, 0.7 (0.1) | 10 weeks, 61, 0.7 (0.1) | 10 weeks, 63, 0.7 (0.1) | 20 weeks, 61, 0.8 (0.3) | 20 weeks, 62, 0.7 (0.3) |
| Other exercise (ParkinsonNet therapists) / Other exercise (standard therapists) |
358, 0.65 (0.20)/ 341, 0.65 (0.22) | NA | 16 weeks, 295, 0.66 (0.20)/ 294, 0.65 (0.23) | NA | 24 weeks, 262, 0.68 (0.21)/ 259, 0.66 (0.23)
| NA |
| SF12 and SF36 Physical Composite Score (range 0‐100) | ||||||
| Canning 2015a (SF12) Gait, balance and functional training vs control |
115, 42.3 (7.6) |
116, 42.9 (7.9) | 26 weeks, 104, 41.3 (8.8) | 26 weeks, 115, 40.2 (7.8) | NA | NA |
| Mirelman 2016 (SF36) Gait, balance and functional training (virtual reality treadmill training) / Gait, balance and functional training (treadmill‐based gait training) |
66, 49 (2.5)/ 64, 44.8 (2.5) | NA | 6 weeks, 66, 52 (2.5)/ 64, 46.5 (2.5) | NA | 6 months, 66, 50.5 (2.5)/ 64, 48 (2.5) | NA |
| SF12 Mental Composite Score (range 0‐100) | ||||||
| Gait, balance and functional training vs Control |
115, 51.6 (6.5) |
116, 50.5 (6.8) | 26 weeks, 104, 51.2 (6.4) | 26 weeks, 115, 50.3 (6.7) | NA | NA |
| Medication Trials | ||||||
| EQ5D Thermometer (0‐100) | ||||||
| Rivastigmine vs placebo |
65, 64 (17) |
65, 65 (17) | 32 weeks, 58, 66 (16) | 32 weeks, 63, 63 (18) | NA | NA |
| EQ 5D Index score (range 0‐1) | ||||||
| Rivastigmine vs placebo |
65, 0.72 (0.19) |
65, 0.71 (0.18) | 32 weeks, 58, 0.66 (0.21) | 32 weeks, 63, 0.66 (0.19) | NA | NA |
| Education plus exercise trials | ||||||
| Parkinson’s Disease Questionnaire 39 (PDQ39) (range 0‐100)* | ||||||
| Resistance training / Gait, balance and functional training (movement strategy training) vs Control |
70, 20.8 (13.6)/ 69, 19.4 (12.8) |
71, 22.1 (12.5) | 3 months, 67, 18.9 (13.5)/ 64, 16.9 (14.0) | 3 months, 54, 18.5 (12.6) | 14 months, 67, 20.0 (13.6)/ 66, 20.8 (14.1) | 14 months, 57, 24.1 (13.1) |
| Gait, balance and functional training plus education vs Control |
67, 23 (14) |
66, 24 (15) | 6 weeks, 62, 21 (14) | 6 weeks, 58, 20 (14) | 58 weeks, 55, 22 (13) | 58 weeks, 53, 22 (14) |
| EQ5D Thermometer (0‐100) | ||||||
| Resistance training / Gait, balance and functional training (movement strategy training) vs Control |
70, 74.1 (16.7)/ 69, 73.9 (15.9) |
71, 72.7 (14.6) | 3 months, 67, 71.8 (16.4)/ 64, 76.5 (16.4) | 3 months, 54, 74.7 (16.0) | 14 months, 67, 75.4 (14.1)/ 66, 75.0 (13.5) | 14 months, 57, 72.8 (16.0) |
| Gait, balance and functional training vs Control |
67, 73 (15) |
66, 72 (16) | 6 weeks, 62, 68 (15) | 6 weeks, 58, 76 (12) | 58 weeks, 55, 72 (17) | 58 weeks, 53, 71 (14) |
| EQ5D Index score (range 0‐1) | ||||||
| Gait, balance and functional training vs Control |
67, 0.67 (0.27) |
66, 0.63 (0.28) | 6 weeks, 62, 0.66 (0.29) | 6 weeks, 58, 0.65 (0.27) | 58 weeks, 55, 0.67 (0.25) | 58 weeks, 53, 0.64 (0.3) |
NA: not applicable
*High score = worse quality of life
** Median and interquartile range reported by trial authors and converted to mean and standard deviation by review authors: Volpe 2014a using technique described by Wan 2014 and Goodwin 2011 using the technique described in the Cochrane Handbook (Higgins 2017). Conversion techniques differed due to the different sample sizes in the trials.
PDQ8 = Parkinson’s Disease Questionnaire 8
| Study ID, (source if not primary reference), sample, comparison, type of evaluation | Intervention(s) and comparator (n in analyses) | Perspectives, type of currency, price year, time horizon | Cost items measured | Intervention costs per participant | Healthcare service costs per participant | Incremental cost per fall prevented/ per QALY gained |
| Exercise trials | ||||||
| People with PD who had fallen at least once in the past year or were at risk of falls. Gait, balance and functional training vs control Evaluated with cost‐effectiveness analyses | Exercise (balance, lower limb strength, and when required cueing), 3 X week, 24 weeks, with 6‐10 sessions supervised either individually or in a group setting (n = 113) vs usual care control (n = 113) | Health system perspective, Australian dollar, 2012, During 6‐month trial period
| Intervention costs (staff time, travel, equipment) Health service use costs (hospital, medical, allied health) Medication costs | $A1,010 (€642) | Exercise group $A4,604 (€2,925) Control group $A3,920 (€2,491) | Cost per fall prevented $A574 (€365) Cost per QALY gained $A338,800 (€215,277) |
| Chivers Seymour 2019 (Ashburn 2019, Xin 2020) People with PD who had fallen at least once in the past year. Gait, balance and functional training vs control Evaluated with cost‐effectiveness analyses | Exercise (balance and lower limb strengthening exercises, plus strategies for preventing falls and reducing freezing of gait), 30 min per day for 6 months, including 12 x 1‐1.5 hour supervised sessions with a physiotherapist (n = 238) vs usual care control (n = 236) | United Kingdom National Health Service and Personal Social Services perspectives, Pound Stirling, 2016, During 6 month intervention period | Intervention costs (physiotherapist salaries, training, travel, equipment and consumables) Health service use (hospital, primary care, social service) Medication costs collected but not included in analyses | £650 (€765) | Exercise group £3,137 (€3,905) Control group £3,069 (€3,613) | Cost per QALY gained £120,659 (€142,063) |
| People with PD, both fallers and non‐fallers. Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) Evaluated with cost analysis | Virtual reality balance training (using Nintendo Wii Fit system, Nintendo Co., Ltd., Kyoto, Japan) delivered via telehealth (using Skype, Microsoft, USA), delivered in pairs (n = 36) vs sensory‐integration balance training delivered in‐person, individually (n = 34), both interventions 50 mins, 3 X week, 7 weeks | Cost of rehabilitation perspective, Euros, Price year not reported, During assessments and 7 week intervention period | Direct costs (personnel for screening, assessments and intervention, plus resource utilisation). Indirect costs (utilities, facilities) | Virtual reality via telehealth balance training (delivered in pairs) €383.55 sensori‐integration balance training (delivered individually) €602.10 | Not reported | Not reported |
| People with PD and 2 or more falls in the preceding year. Gait, balance and functional training vs control Evaluated with cost‐effectiveness analyses | Exercise (balance, lower limb and trunk strength, 1 X week supervised group and 2 X week independent at home for 10 weeks (n=48) vs usual care control (n = 45). Economic analyses conducted with intervention n = 48 and control n = 45 | United Kingdom National Health Service and Personal Social Services perspectives, Pound sterling, 2008/9, During 20 weeks (10 weeks intervention and 10 weeks follow‐up) | Intervention costs (staff time, travel, equipment, venue hire) Health service use (hospital, primary care, social service) Medication costs | £76 (€89) | Exercise group Health care cost £1,198 (€1,410) Health and social care cost £1,444 (€1,700) Control group Health care cost £1,320 (€1,554) Health and social care cost £1,479 (€1,741) | Cost per QALY gained for total health care costs ‐£4,885 (‐€5,752) Cost per QALY gained for combined total health care and social care costs ‐£1,358 (‐€1,599) |
| People with PD, both fallers and non‐fallers. 3D exercise (Tai Chi) / resistance training (functional strength) vs control Evaluated with cost‐effectiveness analyses
| Tai Chi (n=65) vs resistance training (n=65) vs stretching (control) (n=65), all group classes for 60 minutes, 2 X week, 24 weeks | Societal perspective, United States dollar, 2011, During 9 months (6 months intervention and 3 months follow‐up) | Intervention costs (program promotion, recruitment, staff time, insurance, equipment, room hire, printed materials) Non‐intervention costs (PD medication, physical therapy, medical treatment for falls, participant travel) | Tai chi $US1,080 (€952) Resistance $US1,186 (€1,046) Stretching $US1,155 (€1,019)
| PD medication, physical therapy, medical treatment for falls and participant travel costs: Tai chi $US272 (€240) Resistance $US310 (€273) Stretching $US726 (€640)
| Tai chi vs stretching (control): Cost per fall prevented ‐$US175 (‐€154) Cost per QALY gained ‐$US3,394 (‐€2,993)
Resistance vs Tai Chi: Cost per fall prevented $US100 (€88) Cost per QALY gained $US1,236 (€1,090) |
| People with PD, both fallers and non‐fallers. Other exercise (ParkinsonNet therapists) vs Other exercise (standard therapists) Evaluated with cost analysis | Treatment from ParkinsonNet trained physiotherapists (n=343 to 350)* vs usual care (treatment from physiotherapists without specific PD training) (n=332‐340)*, both groups 24 weeks intervention period | Societal perspective, Euro, Price year not reported, but data collected 2005‐2007, During 24 weeks intervention | Health care costs (physiotherapy, medication, consultation, day‐hospital rehabilitation, admission to hospital, home‐care (paid services), informal care, costs due to lost productivity of the care‐partner). | Physiotherapy cost: ParkinsonNet group €297 Usual care group €310
| Excluding physiotherapy: ParkinsonNet group €2,674 Usual care group €3,424
| Not calculated |
| Exercise plus education trial | ||||||
| People with PD, both fallers and non‐fallers. Gait, balance and functional training plus education vs control Evaluated with cost analysis | Exercise (strength training (lower limb and trunk), movement strategy training) and falls prevention education, 1 X week 60 mins supervised and 1 X week 60 mins independent practice for 6 weeks (n=67) vs Life Skills program (control) (n=66) | Health system perspective, Australian dollar, 2016, During 12 months follow‐up | Intervention costs (travel, home visits, therapist training, equipment). Life skills control intervention was considered as a placebo and therefore had no costs attributed to it. Medical costs associated with falling events (medical, medical ancillary, diagnostic and hospitalisation costs) | $A1,596 (€1,013) | Not reported | Not calculated as there was no difference between the groups |
*Different participant numbers for different cost components
Where costs were reported in a currency other than EUR, the cost was converted to EUR (€) on December 23, 2021.
QALY = quality‐adjusted life‐year
| Study ID and comparison | Information related to adverse events |
| Exercise trials | |
| Gait, balance and functional training vs Control | No participants fell while performing the exercise program. |
| Gait, balance and functional training vs Control | Two participants had non‐injurious falls during unsupervised exercise at home. |
| Gait, balance and functional training vs Control | No participants fell while performing the exercise program, and no adverse events were associated with the intervention. 0‐6 months hospitalisations: 9 PDSAFE exercise group participants (1 participant with 2 hospitalisations); 20 control group participants. 6‐12 months hospitalisations: 18 PDSAFE exercise group participants (2 participants with 2 hospitalisations); 21 control group participants (2 participants with 2 hospitalisations). |
| Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | No adverse events were reported during the study. |
| Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | No adverse events or safety concerns were reported during the study. |
| 3D exercise (Tai Chi) vs Control | Not reported |
| Gait, balance and functional training vs Control | No adverse events occurred during the exercise sessions. |
| Gait, balance and functional training (cueing training) / Gait, balance and functional training (treadmill‐based gait training) | No adverse events during the intervention. |
| 3D exercise (Tai Chi) / Resistance training vs Control | Tai‐chi (n=65): 3 in class events ‐ 2 falls, 1 muscle soreness or pain; 24 out of class events ‐ 19 falls, 4 low back pain, 1 ankle sprain. Functional strength training (n=65): 14 in class events ‐ 4 falls, 4 muscle soreness or pain, 3 dizziness or faintness, 3 symptoms of hypotension; 41 out of class events ‐ 31 falls, 3 chest pain, 1 hypotension, 4 low back pain, 2 ankle sprains. Stretching (n=65): 9 in‐class events ‐ 5 falls, 1 muscle soreness or pain, 2 dizziness or faintness, 1 symptoms of hypotension; 36 out of class events ‐ 26 falls, 2 chest pain, 2 hypotension, 5 low back pain, 1 ankle sprain. Nb ‐ out of class events are those that occurred during habitual activity or during an assessment. Participants did not perform any intervention outside the class. |
| Gait, balance and functional training vs Control | Not reported |
| Gait, balance and functional training (virtual reality treadmill training) vs Gait, balance and functional training (treadmill‐based gait training) | No serious adverse events during training. Adverse events other than those that occurred during intervention were recorded for both groups, but were not reported separately for the participants with Parkinson's disease. |
| Other exercise (ParkinsonNet therapists) / Other exercise (standard therapists) | None reported, though not collected systematically. |
| Resistance training (muscle power training) vs Control | Power training (n=20): 1 exacerbation of pre‐existing low back pain, 1 pelvic fracture unrelated to the intervention, and 6 participants required modification to training loads due to transient pain, joint inflammation or illness. Control low intensity exercise (n=20): 2 participants had exacerbations of pre‐existing hernias, though this was not attributable to the low intensity exercise. |
| Gait, balance and functional training (treadmill training at high frequency) vs Gait balance and functional training (treadmill training at intermediate frequency) vs Gait, balance and functional training (treadmill training at low frequency) | Not reported |
| Gait, balance and functional training (Gait and cognitive training practised together) vs Gait, balance and functional training (Gait and cognitive training practised separately) | Not reported |
| Gait, balance and functional training vs Control | Not reported |
| Gait, balance and functional training (best side therapy) / Gait, balance and functional training (worst side therapy) / Gait, balance and functional training (standard therapy) | Not reported |
| Gait, balance and functional training (with a balance pad) / Gait, balance and functional training (without a balance pad) vs Control | Not reported |
| Gait, balance and functional training / Resistance training | No adverse events related to the intervention in either group. |
| Gait, balance and functional training / Flexibility exercise | Not reported |
| Gait, balance and functional training vs Control | Adverse events were reported for the intervention group. Six participants ceased the stepping training: two ceased exercise due to it exacerbating pre‐existing lower back pain; two died; one sustained a knee injury from a fall unrelated to the intervention; one ceased for personal reasons. Additionally, one participant experienced a non‐injurious fall while undertaking the intervention and eight participants reported an increase in pre‐existing pain (e.g. lower back pain, knee pain, foot pain) but felt that the exacerbation was unrelated to the intervention. |
| Gait, balance and functional training (rhythmic auditory stimulation training throughout intervention period) vs Gait, balance and functional training (rhythmic auditory stimulation training with no training in middle 8 weeks of intervention period) | Participants who dropped out did so for reasons unrelated to adverse events. |
| Gait, balance and functional training (with proprioceptive stabiliser) / Gait, balance and functional training (without proprioceptive stabiliser) | No major adverse event related to the intervention. |
| Gait, balance and functional training (hydrotherapy) / Gait, balance and functional training (land‐based therapy) | Not reported |
| Gait, balance and functional training vs Control | No adverse events related to the intervention. |
| Medication trials | |
| Donepezil vs placebo | Donepezil (n=23): Eight participants (35%) reported 16 side effects (e.g. dehydration, gastrointestinal upset, headache, sleep disturbance, muscle cramps, orthostatic hypotension, weight loss). Placebo (n=23): Five participants (22%) reported 6 side effects (e.g. gastrointestinal upset, headache, sleep disturbance). These side effects were reported to be transient in most cases. |
| Rivastigmine vs placebo | Rivastigmine (n=64): 187 adverse events (excluding falls) Placebo (n=65): 122 adverse events (excluding falls) Adverse events included cardiac disorders, endocrine disorders, gastrointestinal disorders, general disorders and administration site disorders, immune system disorders, infections and infestations, injury, poisoning and procedural complications, investigations, metabolism and nutrition disorders, musculoskeletal and connective tissue disorders, neoplasms benign, malignant and unspecified, nervous system disorders, psychiatric disorders, renal and urinary disorders, respiratory, thoracic and mediastinal disorders, skin and subcutaneous tissue disorders, surgical medical procedures, vascular disorders. About one third of participants in the rivastigmine group complained of nausea. Most adverse events were categorised as mild and were considered to be unrelated to the intervention. There were 27 adverse events that were classified as serious; 14 in the rivastigmine group and 13 in the placebo group. Two of these events in the rivastigmine group were considered to be probably related to the rivastigmine. Twenty‐three participants in the rivastigmine group and 19 participants in the placebo group stopped taking the trial medication due to adverse events. |
| Rivastigmine vs placebo | Two participants withdrew due to adverse reactions, however details not provided. |
| Education trial | |
| Personalised education vs control (standardised printed information) | Not reported |
| Exercise plus education trials | |
| Gait, balance and functional training plus education vs Control | Not reported |
| Resistance training / Gait, balance and functional training (movement strategy training) vs Control | Functional strength training group (n=70): 25 occasions of new muscle soreness lasting > 24 hours Movement strategy training group (n=69): 11 occasions of new muscle soreness lasting > 24 hours, 1 fall and 2 occasions of dizziness during the intervention. |
| Gait, balance and functional training vs Control | No adverse events related to the intervention. |
In the exercise studies, a RaR for the rate of falls was reported in 10 studies, and could be calculated in an additional 14. There was one study where the rate of falls was reported for some comparisons but required calculation in other comparisons (Li 2012). The RR for the number of people experiencing a fall was reported in four studies and could be calculated in an additional nine. Data to calculate the risk of fractures (number sustaining one or more fall‐related fractures) were reported in six studies and a further two studies reported there were no fractures in either group (Li 2012; Volpe 2014b). Six exercise studies reported an economic analysis related to the cost of the intervention and/or falls outcomes.
Information regarding adverse events related to the exercise intervention was provided by 15 studies. Only three of these studies (Li 2012; Mirelman 2016; Paul 2014) reported adverse events more broadly and monitored for adverse events using the same methods in all groups over the entire study period. Of these three, one included participants with and without PD and did not report these data separately for the PD group participants (Mirelman 2016). Ten studies did not report whether there were adverse events.
Health‐related quality of life was reported in 12 exercise studies, with one of these studies reporting more than one quality of life outcome. The most commonly reported outcome was the Parkinson’s Disease Questionnaire, with the 39 item (PDQ39) questionnaire reported by five studies and the eight‐item (PDQ8) questionnaire reported by three studies. The EQ5D was reported in three studies, with one study reporting the EQ5D thermometer score and two studies reporting the EQ5D index score. The Physical Composite Score from the SF12 was reported in one study and from the SF36 in an additional study. The Mental Composite Score from the SF12 was reported in one study.
Of the three cholinesterase inhibitor versus placebo studies, one reported the rate of falls (Henderson 2016), and this variable could be calculated in the remaining two (Chung 2010; Li 2015a). One study reported the risk of falling (Li 2015a), with this calculated in the remaining two. One of these studies reported data related to fractures, however a risk ratio (RR) could not be calculated as there were no events (Chung 2010). Two studies monitored adverse events using the same methods in all groups over the entire study period and reported enough data to enable calculation of the rate of adverse events excluding falls (Chung 2010; Henderson 2016). Health‐related quality of life was reported in the form of the EQ5D thermometer and index score in one study (Henderson 2016). None of these studies reported an economic analysis related to fall outcomes.
The education study reported an odds ratio for the risk of falling but did not report rate of falls, risk of fractures, adverse events, quality of life or economic data (Ward 2004).
In the exercise plus education studies, a RaR ratio for the rate of falls was reported in two studies (Morris 2015; Morris 2017) and a RR for the number of people who fell at least once was reported in all three. One of these studies compared two intervention groups and a control group and both the risk of falling and rate of falls was reported for two comparisons but required calculation for a third comparison (Morris 2015). Two studies reported data to calculate the risk of sustaining one or more fall‐related fractures as well as health‐related quality of life at post‐test and at follow‐up (Morris 2015; Morris 2017). These studies also reported information about adverse events related to the intervention, and one study reported information about the cost of the intervention (Morris 2017).
Excluded studies
There were four studies that initially appeared to meet the inclusion criteria but were subsequently excluded (see Characteristics of excluded studies). Two exercise studies were excluded, one because it did not meet the inclusion criteria for the types of outcome measures (Kurlan 2015), and the other because it was a randomised cross‐over trial where falls data were not collected during the control period (Sparrow 2016). Another study (Hill 2015), investigated the effect of inpatient and staff education and included participants with a wide range of diagnoses. Data for just the participants with PD were not available. The fourth excluded study explored the effect of sunlight exposure in increasing 25‐hydroxyvitamin D and reducing hip fractures in people with PD (Sato 2011). This study was excluded as the integrity of the data has been questioned (Bolland 2016), and the publication of the study has been retracted by the journal.
Ongoing studies
We identified 30 ongoing studies; 20 trialling exercise interventions, one trialling medication, three trialling deep brain stimulation, one trialling deep brain stimulation plus physiotherapy, three trialling a model of care, one trialling a multifactorial intervention (environmental modification, exercise and behavioural strategies), and one trialling osteopathic manipulative medicine (see Characteristics of ongoing studies).
Studies that are currently open to recruitment include: 14 exercise studies (NCT04300023; NCT04108741; NCT03972969; NCT04946812; NCT04897256; NCT04874051; NCT04848077; NCT04665869; NCT04634331; DRKS00024982; ChiCTR2000038852; NCT05172661 – by invitation only; ACTRN12620001135909; NCT04613141 – by invitation only), one medication study (NCT04226248), one deep brain stimulation study (NCT04408573), one deep brain stimulation plus physiotherapy study (NCT04953637), two model of care studies (NCT04694443; NCT04555720 – by invitation only), and the multifactorial intervention study (ACTRN12619000415101). Two exercise studies (NCT04389138; NCT04300348) and 1onemodel of care study (NCT05127057) were not yet recruiting.
The exercise studies have a median target sample size of 48 (range 16 to 452) and two of the studies (10%) specified a history of falls or increased fall risk as an inclusion criterion. Seventeen studies are investigating forms of gait, balance and functional training, with two of these studies investigating treadmill training with virtual reality versus treadmill training alone (NCT04108741; NCT03727529); three investigating structured exercise programs versus control (NCT04389138; NCT03972969; ACTRN12620001135909), two investigating exercise in a virtual reality environment versus exercise alone (NCT04874051; NCT04634331); two investigating balance plus cognitive dual task training versus balance training alone (NCT05172661; ChiCTR2000038852); one investigating split belt treadmill training compared to usual treadmill training (NCT04946812); one investigating walking with haptic feedback plus an exercise program versus control (NCT04613141); one investigating walking with auditory feedback plus an exercise program versus the same intervention without the feedback (NCT04300348); one investigating a combined brisk walking and balance program versus flexibility and strength exercise (NCT04665869); one investigating walking with a robotic device versus control (NCT03751371); one investigating volitional and reactive step training using an exergame as well as slip and trip training versus control (ACTRN12618001515280); one investigating exercises focused on turning versus control (NCT04897256) and one investigating exercise including eye movement training versus exercise alone (DRKS00024982). Of the remaining studies, one is investigating home‐based cycling versus control (NCT04300023), one is investigating different proportional increases in daily step count supported via a smartphone app (NCT04848077), and one is investigating muscle power training versus control (RBR‐5w2sqt).
The medication study is investigating a cholinesterase inhibitor (rivastigmine) versus placebo (NCT04226248), has a target sample size of 600 and specifies a history of falls or increased fall risk as an inclusion criterion.
The deep brain stimulation studies have a median target sample size of 15 (range 10 to 30) and none of them specify fall risk as part of the inclusion criteria. One study is investigating cyclic stimulation versus continuous stimulation (NCT04408573), one is investigating flexible subthalamic nucleus stimulation versus standard subthalamic nucleus stimulation (NCT04116177), and the third is investigating segmented (steered) contacts versus a contact combination in ring mode (NCT04093544). The deep brain stimulation plus physiotherapy study has a target sample size of 60 and does not include fall risk in the inclusion criteria (NCT04953637). It is comparing deep brain stimulation plus physiotherapy targeting gait and balance with deep brain stimulation plus encouragement to be active.
The three studies trialling models of care are all comparing a care model with control usual care. They have a median target sample size of 200 (range 76 to 214) and none of them specify fall risk as part of the inclusion criteria. One study is trialling a multicomponent model of care including case management, information technology infrastructure and empowerment of patients, care‐partners and therapists (NCT05127057). The second is trialling multidisciplinary telehealth in conjunction with standard in‐person consultations (NCT04694443). The third study is trialling interdisciplinary care including the development of a treatment plan (NCT04555720).
The multifactorial intervention versus control study has a target sample size of 40, and includes a history of falls in the inclusion criteria (ACTRN12619000415101). The study of osteopathic manipulative medicine versus control has a target sample size of 50, and does not include a history of falls as part of the inclusion criterion (NCT02107638).
Risk of bias in included studies
The results of the risk of bias assessment for each included study is shown in the Characteristics of included studies, in Figure 2 and Figure 3.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Allocation
In the exercise studies, we judged the risk of bias in the generation of the allocation sequence to be low in 72% (n = 18/25) and unclear in 28% (n= 7/25) of studies. The method of concealment of the allocation sequence prior to group assignment was assessed to be at low risk of bias in 52% (n = 13/25) and unclear risk in 48% (12/25). In the three medication studies, the risk of selection bias was low in one study (33%) and unclear in the remaining two (67%) for these items. The education study had low risk of bias for random sequence generation and unclear risk for allocation concealment. In the three exercise plus education studies we judged the risk of bias in the generation of the allocation sequence to be low in all (100%), but the method of concealment of the allocation sequence prior to group assignment was assessed to be at low risk of bias in two studies (67%) and unclear in the other (33%).
Blinding
Blinding of participants and personnel
In most exercise intervention studies (92%, n = 23/25), the education study and the three exercise plus education studies, it was not possible to blind the participants or personnel to group allocation. The risk of performance bias was assessed as unclear in these studies, as the effect of awareness of group allocation in an exercise and/or education study is unclear. The remaining two exercise studies (8%) were able to blind participants and personnel to group allocation and were therefore assessed as at low risk of performance bias. Two (67%) of the medication studies described how blinding of participants and personnel was ensured, and so were assessed as being at low risk of bias. The remaining study did not describe how blinding was achieved and therefore the risk of bias was assessed as unclear.
Blinding of outcome assessment
We assessed the risk of bias for blinding of outcome assessment (detection bias) separately for the falls outcomes and for the fractures outcome.
1. Rate of falls and risk of falling
In the exercise studies, the risk of detection bias related to the measurement of falls outcomes was judged as low in 24% (n = 6/25), unclear in 68% (n = 17/25) and high in 8% (n = 2/25). The risk of bias was low in all three medication studies, since interventions were placebo matched and personnel collecting outcomes were blinded to group allocation. The risk of detection bias for the falls outcomes was unclear in the education study and in one of the exercise plus education studies (33%). It was judged as being at low risk of bias for the remaining two exercise plus education studies (67%).
2. Risk of one or more fall‐related fractures
Eight exercise studies, one medication study and two exercise plus education studies reported data relating to fractures. The risk of detection bias relating to the methods of ascertainment of fractures was unclear in all these studies.
Incomplete outcome data
We assessed the risk of bias for incomplete outcome data (attrition bias) separately for the rate of falls the risk of falling.
1. Rate of falls
The risk of attrition bias in the exercise studies for data relating to the rate of falls were assessed as low in 76% (n = 19/25 studies), unclear in 12% (n = 3/25) and high in the remaining 12% (n = 3/25) of studies. In the medication studies, it was assessed as low in one study (33%), unclear in one study (33%) and high in one study (33%). In the two exercise and education studies that reported rate of falls, the risk of attrition bias was assessed as low in one and high in the other.
2. Risk of falling
The risk of attrition bias in the exercise studies where data were reported relating to the risk of falling (number of people who fell at least once) was assessed as low in 73% (n = 11/15), unclear in 20% (n = 3/15) and high risk of bias in 7% (n = 1/15). In the medication studies, it was assessed as low in two studies (67%), and unclear in one study (33%). The risk of attrition bias in the risk of falling data in the education study was unclear, and in the three exercise plus education studies was low.
Selective reporting
We assessed the risk of bias due to selective reporting of the outcomes included in this review. In the exercise studies, the risk of bias was assessed as low in 44% (n = 11/25), unclear in 48% (n = 12/25) and high in 8% (n = 2/25). In the three medication studies, the risk of bias from selective reporting was unclear in two (67%) studies and high in one (25%) study. In the education study, the risk of bias due to selective reporting was unclear. In all three exercise plus education studies, the risk was assessed as low.
Bias in the recall of falls due to less reliable methods of ascertainment
We assessed the risk of bias in the recall of falls in the exercise studies as being low in 52% (n = 13/25), and unclear in the remaining 48% (n = 12/25). In the medication studies, the risk of bias in the recall of falls was assessed as low in two (67%) studies and unclear in the remaining study (33%). The education study was assessed as having a high risk of bias in the recall of falls as ascertainment of falls relied on participant recall from the prior two months. Two of the exercise plus education studies was assessed as having a low risk of bias (67%), and the risk of bias in the other was unclear (33%).
Other potential sources of bias
In undertaking the GRADE assessment, we downgraded the certainty of evidence based on the risk of bias for the following comparisons.
1. Health‐related quality of life for exercise versus control immediately after the intervention.
2. Health‐related quality of life for exercise versus control at follow‐up.
3. Number of people who fell at least once outcome for cholinesterase inhibitors versus placebo.
4. Number of people who fell at least once outcome for education versus usual care.
5. Rate of falls outcome for exercise plus education versus control.
6. Health‐related quality of life for exercise plus education versus control immediately after the intervention.
7. Health‐related quality of life for exercise plus education versus control at follow‐up.
Further details are provided in the summary of findings tables: summary of findings Table 1 (exercise compared to control); summary of findings Table 2 (cholinesterase inhibitor compared to placebo); summary of findings Table 3 (education compared to control) and summary of findings Table 4 (exercise plus education compared to control).
Effects of interventions
See: Summary of findings 1 Summary of findings for exercise compared to control; Summary of findings 2 Summary of findings for cholinesterase inhibitors compared to placebo; Summary of findings 3 Summary of findings for education compared to control; Summary of findings 4 Summary of findings for exercise plus education compared to control
Effects of exercise interventions
Exercise interventions versus control
See: summary of findings Table 1.
Rate of falls (falls per person‐year)
Compared to a control intervention (i.e. usual care or an intervention not expected to have an effect on falls, such as ‘sham’ exercise or upper limb exercise), exercise (all types combined) probably reduces the rate of falls by 26% (RaR 0.74, 95% confidence interval (CI) 0.63 to 0.87; 1456 participants, 12 studies, I2 = 30%; moderate‐certainty evidence; Analysis 1.1).
Subgroup analysis by exercise type (based on ProFaNE categories, see Table 3 and Table 5) did not find a difference in the effects of different types of exercise on fall rates (test for subgroup differences: Chi2 = 4.92, df = 2, P = 0.09, I2 = 59.3%; Analysis 1.2). Studies of gait, balance and functional training versus control had a RaR of 0.80 (95% CI 0.67 to 0.95; 1146 participants, 9 studies, I2 = 24%); studies of resistance training versus control had a RaR of 0.72 (95% CI 0.55 to 0.94; 136 participants, 2 studies, I2 = 0%); and studies of 3D exercise (Tai Chi) versus control had a RaR of 0.41 (95% CI 0.23 to 0.72; 174 participants, 2 studies, I2 = 0%).
Subgroup analysis by the proportion of exercise sessions that were supervised by a therapist (see features of exercise interventions in Table 5) found a difference in the effect of exercise (test for subgroup differences: Chi2 = 5.95, df = 1, P = 0.01, I2 = 83.2%; Analysis 1.3) with a greater reduction in the rate of falls in studies where participants were fully supervised during exercise (RaR 0.56, 95% CI 0.41 to 0.77; 373 participants, 5 studies, I2 = 21%) compared with studies where participants were not fully supervised (RaR 0.85, 95% CI 0.75 to 0.97; 1083 participants, 7 studies, I2 = 0%).
Subgroup analysis by fall risk at baseline (higher fall risk participants compared with unspecified fall risk participants) did not find a difference in the effect of exercise on fall rates in studies with different inclusion criteria (test for subgroup differences: Chi2 = 0.03, df = 1, P = 0.86, I2 = 0%; Analysis 1.4). Studies where all participants were at a high risk of falls (past falls history or identified fall risk factors) had a RaR of 0.73 (95% CI 0.59 to 0.91; 1082 participants, 7 studies, I2 = 48%) whereas studies that did not use fall risk as an inclusion criterion had a RaR of 0.71 (95% CI 0.56 to 0.90; 374 participants, 5 studies, I2 = 0%).
Most studies included participants with varying disease severity and methods for classifying disease severity varied between studies. Therefore, we performed a subgroup analysis using data from two studies (Canning 2015a; Chivers Seymour 2019) that reported subgroup analyses based on disease severity for the rate of falls (raw data reported in Table 12). Canning 2015 reported a lower disease severity (Unified Parkinson's Disease Rating Scale (UPDRS) motor score 26 or under, equivalent to an MDS‐UPDRS motor score of 33 or under (Hentz 2015)), and a higher disease severity (UPDRS motor score 27 or over, equivalent to an MDS‐UPDRS motor score of 34 or over (Hentz 2015)). Chivers Seymour 2019 reported three subgroups: low disease severity (MDS‐UPDRS motor score 22 or lower); moderate disease severity (MDS‐UPDRS motor score 23 to 38), and higher disease severity (MDS‐UPDRS motor score 39 and over) (data in Ashburn 2019). Due to the differing disease severity cut points, we pooled the low and moderate disease severity subgroups (lower disease severity) and compared them with the higher disease severity subgroups. Results showed there may be a differential intervention effect by disease severity (test for subgroup differences: Chi2 = 7.67, df = 1, P = 0.006, I2 = 87%) with an increase in fall rates with exercise in the higher disease severity subgroup (RaR 1.47, 95% CI 1.11 to 1.94; participant numbers not reported, I2 = 0%), and a slight decrease in fall rates with exercise in the lower disease severity subgroups (RaR 0.65, 95% CI 0.39 to 1.08; participant numbers not reported, I2 = 76%; Analysis 1.5). Notably, both these studies provided minimal physiotherapy supervision (Canning 2015a 13%; Chivers Seymour 2019 7%) and the exercise was performed either wholly (Chivers Seymour 2019), or mostly (Canning 2015a) at home.
| Study ID and comparison | Subgroup definition, number of participants (n) | Intervention lower severity group: falls per person year | Intervention higher severity group: falls per person year | Intervention lower severity group: number (%) of fallers | Intervention higher severity group: number (%) of fallers | Control lower severity group: falls per person year | Control higher severity group: falls per person year | Control lower severity group: number (%) of fallers | Control higher severity group: number (%) of fallers | Length of falls monitoring |
| Gait, balance and functional training vs Control
| Lower disease severity: Hoehn and Yahr stages 2 and 3, n = 96 Higher disease severity: Hoehn and Yahr stage 4, n = 30 | NA | NA | 31 (66%) | 15 (94%) | NA | NA | 37 (76%) | 12 (86%) | 6 months |
| Gait, balance and functional training vs Control
| Lower disease severity: UPDRS motor score ≤ 26, n = 122 Higher disease severity: UPDRS motor score ≥ 27, n = 109 | ND | ND | ND | ND | ND | ND | ND | ND | 6 months |
| Chivers Seymour 2019 (data reported in Ashburn 2019) Gait, balance and functional training vs Control | Lower disease severity: includes both the low disease severity subgroup ‐ MDS‐UPDRS motor score ≤ 22, n = 152 and moderate disease severity subgroup ‐ MDS‐UPDRS motor score 23 ‐ 28, n = 155 Higher disease severity: MDS‐UPDRS motor score ≥ 39, n = 152 | ND | ND | NA | NA | ND | ND | NA | NA | 6 months |
ND: no useable data; NA: not applicable (not reported as an outcome in the trial).
Number of people who experienced one or more falls (risk of falling)
Compared to a control intervention, exercise (all types combined) probably slightly reduces the number of people experiencing one or more falls by 10% (risk ratio (RR) 0.90, 95% CI 0.80 to 1.00, P = 0.05; 932 participants, 9 studies, I2 = 0%; moderate‐certainty evidence; Analysis 2.1). There was one study (Martin 2015) where all participants in both groups fell, and so these data could not be included in the meta‐analyses (Higgins 2017).
Subgroup analysis by exercise type (based on ProFaNE categories, see Table 3 and Table 5) did not show a difference in the effects of different types of exercise on the number of people who fell at least once (test for subgroup differences: Chi2 = 3.14, df = 2, P = 0.21, I2 = 36.2%; Analysis 2.2). Studies of gait, balance and functional training versus control had a RR of 0.92 (95% CI 0.81 to 1.04; 622 participants, 6 studies, I2 = 0%); studies of resistance training versus control had a RR of 0.87 (95% CI 0.43 to 1.74; 136 participants, 2 studies, I2 = 65%); and studies of 3D exercise (Tai Chi) versus control had a RR of 0.59 (95% CI 0.36 to 0.95; 174 participants, 2 studies, I2 = 12%).
Subgroup analysis by the proportion of exercise sessions that were supervised by a therapist (see features of exercise interventions in Table 5) did not show a difference in the effect of exercise on the number of people experiencing one or more falls in studies where participants were fully supervised during exercise (RR 0.75, 95% CI 0.53 to 1.06; 328 participants, 4 studies, I2 = 36%) compared with studies where participants were not fully supervised (RR 0.92, 95% CI 0.82 to 1.04; 604 participants, 5 studies, I2 = 0%); test for subgroup differences Chi2 = 1.24, df = 1, P = 0.27, I2 = 19.3%; Analysis 2.3).
Subgroup analysis by fall risk at baseline did not show a difference in the effect of exercise on the number of people experiencing one or more falls where all participants were at a high risk of falls (past falls history or identified fall risk factors; RR 0.89, 95% CI 0.76 to 1.04; 576 participants, 5 studies, I2 = 15%) compared with studies that did not use fall risk as an inclusion criterion (RR 0.86, 95% CI 0.67 to 1.11; 356 participants, 4 studies, I2 = 0%; test for subgroup differences: Chi2 = 0.06, df = 1, P = 0.81, I2 = 0%; Analysis 2.4).
As for the rate of falls, most studies were not able to be included in subgroup analysis on the effect of exercise on the risk of falls by disease severity. However, we pooled data from two studies (Ashburn 2007; Canning 2015a) that reported subgroup analyses based on disease severity (raw data presented in Table 12). Canning 2015a reported a lower disease severity (UPDRS motor score 26 or under) and a higher disease severity (UPDRS motor score 27 or over. Ashburn 2007 reported a lower disease severity (Hoehn and Yahr stage 2 or 3), and a higher disease severity (Hoehn and Yahr stage 4) subgroup. Results showed there may be a differential intervention effect by disease severity (test for subgroup differences: Chi2 = 8.14, df = 1, P = 0.004, I2 = 87.7%; Analysis 2.5). The results show there may be a slight reduction in the number of people who experienced one or more falls with exercise in the lower disease severity subgroup (RR 0.78, 95% CI 0.62 to 0.98; 218 participants; I2 = 31%), but there may be a slight increase in the proportion of people who fell at least once with exercise in the higher disease severity subgroup (RR 1.19, 95% CI 1.00 to 1.41; 139 participants; I2 = 0%). Notably, both these studies provided minimal physiotherapy supervision (Ashburn 2007 18%; Canning 2015a 13%) and the exercise was performed either wholly (Ashburn 2007) or mostly (Canning 2015a) at home.
Number of people who experienced one or more fall‐related fractures
We are uncertain of the finding that exercise may make little or no difference in the number of people experiencing one or more fall‐related fractures compared to control (RR 0.57, 95% CI 0.28 to 1.17; 989 participants, 5 studies, I2 = 0%; very low‐certainty evidence; Analysis 3.1).
Health‐related quality of life
Immediately post intervention, exercise interventions compared to control may slightly improve health‐related quality of life (standardised mean difference (SMD) ‐0.17, 95% CI ‐0.36 to 0.01; 951 participants, 5 studies; I2 = 48%; low certainty evidence; Analysis 4.1). When the SMD is converted back to a mean difference (MD) in the PDQ39, the difference is ‐2.6 (95% CI ‐5.5 to 0.2), showing the MD exceeds the minimally important difference (MID) of ‐1.6 (Peto 2001), however the 95% CI includes scores both larger and smaller than the MID.
We are uncertain of the finding that exercise improves health‐related quality of life at follow‐up (range 20 weeks to 12 months; SMD ‐0.27, 95% CI ‐0.46 to – 0.08; 429 participants, 3 studies; I2 = 0%; very low‐certainty evidence; Analysis 4.2). When the SMD is converted back to a mean difference (MD) in the PDQ39, the difference is ‐4.1 (95% CI ‐7.0 to – 1.2), which exceeds the minimally important difference of ‐1.6 (Peto 2001).
Exercise versus exercise
The results of studies comparing different types of exercise are presented for rate of falls in Analysis 5.1, for the number of people experiencing one or more falls in Analysis 6.1 and for health‐related quality of life in Analysis 7.1 (post intervention) and Analysis 7.2 (follow‐up). We did not undertake any meta‐analyses for these outcomes due to the substantial variability between exercise programs.
Some studies did find greater effects of one exercise compared to another. Treadmill walking in a virtual reality environment was found to reduce the rate of falls and improve health‐related quality of life compared to treadmill walking alone (Mirelman 2016, n = 130). Additionally, Li 2012 (n = 130) found a reduction in the number of people who fell at least once and improved health‐related quality of life following Tai Chi classes compared to functional resistance training.
The remaining studies showing effects were relatively small (range 27 to 70 participants) and their results require confirmation in different, larger studies. Gandolfi 2017 found that home‐based balance exercises using video games and delivered via telerehabilitation reduced the rate of falls compared to facility‐based balance training without the video games. Ricciardi 2015 compared standard strength, balance and gait training exercise with the same exercise targeting the more affected side, and with the same exercise targeting the less affected side. Results suggested that standard training led to a greater reduction in falls compared to training focused on the less affected side, but there were no other between group differences. Sedaghati 2016 found that balance and gait training with a ‘balance pad’ (i.e. foam mat) led to a greater reduction in the rate of falls than the same exercises without the ‘balance pad’. Similarly, Volpe 2014a reported that balance training with external perturbations was more effective in reducing the rate of falls if it was conducted while participants wore an active proprioceptive stabiliser (a device providing focal vibrations on the 7th cervical vertebra and both soleus tendons) compared to wearing inactive (placebo) devices. Smania 2010 found greater effects on the rate of falls from balance exercises compared to flexibility and co‐ordination exercises not targeting balance.
One study reported one fall‐related fracture in each group when comparing gait, balance and functional training with resistance training (Shen 2015).
Adverse events
Details regarding adverse events are presented in Table 11. Adverse events related to the exercise intervention were reported in 15 studies (2311 participants), with four of these reporting minor adverse events (Canning 2015a; Li 2012; Paul 2014; Song 2018). Canning 2015a reported that two participants experienced non‐injurious falls while undertaking unsupervised exercise at home. Li 2012 reported 26 in‐class adverse events including: two falls and one muscle soreness or pain in the Tai Chi group; four falls and four muscle soreness or pain in the functional strength training group; five falls and one muscle soreness or pain the control (stretching) group. The remaining in‐class adverse events were dizziness/faintness or symptoms of hypotension (six in the functional strength training group and three in the control group). Paul 2014 reported that in the muscle power training group there was one participant who experienced an exacerbation of pre‐existing low back pain and six participants who required modification to training loads due to transient pain, joint inflammation or illness. Song 2018 reported the stepping intervention exacerbated pre‐existing low back pain in two participants, and one participant sustained a non‐injurious fall while performing the stepping exercise. The remaining studies either reported there were no falls during the intervention (Ashburn 2007), no adverse events (Chivers Seymour 2019; Gandolfi 2017; Gandolfi 2019; Goodwin 2011; Harro 2014; Munneke 2010; Shen 2015; Wong‐Yu 2015), or no serious adverse events (Mirelman 2016; Volpe 2014a) related to the intervention.
Adverse events not attributable to the exercise intervention were monitored equally in all groups and reported in three studies (Li 2012; Mirelman 2016; Paul 2014), though one of these studies included participants with and without PD and did not report these data separately for the PD group participants (Mirelman 2016). Li 2012 reported all adverse events were minor to moderate and included falls, muscle soreness and pain, hypotension, chest pain, low back pain and ankle sprain. There were 27 adverse events occurring in the Tai Chi group, 55 in the resistance training group, and 45 in the control (stretching) group. Paul 2014 reported a pelvic fracture in one muscle power training participant and exacerbations of hernias in two control (sham exercise) participants.
Economic analysis
Six exercise studies reported costs or cost‐effectiveness data related to fall outcomes (Table 10) (Canning 2015a; Chivers Seymour 2019; Gandolfi 2017; Goodwin 2011; Li 2012; Munneke 2010). These included intervention costs, healthcare service costs and/or results of study‐based incremental costs per fall prevented/quality‐adjusted life‐year (QALY) gained. We were unable to compare incremental cost‐effectiveness ratios (ICERs) as the perspectives taken, the cost items measured, and the type of healthcare resources included in the calculations varied. Nonetheless, results from the three studies that delivered exercise at a relatively low cost and took an extensive health system perspective (Canning 2015a; Chivers Seymour 2019; Goodwin 2011) reported ICERs suggesting that exercise may be cost‐effective in preventing falls and improving health. For example, the Canning 2015a study reported a cost of $A574 per fall prevented (Farag 2016) and the Goodwin 2011 study reported total healthcare costs of ‐£4,885 per quality‐adjusted life‐year (QALY) gained (Fletcher 2012).
Effects of medication interventions versus placebo
See: summary of findings Table 2.
Two different cholinesterase inhibitors were trialled in comparison to placebo: rivastigmine and donepezil.
Rate of falls (falls per person year)
Cholinesterase inhibitors may reduce the rate of falls by 50% when compared to a placebo medication (RaR 0.50, 95% CI 0.44 to 0.58; 229 participants, 3 studies, I2 = 3%; low‐certainty evidence; Analysis 8.1). Subgroup analyses indicated that there was no difference in the effect on fall rates between rivastigmine and donepezil (test for subgroup differences: Chi2 = 0.22, df = 1, P = 0.64, I2 = 0%; Analysis 8.2; random effects meta‐analysis). We were unable to conduct any subgroup analyses based on fall risk at baseline or disease severity. All three studies included participants at high risk of falls, with two studies (Henderson 2016; Chung 2010) specifying a history of falls in the inclusion criteria and the third study (Li 2015a) including only participants with cognitive impairment, which is known to be a risk factor for falls in people with PD (Latt 2009; van der Marck 2014). None of the studies reported a subgroup analysis falls RaR for disease severity, however one study (Chung 2010) reported an observation that the five participants with the most frequent falls showed the most improvement after six weeks on donepezil (19 participants, no statistics provided).
Number of people who experienced one or more falls (risk of falling)
We are uncertain of the finding of little or no difference in the number of people experiencing one or more falls following a cholinesterase inhibitor compared to placebo (RR 1.01, 95% CI 0.90 to 1.14; 230 participants, 3 studies, I2 = 72%; very low‐certainty evidence; Analysis 9.1). Subgroup analyses indicated that there was no difference in the effect on the risk of falls between rivastigmine and donepezil (test for subgroup differences: Chi2 = 1.08, df = 1, P = 0.30, I2 = 7%; Analysis 9.2; random‐effects meta‐analysis). As for the rate of falls, we were unable to conduct any subgroup analyses based on fall risk at baseline or disease severity.
Number of people who experienced one or more fall‐related fractures
There were insufficient data from the cholinesterase inhibitor versus placebo studies to pool for the number of people sustaining one or more fall‐related fractures. One study reported no fractures in either group (Chung 2010), and the remaining two studies did not report fractures as an outcome (Henderson 2016; Li 2015a).
Health‐related quality of life
We are uncertain whether cholinesterase inhibitors make little or no difference to health‐related quality of life compared to a placebo immediately post intervention. One study reported two health‐related quality of life outcomes (EQ5D thermometer score MD 3.00, 95% CI ‐3.06 to 9.06; EQ5D index score MD ‐0.01, 95% CI ‐0.08 to 0.07; 121 participants, 1 study; very low‐certainty evidence; Analysis 10.1 and Analysis 10.2). The minimally important difference for the EQ5D index score is about 0.07 (95% CI 0.01 to 0.14) (Walters 2005).
None of the studies of medication interventions measured health‐related quality of life at follow‐up.
Rate of adverse events excluding falls (adverse events per person‐year)
Details regarding adverse events are in Table 11. Two of the medication studies (Chung 2010; Henderson 2016) reported adverse events. Most adverse events were considered to be mild and transient in nature. However, Henderson 2016 reported 27 serious adverse events (14 in the rivastigmine group and 13 in the placebo group), with two of these events (both a worsening of PD impairments) in the rivastigmine group considered likely to be related to the rivastigmine intervention.
Meta‐analysis shows that cholinesterase inhibitors may increase the rate of adverse events excluding falls by 60% when compared to a placebo medication (RaR 1.60, 95% CI 1.28 to 2.01; 175 participants, 2 studies, I2 = 16%; low‐certainty evidence; Analysis 11.1).
Economic analysis
None of the medication studies reported an economic analysis.
Effects of education versus usual care
See: summary of findings Table 3.
The single included study of an education intervention compared to usual care only provided data related to this review for the number of people who experienced one or more falls (risk of falling).
Number of people who experienced one or more falls (risk of falling)
We are uncertain whether education increases the number of people experiencing one or more falls (RR 10.89, 95% CI 1.26 to 94.03; 53 participants, 1 study; very low certainty evidence; Analysis 12.1).
Effects of exercise plus education interventions
Exercise plus education versus control
See: summary of findings Table 4.
Rate of falls (falls per person‐year)
We are uncertain whether exercise plus education compared to a control intervention makes little or no difference to the rate of falls (RaR 0.46, 95% CI 0.12 to 1.85; 320 participants, 2 studies, I2 = 87%; very low certainty evidence; Analysis 13.1).
Number of people who experienced one or more falls (risk of falling)
Exercise plus education compared to a control intervention may make little or no difference to the number of people experiencing one or more falls (RR 0.89, 95% CI 0.75 to 1.07; 352 participants, 3 studies, I2 = 0%; low‐certainty evidence; Analysis 14.1).
It was not possible to perform subgroup analyses based on the per cent of exercise supervision, as all studies utilised exercise programs with 50% or less of the exercise supervised. None of the included studies used fall risk as an inclusion criterion. One study (Cattaneo 2019) did not provide any information about disease severity, with the remaining two including predominantly people with mild to moderate disease, so subgroup analyses based on these factors was also not possible.
Number of people who experienced one or more fall‐related fractures
We are uncertain whether an exercise plus education intervention changes the number of people experiencing one or more fall‐related fractures (RR 1.45, 95% CI 0.40 to 5.32; 320 participants; very low‐certainty evidence; Analysis 15.1).
Two studies reported more than one health‐related quality of life outcomes (PDQ39, EQ5D visual analogue scale and the EQ5D Index Score) (Morris 2015; Morris 2017). We are uncertain whether an exercise plus education intervention makes little or no difference to health‐related quality of life after the intervention and at follow‐up. Results for the PD‐specific tool (the PDQ39) shows little or no change (post intervention MD 0.05, 95% CI ‐3.12 to 3.23, 305 participants, 2 studies, very low‐certainty evidence, Analysis 16.1; follow‐up MD ‐2.25, 95% CI ‐5.45 to 0.96, 299 participants, 2 studies, very low‐certainty evidence, Analysis 16.2). The minimally important difference for the PDQ39 is about ‐1.6 (Peto 2001).
Health‐related quality of life
We are uncertain whether exercise plus education makes little or no difference to health‐related quality of life immediately post intervention (PDQ39 MD 0.05, 95% CI ‐3.12 to 3.23; 305 participants, 2 studies; I2 = 0%; very low‐certainty evidence; Analysis 16.1) or at 12 months follow‐up (PDQ39 MD ‐2.25, 95% CI ‐5.45 to 0.96; 299 participants, 2 studies; I2 = 0; very low‐certainty evidence; Analysis 16.2).
Exercise plus education versus exercise plus education
One study (Morris 2015) compared two different types of exercise; gait, balance and functional training in the form of movement strategy training versus resistance training in the form of functional resistance training with weighted vests and resistance bands. Both exercise interventions were delivered with the same fall‐prevention education. The results for the rate of falls are presented in Analysis 17.1, for the number of people experiencing one or more falls in Analysis 18.1, for the number of people sustaining fall‐related fracture in Analysis 19.1 and for health‐related quality of life in Analysis 20.1 (post intervention) and Analysis 20.2 (follow‐up).
This study found that resistance training plus education reduced the rate of falls compared to movement strategy training, but there was no effect on the number of people who fell at least once, the number of people experiencing a fall‐related fracture or on health‐related quality of life (Morris 2015, n = 136).
Adverse events
Two studies of an exercise plus education intervention reported information related to adverse events (Table 11). Morris 2015 reported one fall and two occasions of dizziness during movement strategy training intervention along with 36 occasions of new muscle soreness lasting more than 24 hours (11 in the movement strategy training group and 25 in the functional strength training group). The remaining study stated that there were no adverse events related to the intervention (Morris 2017).
Economic analysis
One study of an exercise plus education intervention provided information regarding the cost of the intervention (Table 10). However, costs per fall prevented were not calculated as there was no reduction in falls in this study (Morris 2017).
Sensitivity analyses
We conducted sensitivity analyses for the pooled falls outcomes for exercise versus control, cholinesterase inhibitor versus placebo and exercise plus education versus control. A summary of these results is in Table 1 and Table 2. The results of the sensitivity analyses can be seen in Analyses 21 to 30, as per the list below.
1. Sensitivity analysis 1, removing studies with high risk of bias in any item, presented in Analysis 21.1 to Analysis 21.6.
2. Sensitivity analysis 2, removing studies with unclear or high risk of bias on random sequence generation, presented in Analysis 22.1 to Analysis 22.4.
3. Sensitivity analysis 3, removing studies with unclear or high risk of bias on allocation concealment, presented in Analysis 23.1 to Analysis 23.5.
4. Sensitivity analysis 4, removing studies with unclear or high risk of bias on assessor blinding, presented in Analysis 24.1 to Analysis 24.4.
5. Sensitivity analysis 5, removing studies with unclear or high risk of bias on incomplete outcome data, presented in Analysis 25.1 to Analysis 25.5.
6. Sensitivity analysis 6, removing studies with less than three months falls monitoring, presented in Analysis 26.1 and Analysis 26.2.
7. Sensitivity analysis 7, removing the comparisons responsible for high levels of heterogeneity, presented in Analysis 27.1 and Analysis 27.2.
8. Sensitivity analysis 8, fixed‐effect meta‐analysis, presented in Analysis 28.1 to Analysis 28.4.
9. Sensitivity analysis 9, random effects meta‐analysis, presented in Analysis 29.1 and Analysis 29.2.
10. Sensitivity analysis 10, reclassifying studies that utilised functional strength training from resistance exercise to gait, balance and functional training, presented in Analysis 30.1 and Analysis 30.2.
As shown in Table 1 and Table 2, generally these sensitivity analyses made little difference to the results of the primary pooled analyses, indicating that overall the review’s methods and findings are robust. The exception to this was the cholinesterase inhibitor versus placebo, number of people who fell at least once and exercise plus education versus control, number of falls outcomes.
In the cholinesterase inhibitor versus placebo number of people who fell at least once outcome, removing studies with a high risk of bias in any item (Sensitivity analysis 1) removed the two largest of the three included studies, and resulted in a much lower risk ratio (all studies RR 1.01, 95% CI 0.90 to 1.14; 249 participants, 3 studies; Analysis 9.1 versus Sensitivity analysis 1 RR 0.31, 95% CI 0.12 to 0.78; 81 participants, 1 study; Analysis 21.4). The remaining smaller study (Chung 2010) had a much greater effect size in favour of medication and much wider confidence intervals than the other two studies. Consequently, the certainty of the evidence for this comparison was downgraded (see summary of findings Table 2).
In the exercise plus education versus control rate of falls outcome (all studies RaR 0.46, 95% CI 0.12 to 1.85; 320 participants, 2 studies; Analysis 13.1), results were substantially changed by: i) removing studies with a high or unclear risk of bias on assessor blinding (Sensitivity analysis 4 RaR 0.24, 95% CI 0.10 to 0.61; 196 participants, 1 study; Analysis 24.3); ii) removing comparisons responsible for the high level of heterogeneity (Sensitivity analysis 7 RaR 0.24, 95% CI 0.10 to 0.61; 196 participants, 1 study; Analysis 27.2), and conducting a fixed effects meta‐analysis (Sensitivity analysis 8 RaR 0.54, 95% CI 0.33 to 0.89; 320 participants, 2 studies; Analysis 28.3). These sensitivity analyses changed the result from indicating little or no difference to the number of falls to indicating a reduction in the rate of falls (Table 1). The certainty of the evidence for this comparison was therefore downgraded (see summary of findings Table 4).
Heterogeneity
There was substantial heterogeneity in this review’s primary analysis for the effect of cholinesterase inhibitors versus placebo on the risk of falling (Chi2 = 7.23, df = 2, P = 0.03, I2 = 72%; Analysis 9.1). We were unable to use our pre‐specified subgroup analyses to explore this heterogeneity. However, removal of one of the three studies (Li 2015a) reduces the heterogeneity to a level where it is unlikely to be important (Chi2 = 0.78, df = 1, P = 0.38, I2 = 0%; Analysis 27.1) with minimal change to the RR (all studies RR 1.01, 95% CI 0.90 to 1.14; Li 2015a excluded RR 1.03, 95% CI 0.92 to 1.16). It is possible that this heterogeneity may be, in part, due to the differing inclusion criteria of these three studies, with Li 2015a including only participants with cognitive impairment (including dementia), while Henderson 2016 excluded participants with dementia and Chung 2010 excluded participants with substantial cognitive impairment (Mini‐mental State Examination score < 25). Further research is required to explore potential sources of heterogeneity in this outcome. However, given the overall stability of the results, we consider the meta‐analyses we have undertaken to be appropriate.
There was also considerable heterogeneity in this review’s primary analysis for the effect of exercise plus education versus control on the rate of falls (Chi2 = 15.16, df = 2, P = 0.0005, I2 = 87%; Analysis 13.1). Removal of one study (Morris 2017) reduced the heterogeneity to a moderate level (Chi2 = 1.96, df = 1, P = 0.16, I2 = 49%; Analysis 27.2), but altered the result from indicating little or no difference to a reduction in the number of falls (all studies RaR 0.46, 95% CI 0.12 to 1.85; Morris 2017 excluded RaR 0.24, 95% CI 0.10 to 0.61). The lack of stability in this result led to the downgrading of the certainty of this evidence. We were unable to conduct the pre‐planned subgroup analyses to explore this heterogeneity. However, the authors of Morris 2017 suggested the lack of effect on falls seen in this trial could be due to an insufficient dose of exercise plus education. As for the cholinesterase inhibitor outcome, further research is required to explore potential sources of heterogeneity in this outcome.
There was no evidence of important heterogeneity in the remaining exercise versus control, cholinesterase inhibitor versus placebo or exercise plus education versus control primary outcomes.
Small sample bias
Funnel plots were generated for the exercise versus control comparisons for both the rate of falls and the number of people who fell at least once (Figure 4 and Figure 5). These plots do show some asymmetry. However, we did not consider it sufficient to downgrade the evidence for these outcomes. There were too few comparisons to warrant generating funnel plots for the other outcomes.
Funnel plot of comparison: 1 Exercise vs control (rate of falls), outcome: 1.1 Rate of falls.
Funnel plot of comparison: 2 Exercise vs control (number of fallers), outcome: 2.1 Number of fallers.
Discussion
Summary of main results
This review explores interventions to prevent falls in people with Parkinson's disease (PD) and includes 25 studies of exercise (2700 participants), three studies of medication (242 participants), one study of education (53 participants with PD) and three studies of exercise combined with education (375 participants). All studies were of a single intervention, except for the three studies that investigated exercise combined with education.
Exercise versus control
There is moderate‐certainty evidence that exercise programs probably reduce the rate of falls (reported in 12 studies) in people with PD, and that the number of people experiencing one or more falls (reported in 9 studies) is probably slightly reduced (see summary of findings Table 1). For the rate of falls, there was an illustrative rate of 8250 falls per 1000 person‐years in people with PD in the control group, with 2145 (26%) fewer falls per 1000 person‐years in the exercise group (95% confidence interval (CI) 1072 (13%) to 3052 (37%) fewer). For the number of people who fell at least once, there was an illustrative risk of 634 fallers per 1000 people with PD in the control group, with 63 (10%) fewer fallers per 1000 people with PD in the exercise group (95% CI 127 (20%) to 0 (0%) fewer). The larger benefit on the rate of falls compared to the number of people who fell at least once suggests that while exercise probably reduces the number of falls people with PD experience, it often does not eliminate falls altogether.
The test for subgroup differences when grouped by exercise type did not show any subgroup differences for the rate of falls or the number of people who fell at least once compared to a control intervention. However, the exercise intervention delivered in most comparisons was gait, balance and functional training, (10 (71%) for the rate of falls and 6 (60%) for the number of people who fell at least once) meaning there were unlikely to be sufficient numbers of studies of alternative intervention types to find a difference if one exists. Subgroup analyses based on the baseline fall risk also did not find an effect on either fall outcome.
Subgroup analysis suggested that exercise programs that are fully supervised by a therapist may reduce the number of falls more so than exercise that was partially supervised; though this was not found for the number of people who fell at least once. Improved results with supervision could be due to several factors, such as feedback on exercise performance, encouragement and increased exercise intensity and challenge. However, fully supervised exercise is not sustainable in the context of a long‐term, neurodegenerative condition. Further work is needed to design and explore methods of identifying the appropriate level of supervision required by individuals with PD to achieve optimal outcomes throughout their disease course. In addition, identifying methods of optimising semi‐supervised exercise and service delivery aimed at fall prevention, such as using intermittent, in‐person, supervised therapy interspersed with therapy supported by telehealth (Pelicioni 2020) and/or feedback‐based technology (Canning 2020) is required.
Pooling of reported subgroups based on disease severity (two randomised controlled trials (RCTs) for each fall outcome) showed differences suggesting that exercise interventions may reduce the rate of falls and the number of people who fell at least once in participants with lower disease severity, but increase them in people with higher disease severity. There is no clear explanation for this, however there is evidence that people with more advanced disease may adhere to prescribed exercise, but compensate by reducing other exercise, which could result in an inadequate dose of exercise overall (Canning 2015a). In contrast, it is possible that improvements in mobility in the more severely affected group leads to people having more exposure to situations where they are at risk of falls (Canning 2015a; Del Din 2020). This issue requires investigation taking a precision medicine approach (Canning 2020; Nonnekes 2018), where interventions are more specifically targeted to the individual’s clinical presentation, risk factors, lifestyle and environment. In addition, analysis of fall rates relative to activity exposure will contribute to further understanding of the effectiveness of interventions designed to reduce falls (Del Din 2020).
Exercise may slightly improve health‐related quality of life compared to control immediately after the intervention (low‐certainty evidence), with conversion of the pooled result to the PDQ39 score showing that the mean difference (‐2.6) may be greater than the minimally important difference (‐1.6) (Peto 2001), though the 95% CI included scores that were smaller than the minimally important difference (‐5.5 to 0.2). However, we are uncertain whether exercise improves health‐related quality of life at follow‐up (range 20 weeks to 12 months; very low‐certainty evidence). We are also uncertain whether exercise makes little or no difference to the number of people sustaining a fall‐related fracture (very low‐certainty evidence).
Exercise versus control and exercise versus exercise
Most exercise studies (15) monitored adverse events related to the exercise intervention. Minor adverse events related to the exercise intervention were reported in four studies, primarily non‐injurious falls, excessive muscle soreness, or pain, dizziness or hypotension. Nine studies reported that there were no adverse events related to the intervention, and two reported that there were no serious adverse events. Only three studies additionally monitored for adverse events unrelated to the intervention using the same methods in all groups across the entire study period (Li 2012; Mirelman 2016; Paul 2014), though two additional studies also mentioned non‐intervention‐related adverse events (Chivers Seymour 2019; Song 2018). Overall, these results suggest that exercise is likely to be a low‐risk intervention.
Six exercise studies included in this review reported an economic evaluation. Four of these gave an indication of value for money for the interventions tested, however there were variations in the methods used which made it difficult to compare studies. There was some evidence that exercise for fall prevention in people with PD can be cost‐effective during the study period and a short time beyond. The relative cost‐effectiveness of different fall‐prevention intervention approaches in people with PD requires further exploration.
Medication versus placebo
There is low‐certainty evidence that cholinesterase inhibitors may reduce the rate of falls (reported in three RCTs) compared to placebo medication (see summary of findings Table 2). Based on an illustrative rate of 28,800 falls per 1000 person‐years in the placebo group, there were 14,400 (50%) fewer falls per 1000 person‐years in the cholinesterase inhibitor group (95% CI 12,096 (42%) to 16,128 (56%) fewer). However, we are uncertain whether this medication makes little or no difference to the number of people who fell at least once and to health‐related quality of life immediately after the intervention (very low‐certainty evidence).
We were unable to conduct the pre‐planned subgroup analyses based on fall risk at baseline or disease severity as all three studies included participants at high risk of falls, and all participants were similar in terms of disease severity.
There is low‐certainty evidence that cholinesterase inhibitors may increase the rate of non fall‐related adverse events (reported in two RCTs) compared to placebo medication. Based on an illustrative rate of 1970 adverse events per 1000 person‐years in the placebo group, there were 1182 (60%) more adverse events per 1000 person‐years in the cholinesterase inhibitor group (95% CI 552 (28%) to 1990 (101%) more). Most adverse events were mild and transient in nature, such as nausea and headache.
Education versus control
There was only one study of an education intervention compared to usual care, which provided data only for the number of people who fell at least once (summary of findings Table 3). This study provided very low‐certainty evidence; hence we are uncertain of the finding that education increases the number of people with PD who fall. The very wide confidence interval means that these data are not informative.
Exercise plus education versus control and exercise plus education versus exercise plus education
We are uncertain whether exercise plus education compared to control makes little or no difference to the rate of falls, the number of people sustaining fall‐related fractures and health‐related quality of life (all very low ‐certainty evidence, see summary of findings Table 4). Exercise plus education may make little or no difference to the number of people experiencing one or more falls (low‐certainty evidence). Based on an illustrative risk of 672 fallers per 1000 people with PD in the control group, there may be 74 (11%) fewer fallers per 1000 people with PD in the exercise plus education group (95% CI 168 (25%) fewer to 47 (7%) more).
Two of the three exercise plus education studies reported adverse events related to the exercise intervention, with one study reporting minor adverse events (Morris 2015) and the other reporting there were no adverse events (Morris 2017). This concurs with the result from the exercise studies, further supporting that exercise may be a low risk intervention.
Overall completeness and applicability of evidence
Trial design and participants
Of the 32 studies included in this review, 25 were of exercise interventions, three were medication interventions, one was an education intervention and three were exercise plus education interventions. Overall, most participants had mild to moderate PD, though the participants in the medication trials had greater disease severity than the trials of the other interventions.
In the exercise studies, 13 studies (52%) recruited participants with a recent history of falls or one or more risk factors for falling. One study (4%) recruited only participants with no recent fall history. Most participants in the exercise studies had mild to moderate disease severity, and minimal or no cognitive impairment, with only one study including people with mild cognitive impairment. These factors combined suggest that overall, many of the participants included in this review were at relatively low risk of falls and the results of this review are unlikely to be applicable to people with a high risk of falls, moderately severe to severe disease or with substantially impaired cognition.
In medication studies, three studies compared a cholinesterase inhibitor with a placebo. Two of these studies recruited only participants with a history of falls, but had vastly different inclusion criteria. One study included both occasional and frequent fallers, requiring one or more falls in the prior year, and excluded people with dementia. Another included only frequent fallers, requiring two or more falls or near falls each week without freezing of gait, and excluded those with cognitive impairment. The remaining study included only people with cognitive impairment, including dementia. Given cognitive impairment is a known risk factor for falls (Allcock 2009; Latt 2009; Paul 2013), the participants in this study were at increased risk of falls. This suggests that these results can be applied to people with PD who are at risk of falls, including people with impaired cognition.
The single education intervention trial did not report any information regarding disease severity and included both people with and without a history of falls. They also included people with some level of cognitive impairment, excluding only those with dementia.
The three studies of exercise plus education included both fallers and non‐fallers and excluded people with cognitive impairment. Two studies reported information related to disease severity, with most participants having mild to moderate disease. Therefore, the results of this review for exercise plus education interventions are unlikely to be applicable to people with moderately severe to severe disease or with substantially impaired cognition.
The illustrative fall rates and fall risk based on control/placebo group fall rates and risk varied between exercise, medication and exercise plus education studies. The illustrative fall risk (i.e. number of people who fell at least once) varied from 634 fallers per 1000 people in the exercise studies, to 672 per 1000 people in the exercise plus education study, and 774 fallers per 1000 people in the cholinesterase inhibitor studies. While somewhat variable, these values are broadly similar to the previously reported average of 60.5% (i.e. 605 fallers per 1000 people) of people with PD falling in any one year (Allen 2013). The illustrative fall rates (i.e. number of falls) had a higher variability, ranging from 8250 falls per 1000 people per year in the exercise studies, to 16,400 falls per 1000 people per year in the exercise plus education studies and up to 28,800 falls per 1000 people per year in the cholinesterase inhibitor studies. Even greater variability than this was reported in a previously published review (4700 to 67,600 falls per 1000 people per year reported in Allen 2013). The variability in the illustrative fall risk for both fall measures reflects, at least in part, the varying inclusion criteria of the different studies.
Setting
Around half the exercise studies included in this review were conducted at a facility and fully supervised, including supervised group exercise. Of the remaining studies, five included both facility and home‐based exercise, and five were solely home‐based, with solely home‐based trials having less than 50% of sessions supervised and some trials reporting as little as 5% of sessions supervised. The subgroup analysis comparing studies with 100% supervision with those with < 100% supervision found subgroup differences for the rate of falls, suggesting that exercise interventions for people with PD may be more effective in reducing falls if they are fully supervised. However, it is unlikely that fully‐supervised exercise will be cost‐effective in the long term. Therefore, identifying individuals who can exercise effectively using a semi‐supervised model has the potential to improve sustainability.
The single study of an education intervention included an individual home‐visit and a follow‐up phone call. One of the three studies of exercise plus education interventions involved home‐based exercise, while the others used a combination of facility and home‐based. Two studies provided supervision of 50% of the exercise, and also provided the education individually to participants alongside the exercise. The other study provided three supervised exercise sessions at a facility, followed by fully home‐based and unsupervised exercise. This study provided a single, one hour group education session at a facility before participants began the exercise program.
Interventions
We classified the exercise interventions according to the ProFaNE guidelines (Lamb 2011; Table 3 and Table 5). Most studies were categorised as gait, balance and functional training, with few studies of resistance training, 3D exercise, flexibility exercise or other exercise. Subgroup analyses for rate of falls and number of people who fell at least once versus control found there was no evidence for one category of exercise being superior to another. However, the small number of studies that were not categorised as gait, balance and functional training meant that our ability to find a difference between exercise interventions, should a difference exist, was limited. In addition, people with PD experience risk factors for falls over and above those attributable to ageing; such as freezing of gait, difficulty performing dual tasks and specific problems with reactive balance (van der Marck 2014). Therefore, exercise programs may include specific exercises designed to address these PD‐specific risk factors, but these details are missed when the exercise is placed in the broader category of gait, balance and functional training. Furthermore, many of the studies included in this review used various combinations of exercise types (e.g. balance, functional strength training and cueing for freezing of gait). These studies arguably reflect programs that are offered in clinical practice, and fit well into the category of gait, balance and functional training. However, other studies trialled specific single interventions, such as cueing (Martin 2015) or step training (Song 2018). Therefore, the use of a broad exercise category to include combinations of exercises and individual exercise as well as PD‐specific and non‐PD specific exercise limited our ability to explore differences between types of exercise.
Some subjectivity in the classification of exercise interventions is also apparent. In particular, we considered that functional strength exercises performed largely in standing using body weight or equipment, such as weighted vests and ankle weights, potentially could have been categorised as gait, balance and functional training rather than resistance exercise. Sensitivity analyses to re‐categorise these studies as gait, balance and functional training for the primary outcomes makes little difference to the test for subgroup differences. However, it should be noted that this re‐classification leaves only one study with a small sample size and wide confidence intervals in the resistance training category (Paul 2014).
The length of the interventions in the exercise studies was short compared to those reported for community dwelling people in the general older population (Sherrington 2019). In the present review, exercise interventions varied from six weeks to six months, with the intervention conducted over 12 weeks or less in 72% of studies. In the aforementioned review of exercise for older people, most exercise programs were 12 weeks or longer, with nearly one third of studies trialling programs of 12 months or more (Sherrington 2019).
Three medication studies compared a cholinesterase inhibitor to a placebo, with two trialling rivastigmine and one trialling donepezil. The length of time that medications were administered in these studies was highly variable, at six weeks, eight months and 12 months. There was no evidence of subgroup differences based on which cholinesterase inhibitor was trialled for either fall outcome.
The education study provided participants with a 12‐month action plan, which included fall‐prevention strategies and was delivered to them in their home by an occupational therapist in a single home visit with a follow‐up phone call.
The three studies of exercise plus education used differing approaches. One study utilised a single one‐hour falls‐prevention education session delivered to small groups of participants (two to four) by a physiotherapist. This was followed by three individual exercise sessions where the participant was taught mobility and balance exercises and asked to perform them on their own two to three times per week at home for two months. The remaining two studies both utilised individual functional progressive resistance training and movement strategy training, though one trialled these individually in two separate intervention groups for eight weeks, while the other combined these exercise interventions for six weeks. Both these studies incorporated falls prevention education into one weekly supervised session, with the other exercise session performed by participants independently.
Outcomes
We extracted data for the rate of falls, number of people who fell at least once, number of people sustaining a fall‐related fracture, health‐related quality of life, rate of adverse events and economic evaluations related to fall outcomes. Most studies of exercise versus control intervention and all the medication versus placebo studies reported both the rate of falls and the number of people who fell at least once. However, less than half of the exercise versus control studies reporting rate of falls also reported the number of people sustaining a fall‐related fracture and/or health‐related quality of life at post intervention. Similarly, only one medication study reported fracture data and health‐related quality of life at post‐test. The education study reported the risk of falling but no other data relevant to this review. All the exercise and education studies reported the number of people who fell at least once, however only two studies reported the rate of falls, the number of people sustaining a fall‐related fracture and health‐related quality of life.
There were some inconsistencies in the way studies defined and collected falls data. Most, but not all studies, defined a fall according to the definition developed by ProFaNE; that is “an unexpected event in which the participant comes to rest on the ground, floor, or lower level” (Lamb 2005) while some studies applied the more stringent criterion of “without overwhelming external force or a major internal event” (Gibson 1987). Some studies omitted to provide any clear definition, and some did not use the ProFaNE recommended protocol for ascertaining falls data (i.e. daily recording of falls with follow‐up at least monthly by researchers blinded to group allocation) (Lamb 2005). While collecting falls data in this way can be burdensome and resource intensive (Iliffe 2015), relying on recall is likely to result in an under‐reporting of falls compared to data that are recorded daily and returned monthly (Hannan 2010). Notably, most studies in this review relied on recall over the prior 6 to 12 months for baseline fall measures. One study (Chivers Seymour 2019) collected baseline fall data prospectively for three months using falls diaries, providing shorter‐term but more accurate baseline fall data. Comparability of studies would be enhanced by the adoption of a standard falls definition and method for ascertaining falls data which may be automated (van der Marck 2011). In the future, further automation of fall detection is likely to be achieved by the use of body worn sensors to monitor falls in daily life, potentially increasing the robustness of falls data (Silva de Lima 2020).
While nearly all exercise studies reported on adverse events related to the intervention, very few measured adverse events in the same way in all groups throughout the study period. This contrasts with the medication studies, where the rate of adverse events in the medication and placebo group were reported in all except one study. The lack of rigorous adverse event monitoring in the exercise studies could be due to lack of resources coupled with the high burden of reporting on participants, who are often also required to keep records of completed exercise along with falls diaries. In contrast, adverse event reporting in medication studies is viewed as a routine component of study protocols and resources allocated accordingly. Additionally, researchers of exercise interventions may consider that the relationship between any particular adverse event and the exercise intervention is more clear‐cut than in medication studies, and therefore safety of the intervention can be surmised from collecting only adverse events that are directly related to the intervention (e.g. injuries or falls when exercising). While exercise interventions generally appear to involve low risk to participants, more consistent monitoring of adverse events is required to provide stronger evidence of this safety.
Economic evaluations related to the cost of the intervention and/or fall outcomes were reported in six exercise studies and one exercise plus education study. Three of the exercise studies reported the cost per fall prevented and/or quality‐adjusted life‐year (QALY) gained. However, these evaluations used a variety of methods, perspectives, time horizons and cost items, making it difficult to compare economic results across studies and intervention types.
Ongoing studies
The design of the 30 identified ongoing studies may help to guide future research priorities for people with PD. Twenty ongoing studies will evaluate the effectiveness of exercise interventions, however only three of these studies have a target sample size of over 100 participants, indicating that most of these studies will be underpowered to find an effect on falls. Most of the exercise studies are exploring an exercise intervention that can be classified as gait, balance and functional training. Around half of the studies are evaluating the relative impact of different exercise programs. Only one exercise study has registered adverse events as an outcome, and these adverse events will be measured only during the intervention (NCT03751371). Additionally, only one ongoing study is exploring the effect of a multifactorial intervention including exercise, where the exercise is combined with environmental modification and behavioural strategies (ACTRN12619000415101).
The ongoing medication study is a phase III trial powered to find an effect on falls. This large‐scale study is comparing rivastigmine with a placebo medication and has a target sample size of 600 participants (NCT04226248).
Several ongoing studies are exploring interventions that were not included in this review, as we did not find any published studies that met the inclusion criteria. In addition to the multifactorial intervention, three small randomised cross‐over trials are investigating the effect of differing regimens of deep brain stimulation on falls in people with PD. An additional study is exploring the effect of deep brain stimulation combined with physiotherapy and another the effect of osteopathic manipulative medicine compared to education. None of these studies are powered to find an effect on falls. However, the three studies exploring different models of care have larger sample sizes, with two of these large enough to find an effect on falls (NCT05127057, n = 214; NCT04555720, n = 200).
Of note, none of the ongoing studies specifically identify fall‐related fractures as an outcome measure, few mention adverse events and only one (rivastigmine versus placebo NCT04226248) is planning a cost‐effectiveness analysis. This, along with the under‐powered sample size of most ongoing studies, highlights areas for future research. However, such research will be costly, requiring large numbers of participants.
Quality of the evidence
This review containing 32 studies (3370 participants) provides moderate‐certainty evidence regarding the effect of exercise on falls in people with PD, however the evidence regarding the effect of medication, education alone or education plus exercise is less certain, ranging from low to very low.
We assessed the certainty of the evidence using the GRADE approach, and have summarised the results in four summary of findings tables: summary of findings Table 1 (exercise compared to control); summary of findings Table 2 (cholinesterase inhibitors compared to placebo); summary of findings Table 3 (education compared to control); summary of findings Table 4 (exercise plus education compared to control).
All studies had high or unclear risk of bias in at least one area. Of note is the unclear risk of bias due to knowledge of the allocated interventions (i.e. performance bias) in most studies with an exercise intervention and in all studies incorporating education. In studies where exercise and/or education is compared to usual care or no intervention, it is not possible to blind to participants or personnel regarding whether they are involved in the intervention (exercise/education) group or not. However, the extent to which this knowledge impacts study results is unclear.
The certainty of the evidence was downgraded for indirectness in the exercise versus control and the exercise plus education versus control outcomes. This was because the included participants had overall mild to moderate disease and good cognition. Therefore, they were not representative of the population with PD seeking falls prevention interventions (Domingos 2015), as many of these people have more advanced disease and impaired cognition.
Sensitivity analysis revealed overall stability in the results for the falls outcomes in the exercise versus control, the cholinesterase inhibitor versus placebo and the exercise plus education versus control comparisons (Table 1 and Table 2). This indicates these results are robust despite the variable risk of bias across studies and are largely unchanged by the methodological choices made in undertaking the review.
There were two comparisons of outcomes where sensitivity analysis made a substantial change to the primary analysis: the number of people who fell at least once in the cholinesterase inhibitor versus placebo comparison and the rate of falls in the exercise plus education versus control comparison. The results of these sensitivity analyses led to the downgrading of the certainty of the evidence for these outcomes.
Potential biases in the review process
It is possible that some relevant studies may have been missed from this review. We attempted to minimise the risk of this occurring by comprehensively searching multiple databases, searching for studies in languages other than English, and searching the reference lists of other reviews, grey literature and trial registries. While the literature searches were run in July 2020, we ran a top‐up search in October 2021, which identified two additional small studies (n < 70). We believe the incorporation of these studies will not change the results of this review (Studies awaiting classification). These studies will be considered for inclusion when we update this review. Additionally, pairs of review authors who were blind to each other’s results both performed screening and data extraction for each study. In selecting studies for inclusion, we followed our protocol methods and excluded studies without usable data. There is a chance this could introduce bias due to selective outcome reporting.
Another potential source of bias is in the categorisation of the exercise types. We classified the exercise interventions according to the ProFaNE guidelines (Lamb 2011). We recognise there is some subjectivity in this system, and the category of gait, balance and functional training is very broad. For example, functional strength exercises performed largely in standing using body weight as well as equipment such as weighted vests and ankle weights potentially could have been categorised as resistance exercise or gait, balance and functional training. Nonetheless, sensitivity analyses that explored the effect of reclassifying these resistance exercise interventions as gait, balance and functional training did not make any important differences to the results evaluating subgroup differences based on exercise category. However, this reclassification left only one study in the resistance training category, along with the two studies in the 3D exercise category. Any possible bias in the categorisation of exercise type therefore remains unclear.
It is also possible that our use of falls data from the longest available time‐period (up to 12 months) for each study introduced bias. The length of intervention and follow‐up varied between studies. While we used a commonly‐used approach to combine all the available data (e.g., Sherrington 2019; Cameron 2018), this means that we have combined data that for some studies was collected only or predominantly during the intervention period, with data from other studies which were collected in a non‐intervention follow‐up period. We acknowledge that this is a limitation, as it could be expected that the effect of the intervention may vary over time, and that the amount of falls data returned by participants would reduce over time (Hunter 2018). Future work could explore if the results from this approach vary with results from combining data only from the intervention period and only from a non‐intervention follow‐up period.
Agreements and disagreements with other studies or reviews
This review is the first Cochrane Review to report the effect of interventions to prevent falls in people with PD. For the exercise interventions, this review extends the findings of a review of RCTs reported by Shen 2016. Shen 2016 restricted the type of exercise intervention to those that were aimed at enhancing balance and gait (including gait, balance or strength exercise) compared to a control group. Additionally, in contrast to the present review, Shen 2016 included studies that reported falls as part of monitoring for adverse events (e.g. Nieuwboer 2007). While Shen 2016 had fewer included studies, the results supported the current finding that exercise probably reduces the rate of falls. However, the Shen 2016 review found a greater reduction in falls than the present review (RaR 0.49, 95% CI 0.33 to 0.72, 605 participants, 4 studies). Furthermore, while the present review found evidence that exercise probably slightly reduces the number of people who fell at least once, Shen 2016 did not find this effect (RR 0.94, 95% CI 0.82 to 1.07, 707 participants, 4 studies). These differences in results are likely to be due to the differing inclusion criteria along with the inclusion of more recently published studies in the current review.
Subgroup analyses in this Cochrane Review suggest that exercise interventions may reduce falls in people with milder disease, but may increase them in people with more advanced disease. This result agrees with a previously published narrative review (Hulbert 2019) which reported a reduction in fall rate following exercise in participants with less severe disease, with this reduction no longer apparent when results are combined across the spectrum of disease.
Subgroup analyses in this review suggest that fully supervised exercise may be more effective in reducing the number of falls than exercise that is partially supervised. This is in contrast to a recent review (Flynn 2019) that found home‐based exercise programs with minimal supervision were effective in improving balance‐related activities, while home‐based programs that were fully supervised were not effective. The difference in result may be because most of the studies in the present review were fully supervised at a facility, whereas only home‐based programs were included in the Flynn 2019 review. Notably, the fully supervised home‐based programs included in Flynn 2019 were of a lower dose than the partially supervised home‐based programs, suggesting that the resource requirement involved in providing fully supervised exercise at home could lead to a lower dose of intervention. Given the need for fall‐prevention interventions for people with PD to be sustainable over the long term, this possible interaction between dose, supervision and exercise location warrants consideration.
We are unaware of any published reviews exploring the effects of non‐exercise interventions on falls.
Study flow diagram.
a Ashburn 2019 was identified through contacting researchers in the field.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Funnel plot of comparison: 1 Exercise vs control (rate of falls), outcome: 1.1 Rate of falls.
Funnel plot of comparison: 2 Exercise vs control (number of fallers), outcome: 2.1 Number of fallers.
Comparison 1: Exercise vs control (rate of falls), Outcome 1: Rate of falls
Comparison 1: Exercise vs control (rate of falls), Outcome 2: Rate of falls subgrouped by ProFaNE exercise categories
Comparison 1: Exercise vs control (rate of falls), Outcome 3: Rate of falls ‐ subgrouped by % supervision (100% supervision vs <100% supervision)
Comparison 1: Exercise vs control (rate of falls), Outcome 4: Rate of falls ‐ subgrouped by baseline fall risk (increased fall risk vs fall risk not specified)
Comparison 1: Exercise vs control (rate of falls), Outcome 5: Rate of falls ‐ pooled disease severity subgroup analyses_UPDRS
Comparison 2: Exercise vs control (number of fallers), Outcome 1: Number of fallers
Comparison 2: Exercise vs control (number of fallers), Outcome 2: Number of fallers subgrouped by ProFaNE exercise categories
Comparison 2: Exercise vs control (number of fallers), Outcome 3: Number of fallers ‐ subgrouped by % supervision (100% supervision vs <100% supervision)
Comparison 2: Exercise vs control (number of fallers), Outcome 4: Number of fallers ‐ subgrouped by baseline fall risk (increased fall risk vs fall risk not specified)
Comparison 2: Exercise vs control (number of fallers), Outcome 5: Number of fallers ‐ pooled disease severity subgroup analyses
Comparison 3: Exercise vs control (number of people sustaining one or more fall‐related fractures), Outcome 1: Number of people sustaining one or more fall‐related fractures
Comparison 4: Exercise vs control (health‐related quality of life), Outcome 1: Health‐related quality of life ‐ combined measures post intervention
Comparison 4: Exercise vs control (health‐related quality of life), Outcome 2: Health‐related quality of life ‐ combined measures follow‐up
Comparison 5: Exercise vs exercise (rate of falls), Outcome 1: Rate of falls, different types of exercise compared
Comparison 6: Exercise vs exercise (number of fallers), Outcome 1: Number of fallers, different types of exercise compared
Comparison 7: Exercise vs exercise (health‐related quality of life), Outcome 1: Quality of life ‐ combined measures post intervention, different types of exercise compared
Comparison 7: Exercise vs exercise (health‐related quality of life), Outcome 2: Quality of life ‐ combined measures follow‐up, different types of exercise compared
Comparison 8: Cholinesterase inhibitor vs placebo (rate of falls), Outcome 1: Rate of falls
Comparison 8: Cholinesterase inhibitor vs placebo (rate of falls), Outcome 2: Rate of falls ‐ subgrouped by medication
Comparison 9: Cholinesterase inhibitor vs placebo (number of fallers), Outcome 1: Number of fallers
Comparison 9: Cholinesterase inhibitor vs placebo (number of fallers), Outcome 2: Number of fallers ‐ subgrouped by medication
Comparison 10: Cholinesterase inhibitor vs placebo (health‐related quality of life), Outcome 1: Quality of life EQ5D thermometer post intervention
Comparison 10: Cholinesterase inhibitor vs placebo (health‐related quality of life), Outcome 2: Quality of life EQ5D Index Score post intervention
Comparison 11: Cholinesterase inhibitor vs placebo (rate of adverse events excluding falls), Outcome 1: Rate of adverse events excluding falls
Comparison 12: Education vs usual care (number of fallers), Outcome 1: Number of fallers
Comparison 13: Exercise and education vs control (rate of falls), Outcome 1: Rate of falls
Comparison 14: Exercise and education vs control (number of fallers), Outcome 1: Number of fallers
Comparison 15: Exercise and education vs control (number of people sustaining one or more fall‐related fractures), Outcome 1: Number of people sustaining one or more fall‐related fractures
Comparison 16: Exercise and education vs control (health‐related quality of life), Outcome 1: Health‐related quality of life ‐ Parkinson's Disease Questionnaire (PDQ39) post intervention
Comparison 16: Exercise and education vs control (health‐related quality of life), Outcome 2: Health‐related quality of life ‐ Parkinson's Disease Questionnaire (PDQ39) at follow‐up
Comparison 17: Exercise and education vs exercise and education (rate of falls), Outcome 1: Rate of falls
Comparison 18: Exercise and education vs exercise and education (number of fallers), Outcome 1: Number of fallers
Comparison 19: Exercise and education vs exercise and education (number of people sustaining one or more fall‐related fractures), Outcome 1: Number of people sustaining one or more fall‐related fractures
Comparison 20: Exercise and education vs exercise and education (health‐related quality of life), Outcome 1: Health‐related quality of life ‐ Parkinson's Disease Questionnaire (PDQ39) post intervention
Comparison 20: Exercise and education vs exercise and education (health‐related quality of life), Outcome 2: Health‐related quality of life ‐ Parkinson's Disease Questionnaire (PDQ39) at follow‐up
Comparison 21: Sensitivity analysis 1: excluding studies at a high risk of bias in any item, Outcome 1: Rate of falls ‐ exercise vs control
Comparison 21: Sensitivity analysis 1: excluding studies at a high risk of bias in any item, Outcome 2: Number of fallers ‐ exercise vs control
Comparison 21: Sensitivity analysis 1: excluding studies at a high risk of bias in any item, Outcome 3: Rate of falls ‐ cholinesterase inhibitor vs placebo
Comparison 21: Sensitivity analysis 1: excluding studies at a high risk of bias in any item, Outcome 4: Number of fallers ‐ cholinesterase inhibitor vs placebo
Comparison 21: Sensitivity analysis 1: excluding studies at a high risk of bias in any item, Outcome 5: Rate of falls ‐ exercise and education vs control
Comparison 21: Sensitivity analysis 1: excluding studies at a high risk of bias in any item, Outcome 6: Number of fallers ‐ exercise and education vs control
Comparison 22: Sensitivity analysis 2: excluding studies with unclear or high risk of bias on random sequence generation, Outcome 1: Rate of falls ‐ exercise vs control
Comparison 22: Sensitivity analysis 2: excluding studies with unclear or high risk of bias on random sequence generation, Outcome 2: Number of fallers ‐ exercise vs control
Comparison 22: Sensitivity analysis 2: excluding studies with unclear or high risk of bias on random sequence generation, Outcome 3: Rate of falls ‐ cholinesterase inhibitor vs placebo
Comparison 22: Sensitivity analysis 2: excluding studies with unclear or high risk of bias on random sequence generation, Outcome 4: Number of fallers ‐ cholinesterase inhibitor vs placebo
Comparison 23: Sensitivity analysis 3: excluding studies with unclear or high risk of bias on allocation concealment, Outcome 1: Rate of falls ‐ exercise vs control
Comparison 23: Sensitivity analysis 3: excluding studies with unclear or high risk of bias on allocation concealment, Outcome 2: Number of fallers ‐ exercise vs control
Comparison 23: Sensitivity analysis 3: excluding studies with unclear or high risk of bias on allocation concealment, Outcome 3: Rate of falls ‐ cholinesterase inhibitor vs placebo
Comparison 23: Sensitivity analysis 3: excluding studies with unclear or high risk of bias on allocation concealment, Outcome 4: Number of fallers ‐ cholinesterase inhibitor vs placebo
Comparison 23: Sensitivity analysis 3: excluding studies with unclear or high risk of bias on allocation concealment, Outcome 5: Number of fallers ‐ exercise and education vs control
Comparison 24: Sensitivity analysis 4, excluding studies with unclear or high risk of bias on assessor blinding, Outcome 1: Rate of falls ‐ exercise vs control
Comparison 24: Sensitivity analysis 4, excluding studies with unclear or high risk of bias on assessor blinding, Outcome 2: Number of fallers ‐ exercise vs control
Comparison 24: Sensitivity analysis 4, excluding studies with unclear or high risk of bias on assessor blinding, Outcome 3: Rate of falls ‐ exercise and education vs control
Comparison 24: Sensitivity analysis 4, excluding studies with unclear or high risk of bias on assessor blinding, Outcome 4: Number of fallers ‐ exercise and education vs control
Comparison 25: Sensitivity analysis 5, excluding studies with unclear or high risk of bias on incomplete outcome data, Outcome 1: Rate of falls ‐ exercise vs control
Comparison 25: Sensitivity analysis 5, excluding studies with unclear or high risk of bias on incomplete outcome data, Outcome 2: Number of fallers ‐ exercise vs control
Comparison 25: Sensitivity analysis 5, excluding studies with unclear or high risk of bias on incomplete outcome data, Outcome 3: Rate of falls ‐ cholinesterase inhibitor vs placebo
Comparison 25: Sensitivity analysis 5, excluding studies with unclear or high risk of bias on incomplete outcome data, Outcome 4: Number of fallers ‐ cholinesterase inhibitor vs placebo
Comparison 25: Sensitivity analysis 5, excluding studies with unclear or high risk of bias on incomplete outcome data, Outcome 5: Rate of falls ‐ exercise and education vs control
Comparison 26: Sensitivity analysis 6, excluding studies with less than three months falls monitoring, Outcome 1: Rate of falls ‐ exercise vs control
Comparison 26: Sensitivity analysis 6, excluding studies with less than three months falls monitoring, Outcome 2: Number of fallers ‐ exercise vs control
Comparison 27: Sensitivity analysis 7, excluding comparisons responsible for the high level of heterogeneity, Outcome 1: Number of fallers ‐ cholinesterase inhibitor vs placebo
Comparison 27: Sensitivity analysis 7, excluding comparisons responsible for the high level of heterogeneity, Outcome 2: Rate of falls ‐ exercise and education vs control
Comparison 28: Sensitivity analysis 8, fixed‐effect meta‐analysis, Outcome 1: Rate of falls ‐ exercise vs control
Comparison 28: Sensitivity analysis 8, fixed‐effect meta‐analysis, Outcome 2: Number of fallers ‐ exercise vs control
Comparison 28: Sensitivity analysis 8, fixed‐effect meta‐analysis, Outcome 3: Rate of falls ‐ exercise and education vs control
Comparison 28: Sensitivity analysis 8, fixed‐effect meta‐analysis, Outcome 4: Number of fallers ‐ exercise and education vs control
Comparison 29: Sensitivity analysis 9, random effects meta‐analysis, Outcome 1: Rate of falls ‐ cholinesterase inhibitor vs placebo
Comparison 29: Sensitivity analysis 9, random effects meta‐analysis, Outcome 2: Number of fallers ‐ cholinesterase inhibitor vs placebo
Comparison 30: Sensitivity analysis 10, reclassifying functional resistance training from resistance training to gait, balance and functional training, Outcome 1: Rate of falls ‐ exercise vs control
Comparison 30: Sensitivity analysis 10, reclassifying functional resistance training from resistance training to gait, balance and functional training, Outcome 2: Number of fallers ‐ exercise vs control
| Exercise (all types) compared with control (e.g. usual activities) for preventing falls in people with Parkinson's disease | ||||||
| Patient or population: People with Parkinson's disease Settings: Any Intervention: Exercise of all types Comparison: Control ‐ usual care or a non‐active intervention | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Control | Exercise (all types) | |||||
| Rate of Falls (falls per person‐year) Follow‐up: range 2 weeks to 12 months | All exercise trials population | Rate ratio 0.74 (0.63 to 0.87) | 1456 | ⊕⊕⊕⊝ | Overall, exercise probably reduces the number of falls by 26% (95% CI 37% reduction to 13% reduction).
| |
| 8250 falls per 1000 people
| 6105 falls per 1000 people | |||||
| Number of people who experienced one or more falls Follow‐up: range 2 weeks to 12 months | All exercise trials population | Risk ratio 0.90 (0.80 to 1.00) | 932 | ⊕⊕⊕⊝ | Overall, exercise probably slightly reduces the number of people experiencing one or more falls by 10% (95% CI 20% reduction to no change).
| |
| 634 fallers per 1000 people
| 571 fallers per 1000 people | |||||
| Number of people sustaining one or more fall‐related fractures Follow‐up: range 20 weeks to 12 months | All exercise trials population | Risk ratio 0.57 (0.28 to 1.17) | 989 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the findings that exercise may make little or no difference in the number of people experiencing one or more fall‐related fractures. | |
| 40 people with fracture per 1000
| 23 people with fracture per 1000 | |||||
| Quality of life immediately after the intervention assessed with various measures Follow‐up: range 8 weeks to 6 months A lower score indicates better quality of life | ‐ | The mean quality of life score in the intervention groups was 0.17 standard deviations lower (0.36 lower to 0.01 higher). |
| 951 | ⊕⊕⊝⊝ | Overall, exercise may slightly improve quality of life by 2.6 points in the PDQ39 score (MD = 2.6 lower, 95% CI 5.5 lower to 0.2 higher). Of note is that the 95% CI includes the possibility of both increased and no change in quality of life.
The SMD was converted back to MD using the PDQ39 scale (0‐100), using the pooled SD from the baseline scores of the largest included trial (Chivers Seymour 2019). The MID for the PDQ39 is about 1.6 (Peto 2001).
SMD was calculated from 2 trials using the PDQ39, 1 trial using the PDQ8, 1 trial using the EQ‐5D visual analogue scale and 1 trial using the EQ‐5D index score. |
| Adverse events | Adverse events were reported inconsistently and often only for the exercise group. Three studies reported there were no adverse events related to the exercise intervention and one reported there were no falls during exercise. The remaining four studies reported minor adverse events such as muscle or joint soreness and non‐injurious falls. | Not estimable | 1242 (8 RCTs) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether exercise has an effect on adverse events. | |
| Economic outcomes | We were unable to compare ICERs due to variations in the methods used, however reported ICERs suggest that exercise may be cost‐effective in preventing falls. | Not estimable | 923 (4 RCTs) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether exercise is a cost‐effective intervention for falls prevention. | |
| *The assumed risk is the median control group risk across studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. | ||||||
| aDowngraded due to indirectness as most of the included participants had mild to moderate disease and good cognition. There was no downgrading for risk of bias as most trials had low or unclear risk of bias and the unclear risk of bias (predominantly performance bias and detection bias) unlikely to lower the confidence in the estimation of the effect. bDowngraded due to indirectness as most of the included participants had mild to moderate disease and good cognition. Downgraded by two levels due to imprecision as there was a small number of events and a wide confidence interval. There was no downgrading for risk of bias as most trials had low or unclear risk of bias and the unclear risk of bias (predominantly performance bias and detection bias) unlikely to lower the confidence in the estimation of the effect. cDowngraded by one level due to risk of bias as most trials were at high or unclear risk of bias for performance bias and detection bias as quality of life is a self‐reported measure. Downgraded by a further level due to indirectness as most of the included participants had mild to moderate disease and good cognition. dDowngraded by three levels due to incomplete data.
| ||||||
| Cholinesterase inhibitors compared with placebo medication for preventing falls in people with Parkinson's disease | ||||||
| Patient or population: People with Parkinson's disease Settings: Any Intervention: cholinesterase inhibitor medication (rivastigmine, donepezil) Comparison: placebo medication | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Placebo | Cholinesterase inhibitor | |||||
| Rate of falls (falls per person‐year) Follow‐up: range 12 weeks to 12 months | Cholinesterase inhibitor trial population | Rate ratio 0.50 (0.44 to 0.58) | 229 | ⊕⊕⊝⊝ | Overall, cholinesterase inhibitors may reduce the number of falls by 50% (95% CI 42% reduction to 56% reduction).
| |
| 28,800 falls per 1000
| 14,400 falls per 1000 | |||||
| Number of people who experienced one or more falls Follow‐up: range 12 weeks to 12 months | Cholinesterase inhibitor trial population | Risk rato 1.01 (0.90 to 1.14) | 230 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that cholinesterase inhibitors make little or no difference to the number of people experiencing one or more falls. | |
| 774 fallers per 1000
| 782 fallers per 1000 | |||||
| Number of people sustaining one or more fall‐related fractures Follow‐up: 12 weeks | Reported in one study, with no fractures in either group. | Not estimable | 23 (1 RCT) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether cholinesterase inhibitors make little or no difference to the number of people sustaining one or more fall‐related fractures. | |
| Quality of Life immediately after the intervention (EQ5D Thermometer, scale 0 to 100; and EQ5D Index Score, scale 0‐1, high score is better quality of life) Follow‐up: 8 months | The mean EQ5D thermometer score was 63 and the mean EQ5D Index Score was 0.66 in the placebo group. | In the cholinesterase inhibitor group the mean EQ5D Thermometer Score was 3 points higher (3.06 lower to 9.06 higher) and the mean EQ5D Index Score was 0.01 points lower (0.08 lower to 0.07 higher). |
| 121 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that cholinesterase inhibitors may make little or no difference to health‐related quality of life immediately after the intervention. |
| Rate of adverse events excluding falls (per person year) Follow‐up: range 12 weeks to 8 months
| Cholinesterase inhibitor trial population | Rate ratio 1.60 (1.28 to 2.01) | 175 | ⊕⊕⊝⊝ | Overall, cholinesterase inhibitors may increase the number of non fall‐related adverse events by 60% (95% CI 28% increase to 101% increase).
| |
| 1970 adverse events per 1000
| 3152 adverse events per 1000 | |||||
| Economic outcomes |
|
|
|
| No data reported for this outcome | |
| *The assumed risk is the median control group risk across studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. | ||||||
| aDowngraded by two levels for imprecision due to the relatively small sample size. There was no downgrading for risk of bias as the sensitivity analyses to remove trials at high risk of bias in any item, or high/unclear risk of bias in any domain, made little difference to the result (Table 1). bDowngraded by one level due to risk of bias as results changed when removing the two trials with a high risk of bias in any item (Henderson 2016, Chung 2010) (Table 2). Downgraded an additional two levels due to imprecision because of the relatively small sample size. There was no downgrading for inconsistency as results were essentially unchanged with removal of the comparison responsible for the high heterogeneity (Li 2015a) (Table 2). cDowngraded by two levels for imprecision due to the very small sample size. Downgraded a further one level as only one of the three studies included in the review for this comparison contributed to the outcome. dDowngraded by two levels for imprecision due to the relatively small sample size. Downgraded a further one level as only one of the three studies included in the review for this comparison contributed to the outcome. eDowngraded by two levels for imprecision due to the relatively small sample size. | ||||||
| Sensitivity analysis | Pooled impact of intervention on fall rate, Rate ratio, 95% CI |
| Exercise trials vs control | |
| Primary analysis, all trials, random effects meta‐analysis | 0.74, 0.63 to 0.87; participants = 1456; trials = 12 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.74, 0.61 to 0.90; participants = 1,245; trials = 9 |
| Sensitivity analysis 2, removing trials with unclear or high risk of bias on random sequence generation | 0.90, 0.76 to 1.05; participants = 995; trials = 7 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 0.80, 0.70 to 0.91; participants = 1299; trials = 8 |
| Sensitivity analysis 4, removing trials with unclear or high risk of bias on assessor blinding | 0.92, 0.73 to 1.16; participants = 692; trials = 2 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 0.77, 0.65 to 0.92; participants = 1260; trials = 11 |
| Sensitivity analysis 6, removing trials with less than three months falls monitoring | 0.79, 0.68 to 0.92; participants = 1268; trials = 9 |
| Sensitivity analysis 8, all exercise trials, fixed effects meta‐analysis | 0.79, 0.71 to 0.88; participants = 1456; trials = 12 |
| Primary analysis, subgrouped by exercise type Gait, balance and functional training Resistance training 3D exercise | 0.80, 0.67 to 0.95; participants = 1146; trials = 9 0.72, 0.55 to 0.94; participants = 137; trials = 2 0.41, 0.23 to 0.72; participants = 174; trials = 2 Test for subgroup differences Chi2 = 4.92, df = 2 (P = 0.09), I2 = 59.3% |
| Sensitivity analysis 10, classification of interventions that included functional strength training from resistance training to gait, balance and functional training Gait, balance and functional training Resistance training 3D exercise | 0.78, 0.68 to 0.91; participants = 1244; trials = 10 0.84, 0.28 to 2.53; participants = 38; trials = 1 0.41, 0.23 to 0.72; participants = 174; trials = 2 Test for subgroup differences Chi2 = 4.8, df = 2 (P = 0.09), I2 = 58.3% |
| Medication trials ‐ cholinesterase inhibitor vs placebo | |
| Primary analysis, all trials, fixed effects meta‐analysis | 0.50, 0.44 to 0.58; participants = 229; trials = 3 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.43, 0.32 to 0.56; participants = 81; trials = 1 |
| Sensitivity analysis 2, removing trials with unclear or high risk of bias on random sequence generation | 0.60, 0.38 to 0.96; participants = 129; trials = 1 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 0.60, 0.38 to 0.96; participants = 129; trials = 1 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 0.60, 0.38 to 0.96; participants = 129; trials = 1 |
| Sensitivity analysis 9, all cholinesterase inhibitor trials, random effects meta‐analysis | 0.50, 0.43 to 0.58; participants = 229; trials = 3 |
| Exercise plus education trials vs control | |
| Primary analysis, all trials, random effects meta‐analysis | 0.46, 0.12 to 1.85; participants = 320; trials = 2 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 1.58, 0.74 to 3.40; participants = 124; trials = 1 |
| Sensitivity analysis 4, removing trials with unclear or high risk of bias on assessor blinding | 0.24, 0.10 to 0.61; participants = 196; trials = 1 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 1.58, 0.74 to 3.40; participants = 124; trials = 1 |
| Sensitivity analysis 7, removing the comparison responsible for the high level of heterogeneity (Morris 2017) | 0.24, 0.10 to 0.61; participants = 196; trials = 1 |
| Sensitivity analysis 8, all exercise plus education trials, fixed effects meta‐analysis | 0.54, 0.33 to 0.89; participants = 320; trials = 2 |
| Sensitivity analysis | Pooled impact of intervention on risk of falling, Risk ratio, 95% CI |
| Exercise trials vs control | |
| Primary analysis, all exercise trials, random effects meta‐analysis | 0.90, 0.80 to 1.00; participants = 932; trials = 9 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.87, 0.75 to 1.02; participants = 721; trials = 6 |
| Sensitivity analysis 2, removing trials with unclear or high risk of bias on random sequence generation | 0.89, 0.76 to 1.04; participants = 516; trials = 5 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 0.91, 0.81 to 1.03; participants = 838; trials = 7 |
| Sensitivity analysis 4, removing trials with unclear or high risk of bias on assessor blinding | 0.93, 0.78 to 1.11; participants = 231; trials = 1 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 0.89, 0.79 to 1.00; participants = 736; trials = 8 |
| Sensitivity analysis 6, removing trials with less than three months falls monitoring | 0.89, 0.77 to 1.02; participants = 789; trials = 7 |
| Sensitivity analysis 8, all exercise trials, fixed effects meta‐analysis | 0.90, 0.80 to 1.00; participants = 932; trials = 9 |
| Primary analysis, subgrouped by exercise type Gait, balance and functional training Resistance training 3D exercise | 0.92, 0.81 to 1.04; participants = 622; trials = 6 0.87, 0.43 to 1.74; participants = 136; trials = 2 0.59, 0.36 to 0.95; participants = 174; trials = 2 Test for subgroup differences Chi2 = 3.14, df = 2 (P = 0.21), I2 = 36.2% |
| Sensitivity analysis 10, classification of interventions that included functional strength training from resistance training to gait, balance and functional training Gait, balance and functional training Resistance training 3D exercise | 0.93, 0.83 to 1.05; participants = 720; trials = 7 0.58, 0.30 to 1.13; participants = 38; trials = 1 0.59, 0.36 to 0.95; participants = 174; trials = 2 Test for subgroup differences Chi2 = 5.02, df = 2 (P = 0.08), I2 = 60.1% |
| Medication trials ‐ cholinesterase inhibitor vs placebo | |
| Primary analysis, all trials, fixed effects meta‐analysis | 1.01, 0.90 to 1.14; participants = 230; trials = 3 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.31, 0.12 to 0.78; participants = 81; trials = 1 |
| Sensitivity analysis 2, removing trials with unclear or high risk of bias on random sequence generation | 1.00, 0.87 to 1.15; participants = 130; trials = 1 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 1.00, 0.87 to 1.15; participants = 130; trials = 1 |
| Sensitivity analysis 5, removing trials with unclear or high risk of bias on incomplete outcome data | 1.03, 0.92 to 1.16; participants = 149; trials = 2 |
| Sensitivity analysis 7, removing the comparison responsible for the high level of heterogeneity (Li 2015a) | 1.03, 0.92 to 1.16; participants = 149; trials = 2 |
| Sensitivity analysis 9, all cholinesterase inhibitor trials, random effects meta‐analysis | 0.95, 0.70 to 1.28; participants = 230; trials = 3 |
| Exercise plus education trials vs control | |
| Primary analysis, all trials, random effects meta‐analysis | 0.89, 0.75 to 1.07; participants = 352; trials = 3 |
| Sensitivity analysis 1, removing trials with high risk of bias in any item | 0.84, 0.65 to 1.08; participants = 156; trials = 2 |
| Sensitivity analysis 3, removing trials with unclear or high risk of bias on allocation concealment | 0.90, 0.75 to 1.08; participants = 320, trials = 2 |
| Sensitivity analysis 4, removing trials with unclear or high risk of bias on assessor blinding | 0.93, 0.73 to 1.19; participants = 228, trials = 2 |
| Sensitivity analysis 8, all exercise plus education trials, fixed effects meta‐analysis | 0.89, 0.75 to 1.07; participants = 352; trials = 3 |
| Health education compared with usual care for preventing falls in people with Parkinson's disease | ||||||
| Patient or population: People with Parkinson's disease Settings: Any Intervention: Education about falls prevention Comparison: Usual care | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Usual care | Health education | |||||
| Rate of falls (falls per person‐year) |
|
|
|
| No data reported for this outcome | |
| Number of people who experienced one or more falls Follow‐up: 12 months | All exercise trials population* | Risk ratio 10.89 (1.26 to 94.03) | 53 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that health education increases the number of people who experience one or more falls. | |
| 634 fallers per 1000 people | 6,911 per 1000 | |||||
| Number of people sustaining one or more fall‐related fractures |
|
|
|
| No data reported for this outcome | |
| Quality of life |
|
|
|
| No data reported for this outcome | |
| Adverse events |
|
|
|
| No data reported for this outcome | |
| Economic outcomes |
|
|
|
| No data reported for this outcome | |
| *The assumed risk is the median control group risk across exercise versus control studies, as there were no data to calculate the illustrative risk in the health education trial. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. | ||||||
| aDowngraded three levels due to risk of bias (single study with high risk of bias for method of ascertaining falls (recall bias) and unclear risk for allocation concealment, performance bias, detection bias, attrition bias and reporting bias). Also downgraded for imprecision due to the relatively small sample size and very wide confidence interval. | ||||||
| Exercise (all types) plus education for falls prevention compared with control (e.g. usual activities) for preventing falls in people with Parkinson's disease | ||||||
| Patient or population: People with Parkinson's disease Settings: Any Intervention: Exercise of all types plus fall‐prevention education Comparison: Control ‐ usual care or a non‐active intervention | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Control | Exercise plus education | |||||
| Rate of Falls (falls per person‐year) Follow‐up: 12 months | Exercise plus education trials population | Rate ratio 0.46 (0.12 to 1.85) | 320 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that exercise plus education makes little or no difference to the number of falls. | |
| 16,400 falls per 1000 people | 7,544 per 1000 | |||||
| Number of people who experienced one or more falls Follow‐up: range 6 months to 12 months | Exercise plus education trials population | Risk Ratio 0.89 (0.75 to 1.07) | 352 | ⊕⊕⊝⊝ | Overall, exercise plus education may make little or no difference to the number of people experiencing one or more falls (11% reduction (95% CI 25% reduction to 7% increase)).
| |
| 672 per 1000 | 598 per 1000 | |||||
| Number of people sustaining one or more fall‐related fractures Follow‐up: 12 months | Exercise plus education trials population | Risk ratio 1.45 (0.40 to 5.32) | 320 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that exercise plus education makes little or no difference to the number of people experiencing one or more fall‐related fractures. | |
| 25 per 1000 | 36 per 1000 | |||||
| Quality of life immediately after the intervention assessed with the PDQ39 (range 0 to 100) Follow‐up: 6 weeks A lower score indicates better quality of life | ‐ | The mean PDQ39 in the intervention groups was |
| 305 | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain of the finding that exercise plus education makes little or no difference to health‐related quality of life immediately after the intervention. |
| Adverse events | Adverse events related to the exercise intervention only were reported. One study reported there were no adverse events, while the other reported minor adverse events such as muscle soreness and a fall while exercising. | Not estimable | 343 (2 RCTs) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether exercise plus education has an effect on adverse events. | |
| Economic Outcomes | Costs per fall prevented were not calculated as there was no reduction in falls in this study | Not estimable | 133 (1 RCT) | ⊕⊝⊝⊝ | The evidence is of very low certainty, hence we are uncertain whether exercise plus education is a cost‐effective intervention for falls prevention. | |
| *The assumed risk the median control group risk across studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. | ||||||
| aDowngraded by three levels due to risk of bias as results changed when removing the study with a high risk of bias on assessor blinding (Morris 2017) and for inconsistency due to a high level of heterogeneity, with the result changed when the comparison responsible for the high heterogeneity (Morris 2017) was removed (Table 1). Downgraded for imprecision due to the wide confidence interval and small sample size and indirectness as most of the included participants had mild to moderate disease and good cognition. Additionally downgraded as the result changed when fixed effects analysis was used (Table 1). bDowngraded one level for imprecision due to the relatively small sample size and an additional level for indirectness as most of the included participants had mild to moderate disease and good cognition. There was no downgrading for risk of bias as the sensitivity analyses to remove trials at high risk of bias in any item, or high/unclear risk of bias in any domain, made little difference to the result (Table 2). cDowngraded by two levels for imprecision due to the relatively small sample size, the small number of events and the very wide confidence interval. Downgraded a further level for indirectness as most of the included participants had mild to moderate disease and good cognition. dDowngraded by one level due to risk of bias as the studies included for this outcome were at unclear risk of bias for performance bias and high risk of bias for detection bias as quality of life is a self‐reported measure. Downgraded by one level for imprecision due to the relatively small sample size and wide confidence interval. Downgraded a further level for indirectness as most of the included participants had mild to moderate disease and good cognition. eDowngraded by three levels due to incomplete data and serious risk of bias from reporting bias. fDowngraded by two levels for imprecision due to the small sample size. Downgraded a further level for indirectness as most of the included participants had mild to moderate disease and good cognition. Downgraded a further one level as only one of the three studies included in the review for this comparison contributed to the outcome. | ||||||
| Exercise Category | ProFaNE exercise description | How the criteria were applied in this review* |
| Gait, balance and functional training | Gait training involves specific correction of walking technique (e.g., posture, stride length and cadence) and changes of pace, level and direction. Balance training involves the efficient transfer of bodyweight from one part of the body to another or challenges specific aspects of the balance systems (e.g. vestibular systems). Balance retraining activities range from the re‐education of basic functional movement patterns to a wide variety of dynamic activities that target more sophisticated aspects of balance. Functional training utilises functional activities as the training stimulus, and is based on the theoretical concept of task specificity. All gait, balance and functional training should be based on an assessment of the participant's abilities prior to starting the program; tailoring of the intervention to the individuals abilities; and progression of the exercise program as ability improves. | Selected as the primary exercise category when the majority of the exercise was conducted in standing and when the intervention focus and the majority of time spent was on exercise in this category. Movement strategy training and cueing are included in this category. |
| Resistance training | The term Resistance Training covers all types of weight training.i.e contracting the muscles against a resistance to overload and bring about a training effect in the muscular system. The resistance is an external force, which can be ones own body placed in an unusual relationship to gravity (e.g. prone back extension) or an external resistance (e.g. free weight). All strength/resistance training should be based on an assessment of the participant's abilities prior to starting the program; tailoring of the intervention to the individuals abilities; and progression of the exercise program as ability improves. | Selected as the primary category for interventions where additional resistance was used or where it was clear that overload was sufficient without external resistance and where the intervention focus and the majority of time spent was on exercise in this category. |
| Flexibility | Flexibility training is the planned process by which stretching exercises are practised and progressed to restore or maintain the optimal Range Of Movement (ROM) available to a joint or joints. The ranges of motion used by flexibility programs may vary from restoration/maintenance of the entire physiological range of motion, or alternatively, maintenance of range that is essential to mobility or other functions. | Selected as the primary category for interventions where flexibility training was a stated aim of the intervention and where the intervention focus and the majority of time spent was on exercise in this category. |
| 3D | 3D training involves constant movement in a controlled, fluid, repetitive way through all 3 spatial planes or dimensions (forward and back, side to side, and up and down). Tai Chi and Qi Gong incorporate specific weight transferences and require upright posture and subtle changes of head position and gaze direction. Dance involves a wide range of dynamic movement qualities, speeds and patterns. | Selected as the primary exercise category where the intervention focus and the majority of time was spent on exercise in this category (e.g., Tai Chi or dance). |
| General Physical activity | Physical activity is any bodily movement produced by skeletal muscle contraction resulting in a substantial increase in energy expenditure. Physical activity has occupational, transportation and recreational components and includes pursuits like golf, tennis and swimming. It also includes other activities and pastimes like gardening, cutting wood and carpentry. Physical activity can provide progressive health benefits and is a catalyst for improving health attitudes, health habits and lifestyle. Increasing habitual physical activity should be with specific recommendations as to duration, frequency and intensity if a physical or mental health improvement is indicated. | Selected as the primary category where the intervention focus and the majority of time was spent on exercise in this category (e.g. unstructured physical activity, including unstructured waking). |
| Endurance | Endurance training is aimed at cardiovascular conditioning and is aerobic in nature and simultaneously increases the heart rate and the return of blood to the heart. | Selected as the primary category for interventions where the intervention focus and the majority of time spent was on structured aerobic training (e.g. exercise with a target heart rate range). |
| Other | Other kind of exercises not described. | Selected as the primary category if the intervention did not meet the other categories listed and where the intervention focus and the majority of time was spent in this category. This category included interventions where the exercise was not described in sufficient detail to allocate a category. |
| *Interventions were allocated primary categories using categorisation based on Sherrington 2019. | ||
| Domain | Criteria for judging risk of bias |
| Random sequence generation: selection bias (biased allocation to interventions) due to inadequate generation of a randomised sequence | • Judgement of ’low risk’ if the trial authors described a random component in the sequence generation, e.g. referring to a random number table; using a computer random number generator; coin tossing; shuffling cards or envelopes; throwing dice; drawing of lots; minimisation. • Judgement of ’high risk’ if the trial used a systematic nonrandom method, e.g. date of admission; odd or even date of birth; case record number; clinician judgement; participant preference; patient risk factor score or test results; availability of intervention. • Judgement of ’unclear risk’ if there is insufficient information about the sequence generation process to permit judgement of ’low risk’ or ’high risk’. |
| Allocation concealment: selection bias (biased allocation to interventions) due to inadequate concealment of allocations prior to assignment | • Judgement of ’low risk’ in studies using: ◦ cluster randomisation if allocation of all cluster units performed at the start of the study and individual participant recruitment was completed prior to assignment of the cluster, and the same participants were followed up over time or individual participants were recruited after cluster assignment, but recruitment carried out by a person unaware of group allocation and participant characteristics (e.g. fall history) or individual participants in intervention and control arms were invited by mail questionnaire with identical information. • Judgement of ’high risk’ in studies using: • Judgement of ’unclear risk’ if insufficient information to permit judgement of ’low risk’ or ’high risk’. This is usually the case if the method of concealment is not described or not described in sufficient detail to allow a definite judgement, e.g. if the use of assignment envelopes is described, but it remains unclear whether envelopes were sequentially numbered, opaque and sealed. |
| Blinding of participants and personnel: performance bias due to knowledge of the allocated interventions by participants and personnel carrying out the interventions | • Judgement of ’low risk’ if blinding of participants and personnel implementing the interventions was ensured, and unlikely that the blinding could have been broken. |
| Blinding of outcome assessment: detection bias due to knowledge of the allocated interventions by outcome assessors | • Falls, fallers: • Fractures: |
| Incomplete outcome data: attrition bias due to amount, nature or handling of incomplete outcome data | • Judgement of ’low risk’ if there are no missing outcome data, or less than 20% of outcome data are missing and losses are balanced in numbers across intervention groups with similar reasons for missing data across groups or missing data have been imputed using appropriate methods. |
| Selective reporting: reporting bias due to selective outcome reporting | • Judgement of ’low risk’ if the study protocol is available (i.e., published protocol or trial registry) and all prespecified study outcomes are reported in the prespecified way or the study protocol is unavailable, but it is clear the published report includes all expected outcomes. |
| Method of ascertaining falls: bias in the recall of falls due to unreliable methods of ascertainment | • Judgement of ’low risk’ if the study used some form of concurrent collection of data about falling, e.g. participants given postcards to fill in daily and mail back monthly, calendar to mark monthly, or more frequent, follow‐up by the researchers. |
| We adapted this from Table 8.5.a 'The Cochrane Collaboration's tool for assessing risk of bias’ and Table 8.5.d 'Criteria for judging risk of bias in the 'Risk of bias’ assessment tool’ (Higgins 2017) and from Sherrington 2019. | |
| Study ID | Exercise description | Primary exercise category | Duration of exercise intervention (weeks) | Group/Individual | Location | % supervision* |
| Exercise trials | ||||||
|
| Functional strength, range of movement, balance and walking exercise. | Gait, balance and functional training | 6 | Individual | Home‐based | 18% |
|
| Functional strength, balance and cueing exercise (some participants attended monthly group classes). | Gait, balance and functional training | 24 | Both (most individual but some participants attended monthly exercise classes) | Both (most home‐based (but classes were held at a facility) | 13% |
|
| Functional strength and balance exercise and strategies for fall and freezing avoidance. | Gait, balance and functional training | 26 | Individual | Home‐based | 7% |
| Virtual reality balance training delivered via telehealth
| Gait, balance and functional training | 7 | Group (pairs) | Home‐based | 100% | |
| Sensory‐integration balance training | Gait, balance and functional training | 7 | Individual | Facility‐based | 100% | |
| Trunk‐specific exercise | Gait, balance and functional training | 4 | Individual | Both | Unclear ‐ 100% at facility, number of unsupervised home‐sessions prescribed unclear | |
| General exercise | Gait, balance and functional training | 4 | Individual | Both | Unclear ‐ 100% at facility, number of unsupervised home‐sessions prescribed unclear | |
|
| Tai Chi classes. | 3D (Tai Chi) | 12 | Group | Facility‐based | 100% |
|
| Functional strength and balance exercise. | Gait, balance and functional training | 10 | Both | Both | 33% |
|
| Rhythmic auditory cued overground walking. | Gait, balance and functional training | 6 | Group | Facility‐based | 100% |
|
| Treadmill‐based gait training. | Gait, balance and functional training | 6 | Individual | Facility‐based | 100% |
|
| Tai Chi classes. | 3D (Tai Chi) | 24 | Group | Facility‐based | 100% |
|
| Functional strength exercise with weighted vests and ankle weights. | Resistance training | 24 | Group | Facility‐based | 100% |
|
| Exercises to address freezing of gait and associated falls, and walking using cues. | Gait, balance and functional training | 24 | Individual | Home‐based | 5% |
|
| Treadmill training in a virtual reality environment. | Gait, balance and functional training | 6 | Individual | Facility‐based | 100% |
|
| Treadmill‐based gait training. | Gait, balance and functional training | 6 | Individual | Facility‐based | 100% |
|
| Physiotherapy provided by ParkinsonNet therapists. | Other ‐ ParkinsonNet trained therapists | 24 | Individual | Unclear | ND |
|
| Physiotherapy usual care. | Other ‐ usual therapists | 24 | Individual | Unclear | ND |
|
| Progressive lower limb muscle power training using strength training machines. | Resistance training | 12 | Group (pairs) | Facility‐based | 100% |
| High frequency treadmill training (5 times per week for 10 sessions) | Gait, balance and functional training | 2 | Individual | Facility‐based | 100% | |
| Intermediate frequency treadmill training (3 times per week for 10 sessions) | Gait, balance and functional training | 3.3 | Individual | Facility‐based | 100% | |
| Low frequency treadmill training (2 times per week for 10 sessions) | Gait, balance and functional training | 5 | Individual | Facility‐based | 100% | |
| Gait and cognitive training practised together | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% | |
| Gait and cognitive training practised separately | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% | |
|
| Gait and stepping training. | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% |
|
| Strength, balance and gait training targeting the more affected side. | Gait, balance and functional training | 12 | Unclear | Facility‐based | 100% |
|
| Strength, balance and gait training targeting the less affected side. | Gait, balance and functional training | 12 | Unclear | Facility‐based | 100% |
|
| Functional strength, balance and gait training. | Gait, balance and functional training | 12 | Unclear | Facility‐based | 100% |
|
| Progressive balance and gait training with a balance pad (i.e. foam to stand on). | Gait, balance and functional training | 10 | Unclear | Facility‐based | 100% |
|
| Progressive balance and gait training without a balance pad. | Gait, balance and functional training | 10 | Unclear | Facility‐based | 100% |
|
| Balance and gait training. | Gait, balance and functional training | 12 | Unclear | Both | 55% |
|
| Lower limb resistance training using strength training machines (facility) and functional strength training (home) | Resistance training | 12 | Unclear | Both | 55% |
|
| Balance exercises. | Gait, balance and functional training | 7 | Individual | Facility‐based | 100% |
|
| Flexibility and coordination exercises not aimed at improving balance. | Flexibility | 7 | Individual | Facility‐based | 100% |
| Stepping videogame exercise | Gait, balance and functional training | 12 | Individual | Home‐based | 8% | |
| Gait training with rhythmic auditory stimulation throughout intervention period | Gait, balance and functional training | 24 | Individual | Home‐based | Unclear | |
| Gait training with rhythmic auditory stimulation, with no training in middle 8 weeks of intervention period | Gait, balance and functional training | 16 | Individual | Home‐based | Unclear | |
|
| Balance training using external perturbations wearing a proprioceptive stabiliser. | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% |
|
| Balance training using external perturbations with a sham proprioceptive stabiliser. | Gait, balance and functional training | 8 | Individual | Facility‐based | 100% |
|
| Hydrotherapy with perturbation‐based balance training. | Gait, balance and functional training | 8 | Unclear | Facility‐based | 100% |
|
| Land‐based therapy with perturbation‐based balance training. | Gait, balance and functional training | 8 | Unclear | Facility‐based | 100% |
|
| Strength and balance exercise, including dance and modified Wing Chun martial art. | Gait, balance and functional training | 8 | Both | Both | 40% |
| Exercise plus education trials | ||||||
| Tailored mobility and balance exercises (plus fall prevention education). | Gait, balance and functional training | 8 | Individual | Home‐based | 14% | |
|
| Functional progressive resistance training with weighted vests and resistance bands. | Resistance training | 8 | Individual | Both | 50% |
|
| Movement strategy training. | Gait, balance and functional training | 8 | Individual | Both | 50% |
|
| Functional strength, movement strategy training (plus falls prevention education). | Gait, balance and functional training | 6 | Individual | Home‐based | 50% |
| * % supervision calculated according to the % of exercise sessions supervised. ND: no useable data | ||||||
| Study ID and comparison | Source for rate ratio: rate of falls | Source for risk ratio: number of fallers | Source for risk ratio: number with fractures | Source for risk ratio: number with adverse events |
| Exercise trials | ||||
| Gait, balance and functional training vs Control | 3* | 7 | 7 | NA |
| Gait, balance and functional training vs Control | 1 | 5 | 7 | NA |
| Gait, balance and functional training vs Control | 1a++ | NA | 7 | NA |
| Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | 3‡‡‡ | NA | NA | NA |
| Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | 3‡‡‡ | NA | NA | NA |
| 3D exercise (Tai Chi) vs Control | 3 | 7 | NA | NA |
| Gait, balance and functional training vs Control | 1a++ | 6a | 7 | NA |
| Gait, balance and functional training (cueing training) vs Gait, balance and functional training (treadmill‐based gait training) | 3 | 7 | NA | NA |
| 3D exercise (Tai Chi) vs Resistance training (functional strength) and 3D exercise (Tai Chi) vs Control | 1 | 7 | NE | NA |
| Resistance training (functional strength) vs Control | 3 | 7 | NE | NA |
| Gait, balance and functional training vs Control | 1* | 7 | NA | NA |
| Gait, balance and functional training (virtual reality treadmill training) vs Gait, balance and functional training (treadmill‐based gait training) | 1a | NA | NA | NA |
| Other exercise (ParkinsonNet therapists) vs Other exercise (standard therapists) | 3c | NA | NA | NA |
| Resistance training vs Control | 1** | 5 | 7 | NA |
| Gait, balance and functional training (treadmill training at high frequency) vs Gait balance and functional training (treadmill training at intermediate frequency) vs Gait, balance and functional training (treadmill training at low frequency) | 3‡‡‡ | NA | NA | NA |
| Gait, balance and functional training (Gait and cognitive training practised together) vs Gait, balance and functional training (Gait and cognitive training practised separately) | 3‡‡‡ | NA | NA | NA |
| Gait, balance and functional training vs Control | 3 | 7 | NA | NA |
| Gait, balance and functional training (best side therapy) vs Gait, balance and functional training (worst side therapy) vs Gait, balance and functional training (standard therapy) | 3 | NA | NA | NA |
| Gait, balance and functional training (with a balance pad) vs Gait, balance and functional training (without a balance pad) vs Control | 3 | NA | NA | NA |
| Shen 2015*** Gait, balance and functional training vs Resistance training | 1a | 7 | 7 | NA |
| Gait, balance and functional training vs Flexibility exercise | 3 | NA | NA | NA |
| Gait, balance and functional training vs Control | 1 | 7 | NA | NA |
| Gait, balance and functional training (rhythmic auditory stimulation training throughout intervention period) vs Gait, balance and functional training (rhythmic auditory stimulation training with no training in middle 8 weeks of intervention period) | NA | 7 | NA | NA |
| Gait, balance and functional training (with proprioceptive stabiliser) vs Gait, balance and functional training (without proprioceptive stabiliser) | 3 | NA | NA | NA |
| Gait, balance and functional training (hydrotherapy) vs Gait, balance and functional training (land‐based therapy) | 3 | NA | NE | NA |
| Gait, balance and functional training vs Control | 1 | 6 | NA | NA |
| Medication trials | ||||
| Donepezil vs placebo | 3 | 7 | NE | 3 |
| Rivastigmine vs placebo | 1* | 7 | NA | 3 |
| Rivastigmine vs placebo | 3 | 6 | NA | ND |
| Education trial | ||||
| Personalised education vs control (standardised printed information) | NA | 6a | NA | NA |
| Exercise plus education trials | ||||
| Gait, balance and functional training + education vs Control | NA | 4 | NA | NA |
| Resistance training (functional strength) + education vs Control and Gait, balance and functional training (movement strategy training) + education vs Control | 1 | 5 | 7 | NA |
| Resistance training (functional strength) + education vs Gait, balance and functional training (movement strategy training) + education | 3 | 7 | 7 | NA |
| Gait, balance and functional training + education vs Control | 1 | 5 | 7 | NA |
| ND: no useable data; NA: not applicable (not reported as an outcome in the trial OR not applicable for adverse events for exercise and exercise plus education trials as these were not pooled); NE (no events in either group.) *One participant with excessive number of falls removed from analysis. **Two participants with excessive number of falls assigned a value of 10 falls. ***One participant from the balance group and 2 from the resistance group with excessive number of falls at baseline removed from the analysis. +randomisation stratified by falls history ++adjusted for previous falls +++Incidence rate ratio using Poisson‐Inverse Gaussian regression, with unpublished 95% confidence interval provided by trial authors. ‡0 to 6 months data used as 0 to 12 months not available ‡‡0 to 10 weeks data used for rate ratio as 0 to 20 weeks not available ‡‡‡the separate time periods of falls data were combined Codes for source of rate ratio: 1. Incidence rate ratio reported by trial authors 2. Hazard ratio/relative hazard (multiple events) reported by trial authors 3. Incidence rate ratio calculated by review authors a. Adjusted for confounders by trial authors b. Adjusted for clustering by trial authors c. Adjusted for clustering by review authors Codes for source of risk ratio: 4. Hazard ratio/relative hazard (first fall only) reported by trial authors 5. Relative risk reported by trial authors 6. Odds ratio reported by trial authors 7. Relative risk calculated by review authors a. Adjusted for confounders by trial authors b. Adjusted for clustering by trial authors c. Adjusted for clustering by review authors | ||||
| Study ID and comparison | Intervention group: falls per person year | Intervention group: number (%) of fallers | Intervention group: number (%) of people sustaining one or more fall‐related fractures | Intervention group: non‐fall‐related adverse events per person year | Intervention group: number in analysis | Control group: falls per person year | Control group: number (%) of fallers | Control group: number (%) of people sustaining one or more fall‐related fractures | Control group: non‐fall‐related adverse events per person year | Control group: number in analysis | Length of falls/adverse events monitoring |
| Exercise trials | |||||||||||
| Gait, balance and functional training vs Control
| 6.3 | 46 (73%) | 2 (3%) | NA | Rate of falls and number of fallers: 63 Number sustaining fracture: 67
| 7.9* | 49 (78%) | 6 (9%) | NA | Rate of falls: 62 Number of fallers: 63 Number sustaining fracture: 67 | 6 months |
| Gait, balance and functional training vs Control
| 8.2 | 75 (65%) | 3 (3%) | NA | 115 | 14.0 | 81 (70%) | 4 (3%) | NA | 116 | 6 months |
| Gait, balance and functional training vs Control | 0‐6 months: 6.8 6‐12 months: 5.4 | NA | 0‐6 months: 5 (2%) 6‐12 months: 7 (5%) | NA | 0‐6 months: 231 6‐12 months: 127 | 0‐6 months: 5.4 6‐12 months: 5.6 | NA | 0‐6 months: 9 (4%) 6‐12 months: 3 (2%) | NA | 0‐6 months: 230 6‐12 months: 147 | 12 months, divided into 0‐6 and 6‐12 month time periods |
| Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | 4.0/8.5 | NA | NA | NA | 36/34 | NA | NA | NA | NA | NA | 2 months‡‡ |
| Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | 6.3/2.9 | NA | NA | NA | 19/18 | NA | NA | NA | NA | NA | 2 months‡‡ |
| 3D exercise (Tai Chi) vs Control | 0.6 | 8 (22%) | NA | NA | 37 | 1.3 | 19 (49%) | NA | NA | 39 | 6 months |
| Gait, balance and functional training vs Control | 0‐10 weeks: 93.9 10‐20 weeks: 34.5 | 0‐20 weeks: 52 (85%) | 0‐20 weeks: 0 (0%) | NA | 61 | 0‐10 weeks: 168.1 10‐20 weeks: 155.4 | 0‐20 weeks: 55 (86%) | 0‐20 weeks: 1 (2%) | NA | 64 | 20 weeks |
| Gait, balance and functional training (cueing training) / Gait, balance and functional training (treadmill‐based gait training) | 0.4/1.0 | 2 (20%)/4 (40%) | NA | NA | 10/10 | NA | NA | NA | NA | NA | 6 months |
| 3D exercise (Tai Chi) / Resistance training vs Control | 1.9/4.1 | 19 (29%)/31 (48%) | 0 (0%)/0 (0%) | NA | 65/65 | 5.7 | 26 (40%) | 0 (0%) | NA | 65 | 6 months |
| Gait, balance and functional training vs Control | 166.4 (using fall rate data from week 24‐28)* | 10 (100%) | NA | NA | Fall rate: 9 Number of fallers: 10 | 140.4 (using fall rate data from week 24‐28) | 9 (100%) | NA | NA | Fall rate: 8 Number of fallers: 9 | 6 months |
| Gait, balance and functional training (virtual reality treadmill training) / Gait, balance and functional training (treadmill‐based gait training) | ND | NA | NA | NA | 66/64 | NA | NA | NA | NA | NA | 6 months |
| Other exercise (ParkinsonNet therapists) / Other exercise (standard therapists) | 1.5/1.4 | NA | NA | NA | 329/312 | NA | NA | NA | NA | NA | 24 weeks |
| Resistance training vs Control | 6.5 | 7 (37%) | 1 (5%) | NA | 19 | 11.6 | 12 (63%) | 0 (0%) | NA | 19 | 6 months |
| Gait, balance and functional training (treadmill training at high frequency) vs Gait balance and functional training (treadmill training at intermediate frequency) vs Gait, balance and functional training (treadmill training at low frequency) | Unclear, as timeframe for falls monitoring not reported | NA | NA | NA | 10/10/10 | NA | NA | NA | NA | NA | Unclear |
| Gait, balance and functional training (Gait and cognitive training practised together) vs Gait, balance and functional training (Gait and cognitive training practised separately) | 9.1/8.2 | NA | NA | NA | 10/9 | NA | NA | NA | NA | NA | 2 months‡‡ |
| Gait, balance and functional training vs Control | 23.1 | 5 (56%) | ND | NA | 9 | 37.6 | 6 (67%) | ND | NA | 9 | 2 weeks |
| Gait, balance and functional training (best side therapy) / Gait, balance and functional training (worst side therapy) / Gait, balance and functional training (standard therapy) | 11.1/7.2/4.9 | NA | NA | NA | 9/9/9 | NA | NA | NA | NA | NA | 16 weeks |
| Gait, balance and functional training (with a balance pad) / Gait, balance and functional training (without a balance pad) vs Control | 1.04/4.16 | NA | NA | NA | 15/14 | 7.8 | NA | NA | NA | 15 | 10 weeks |
| Gait, balance and functional training / Resistance training | 0.41/1.02 | 6 (27%)/13 (57%) | 1 (5%)/1 (4%) | NA | Fall rate: 21/21 Number of fallers and fractures: 22/23 | NA | NA | NA | NA | NA | 15 months |
| Gait, balance and functional training / Flexibility exercise | 15.6/49.2 | NA | NA | NA | 28/27 | NA | NA | NA | NA | NA | 1 month |
| Gait, balance and functional training vs Control | 9.4 | 16 (55%) | NA | NA | 29 | 8.6 | 17 (68%) | NA | NA | 25 | 6 months |
| Gait, balance and functional training (rhythmic auditory stimulation training throughout intervention period) vs Gait, balance and functional training (rhythmic auditory stimulation training with no training in middle 8 weeks of intervention period) | NA | 24 (96%)/22 (100%) | NA | NA | 25/22 | NA | NA | NA | NA |
|
|
| Gait, balance and functional training (with proprioceptive stabiliser) / Gait, balance and functional training (without proprioceptive stabiliser) | 11.4/18.6 | NA | NA | NA | 20/20 | NA | NA | NA | NA | NA | 4 months |
| Gait, balance and functional training (hydrotherapy) / Gait, balance and functional training (land‐based therapy) | 3.6/9.6 | NA | 0 (0%)/0 (0%) | NA | 17/17 | NA | NA | NA | NA | NA | 2 months |
| Gait, balance and functional training vs Control | 0.38 | 6 (19%) | NA | NA | 32 | 0.38 | 8 (22%) | NA | NA | 36 | 6 months |
| Medication trials | |||||||||||
| Donepezil vs placebo+ | 47.45 | 18 (95%) | 0 (0%) | 3.0 | Fall rates and number of fallers and fractures:19 Adverse events: 23 | 91.25 | 16 (84%) | 0 (0%) | 1.1 | Fall rates and number of fallers and fractures: 19 Adverse events: 23 | 12 weeks |
| Rivastigmine vs placebo | 16.8* | 56 (86%) | NA | 4.4 | Fall rate and adverse events: 64 Number of fallers and fractures:65 | 28.8 | 56 (86%) | NA | 2.8 | 65 | 8 months |
| Rivastigmine vs placebo | 1.82 | 13 (32%) | NA | ND | 41 | 4.26 | 24 (60%) | NA | ND | 40 | 12 months |
| Education trial | |||||||||||
| Personalised education vs control (standardised printed information) | NA | ND | NA | NA | 27 | NA | ND | NA | NA | 26 | 12 months |
| Education plus exercise trials | |||||||||||
| Gait, balance and functional training plus education vs Control | NA | ND | NA | NA | 15 | NA | ND | NA | NA | 17 | 6 months |
| Resistance training / Gait, balance and functional training (movement strategy training) vs Control | 2.8/6.6 | 36 (52%)/44 (66%) | 3 (4%)/3 (4%) | NA | 69/67 | 18.6 | 37 (63%) | 2 (3%) | NA | 59 | 12 months |
| Gait, balance and functional training plus education vs Control | 21.9 | 39 (61%) | 2 (3%) | NA | 64 | 14.2 | 43 (72%) | 1 (2%) | NA | 60 | 12 months |
| ND: no useable data; NA: not applicable (not reported as an outcome in the trial OR not applicable for adverse events for exercise and education trials as these were not pooled). *outlier with excessive number of falls excluded +randomised cross‐over trial ‡‡the two separate months of falls data were combined | |||||||||||
| Study ID and groups | Intervention group: Number of participants | Intervention group: falls per person year | Intervention group: number (%) of fallers | Control group: number of participants | Control group: falls per person year | Control group: number (%) of fallers | Randomisation stratified by fall history | Timeframe for baseline falls monitoring |
| Exercise trials | ||||||||
| Gait, balance and functional training vs Control | 70 | 60 | 70 (100%) | 72 | 61 | 72 (100%) | No | 12 months, measured retrospectively |
| Gait, balance and functional training vs Control | 115 | 2 | 90 (78%) | 116 | 2 | 90 (78%) | Yes | 12 months, measured retrospectively |
| Gait, balance and functional training vs Control | 3 months prospective: 237 12 months retrospective: 238 | 3 months prospective: 23.6 12 months retrospective: 26 | 12 months retrospective: 238 (100%) | 3 months prospective and 12 months retrospective: 236 | 3 months prospective: 12 12 months retrospective: 19 | 236 (100%) | No | 3 months, measured prospectively prior to commencing intervention and 12 months, measured retrospectively |
| Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | 38/38 | 6.9/22.1 | ND | NA | NA | NA | No | 1 month, unclear if measured prospectively or retrospectively |
| Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | 19/18 | 19.6/7.9 | ND | NA | NA | NA | No | 1 month, unclear if measured prospectively or retrospectively |
| 3D exercise (Tai Chi) vs Control | 40 | ND | ND | 40 | ND | ND | No | ND |
| Gait, balance and functional training vs Control | Rate of falls analysis: 60 Number of fallers analysis: 64 | 137.8 | 55 (86%) | Rate of falls analysis: 62 Number of fallers analysis: 66 | 156.2 | 54 (82%) | No | 10 weeks, measured prospectively prior to commencing intervention |
| Gait, balance and functional training (cueing training) / Gait, balance and functional training (treadmill‐based gait training) | 10/10 | 1/1.4 | 3 (30%)/5 (50%) | NA | NA | NA | No | 6 months, measured retrospectively |
| 3D exercise (Tai Chi) / resistance training (functional strength) vs Control | 65/65 | ND | ND | 65 | ND | ND | No | ND |
| Gait, balance and functional training vs Control | Rate of falls analysis: 11* Number of fallers analysis: 12 | 202.8 | 9 (75%) | 9 | 150.8 | 6 (67%) | No | 5 weeks, measured prospectively from the point of study entry ‐ unclear if this overlaps with the first 3 weeks of the intervention period |
| Gait, balance and functional training (virtual reality treadmill training) / Gait, balance and functional training (treadmill‐based gait training) | 66/64 | 36.5/38.5 | 66 (100%)/64 (100%) | NA | NA | NA | No | 6 months, measured retrospectively |
| Other exercise (ParkinsonNet therapists) / Other exercise (standard therapists) | 358/341 | ND | ND | NA | NA | NA | No | ND |
| Resistance training vs Control | 20 | ND | 5 (25%) | 20 | ND | 7 (35% | No | 12 months, measured retrospectively |
| Gait, balance and functional training (treadmill training at high frequency) vs Gait balance and functional training (treadmill training at intermediate frequency) vs Gait, balance and functional training (treadmill training at low frequency) | 10/10/10 | Unclear, as timeframe for falls monitoring not reported | NA | NA | NA | NA | No | Unclear |
| Gait, balance and functional training (Gait and cognitive training practised together) vs Gait, balance and functional training (Gait and cognitive training practised separately) | 10/9 | 28/7.4 | ND | NA | NA | NA | No | 30 days, measured retrospectively |
| Gait, balance and functional training vs Control | 9 | 66.4 | 5 (56%) | 9 | 66.4 | 6 (67%) | No | 2 weeks, measured prospectively |
| Gait, balance and functional training (best side therapy) / Gait, balance and functional training (worst side therapy) / Gait, balance and functional training (standard therapy) | 9/9/10 | ND | ND | NA | NA | NA | No | ND |
| Gait, balance and functional training (with a balance pad) / Gait, balance and functional training (without a balance pad) vs Control | 15/14 | 6.8/6.7 | ND | 15 | 6.2 | ND | No | 10 weeks, unclear if measured retrospectively or prospectively |
| Gait, balance and functional training / Resistance training | 22/23 | 0.57/0.76** | 9 (41%)/10 (43%) | NA | NA | NA | No | 12 months, measured retrospectively |
| Gait, balance and functional training / Flexibility exercise | 28/27 | 51.6/55.2 | ND | NA | NA | NA | No | 1 month, measured prospectively |
| Gait, balance and functional training vs Control | 31 | ND | 17 (55%) | 29 | ND | 16 (55%) | No | 6 months, measured retrospectively |
| Gait, balance and functional training (rhythmic auditory stimulation training throughout intervention period) vs Gait, balance and functional training (rhythmic auditory stimulation training with no training in middle 8 weeks of intervention period) | 25/22 | 4.5/4.2 | ND | NA | NA | NA | No | 12 months, measured retrospectively |
| Gait, balance and functional training (with proprioceptive stabiliser) / Gait, balance and functional training (without proprioceptive stabiliser) | 20/20 | ND | 16 (80%)/12 (60%) | NA | NA | NA | No | 2 months, measured prospectively |
| Gait, balance and functional training (hydrotherapy) / Gait, balance and functional training (land‐based therapy) | 17/17 | 18/12.6 | 17 (100%)/17 (100%) | NA | NA | NA | No | Rate of falls: 2 months, measured prospectively Number of fallers: 12 months, measured retrospectively |
| Gait, balance and functional training vs Control | 32 | 0 | 0 (0%) | 38 | 0 | 0 (0%) | No | 6 months, measured retrospectively |
| Medication trials | ||||||||
| Donepezil vs placebo | 19 | ND | 19 (100%) | 19 | ND | 19 (100%) | No | Unclear: participants had all fallen or nearly fallen 2 or more times per week, measured retrospectively |
| Rivastigmine vs placebo | 65 | 5.0 | 65 (100%) | 65 | 5.5 | 65 (100%) | No | 12 months, measured retrospectively |
| Rivastigmine vs placebo | 41 | 3.6 | 22 (54%) | 40 | 3.8 | 23 (58%) | No | Unclear |
| Education trial | ||||||||
| Personalised education vs control (standardised printed information) | 27 | ND | ND | 26 | ND | ND | No | ND |
| Exercise plus education trials | ||||||||
| Gait, balance and functional training plus education vs Control | 15 | ND | ND | 17 | ND | ND | No | ND |
| Resistance training (functional strength) / Gait, balance and functional training (movement strategy training) vs Control | 70/69 | ND | 38 (54%)/40 (58%) | 71 | ND | 38 (54%) | No | 12 months, measured retrospectively |
| Gait, balance and functional training plus education vs Control | 67 | ND | 38 (57%) | 66 | ND | 35 (53%) | No | 12 months, measured retrospectively |
| ND: no useable data; NA: not applicable *One participant with excessive number of falls removed from analysis **One participant from the balance group and 2 from the resistance group with excessive number of falls removed from the analysis | ||||||||
| Study ID and comparison | Intervention group baseline n, mean (SD) | Control group baseline n, mean (SD) | Intervention group post timeframe, n, mean (SD) | Control group post timeframe, n, mean (SD) | Intervention group follow‐up timeframe, n, mean (SD) | Control group follow‐up timeframe, n, mean (SD) |
| Exercise trials | ||||||
| Parkinson’s Disease Questionnaire 39 (PDQ39) and 8 (PDQ8) (range 0‐100)* | ||||||
| Gait, balance and functional training vs Control |
115, 28 (13.9) |
116, 30.7 (15.4) | 26 weeks, 104, 29.7 (14.8) | 26 weeks, 115, 32.5 (15.9) | NA | NA |
| Gait, balance and functional training vs Control | 126, 27.4 (14.3) | 153, 28.7 (15.9) | 6 months, 126, 28.3 (15.0) | 6 months, 153, 29.5 (16.5) | 12 months, 77, 29.1 (15.4) | 12 months, 100, 31.7 (15.5) |
| Gandolfi 2017 (PDQ8) Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | 36, 30.7 (15.5)/ 34, 30.5 (16.0) | NA | 7 weeks, 36, 24.1 (14.8)/ 34, 24.2 (15.9) | NA | 11 weeks, 36, 25.8 (14.9)/ 34, 23.9 (13.2)
| NA |
| Gandolfi 2019 (PDQ8) Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | 19, 25.5 (11.8)/ 18, 18.7 (10.8) | NA | 4 weeks, 19, 21.5 (10.0)/ 18, 15.3 (8.6) | NA | 8 weeks, 19, 23.0 (12.6)/ 18, 21.0 (8.8) | NA |
| Gait, balance and functional training (cueing training) / Gait, balance and functional training (treadmill‐based gait training) |
11, 31.1 (14.8)/ 11, 40.1 (17.5 | NA | 6 weeks, 10, 27.5 (17.9)/ 10, 27.4 (10.0) | NA | 3 months, 10, 25.4 (15.0)/ 9, 30.0 (12.9) | NA |
| Li 2012 (PDQ8) 3D exercise (Tai Chi) / Resistance training vs Control |
65, 25.1 (16.8)/ 65, 25.3 (14.7) |
65, 25.2 (16.3) | 6 months, 65, 15.5 (11.4)/ 65, 21.4 (12.7) | 6 months, 65, 25.1 (15.6) | NA | NA |
| Gait, balance and functional training (with proprioceptive stabiliser) / Gait, balance and functional training (without proprioceptive stabiliser) |
20, 62.7 (19.5)/ 20, 61.4 (38.9) | NA | 2 months, 20, 44.0 (22.3)/ 20, 58.5 (37.9) | NA | 4 months, 20, 53.7 (22.3)/ 20, 61.0 (35.1) | NA |
| Gait, balance and functional training (hydrotherapy) / Gait, balance and functional training (land‐based therapy) |
17, 60.3 (19.9)/ 17, 64.4 (28.6) | NA | 2 months 17, 41.9 (20.9)/ 17, 56.4 (26.8) | NA | NA | NA |
| EQ5D Thermometer (0‐100) | ||||||
| Gait, balance and functional training vs Control |
70, 63.1 (17.1) |
71, 64.6 (14.5) | 8 weeks, 67, 61.3 (19.8) | 8 weeks, 66, 61.7 (14.5) | 6 months, 65, 63.0 (18.7) | 6 months, 64, 56.6 (16.9) |
| EQ5D Index score (range 0‐1) | ||||||
| Gait, balance and functional training vs Control |
61, 0.7 (0.1) |
63, 0.7 (0.1) | 10 weeks, 61, 0.7 (0.1) | 10 weeks, 63, 0.7 (0.1) | 20 weeks, 61, 0.8 (0.3) | 20 weeks, 62, 0.7 (0.3) |
| Other exercise (ParkinsonNet therapists) / Other exercise (standard therapists) |
358, 0.65 (0.20)/ 341, 0.65 (0.22) | NA | 16 weeks, 295, 0.66 (0.20)/ 294, 0.65 (0.23) | NA | 24 weeks, 262, 0.68 (0.21)/ 259, 0.66 (0.23)
| NA |
| SF12 and SF36 Physical Composite Score (range 0‐100) | ||||||
| Canning 2015a (SF12) Gait, balance and functional training vs control |
115, 42.3 (7.6) |
116, 42.9 (7.9) | 26 weeks, 104, 41.3 (8.8) | 26 weeks, 115, 40.2 (7.8) | NA | NA |
| Mirelman 2016 (SF36) Gait, balance and functional training (virtual reality treadmill training) / Gait, balance and functional training (treadmill‐based gait training) |
66, 49 (2.5)/ 64, 44.8 (2.5) | NA | 6 weeks, 66, 52 (2.5)/ 64, 46.5 (2.5) | NA | 6 months, 66, 50.5 (2.5)/ 64, 48 (2.5) | NA |
| SF12 Mental Composite Score (range 0‐100) | ||||||
| Gait, balance and functional training vs Control |
115, 51.6 (6.5) |
116, 50.5 (6.8) | 26 weeks, 104, 51.2 (6.4) | 26 weeks, 115, 50.3 (6.7) | NA | NA |
| Medication Trials | ||||||
| EQ5D Thermometer (0‐100) | ||||||
| Rivastigmine vs placebo |
65, 64 (17) |
65, 65 (17) | 32 weeks, 58, 66 (16) | 32 weeks, 63, 63 (18) | NA | NA |
| EQ 5D Index score (range 0‐1) | ||||||
| Rivastigmine vs placebo |
65, 0.72 (0.19) |
65, 0.71 (0.18) | 32 weeks, 58, 0.66 (0.21) | 32 weeks, 63, 0.66 (0.19) | NA | NA |
| Education plus exercise trials | ||||||
| Parkinson’s Disease Questionnaire 39 (PDQ39) (range 0‐100)* | ||||||
| Resistance training / Gait, balance and functional training (movement strategy training) vs Control |
70, 20.8 (13.6)/ 69, 19.4 (12.8) |
71, 22.1 (12.5) | 3 months, 67, 18.9 (13.5)/ 64, 16.9 (14.0) | 3 months, 54, 18.5 (12.6) | 14 months, 67, 20.0 (13.6)/ 66, 20.8 (14.1) | 14 months, 57, 24.1 (13.1) |
| Gait, balance and functional training plus education vs Control |
67, 23 (14) |
66, 24 (15) | 6 weeks, 62, 21 (14) | 6 weeks, 58, 20 (14) | 58 weeks, 55, 22 (13) | 58 weeks, 53, 22 (14) |
| EQ5D Thermometer (0‐100) | ||||||
| Resistance training / Gait, balance and functional training (movement strategy training) vs Control |
70, 74.1 (16.7)/ 69, 73.9 (15.9) |
71, 72.7 (14.6) | 3 months, 67, 71.8 (16.4)/ 64, 76.5 (16.4) | 3 months, 54, 74.7 (16.0) | 14 months, 67, 75.4 (14.1)/ 66, 75.0 (13.5) | 14 months, 57, 72.8 (16.0) |
| Gait, balance and functional training vs Control |
67, 73 (15) |
66, 72 (16) | 6 weeks, 62, 68 (15) | 6 weeks, 58, 76 (12) | 58 weeks, 55, 72 (17) | 58 weeks, 53, 71 (14) |
| EQ5D Index score (range 0‐1) | ||||||
| Gait, balance and functional training vs Control |
67, 0.67 (0.27) |
66, 0.63 (0.28) | 6 weeks, 62, 0.66 (0.29) | 6 weeks, 58, 0.65 (0.27) | 58 weeks, 55, 0.67 (0.25) | 58 weeks, 53, 0.64 (0.3) |
| NA: not applicable *High score = worse quality of life ** Median and interquartile range reported by trial authors and converted to mean and standard deviation by review authors: Volpe 2014a using technique described by Wan 2014 and Goodwin 2011 using the technique described in the Cochrane Handbook (Higgins 2017). Conversion techniques differed due to the different sample sizes in the trials. PDQ8 = Parkinson’s Disease Questionnaire 8 | ||||||
| Study ID, (source if not primary reference), sample, comparison, type of evaluation | Intervention(s) and comparator (n in analyses) | Perspectives, type of currency, price year, time horizon | Cost items measured | Intervention costs per participant | Healthcare service costs per participant | Incremental cost per fall prevented/ per QALY gained |
| Exercise trials | ||||||
| People with PD who had fallen at least once in the past year or were at risk of falls. Gait, balance and functional training vs control Evaluated with cost‐effectiveness analyses | Exercise (balance, lower limb strength, and when required cueing), 3 X week, 24 weeks, with 6‐10 sessions supervised either individually or in a group setting (n = 113) vs usual care control (n = 113) | Health system perspective, Australian dollar, 2012, During 6‐month trial period
| Intervention costs (staff time, travel, equipment) Health service use costs (hospital, medical, allied health) Medication costs | $A1,010 (€642) | Exercise group $A4,604 (€2,925) Control group $A3,920 (€2,491) | Cost per fall prevented $A574 (€365) Cost per QALY gained $A338,800 (€215,277) |
| Chivers Seymour 2019 (Ashburn 2019, Xin 2020) People with PD who had fallen at least once in the past year. Gait, balance and functional training vs control Evaluated with cost‐effectiveness analyses | Exercise (balance and lower limb strengthening exercises, plus strategies for preventing falls and reducing freezing of gait), 30 min per day for 6 months, including 12 x 1‐1.5 hour supervised sessions with a physiotherapist (n = 238) vs usual care control (n = 236) | United Kingdom National Health Service and Personal Social Services perspectives, Pound Stirling, 2016, During 6 month intervention period | Intervention costs (physiotherapist salaries, training, travel, equipment and consumables) Health service use (hospital, primary care, social service) Medication costs collected but not included in analyses | £650 (€765) | Exercise group £3,137 (€3,905) Control group £3,069 (€3,613) | Cost per QALY gained £120,659 (€142,063) |
| People with PD, both fallers and non‐fallers. Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) Evaluated with cost analysis | Virtual reality balance training (using Nintendo Wii Fit system, Nintendo Co., Ltd., Kyoto, Japan) delivered via telehealth (using Skype, Microsoft, USA), delivered in pairs (n = 36) vs sensory‐integration balance training delivered in‐person, individually (n = 34), both interventions 50 mins, 3 X week, 7 weeks | Cost of rehabilitation perspective, Euros, Price year not reported, During assessments and 7 week intervention period | Direct costs (personnel for screening, assessments and intervention, plus resource utilisation). Indirect costs (utilities, facilities) | Virtual reality via telehealth balance training (delivered in pairs) €383.55 sensori‐integration balance training (delivered individually) €602.10 | Not reported | Not reported |
| People with PD and 2 or more falls in the preceding year. Gait, balance and functional training vs control Evaluated with cost‐effectiveness analyses | Exercise (balance, lower limb and trunk strength, 1 X week supervised group and 2 X week independent at home for 10 weeks (n=48) vs usual care control (n = 45). Economic analyses conducted with intervention n = 48 and control n = 45 | United Kingdom National Health Service and Personal Social Services perspectives, Pound sterling, 2008/9, During 20 weeks (10 weeks intervention and 10 weeks follow‐up) | Intervention costs (staff time, travel, equipment, venue hire) Health service use (hospital, primary care, social service) Medication costs | £76 (€89) | Exercise group Health care cost £1,198 (€1,410) Health and social care cost £1,444 (€1,700) Control group Health care cost £1,320 (€1,554) Health and social care cost £1,479 (€1,741) | Cost per QALY gained for total health care costs ‐£4,885 (‐€5,752) Cost per QALY gained for combined total health care and social care costs ‐£1,358 (‐€1,599) |
| People with PD, both fallers and non‐fallers. 3D exercise (Tai Chi) / resistance training (functional strength) vs control Evaluated with cost‐effectiveness analyses
| Tai Chi (n=65) vs resistance training (n=65) vs stretching (control) (n=65), all group classes for 60 minutes, 2 X week, 24 weeks | Societal perspective, United States dollar, 2011, During 9 months (6 months intervention and 3 months follow‐up) | Intervention costs (program promotion, recruitment, staff time, insurance, equipment, room hire, printed materials) Non‐intervention costs (PD medication, physical therapy, medical treatment for falls, participant travel) | Tai chi $US1,080 (€952) Resistance $US1,186 (€1,046) Stretching $US1,155 (€1,019)
| PD medication, physical therapy, medical treatment for falls and participant travel costs: Tai chi $US272 (€240) Resistance $US310 (€273) Stretching $US726 (€640)
| Tai chi vs stretching (control): Cost per fall prevented ‐$US175 (‐€154) Cost per QALY gained ‐$US3,394 (‐€2,993)
Resistance vs Tai Chi: Cost per fall prevented $US100 (€88) Cost per QALY gained $US1,236 (€1,090) |
| People with PD, both fallers and non‐fallers. Other exercise (ParkinsonNet therapists) vs Other exercise (standard therapists) Evaluated with cost analysis | Treatment from ParkinsonNet trained physiotherapists (n=343 to 350)* vs usual care (treatment from physiotherapists without specific PD training) (n=332‐340)*, both groups 24 weeks intervention period | Societal perspective, Euro, Price year not reported, but data collected 2005‐2007, During 24 weeks intervention | Health care costs (physiotherapy, medication, consultation, day‐hospital rehabilitation, admission to hospital, home‐care (paid services), informal care, costs due to lost productivity of the care‐partner). | Physiotherapy cost: ParkinsonNet group €297 Usual care group €310
| Excluding physiotherapy: ParkinsonNet group €2,674 Usual care group €3,424
| Not calculated |
| Exercise plus education trial | ||||||
| People with PD, both fallers and non‐fallers. Gait, balance and functional training plus education vs control Evaluated with cost analysis | Exercise (strength training (lower limb and trunk), movement strategy training) and falls prevention education, 1 X week 60 mins supervised and 1 X week 60 mins independent practice for 6 weeks (n=67) vs Life Skills program (control) (n=66) | Health system perspective, Australian dollar, 2016, During 12 months follow‐up | Intervention costs (travel, home visits, therapist training, equipment). Life skills control intervention was considered as a placebo and therefore had no costs attributed to it. Medical costs associated with falling events (medical, medical ancillary, diagnostic and hospitalisation costs) | $A1,596 (€1,013) | Not reported | Not calculated as there was no difference between the groups |
| *Different participant numbers for different cost components Where costs were reported in a currency other than EUR, the cost was converted to EUR (€) on December 23, 2021. QALY = quality‐adjusted life‐year | ||||||
| Study ID and comparison | Information related to adverse events |
| Exercise trials | |
| Gait, balance and functional training vs Control | No participants fell while performing the exercise program. |
| Gait, balance and functional training vs Control | Two participants had non‐injurious falls during unsupervised exercise at home. |
| Gait, balance and functional training vs Control | No participants fell while performing the exercise program, and no adverse events were associated with the intervention. 0‐6 months hospitalisations: 9 PDSAFE exercise group participants (1 participant with 2 hospitalisations); 20 control group participants. 6‐12 months hospitalisations: 18 PDSAFE exercise group participants (2 participants with 2 hospitalisations); 21 control group participants (2 participants with 2 hospitalisations). |
| Gait, balance and functional training (virtual reality telerehabilitation) vs Gait, balance and functional training (balance training in a facility) | No adverse events were reported during the study. |
| Gait, balance and functional training (trunk‐specific exercises) vs Gait, balance and functional training (general exercises) | No adverse events or safety concerns were reported during the study. |
| 3D exercise (Tai Chi) vs Control | Not reported |
| Gait, balance and functional training vs Control | No adverse events occurred during the exercise sessions. |
| Gait, balance and functional training (cueing training) / Gait, balance and functional training (treadmill‐based gait training) | No adverse events during the intervention. |
| 3D exercise (Tai Chi) / Resistance training vs Control | Tai‐chi (n=65): 3 in class events ‐ 2 falls, 1 muscle soreness or pain; 24 out of class events ‐ 19 falls, 4 low back pain, 1 ankle sprain. Functional strength training (n=65): 14 in class events ‐ 4 falls, 4 muscle soreness or pain, 3 dizziness or faintness, 3 symptoms of hypotension; 41 out of class events ‐ 31 falls, 3 chest pain, 1 hypotension, 4 low back pain, 2 ankle sprains. Stretching (n=65): 9 in‐class events ‐ 5 falls, 1 muscle soreness or pain, 2 dizziness or faintness, 1 symptoms of hypotension; 36 out of class events ‐ 26 falls, 2 chest pain, 2 hypotension, 5 low back pain, 1 ankle sprain. Nb ‐ out of class events are those that occurred during habitual activity or during an assessment. Participants did not perform any intervention outside the class. |
| Gait, balance and functional training vs Control | Not reported |
| Gait, balance and functional training (virtual reality treadmill training) vs Gait, balance and functional training (treadmill‐based gait training) | No serious adverse events during training. Adverse events other than those that occurred during intervention were recorded for both groups, but were not reported separately for the participants with Parkinson's disease. |
| Other exercise (ParkinsonNet therapists) / Other exercise (standard therapists) | None reported, though not collected systematically. |
| Resistance training (muscle power training) vs Control | Power training (n=20): 1 exacerbation of pre‐existing low back pain, 1 pelvic fracture unrelated to the intervention, and 6 participants required modification to training loads due to transient pain, joint inflammation or illness. Control low intensity exercise (n=20): 2 participants had exacerbations of pre‐existing hernias, though this was not attributable to the low intensity exercise. |
| Gait, balance and functional training (treadmill training at high frequency) vs Gait balance and functional training (treadmill training at intermediate frequency) vs Gait, balance and functional training (treadmill training at low frequency) | Not reported |
| Gait, balance and functional training (Gait and cognitive training practised together) vs Gait, balance and functional training (Gait and cognitive training practised separately) | Not reported |
| Gait, balance and functional training vs Control | Not reported |
| Gait, balance and functional training (best side therapy) / Gait, balance and functional training (worst side therapy) / Gait, balance and functional training (standard therapy) | Not reported |
| Gait, balance and functional training (with a balance pad) / Gait, balance and functional training (without a balance pad) vs Control | Not reported |
| Gait, balance and functional training / Resistance training | No adverse events related to the intervention in either group. |
| Gait, balance and functional training / Flexibility exercise | Not reported |
| Gait, balance and functional training vs Control | Adverse events were reported for the intervention group. Six participants ceased the stepping training: two ceased exercise due to it exacerbating pre‐existing lower back pain; two died; one sustained a knee injury from a fall unrelated to the intervention; one ceased for personal reasons. Additionally, one participant experienced a non‐injurious fall while undertaking the intervention and eight participants reported an increase in pre‐existing pain (e.g. lower back pain, knee pain, foot pain) but felt that the exacerbation was unrelated to the intervention. |
| Gait, balance and functional training (rhythmic auditory stimulation training throughout intervention period) vs Gait, balance and functional training (rhythmic auditory stimulation training with no training in middle 8 weeks of intervention period) | Participants who dropped out did so for reasons unrelated to adverse events. |
| Gait, balance and functional training (with proprioceptive stabiliser) / Gait, balance and functional training (without proprioceptive stabiliser) | No major adverse event related to the intervention. |
| Gait, balance and functional training (hydrotherapy) / Gait, balance and functional training (land‐based therapy) | Not reported |
| Gait, balance and functional training vs Control | No adverse events related to the intervention. |
| Medication trials | |
| Donepezil vs placebo | Donepezil (n=23): Eight participants (35%) reported 16 side effects (e.g. dehydration, gastrointestinal upset, headache, sleep disturbance, muscle cramps, orthostatic hypotension, weight loss). Placebo (n=23): Five participants (22%) reported 6 side effects (e.g. gastrointestinal upset, headache, sleep disturbance). These side effects were reported to be transient in most cases. |
| Rivastigmine vs placebo | Rivastigmine (n=64): 187 adverse events (excluding falls) Placebo (n=65): 122 adverse events (excluding falls) Adverse events included cardiac disorders, endocrine disorders, gastrointestinal disorders, general disorders and administration site disorders, immune system disorders, infections and infestations, injury, poisoning and procedural complications, investigations, metabolism and nutrition disorders, musculoskeletal and connective tissue disorders, neoplasms benign, malignant and unspecified, nervous system disorders, psychiatric disorders, renal and urinary disorders, respiratory, thoracic and mediastinal disorders, skin and subcutaneous tissue disorders, surgical medical procedures, vascular disorders. About one third of participants in the rivastigmine group complained of nausea. Most adverse events were categorised as mild and were considered to be unrelated to the intervention. There were 27 adverse events that were classified as serious; 14 in the rivastigmine group and 13 in the placebo group. Two of these events in the rivastigmine group were considered to be probably related to the rivastigmine. Twenty‐three participants in the rivastigmine group and 19 participants in the placebo group stopped taking the trial medication due to adverse events. |
| Rivastigmine vs placebo | Two participants withdrew due to adverse reactions, however details not provided. |
| Education trial | |
| Personalised education vs control (standardised printed information) | Not reported |
| Exercise plus education trials | |
| Gait, balance and functional training plus education vs Control | Not reported |
| Resistance training / Gait, balance and functional training (movement strategy training) vs Control | Functional strength training group (n=70): 25 occasions of new muscle soreness lasting > 24 hours Movement strategy training group (n=69): 11 occasions of new muscle soreness lasting > 24 hours, 1 fall and 2 occasions of dizziness during the intervention. |
| Gait, balance and functional training vs Control | No adverse events related to the intervention. |
| Study ID and comparison | Subgroup definition, number of participants (n) | Intervention lower severity group: falls per person year | Intervention higher severity group: falls per person year | Intervention lower severity group: number (%) of fallers | Intervention higher severity group: number (%) of fallers | Control lower severity group: falls per person year | Control higher severity group: falls per person year | Control lower severity group: number (%) of fallers | Control higher severity group: number (%) of fallers | Length of falls monitoring |
| Gait, balance and functional training vs Control
| Lower disease severity: Hoehn and Yahr stages 2 and 3, n = 96 Higher disease severity: Hoehn and Yahr stage 4, n = 30 | NA | NA | 31 (66%) | 15 (94%) | NA | NA | 37 (76%) | 12 (86%) | 6 months |
| Gait, balance and functional training vs Control
| Lower disease severity: UPDRS motor score ≤ 26, n = 122 Higher disease severity: UPDRS motor score ≥ 27, n = 109 | ND | ND | ND | ND | ND | ND | ND | ND | 6 months |
| Chivers Seymour 2019 (data reported in Ashburn 2019) Gait, balance and functional training vs Control | Lower disease severity: includes both the low disease severity subgroup ‐ MDS‐UPDRS motor score ≤ 22, n = 152 and moderate disease severity subgroup ‐ MDS‐UPDRS motor score 23 ‐ 28, n = 155 Higher disease severity: MDS‐UPDRS motor score ≥ 39, n = 152 | ND | ND | NA | NA | ND | ND | NA | NA | 6 months |
| ND: no useable data; NA: not applicable (not reported as an outcome in the trial). | ||||||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Rate of falls Show forest plot | 12 | 1456 | Rate Ratio (IV, Random, 95% CI) | 0.74 [0.63, 0.87] |
| 1.2 Rate of falls subgrouped by ProFaNE exercise categories Show forest plot | 12 | 1456 | Rate Ratio (IV, Random, 95% CI) | 0.74 [0.63, 0.87] |
| 1.2.1 Gait, balance and functional training vs Control | 9 | 1146 | Rate Ratio (IV, Random, 95% CI) | 0.80 [0.67, 0.95] |
| 1.2.2 Resistance training vs control | 2 | 136 | Rate Ratio (IV, Random, 95% CI) | 0.72 [0.55, 0.94] |
| 1.2.3 3D exercise (Tai Chi) vs Control | 2 | 174 | Rate Ratio (IV, Random, 95% CI) | 0.41 [0.23, 0.72] |
| 1.3 Rate of falls ‐ subgrouped by % supervision (100% supervision vs <100% supervision) Show forest plot | 12 | Rate Ratio (IV, Random, 95% CI) | Subtotals only | |
| 1.3.1 100% supervision | 5 | 373 | Rate Ratio (IV, Random, 95% CI) | 0.56 [0.41, 0.77] |
| 1.3.2 < 100% supervision | 7 | 1083 | Rate Ratio (IV, Random, 95% CI) | 0.85 [0.75, 0.97] |
| 1.4 Rate of falls ‐ subgrouped by baseline fall risk (increased fall risk vs fall risk not specified) Show forest plot | 12 | Rate Ratio (IV, Random, 95% CI) | Subtotals only | |
| 1.4.1 Higher fall risk participants | 7 | 1082 | Rate Ratio (IV, Random, 95% CI) | 0.73 [0.59, 0.91] |
| 1.4.2 Unspecified fall risk participants | 5 | 374 | Rate Ratio (IV, Random, 95% CI) | 0.71 [0.56, 0.90] |
| 1.5 Rate of falls ‐ pooled disease severity subgroup analyses_UPDRS Show forest plot | 2 | Rate Ratio (IV, Random, 95% CI) | Subtotals only | |
| 1.5.1 Higher disease severity participants | 2 | Rate Ratio (IV, Random, 95% CI) | 1.47 [1.11, 1.94] | |
| 1.5.2 Lower disease severity participants | 2 | Rate Ratio (IV, Random, 95% CI) | 0.65 [0.39, 1.08] | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 Number of fallers Show forest plot | 9 | 932 | Risk Ratio (IV, Random, 95% CI) | 0.90 [0.80, 1.00] |
| 2.2 Number of fallers subgrouped by ProFaNE exercise categories Show forest plot | 9 | 932 | Risk Ratio (IV, Random, 95% CI) | 0.90 [0.80, 1.00] |
| 2.2.1 Gait, balance and functional training vs Control | 6 | 622 | Risk Ratio (IV, Random, 95% CI) | 0.92 [0.81, 1.04] |
| 2.2.2 Resistance training vs control | 2 | 136 | Risk Ratio (IV, Random, 95% CI) | 0.87 [0.43, 1.74] |
| 2.2.3 3D exercise (Tai Chi) vs control | 2 | 174 | Risk Ratio (IV, Random, 95% CI) | 0.59 [0.36, 0.95] |
| 2.3 Number of fallers ‐ subgrouped by % supervision (100% supervision vs <100% supervision) Show forest plot | 9 | Risk Ratio (IV, Random, 95% CI) | Subtotals only | |
| 2.3.1 100% supervision | 4 | 328 | Risk Ratio (IV, Random, 95% CI) | 0.75 [0.53, 1.06] |
| 2.3.2 < 100% supervision | 5 | 604 | Risk Ratio (IV, Random, 95% CI) | 0.92 [0.82, 1.04] |
| 2.4 Number of fallers ‐ subgrouped by baseline fall risk (increased fall risk vs fall risk not specified) Show forest plot | 9 | Risk Ratio (IV, Random, 95% CI) | Subtotals only | |
| 2.4.1 Higher fall risk participants | 5 | 576 | Risk Ratio (IV, Random, 95% CI) | 0.89 [0.76, 1.04] |
| 2.4.2 Unspecified fall risk participants | 4 | 356 | Risk Ratio (IV, Random, 95% CI) | 0.86 [0.67, 1.11] |
| 2.5 Number of fallers ‐ pooled disease severity subgroup analyses Show forest plot | 2 | Risk Ratio (IV, Random, 95% CI) | Subtotals only | |
| 2.5.1 Higher disease severity participants | 2 | 139 | Risk Ratio (IV, Random, 95% CI) | 1.19 [1.00, 1.41] |
| 2.5.2 lower disease severity participants | 2 | 218 | Risk Ratio (IV, Random, 95% CI) | 0.78 [0.62, 0.98] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 3.1 Number of people sustaining one or more fall‐related fractures Show forest plot | 5 | 989 | Risk Ratio (IV, Random, 95% CI) | 0.57 [0.28, 1.17] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 4.1 Health‐related quality of life ‐ combined measures post intervention Show forest plot | 5 | 951 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.17 [‐0.36, 0.01] |
| 4.2 Health‐related quality of life ‐ combined measures follow‐up Show forest plot | 3 | 429 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.27 [‐0.46, ‐0.08] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 5.1 Rate of falls, different types of exercise compared Show forest plot | 14 | Rate Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 5.1.1 Gait, balance and functional training vs gait, balance and functional training | 10 | Rate Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 5.1.2 Gait, balance and functional training vs resistance training | 1 | Rate Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 5.1.3 Gait, balance and functional training vs flexibility | 1 | Rate Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 5.1.4 3D exercise (Tai Chi) vs resistance training | 1 | Rate Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 5.1.5 Other exercise vs Other exercise | 1 | Rate Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 6.1 Number of fallers, different types of exercise compared Show forest plot | 4 | Risk Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 6.1.1 Gait, balance and functional training vs gait, balance and functional training | 2 | Risk Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 6.1.2 Gait, balance and functional training vs resistance training | 1 | Risk Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 6.1.3 3D exercise (Tai Chi) vs resistance training | 1 | Risk Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 7.1 Quality of life ‐ combined measures post intervention, different types of exercise compared Show forest plot | 8 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 7.1.1 Gait, balance and functional training vs Gait, balance and functional training | 6 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 7.1.2 3D exercise (Tai Chi) vs Resistance exercise | 1 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 7.1.3 Other exercise vs Other exercise | 1 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 7.2 Quality of life ‐ combined measures follow‐up, different types of exercise compared Show forest plot | 5 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 7.2.1 Functional gait, balance and strength training vs Functional gait, balance and strength training | 5 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 8.1 Rate of falls Show forest plot | 3 | 248 | Rate Ratio (IV, Fixed, 95% CI) | 0.50 [0.44, 0.58] |
| 8.2 Rate of falls ‐ subgrouped by medication Show forest plot | 3 | 248 | Rate Ratio (IV, Random, 95% CI) | 0.50 [0.43, 0.58] |
| 8.2.1 Rivastigmine vs placebo | 2 | 210 | Rate Ratio (IV, Random, 95% CI) | 0.48 [0.35, 0.66] |
| 8.2.2 Donepezil vs placebo | 1 | 38 | Rate Ratio (IV, Random, 95% CI) | 0.52 [0.44, 0.62] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 9.1 Number of fallers Show forest plot | 3 | 249 | Risk Ratio (IV, Fixed, 95% CI) | 1.01 [0.90, 1.14] |
| 9.2 Number of fallers ‐ subgrouped by medication Show forest plot | 3 | Risk Ratio (IV, Random, 95% CI) | 0.95 [0.70, 1.28] | |
| 9.2.1 Rivastigmine vs placebo | 2 | Risk Ratio (IV, Random, 95% CI) | 0.61 [0.20, 1.90] | |
| 9.2.2 Donepezil vs placebo | 1 | Risk Ratio (IV, Random, 95% CI) | 1.13 [0.90, 1.40] | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 10.1 Quality of life EQ5D thermometer post intervention Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 10.1.1 Rivastigmine vs placebo | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 10.2 Quality of life EQ5D Index Score post intervention Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 10.2.1 Rivastigmine vs placebo | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 11.1 Rate of adverse events excluding falls Show forest plot | 2 | 175 | Rate Ratio (IV, Fixed, 95% CI) | 1.60 [1.28, 2.01] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 12.1 Number of fallers Show forest plot | 1 | Risk Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 13.1 Rate of falls Show forest plot | 2 | 320 | Rate Ratio (IV, Random, 95% CI) | 0.46 [0.12, 1.85] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 14.1 Number of fallers Show forest plot | 3 | 352 | Risk Ratio (IV, Random, 95% CI) | 0.89 [0.75, 1.07] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 15.1 Number of people sustaining one or more fall‐related fractures Show forest plot | 2 | 320 | Risk Ratio (IV, Random, 95% CI) | 1.45 [0.40, 5.32] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 16.1 Health‐related quality of life ‐ Parkinson's Disease Questionnaire (PDQ39) post intervention Show forest plot | 2 | 305 | Mean Difference (IV, Random, 95% CI) | 0.05 [‐3.12, 3.23] |
| 16.2 Health‐related quality of life ‐ Parkinson's Disease Questionnaire (PDQ39) at follow‐up Show forest plot | 2 | 299 | Mean Difference (IV, Random, 95% CI) | ‐2.25 [‐5.45, 0.96] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 17.1 Rate of falls Show forest plot | 1 | Rate Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 17.1.1 Gait, balance and functional training plus education vs resistance training plus education | 1 | Rate Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 18.1 Number of fallers Show forest plot | 1 | Risk Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 18.1.1 Gait, balance and functional training plus education vs resistance training plus education | 1 | Risk Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 19.1 Number of people sustaining one or more fall‐related fractures Show forest plot | 1 | Risk Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| 19.1.1 Gait, balance and functional training plus education vs resistance training plus education | 1 | Risk Ratio (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 20.1 Health‐related quality of life ‐ Parkinson's Disease Questionnaire (PDQ39) post intervention Show forest plot | 1 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 20.1.1 Gait, balance and functional training plus education vs resistance training plus education | 1 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 20.2 Health‐related quality of life ‐ Parkinson's Disease Questionnaire (PDQ39) at follow‐up Show forest plot | 1 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 20.2.1 Gait, balance and functional training plus education vs resistance training plus education | 1 | Std. Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 21.1 Rate of falls ‐ exercise vs control Show forest plot | 9 | 1245 | Rate Ratio (IV, Random, 95% CI) | 0.74 [0.61, 0.90] |
| 21.2 Number of fallers ‐ exercise vs control Show forest plot | 6 | 721 | Risk Ratio (IV, Random, 95% CI) | 0.87 [0.75, 1.02] |
| 21.3 Rate of falls ‐ cholinesterase inhibitor vs placebo Show forest plot | 1 | 81 | Rate Ratio (IV, Fixed, 95% CI) | 0.43 [0.32, 0.56] |
| 21.3.1 Rivastigmine vs placebo | 1 | 81 | Rate Ratio (IV, Fixed, 95% CI) | 0.43 [0.32, 0.56] |
| 21.4 Number of fallers ‐ cholinesterase inhibitor vs placebo Show forest plot | 1 | 81 | Risk Ratio (IV, Fixed, 95% CI) | 0.31 [0.12, 0.78] |
| 21.4.1 Rivastigmine vs placebo | 1 | 81 | Risk Ratio (IV, Fixed, 95% CI) | 0.31 [0.12, 0.78] |
| 21.5 Rate of falls ‐ exercise and education vs control Show forest plot | 1 | 124 | Rate Ratio (IV, Random, 95% CI) | 1.58 [0.74, 3.40] |
| 21.6 Number of fallers ‐ exercise and education vs control Show forest plot | 2 | 156 | Risk Ratio (IV, Random, 95% CI) | 0.84 [0.65, 1.08] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 22.1 Rate of falls ‐ exercise vs control Show forest plot | 7 | 995 | Rate Ratio (IV, Random, 95% CI) | 0.90 [0.76, 1.05] |
| 22.2 Number of fallers ‐ exercise vs control Show forest plot | 5 | 516 | Risk Ratio (IV, Random, 95% CI) | 0.89 [0.76, 1.04] |
| 22.3 Rate of falls ‐ cholinesterase inhibitor vs placebo Show forest plot | 1 | 129 | Rate Ratio (IV, Fixed, 95% CI) | 0.60 [0.38, 0.96] |
| 22.3.1 Rivastigmine vs placebo | 1 | 129 | Rate Ratio (IV, Fixed, 95% CI) | 0.60 [0.38, 0.96] |
| 22.4 Number of fallers ‐ cholinesterase inhibitor vs placebo Show forest plot | 1 | 130 | Risk Ratio (IV, Fixed, 95% CI) | 1.00 [0.87, 1.15] |
| 22.4.1 Rivastigmine vs placebo | 1 | 130 | Risk Ratio (IV, Fixed, 95% CI) | 1.00 [0.87, 1.15] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 23.1 Rate of falls ‐ exercise vs control Show forest plot | 8 | 1299 | Rate Ratio (IV, Random, 95% CI) | 0.80 [0.70, 0.91] |
| 23.2 Number of fallers ‐ exercise vs control Show forest plot | 7 | 838 | Risk Ratio (IV, Random, 95% CI) | 0.91 [0.81, 1.03] |
| 23.3 Rate of falls ‐ cholinesterase inhibitor vs placebo Show forest plot | 1 | 129 | Rate Ratio (IV, Fixed, 95% CI) | 0.60 [0.38, 0.96] |
| 23.3.1 Rivastigmine vs placebo | 1 | 129 | Rate Ratio (IV, Fixed, 95% CI) | 0.60 [0.38, 0.96] |
| 23.4 Number of fallers ‐ cholinesterase inhibitor vs placebo Show forest plot | 1 | 130 | Risk Ratio (IV, Fixed, 95% CI) | 1.00 [0.87, 1.15] |
| 23.4.1 Rivastigmine vs placebo | 1 | 130 | Risk Ratio (IV, Fixed, 95% CI) | 1.00 [0.87, 1.15] |
| 23.5 Number of fallers ‐ exercise and education vs control Show forest plot | 2 | 320 | Risk Ratio (IV, Random, 95% CI) | 0.90 [0.75, 1.08] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 24.1 Rate of falls ‐ exercise vs control Show forest plot | 2 | 692 | Rate Ratio (IV, Random, 95% CI) | 0.92 [0.73, 1.16] |
| 24.2 Number of fallers ‐ exercise vs control Show forest plot | 1 | 231 | Risk Ratio (IV, Random, 95% CI) | 0.93 [0.78, 1.11] |
| 24.3 Rate of falls ‐ exercise and education vs control Show forest plot | 1 | 196 | Rate Ratio (IV, Random, 95% CI) | 0.24 [0.10, 0.61] |
| 24.4 Number of fallers ‐ exercise and education vs control Show forest plot | 2 | 228 | Risk Ratio (IV, Random, 95% CI) | 0.93 [0.73, 1.19] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 25.1 Rate of falls ‐ exercise vs control Show forest plot | 11 | 1260 | Rate Ratio (IV, Random, 95% CI) | 0.77 [0.65, 0.92] |
| 25.2 Number of fallers ‐ exercise vs control Show forest plot | 8 | 736 | Risk Ratio (IV, Random, 95% CI) | 0.89 [0.79, 1.00] |
| 25.3 Rate of falls ‐ cholinesterase inhibitor vs placebo Show forest plot | 1 | 129 | Rate Ratio (IV, Fixed, 95% CI) | 0.60 [0.38, 0.96] |
| 25.3.1 Rivastigmine vs placebo | 1 | 129 | Rate Ratio (IV, Fixed, 95% CI) | 0.60 [0.38, 0.96] |
| 25.4 Number of fallers ‐ cholinesterase inhibitor vs placebo Show forest plot | 2 | 168 | Risk Ratio (IV, Fixed, 95% CI) | 1.03 [0.92, 1.16] |
| 25.4.1 Rivastigmine vs placebo | 1 | 130 | Risk Ratio (IV, Fixed, 95% CI) | 1.00 [0.87, 1.15] |
| 25.4.2 Donepezil vs placebo | 1 | 38 | Risk Ratio (IV, Fixed, 95% CI) | 1.13 [0.90, 1.40] |
| 25.5 Rate of falls ‐ exercise and education vs control Show forest plot | 1 | 124 | Rate Ratio (IV, Random, 95% CI) | 1.58 [0.74, 3.40] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 26.1 Rate of falls ‐ exercise vs control Show forest plot | 9 | 1268 | Rate Ratio (IV, Random, 95% CI) | 0.79 [0.68, 0.92] |
| 26.2 Number of fallers ‐ exercise vs control Show forest plot | 7 | 789 | Risk Ratio (IV, Random, 95% CI) | 0.89 [0.77, 1.02] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 27.1 Number of fallers ‐ cholinesterase inhibitor vs placebo Show forest plot | 2 | 168 | Risk Ratio (IV, Fixed, 95% CI) | 1.03 [0.92, 1.16] |
| 27.1.1 Rivastigmine vs placebo | 1 | 130 | Risk Ratio (IV, Fixed, 95% CI) | 1.00 [0.87, 1.15] |
| 27.1.2 Donepezil vs placebo | 1 | 38 | Risk Ratio (IV, Fixed, 95% CI) | 1.13 [0.90, 1.40] |
| 27.2 Rate of falls ‐ exercise and education vs control Show forest plot | 1 | 196 | Rate Ratio (IV, Random, 95% CI) | 0.24 [0.10, 0.61] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 28.1 Rate of falls ‐ exercise vs control Show forest plot | 12 | 1456 | Rate Ratio (IV, Fixed, 95% CI) | 0.79 [0.71, 0.88] |
| 28.2 Number of fallers ‐ exercise vs control Show forest plot | 9 | 932 | Risk Ratio (IV, Fixed, 95% CI) | 0.90 [0.80, 1.00] |
| 28.3 Rate of falls ‐ exercise and education vs control Show forest plot | 2 | 320 | Rate Ratio (IV, Fixed, 95% CI) | 0.54 [0.33, 0.89] |
| 28.4 Number of fallers ‐ exercise and education vs control Show forest plot | 3 | 352 | Risk Ratio (IV, Fixed, 95% CI) | 0.89 [0.75, 1.07] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 29.1 Rate of falls ‐ cholinesterase inhibitor vs placebo Show forest plot | 3 | 248 | Rate Ratio (IV, Random, 95% CI) | 0.50 [0.43, 0.58] |
| 29.1.1 Rivastigmine vs placebo | 2 | 210 | Rate Ratio (IV, Random, 95% CI) | 0.48 [0.35, 0.66] |
| 29.1.2 Donepezil vs placebo | 1 | 38 | Rate Ratio (IV, Random, 95% CI) | 0.52 [0.44, 0.62] |
| 29.2 Number of fallers ‐ cholinesterase inhibitor vs placebo Show forest plot | 3 | 249 | Risk Ratio (IV, Random, 95% CI) | 0.95 [0.70, 1.28] |
| 29.2.1 Rivastigmine vs placebo | 2 | 211 | Risk Ratio (IV, Random, 95% CI) | 0.61 [0.20, 1.90] |
| 29.2.2 Donepezil vs placebo | 1 | 38 | Risk Ratio (IV, Random, 95% CI) | 1.13 [0.90, 1.40] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 30.1 Rate of falls ‐ exercise vs control Show forest plot | 12 | 1456 | Rate Ratio (IV, Random, 95% CI) | 0.74 [0.63, 0.87] |
| 30.1.1 Gait, balance and functional training vs Control | 10 | 1244 | Rate Ratio (IV, Random, 95% CI) | 0.78 [0.68, 0.91] |
| 30.1.2 Resistance training vs control | 1 | 38 | Rate Ratio (IV, Random, 95% CI) | 0.84 [0.28, 2.53] |
| 30.1.3 3D exercise (Tai Chi) vs Control | 2 | 174 | Rate Ratio (IV, Random, 95% CI) | 0.41 [0.23, 0.72] |
| 30.2 Number of fallers ‐ exercise vs control Show forest plot | 9 | 932 | Risk Ratio (IV, Random, 95% CI) | 0.90 [0.80, 1.00] |
| 30.2.1 Gait, balance and functional training vs Control | 7 | 720 | Risk Ratio (IV, Random, 95% CI) | 0.93 [0.83, 1.05] |
| 30.2.2 Resistance training vs control | 1 | 38 | Risk Ratio (IV, Random, 95% CI) | 0.58 [0.30, 1.13] |
| 30.2.3 3D exercise (Tai Chi) vs control | 2 | 174 | Risk Ratio (IV, Random, 95% CI) | 0.59 [0.36, 0.95] |
