Chlorpromazine versus placebo for schizophrenia
Abstract
Background
Chlorpromazine, formulated in the 1950s, remains a benchmark treatment for people with schizophrenia.
Objectives
To review the effects of chlorpromazine compared with placebo, for the treatment of schizophrenia.
Search methods
We searched the Cochrane Schizophrenia Group’s Trials Register (15 May 2012). We also searched references of all identified studies for further trial citations. We contacted pharmaceutical companies and authors of trials for additional information.
Selection criteria
We included all randomised controlled trials (RCTs) comparing chlorpromazine with placebo for people with schizophrenia and non‐affective serious/chronic mental illness irrespective of mode of diagnosis. Primary outcomes of interest were death, violent behaviours, overall improvement, relapse and satisfaction with care.
Data collection and analysis
We independently inspected citations and abstracts, ordered papers, re‐inspected and quality assessed these. We analysed dichotomous data using risk ratio (RR) and estimated the 95% confidence interval (CI) around this. We excluded continuous data if more than 50% of participants were lost to follow‐up. Where continuous data were included, we analysed this data using mean difference (MD) with a 95% confidence interval. We used a fixed‐effect model.
Main results
We inspected over 1100 electronic records. The review currently includes 315 excluded studies and 55 included studies. The quality of the evidence is very low. We found chlorpromazine reduced the number of participants experiencing a relapse compared with placebo during six months to two years follow‐up (n=512, 3 RCTs, RR 0.65 CI 0.47 to 0.90), but data were heterogeneous. No difference was found in relapse rates in the short, medium or long term over two years, although data were also heterogeneous. We found chlorpromazine provided a global improvement in a person's symptoms and functioning (n=1164, 14 RCTs, RR 0.71 CI 0.58 to 0.86). Fewer people allocated to chlorpromazine left trials early ( n=1831, 27 RCTs, RR 0.64 CI 0.53 to 0.78) compared with placebo. There are many adverse effects. Chlorpromazine is clearly sedating (n=1627, 23 RCTs, RR 2.79 CI 2.25 to 3.45), it increases a person's chances of experiencing acute movement disorders (n=942, 5 RCTs, RR 3.47 CI 1.50 to 8.03) and parkinsonism (n=1468, 15 RCTs, RR 2.11 CI 1.59 to 2.80). Akathisia did not occur more often in the chlorpromazine group than placebo. Chlorpromazine clearly causes a lowering of blood pressure with accompanying dizziness (n=1488, 18 RCTs, RR 2.38 CI 1.74 to 3.25) and considerable weight gain (n=165, 5 RCTs, RR 4.92 CI 2.32 to 10.43).
Authors' conclusions
The results of this review confirm much that clinicians and recipients of care already know but aim to provide quantification to support clinical impression. Chlorpromazine's global position as a 'benchmark' treatment for psychoses is not threatened by the findings of this review. Chlorpromazine, in common use for half a century, is a well‐established but imperfect treatment. Judicious use of this best available evidence should lead to improved evidence‐based decision making by clinicians, carers and patients.
PICOs
Plain language summary
Chlorpromazine versus placebo for schizophrenia
For previous plain language summary please see Appendix 3.
People with schizophrenia often hear voices or see things (hallucinations) and have strange beliefs (delusions). The main treatment for these symptoms of schizophrenia is antipsychotic drugs. Chlorpromazine was one of the first drugs discovered to be effective in the treatment of schizophrenia during the 1950s. It remains one of the most commonly used and inexpensive treatments even today. However, being an older drug (‘typical’ or first generation) it also has serious side effects, such as blurred vision, a dry mouth, tremors or uncontrollable shaking, depression, muscle stiffness and restlessness.
An update search was carried out in 2012 and the review now includes 55 studies that assess the effects of chlorpromazine in treating schizophrenia compared with no active treatment (‘dummy’ treatment or placebo). Evidence was, in the main, rated by the review authors as low quality. There is some evidence to suggest that chlorpromazine reduces relapse and improves people’s mental health, symptoms and functioning. However, the side effects of chlorpromazine are severe and debilitating. Chlorpromazine causes sleepiness and sedation. It also causes movement disorders (such as tremors and uncontrollable shaking), considerable weight gain and lowering of blood pressure with accompanying dizziness.
Chlorpromazine is low‐cost and widely available. Despite its many side effects, chlorpromazine is likely to remain a benchmark drug and one of the most widely used treatments for schizophrenia worldwide.
It should be noted that the quality of evidence from the 55 included studies was low and in addition to this, 315 studies were excluded because of flaws in the reporting of information or data and in research design and methods. Larger, better conducted and reported trials should focus on important outcomes such as quality of life, levels of satisfaction, relapse, hospital discharge or admission and number of violent incidents.
Authors' conclusions
Summary of findings
| CHLORPROMAZINE versus PLACEBO for schizophrenia | ||||||
| Patient or population: patients with schizophrenia | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control | CHLORPROMAZINE versus PLACEBO | |||||
| Death | See comment | See comment | Not estimable | 14 | ⊕⊝⊝⊝ | One study specifically reported mortality and there were no deaths in either the chlorpromazine or placebo group. There were no reports of death in any other study. |
| Relapse | 710 per 1000 | 461 per 1000 | RR 0.65 | 512 | ⊕⊝⊝⊝ | 2 trials report this outcome at 0‐8 weeks follow‐up, 3 trials at 6 months to 2 years, and 2 trials 2‐5 years, none showed a significant difference. |
| Global state: no overall improvement | 897 per 1000 | 637 per 1000 | RR 0.71 (0.58 to 0.90) | 1164 | ⊕⊝⊝⊝ | 13 trials also reported on this outcome at 0‐8 weeks follow‐up and showed a significant result in favour of chlorpromazine. |
| Leaving the study early | 200 per 1000 | 128 per 1000 | RR 0.64 | 1831 | ⊕⊕⊝⊝ | 17 trials reported on this outcome at 0‐8 weeks follow‐up and showed significant results in favour of chlorpromazine. 2 trials reported on this outcome at 6 months to 2 years follow‐up, and 1 trial at 2‐5 years, and there was no significant difference. |
| Satisfaction with treatment ‐ not reported | See comment | See comment | Not estimable | ‐ | See comment | No studies reported on this outcome. |
| Behaviour: deteriorated/ disturbed/un‐cooperative | 471 per 1000 | 231 per 1000 | RR 0.49 | 1040 | ⊕⊝⊝⊝ | 2 trials reported on this outcome at 0‐8 weeks follow‐up and found no significant difference. |
| Cost of care ‐ not reported | See comment | See comment | Not estimable | ‐ | See comment | No studies reported on this outcome. |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. 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 | ||||||
| 1 Serious risk of bias: the study had an unclear risk of bias for random sequence generation, allocation concealment and other bias as the drugs were provided by a pharmaceutical company. | ||||||
Background
Description of the condition
Approximately 24 million people currently suffer from schizophrenia (WHO 1998), the majority of whom live in low‐ or middle‐income countries. Chlorpromazine remains one of the most commonly used and inexpensive treatments for people with schizophrenia (Odejide 1982), despite its well‐documented adverse effects and the advent of a new generation of antipsychotic drugs. It is one of the essential drugs listed by the World Health Organization (WHO 2003). Chlorpromazine is commonly prescribed in India, and in South East Asia, the older generation of antipsychotics are used to treat the majority of people with schizophrenia (Chong 2004). In 2003, chlorpromazine was the most frequently prescribed of the first generation 'typical' antipsychotic drugs in the UK at a time when the 'typical' group of antipsychotics accounted for 44% of all antipsychotic prescriptions (PPA, 2003).
Description of the intervention
Chlorpromazine, a drug developed in 1951 for reducing allergic reactions (an antihistamine) began to be used as part of a cocktail of drugs in order to induce a state of 'artificial hibernation' for surgical procedures (Laborit 1951). Its ability to reduce psychic stress led researchers to demonstrate its effectiveness for treating certain psychiatric disorders (Delay 1952). Chlorpromazine was hailed as a major discovery for schizophrenia, an illness for which few treatment options existed (Davis 1978). The impact of this drug has been so great, that according to one author, it has been heralded as the second revolution in psychiatry (the first being psychoanalysis) (Grozier 1973). Chlorpromazine is the first of many drugs to be classed as a 'neuroleptic' (literally translated: to grasp the nerve), a term coined by two of its first protagonists (Delay 1952). Early trials of chlorpromazine for schizophrenia indicated, that in comparison with placebo, it hastened clinical recovery, facilitated improvements in social functioning and was effective at preventing relapse.
How the intervention might work
The antipsychotic effect of chlorpromazine results from its action on particular areas within specific cells of the brain (Sedvall 1995). It is thought to affect how receptive these cells are to dopamine. However, chlorpromazine is not specific to one site of action within the body. Consequently, it is known to cause adverse effects ranging from dry mouth, blurred vision and urinary retention as well as restlessness, tremors, facial rigidity, shuffling gait and repetitive movements of the face and/or trunk which can be difficult to reverse (APA 1992). Chlorpromazine has also been linked to depressive symptoms that may be caused by the drug itself (neuroleptic dysphoria, Awad 1993). In addition, the use of chlorpromazine has been associated with a potentially fatal disturbance of blood pressure, temperature and muscle control (neuroleptic malignant syndrome, APA 1994).
Why it is important to do this review
There are questions relating to the differential response to drugs between certain groups of people with schizophrenia. For example, there may be differences in the effects of treatment for men and women (Hambrecht 1992; Kendler 1995; Szymanski 1995), for children, adults or the elderly (Kaplan 1990; Rosen 1990), or for people who are experiencing their first episode as opposed to those with a longer illness duration (Hill 1992; Szymanski 1996). When prescribing drugs for schizophrenia dosage is important in order to obtain optimal response with minimal adverse effects (Bollini 1994; Kane 1985). There also remains debate about the applicability of research findings to the 'real world' of clinicians (Jenicek 1990). For example, trials undertaken on highly selected groups of people with schizophrenia may be of very limited use in the 'everyday' situation. We attempted to investigate whether, for the primary outcomes of interest (see: Methods, Types of outcome measures), a real difference exists for those with diagnoses of schizophrenia made with operational 'checklists' as opposed to those with less rigorous diagnoses. A final question we posed was whether the effects of chlorpromazine were different between patients treated recently (1990‐2002) to those treated in earlier decades (1951‐1989).
New trials often use chlorpromazine as the 'benchmark' or 'control' drug rather than a placebo when a new treatment is being evaluated. The aim of this review is to evaluate this 'benchmark' in comparison to placebo. This is an update of a Cochrane Review first published in 1998, Issue 1 of The Cochrane Database of Systematic Reviews (Thornley 1998a) and updated in 2003 (Thornley 2003) and 2007 (Adams 2007).
Objectives
To review the effects of chlorpromazine compared with placebo, for the treatment of schizophrenia.
It was expected that several subgroup analyses could be undertaken within this review (see Subgroup analysis and investigation of heterogeneity).
Methods
Criteria for considering studies for this review
Types of studies
We sought all relevant randomised controlled trials. Where a trial was described as 'double blind' but it was implied that the study was randomised, we included these trials in a sensitivity analysis. If their inclusion did not result in a substantive difference, they remained in the analyses. If their inclusion did result in statistically significant differences, we did not add the data from these lower quality studies to the results of the better quality trials, but presented these within a subcategory. We excluded quasi‐randomised studies, such as those allocating by alternate days of the week.
Types of participants
We included people with schizophrenia and other types of schizophrenia‐like psychoses (schizophreniform and schizoaffective disorders) however diagnosed, irrespective of age, sex or severity of illness.
Types of interventions
1. Chlorpromazine: any dose or mode of administration (oral or by injection)
2. Placebo (active or inactive) or no treatment
Types of outcome measures
We categorised outcomes as short term (zero to eight weeks), medium term (nine weeks to six months) and long term (six months to two years).
Primary outcomes
We classified these outcomes as primary outcomes for the 2002 update to help minimise the potential for multiple statistical testing that could be undertaken within sensitivity analyses. We tried to choose these on the grounds of clinical importance and were helped in this by inclusion of a new co‐reviewer, JR, who was not as familiar with the data as the previous authors. We have used the same outcomes but rearranged into new sub‐headings for this 2012 update.
1. Death ‐ suicide and natural causes
2. Relapse ‐ as defined by each study
3. Global state
3.1 Overall improvement*
4. Leaving the study early
5. Satisfaction with treatment ‐ participant/carer
6. Behaviour
6.1 Specific behaviours (e.g. aggressive or violent behaviour)
7. Economic
7.1 Cost of care
Secondary outcomes
1. Global state
1.1 Duration of hospital stay
1.2 Re‐admission
1.3 Severity of Illness
2. Mental state
2.1 General symptoms
2.2 Specific symptoms
2.2.1 Positive symptoms (delusions, hallucinations, disordered thinking)
2.2.2 Negative symptoms (avolition, poor self‐care, blunted affect)
2.2.3 Mood ‐ depression
3. Behaviour
3.1 General behaviour
3.2 Social functioning
3.3 Employment status during trial (employed/unemployed)
3.4 Occurrence of violent incidents (to self, others or property)
4. Adverse effects
4.1 General
4.2 Specific
4.2.2 Movement disorders (extrapyramidal side effects, specifically tardive dyskinesia and neuroleptic malignant syndrome)
4.2.3 Sedation
4.2.4 Dry mouth
5. 'Summary of findings' table
We used the GRADE approach to interpret findings (Schünemann 2008) and GRADE profiler (GRADEPRO) to import data from Review Manager 5 (Review Manager) to create a 'Summary of findings' table. These tables provide outcome‐specific information concerning the overall quality of evidence from each included study in the comparison, the magnitude of effect of the interventions examined, and the sum of available data on all outcomes we rate as important to patient‐care and decision making. We have selected the following main outcomes for inclusion in the 'Summary of findings' table.
-
Death ‐ suicide and natural causes
-
Overall improvement
-
Relapse ‐ as defined by each study
-
Leaving the study early
-
Satisfaction with treatment ‐ participant/carer
-
Specific behaviours (e.g. aggressive or violent behaviour)
-
Cost of care
Search methods for identification of studies
Electronic searches
For previous searches please see Appendix 2.
1. Cochrane Schizophrenia Group Trials Register (May 2012)
The Trials Search Co‐ordinator searched the Cochrane Schizophrenia Group’s Trials Register (15 May 2012).
The Cochrane Schizophrenia Group’s Trials Register is compiled by systematic searches of major databases, handsearches of relevant journals and conference proceedings (see group module).
Searching other resources
1. Reference searching
We inspected the references of all identified studies for further studies.
2. Personal contact
For this update, we did not contact the first author of each included study for information regarding unpublished trials.
3. Pharmaceutical companies
In previous versions of this review, we contacted pharmaceutical companies for any unpublished and published trials. Approaches have been made to Rhone Poulenc Rorer, the original developers of chlorpromazine, for access to archive material. Dr R.A Pargiter of Hobart, Tasmania very kindly donated a series of reports from May and Baker (the pharmaceutical company which originally produced chlorpromazine) that listed presentations of work relevant to chlorpromazine and schizophrenia, dating from 1955 to 1973. We (BT, CEA and JR) handsearched these for further studies.
Data collection and analysis
Methods used in data collection and analysis for this 2012 update are below; for previous methods please see Appendix 3.
Selection of studies
For this 2012 update, the Cochrane Schizophrenia group provided Enhance Reviews a database of relevant abstracts; the Enhance Reviews team inspected full articles of the abstracts meeting the inclusion criteria.
Data extraction and management
1. Extraction
For this 2012 update, two members of the Enhance Reviews team extracted data from included studies. In addition, Jun Xia (JX) extracted data for all Chinese studies. We extracted data presented only in graphs and figures whenever possible. In the previous versions of the review, when further information was necessary, we contacted authors of studies in order to obtain missing data or for clarification. If studies were multi‐centre, where possible, we extracted data relevant to each component centre separately.
2. Management
2.1 Forms
We extracted data onto standard, simple forms, created in a web‐based software (www.systematic‐review.ca).
2.2 Scale‐derived data
We included continuous data from rating scales only if:
a. the psychometric properties of the measuring instrument have been described in a peer‐reviewed journal (Marshall 2000); and
b. the measuring instrument had not been written or modified by one of the trialists for that particular trial.
Ideally, the measuring instrument should either be i. a self‐report or ii. completed by an independent rater or relative (not the therapist). We realise that this is not often reported clearly; we have noted whether or not this is the case in Description of studies.
2.3 Endpoint versus change data
There are advantages of both endpoint and change data. Change data can remove a component of between‐person variability from the analysis. On the other hand, calculation of change needs two assessments (baseline and endpoint), which can be difficult in unstable and difficult to measure conditions such as schizophrenia. We decided primarily to use endpoint data, and only use change data if the former were not available. We combined endpoint and change data in the analysis as we used mean differences (MD) rather than standardised mean differences throughout (Higgins 2011, Chapter 9.4.5.2).
2.4 Skewed data
Continuous data on clinical and social outcomes are often not normally distributed. To avoid the pitfall of applying parametric tests to non‐parametric data, we aimed to apply the following standards to all data before inclusion:
a) standard deviations (SDs) and means are reported in the paper or obtainable from the authors;
b) when a scale starts from the finite number zero, the SD, when multiplied by two, is less than the mean (as otherwise the mean is unlikely to be an appropriate measure of the centre of the distribution (Altman 1996));
c) if a scale started from a positive value (such as the Positive and Negative Syndrome Scale (PANSS, Kay 1986) which can have values from 30 to 210), we modified the calculation described above to take the scale starting point into account. In these cases skew is present if 2 SD > (S‐S min), where S is the mean score and S min is the minimum score.
Endpoint scores on scales often have a finite start and end point and these rules can be applied. We entered skewed endpoint data from studies of fewer than 200 participants in additional tables rather than into an analysis. Skewed data pose less of a problem when looking at the mean if the sample size is large; we entered such endpoint data into syntheses.
When continuous data are presented on a scale that includes a possibility of negative values (such as change data), it is difficult to tell whether data are skewed or not, we entered skewed change data into analyses regardless of the size of the study.
2.5 Common measure
To facilitate comparison between trials, we intended to convert variables that can be reported in different metrics, such as days in hospital (mean days per year, per week or per month) to a common metric (e.g. mean days per month).
2.6 Conversion of continuous to binary
Where possible, we made efforts to convert outcome measures to dichotomous data. This can be done by identifying cut‐off points on rating scales and dividing participants accordingly into 'clinically improved' or 'not clinically improved'. It is generally assumed that if there is a 50% reduction in a scale‐derived score such as the Brief Psychiatric Rating Scale (BPRS, Overall 1962) or the PANSS (Kay 1986), this could be considered as a clinically significant response (Leucht 2005; Leucht 2005a). If data based on these thresholds were not available, we used the primary cut‐off presented by the original authors.
2.7 Direction of graphs
Where possible, we entered data in such a way that the area to the left of the line of no effect indicated a favourable outcome for chlorpromazine.
Assessment of risk of bias in included studies
For this 2012 update, two members of the Enhance Reviews team worked independently by using criteria described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) to assess trial quality for the new included studies and all previously included studies. This new set of criteria is based on evidence of associations between overestimate of effect and high risk of bias of the article such as sequence generation, allocation concealment, blinding, incomplete outcome data and selective reporting.
Where inadequate details of randomisation and other characteristics of trials were provided, we did not contact authors of the studies in order to obtain additional information.
We have noted the level of risk of bias in both the text of the review and in the summary of findings Table for the main comparison.
Measures of treatment effect
1. Binary data
For binary outcomes we calculated a standard estimation of the risk ratio (RR) and its 95% confidence interval (CI). It has been shown that RR is more intuitive (Boissel 1999) than odds ratios and that odds ratios tend to be interpreted as RR by clinicians (Deeks 2000).
2. Continuous data
For continuous outcomes we estimated mean difference (MD) between groups. We would prefer not to calculate effect size measures (standardised mean difference (SMD)). However, if scales of very considerable similarity were used, we presumed there was a small difference in measurement, and we would have calculated effect size and transformed the effect back to the units of one or more of the specific instruments.
Unit of analysis issues
1. Cluster trials
Studies increasingly employ 'cluster randomisation' (such as randomisation by clinician or practice), but analysis and pooling of clustered data poses problems. Authors often fail to account for intra‐class correlation in clustered studies, leading to a 'unit of analysis' error (Divine 1992) whereby P values are spuriously low, confidence intervals unduly narrow and statistical significance overestimated. This causes type I errors (Bland 1997; Gulliford 1999).
We did not include any cluster trials in this review. If we had, where clustering was not accounted for in primary studies, we would have presented data in a table, with a (*) symbol to indicate the presence of a probable unit of analysis error. Where clustering was incorporated into the analysis of primary studies, we would have presented these data as if from a non‐cluster randomised study, but adjusted for the clustering effect.
In subsequent versions of this review we will seek to contact first authors of studies to obtain intra‐class correlation coefficients (ICC) for their clustered data and to adjust for this by using accepted methods (Gulliford 1999).
We have sought statistical advice and have been advised that the binary data as presented in a report should be divided by a 'design effect'. This is calculated using the mean number of participants per cluster (m) and the ICC [Design effect=1+(m‐1)*ICC] (Donner 2002). If the ICC was not reported it was assumed to be 0.1 (Ukoumunne 1999).
If cluster studies have been appropriately analysed taking into account ICCs and relevant data documented in the report, synthesis with other studies would have been possible using the generic inverse variance technique.
2. Cross‐over trials
A major concern of cross‐over trials is the carry‐over effect. It occurs if an effect (e.g. pharmacological, physiological or psychological) of the treatment in the first phase is carried over to the second phase. As a consequence, on entry to the second phase the participants can differ systematically from their initial state despite a wash‐out phase. For the same reason cross‐over trials are not appropriate if the condition of interest is unstable (Elbourne 2002). As both effects are very likely in severe mental illness, we only used data of the first phase of cross‐over studies.
3. Studies with multiple treatment groups
Where a study involved more than two treatment arms, if relevant, we presented the additional treatment arms in the comparisons. If data were binary, we simply added these and combined within the two‐by‐two table. If data were continuous, we combined data following the formula in section 7.7.3.8 (Combining groups) of the Cochrane Handbook for Systemic reviews of Interventions (Higgins 2011). Where the additional treatment arms were not relevant, we did not reproduce these data.
Dealing with missing data
1. Overall loss of credibility
At some degree of loss of follow‐up, data must lose credibility (Xia 2009). We chose that, for any particular outcome, should more than 50% of data be unaccounted for, we did not reproduce these data or use them within analyses. If, however, more than 50% of those in one arm of a study were lost, but the total loss was less than 50%, we marked such data with (*) to indicate that such a result may well be prone to bias.
2. Binary
In the case where attrition for a binary outcome was between 0% and 50%, we presented data for the total number of participants randomised for studies that used an intention‐to‐treat (ITT) analysis; where studies did not use an ITT analysis, we presented completer only data.
3. Continuous
3.1 Attrition
In the case where attrition for a continuous outcome was between 0% and 50%, and data only from people who completed the study to that point were reported, we presented and used these data.
3.2 Standard deviations
If standard deviations (SDs) were not reported, we first tried to obtain the missing values from the authors. If not available, where there are missing measures of variance for continuous data, but an exact standard error (SE) and confidence intervals (CIs) available for group means, and either P value or T value available for differences in mean, we can calculate them according to the rules described in the Cochrane Handbook for Systemic reviews of Interventions (Higgins 2011). When only the SE is reported, SDs are calculated by the formula SD=SE * square root (n). Chapters 7.7.3 and 16.1.3 of the Cochrane Handbook for Systemic reviews of Interventions (Higgins 2011) present detailed formulae for estimating SDs from P values, T or F values, CIs, ranges or other statistics. If these formulae do not apply, we would calculate the SDs according to a validated imputation method which is based on the SDs of the other included studies (Furukawa 2006). Although some of these imputation strategies can introduce error, the alternative would be to exclude a given study’s outcome and thus to lose information. We did not impute any SDs, if we had we would have examined the validity of the imputations in a sensitivity analysis excluding imputed values.
3.3 Last observation carried forward
We anticipated that in some studies the method of last observation carried forward (LOCF) would be employed within the study report. As with all methods of imputation to deal with missing data, LOCF introduces uncertainty about the reliability of the results (Leucht 2007). Therefore, where LOCF data have been used in the trial, if less than 50% of the data have been assumed, we reproduced these data and indicated that they are the product of LOCF assumptions.
Assessment of heterogeneity
1. Clinical heterogeneity
We considered all included studies initially, without seeing comparison data, to judge clinical heterogeneity. We simply inspected all studies for clearly outlying people or situations which we had not predicted would arise. When such situations or participant groups arose, we fully discussed these.
2. Methodological heterogeneity
We considered all included studies initially, without seeing comparison data, to judge methodological heterogeneity. We simply inspected all studies for clearly outlying methods which we had not predicted would arise. When such methodological outliers arose, we fully discussed these.
3. Statistical heterogeneity
3.1 Visual inspection
We visually inspected graphs to investigate the possibility of statistical heterogeneity.
3.2 Employing the I2 statistic
We investigated heterogeneity between studies by considering the I2 method alongside the Chi2 P value. The I2 provides an estimate of the percentage of inconsistency thought to be due to chance (Higgins 2003). The importance of the observed value of I2 depends on i. magnitude and direction of effects and ii. strength of evidence for heterogeneity (e.g. P value from Chi2 test, or a confidence interval for I2). An I2 estimate greater than or equal to around 50% accompanied by a statistically significant Chi2 statistic was interpreted as evidence of substantial levels of heterogeneity (Higgins 2011). When substantial levels of heterogeneity were found in the primary outcome, we explored reasons for the heterogeneity (Subgroup analysis and investigation of heterogeneity). If data were heterogeneous we used a random‐effects model.
Assessment of reporting biases
Reporting biases arise when the dissemination of research findings is influenced by the nature and direction of results (Egger 1997). These are described in Section 10 of the Cochrane Handbook for Systemic reviews of Interventions (Higgins 2011). We are aware that funnel plots may be useful in investigating reporting biases but are of limited power to detect small‐study effects. We did not use funnel plots for outcomes where there were 10 or fewer studies, or where all studies were of similar sizes. In other cases, where funnel plots were possible, we sought statistical advice in their interpretation.
Data synthesis
We understand that there is no closed argument for preference for use of fixed‐effect or random‐effects models. The random‐effects method incorporates an assumption that the different studies are estimating different, yet related, intervention effects. This often seems to be true to us and the random‐effects model takes into account differences between studies even if there is no statistically significant heterogeneity. There is, however, a disadvantage to the random‐effects model: it puts added weight onto small studies, which often are the most biased ones. Depending on the direction of effect, these studies can either inflate or deflate the effect size. We chose the fixed‐effect model for all analyses. The reader is, however, able to choose to inspect the data using the random‐effects model.
Subgroup analysis and investigation of heterogeneity
1. Subgroup analyses ‐ only primary outcomes
1.1 Clinical state, stage or problem
We proposed to undertake this review and provide an overview of the effects of chlorpromazine for people with schizophrenia in general. In addition, however, we tried to report data on subgroups of people in the same clinical state, stage and with similar problems.
We also undertook subgroup analyses comparing the results for the following:
-
men versus women; under 18 years of age versus 18‐65 years old versus older than 65;
-
acutely ill people (< one‐month in duration) versus people who have been ill for longer;
-
high dose (> 501 mg/day) versus low doses (1‐500 mg/day);
-
people diagnosed according to any operational criteria versus those who have not been diagnosed using operational criteria;
-
studies published before 1990 versus studies published between 1990 and the present.
2. Investigation of heterogeneity
If inconsistency was high, we have reported this. First, we investigated whether data had been entered correctly. Second, if data were correct, we visually inspected the graph and successively removed outlying to see if homogeneity was restored.
When unanticipated clinical or methodological heterogeneity was obvious we simply stated hypotheses regarding these for future reviews or versions of this review. We do not anticipate undertaking analyses relating to these.
Sensitivity analysis
We applied sensitivity analyses to the primary outcomes of this review.
1. Implication of randomisation
We aimed to include trials in a sensitivity analysis if they were described in some way so as to imply randomisation. For the primary outcomes we included these studies and if there was no substantive difference when the implied randomised studies were added to those with better description of randomisation, then we entered all data from these studies.
2. Assumptions for lost binary data
Where assumptions had to be made regarding people lost to follow‐up (see Dealing with missing data), we compared the findings of the primary outcomes when we use our assumption/s and when we used data only from people who completed the study to that point. A sensitivity analysis was undertaken to test how prone results were to change when completer‐only data only were compared to the imputed data using the above assumption. If there was a substantial difference, we reported results and discussed them but continued to employ our assumption.
3. Risk of bias
We analysed the effects of excluding trials that were judged to be at high risk of bias across one or more of the domains of randomisation (implied as randomised with no further details available): allocation concealment, blinding and outcome reporting for the meta‐analysis of the primary outcome. If the exclusion of trials at high risk of bias did not substantially alter the direction of effect or the precision of the effect estimates, then we included data from these trials in the analysis.
4. Imputed values
Had we included any cluster‐randomised trials, we would have undertaken a sensitivity analysis to assess the effects of including data from trials where we used imputed values for ICC in calculating the design effect in cluster‐randomised trials.
If we noted substantial differences in the direction or precision of effect estimates in any of the sensitivity analyses listed above, we did not pool data from the excluded trials with the other trials contributing to the outcome, but presented them separately.
Results
Description of studies
Please see Characteristics of included studies and Characteristics of excluded studies.
Results of the search
The original searches in 1995 yielded over 600 references, and after initial appraisal we selected 201 studies for further inspection of the full papers. An additional 50 papers were also identified from the reference lists. We were able to include 45 relevant studies. For the July 2002 update, we found 106 citations and were able to include just one additional study, and four additional reports of already included studies. In addition, May and Baker's files of reports of chlorpromazine studies, kindly supplied by Dr Pargiter of Hobart, Tasmania, were handsearched by BT, CA and JR. These files yielded 37 more records that met the review criteria. Three of these reports were previously unknown studies that could be included. Thirty‐four were excluded. We are very grateful to Dr Pargiter, who kept those files for so long and then, knowing that they could at last be usefully employed, donated them. For the January 2007 update search, we found 317 references from 142 studies. We were able to include one additional study (Xiong 1994).
The May 2012 update search identified a total of 21 new relevant studies. Five trials met the inclusion criteria for this review and there are now 55 studies included in the latest version of this review (see Figure 1). A total of 315 studies are now excluded. There are no studies awaiting assessment and no ongoing studies have been identified.
Study flow diagram.
Included studies
We included 55 studies.
1. Methods
Four included studies were cross‐over trials (Baker 1959; Letemendia 1967; Nishikawa 1982; Shepherd 1956), two had a factorial design (Hamilton 1960; Hogarty 1973), and the remainder were parallel studies. All studies were either stated to be randomised or implied randomisation.
2. Length of trials
The most common study length was six to 12 weeks but the range was considerable with two trials being just over 24 hours in duration, whilst the two longest were over a period of three years.
3. Participants
It was reported in all trials that the participants suffered from schizophrenia (with the exception of Vaughan 1955 who randomised people with mental illnesses who were 'chronic and intractable' with motor restlessness, psychomotor agitation, and excitement and Hankoff 1962 who did not clearly state the diagnoses but included psychiatric outpatients who were 'schizophrenic and non‐schizophrenic', with the majority having schizophrenia). Only 14 of the 55 trials described the diagnostic criteria used, or the symptoms required for people to be included. Otherwise, entry to most included studies was on a clinical diagnosis of schizophrenia. A total of 5506 participants are now included in this review.
4. Setting
Most studies were hospital‐based with only a few of the studies being undertaken in the community.
5. Study size
The mean number of participants was 99, ranging from 21 (Payne 1960) to 838 (Prien 1968).
6. Intervention
6.1 Chlorpromazine: The doses of chlorpromazine in these studies ranged from 25 mg/day (Reschke 1974) to 2400 mg/day (Dean 1958). The mean dose was 574 mg/day (SD 446).
6.2 Placebo: All trials compared chlorpromazine with placebo or no treatment. Kurland 1961 used a 'positive' placebo (phenobarbital) and a 'negative' placebo, the results of which were combined in this review. Clark 1968a randomised participants to placebo and a no‐drug group, which were also combined in this review. Prien 1968, however, randomised to a placebo group and a 'routine conventional hospital treatment' group. These groups were not combined because the latter had the opportunity to receive any medication that the treating physicians felt appropriate (presumably including chlorpromazine). Prien 1968 also included two arms, one with lower doses and the other with higher doses. We pooled data from these arms in the main analysis and conducted subgroup analyses.
6.3 Other drug treatment arms: Thirty‐eight of the trials also included at least one more drug treatment arm in addition to placebo and chlorpromazine. Data were not included from these treatment arms.
7. Outcomes
The following outcomes were reported by the included studies: death, relapse, global impression, mental state, behaviour, leaving the study early and adverse effects. None of the included studies attempted to quantify levels of satisfaction, or quality of life and there is no evidence of any direct economic evaluation of chlorpromazine. Most outcomes analysed were dichotomous, and presented as such, or were ordinal data that could be dichotomised.
7.1 Outcome scales
The following scales provided continuous data for the analysis.
7.1.2 Mental state
i. Brief Psychiatric Rating Scale (Overall 1962)
A brief rating scale used to assess the severity of a range of psychiatric symptoms, including psychotic symptoms. The original scale has 16 items, but a revised 18‐item scale is commonly used. Each item is defined on a seven‐point scale varying from 'not present' to 'extremely severe', scoring from zero to six or one to seven. Scores can range from zero to 126, with high scores indicating more severe symptoms. Tetreault 1969 reported data from this scale.
ii. Global impression
4.7.2.1 Clinical Global Impression (Guy 1976)
A rating instrument commonly used in studies on schizophrenia that enables clinicians to quantify severity of illness and overall clinical improvement. A seven‐point scoring system is usually used with low scores indicating decreased severity and/or greater recovery. Borison 1991 reported data from this scale.
7.1.3 Behaviour
i. Modified Rosenthal Rating Scale (Rosenthal 1963)
A scale for nurses to rate the behaviour of psychiatric patients. Lower scores indicate improved behaviour. Tetreault 1969 reported data from this scale.
ii. Parkside Behaviour Rating Scale (Schmidt 1957)
A rating scale in which six behavioural characteristics are rated on a five‐point scale. The worst possible behaviour would carry a rating of six points, as against a maximum of 30 points for unproblematic behaviour. Baker 1959 reported data from this scale.
iii. Fergus Falls rating scale (Lucero 1951)
The L‐M Fergus Falls Behavior Rating Scale is a method of rating the behaviour of patients in mental hospitals, which measures 11 aspects of behaviour, and the changes in one patient over a length of time. Ramu 1999a reported data from this scale.
7.1.4. Adverse effects
i. Extrapyramidal Bilan scale (Tetreault 1969a)
A nine‐item rating scale for use by neurologists, to measure severity of symptoms such as facial mask, tremor, rigidity, akathisia, dystonia, dyskinesia and others. Each item can be scored from zero to three, such that the overall score can range from zero (no symptoms) to a possible 27 (severe symptoms of all types). Tetreault 1969 reported data from this scale.
Excluded studies
We have now excluded 315 studies. The studies listed in the 'Excluded studies' section had to be inspected in hard copy in order to make the final decision. Nearly half were not randomised, did not imply randomisation or did not describe the allocation procedure used. In several studies, participants were not suffering from schizophrenia. Another sizeable proportion of the trials did not compare chlorpromazine with placebo, but in combination with other treatments. A few were chlorpromazine withdrawal studies investigating the effects of instigation of treatment, which are not relevant to this review. We will include these withdrawal studies in a later review. Eighty‐eight studies had no usable outcomes. Either data did not have clear clinical implications, for example EEG recordings, or genuinely relevant clinical data were not adequately reported. Frequently the numbers of participants in each group were not specified, means or standard deviations were not given or data were not reported from individual arms of cross‐over studies.
Awaiting assessment
No studies are currently awaiting assessment.
Ongoing studies
We identified no ongoing studies.
Risk of bias in included studies
Please also see Figure 2 and Figure 3.
'Risk of bias' graph: review authors' judgements about each 'Risk of bias' item presented as percentages across all included studies.
'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study.
Allocation
Only four studies described the methods used to generate random allocation. Two studies (Hine 1958; Letemendia 1967) randomised by the toss of a coin, and two used tables of random numbers (Hamill 1975; Tetreault 1969a). Two studies (Cole 1964; Hall 1955) described some form of allocation concealment (sealed envelopes). For the other 53 studies, readers are given little assurance that bias was minimised during the allocation procedure, yet 24 (24/53, 45%) reported that the participants allocated to each treatment group were very similar. One study, Cooper 2000, reported that participants were randomly assigned in blocks of six and Weckowicz 1960 reported that participants were divided into three matched groups. For the remaining studies it is improbable that such equal numbers could have been obtained unless block randomisation was used, yet 24 out of 53 studies (45%) had exactly the same numbers in the chlorpromazine and placebo groups.
Blinding
Thirty‐five studies had a low risk of bias for performance bias and described the methods used to ensure blinding of participants and personnel. Twenty studies had an unclear risk of bias. Nine studies stated that outcome assessors were blinded and were rated a low risk of bias, the remainder had an unclear risk of bias. Two studies (Grygier 1958; Hall 1955) tested how successful their attempts at blinding were. Three studies (Clark 1970b; Hamill 1975; Simon 1958) gave no indication that blinding had been attempted.
Incomplete outcome data
Twenty‐nine studies were rated as low risk of bias for incomplete outcome data and 19 studies had an unclear risk of bias. Seven studies were rated as having a high risk of bias.
Selective reporting
Only five studies had a low risk of bias for selective reporting. Three studies had an unclear risk of bias and 47 of the studies were rated as high risk of bias for selective reporting.
Other potential sources of bias
Eight trials were subject to other biases as they were either partly or fully funded by the pharmaceutical industry; in Borison 1991 two of the trialists are in prison for research fraud. Twelve studies were of low risk of bias for other potential sources of bias and the remainder had an unclear risk of bias.
Effects of interventions
See: Summary of findings for the main comparison CHLORPROMAZINE versus PLACEBO for schizophrenia
We used risk ratios (RR) for dichotomous data and mean differences (MD) for continuous data, with their respective 95% confidence intervals (CIs) throughout.
1. Comparison: CHLORPROMAZINE versus PLACEBO
We categorised outcomes as short term (up to eight weeks), medium term (nine weeks to six months) and long term (six months to two years).
1.1 Death
We found only one small trial (n=14) that specifically reported mortality (Baker 1959); there were no deaths in either the chlorpromazine or placebo group (Analysis 1.1). We found no reports of death in any study and currently over 5506 people have been included in trials relating to this review (of which 1741 were given chlorpromazine).
1.2 Relapse
We found short‐term (n=74, 2 RCTs) and medium‐term (n=809, 4 RCTs) data did not show a significant difference in rates of relapse (Analysis 1.2), but with significant heterogeneity (I2=78% and 96%, respectively). Removing the studies with results that were causing this heterogeneity, as judged by visual inspection (Prien 1968; Spohn 1977) eliminates this heterogeneity. We found longer‐term data (six months to two years) favoured the chlorpromazine group (n=512, 3 RCTs, RR 0.65 CI 0.47 to 0.90), but the two long‐term studies lasting two to five years (Hogarty 1973; Nishikawa 1982) showed no difference (n=394, 2 RCTs), again with significant heterogeneity (I2 = 72% and 84%, respectively). In this case, the larger trials (Hogarty 1973; Prien 1968) show a better effect for chlorpromazine and it may well be that the smaller trials are the outlying ones. However, Prien 1968 includes a high‐dose treatment arm (2000 mg/day of chlorpromazine), which may explain some of the heterogeneity in the results, see the subgroup analysis (Analysis 1.23) below.
1.3 Global state
1.3.1 No overall improvement
We found short‐term global state data ('no overall improvement' ‐ psychiatrist‐rated ; Analysis 1.3) significantly favoured chlorpromazine (n=728, 13 RCTs, RR 0.61 CI 0.46 to 0.82) compared with placebo. Medium‐term data up to six months also favoured chlorpromazine (n=1164, 14 RCTs, RR 0.71 CI 0.58 to 0.86). There was significant heterogeneity at both short term and medium term (I2=69% and 81%, respectively). There were no obviously outlying trials for this outcome at short term, so none were removed from the analysis. For medium term, removal of the largest trial, Prien 1968, restores homogeneity, an effect that does not appear in the subgroup analysis for high versus low dose, see the subgroup analysis (Analysis 1.23) below, and again, it may be the smaller trials that are outliers.
Nurse‐rated global state 'no overall improvement' scores (Analysis 1.4) were equivocal at short‐term assessment in one small study (Weckowicz 1960) (n=29, RR 0.91 CI 0.65 to 1.27). However, scores from one research group (Clark 1970a; Clark 1972; Clark 1977) favoured chlorpromazine at medium‐term assessment (n=84, 3 RCTs, RR 0.48 CI 0.35 to 0.64). Similar data were recorded in continuous form in only one small study (Borison 1991) and results were equivocal (n=19, 1 RCT; Analysis 1.5).
1.3.2 Severity of illness
We found estimates by psychiatrists for the severity of illness (Analysis 1.6) were equivocal at short‐term assessments (n=44, 1 RCT) however, medium‐term data showed significantly greater improvement in the chlorpromazine group (n=694, 3 RCTs, RR 0.80 CI 0.74 to 0.86) compared with placebo. Nurse‐rated severity of illness scores (Analysis 1.7) also favoured the chlorpromazine group (medium term, n=66, 2 RCTs, RR 0.63 CI 0.45 to 0.90) compared with placebo.
1.4 Leaving the study early
People allocated to chlorpromazine are more likely to remain in the study than participants given placebo (Analysis 1.8), in both short‐ (n=1065, 17 RCTs, RR 0.76 CI 0.63 to 0.92) and medium‐term studies (n=1831, 27 RCTs, RR 0.64 CI 0.53 to 0.78). The short‐term studies showed some heterogeneity (I2 = 51%). When we analyse the data using random‐effects, the result becomes non‐significant (n=1065, 17 RCTs, RR 0.80 CI0.58 to 1.10). Removing the study with results that were causing this heterogeneity, as judged by visual inspection (Cole 1964, which used very high doses of chlorpromazine in one arm of the trial) eliminates this heterogeneity. However, we did not find any significant differences in attrition rates from the comparatively large studies (Engelhardt 1960, Hogarty 1973, n=492), which were conducted for up to two years. Also, longer‐term data (Hogarty 1973) did not demonstrate a significant difference in retention rates.
1.5 Mental state
There are only a few studies with usable data relating to mental state.
1.5.1 Improved (50% reduction in BPRS)
We found no short‐term difference in mental state using a cut‐off point of at least a 50% decline in score to indicate 'improvement' (Cooper 2000, n=106; Analysis 1.9).
1.5.2 Average endpoint score (BPRS)
What continuous data there are favour chlorpromazine at short‐ (n=49, 2 RCTs, MD ‐4.82 CI ‐8.48 to ‐1.15) and medium‐term assessments (Tetreault 1969, n=30, MD ‐7.70 CI ‐14.77 to ‐0.63) (Analysis 1.10).
1.5.3 Average change score (BPRS)
See Table 1.
| Study | Chlorpromazine mean | Chlorpromazine SD | Chlorpromazine N | Placebo mean | Placebo SD | Placebo N |
| ‐4.3 | 19.1 | 53 (LOCF) | ‐2.9 | 16.3 | 53 (LOCF) |
LOCF ‐ last observation carried forward
1.6 Behaviour
1.6.1 Deteriorated/disturbed/un‐cooperative
We found participants did not differ significantly in experiencing a worsening in their behaviour (Analysis 1.11) at short‐term assessment (n=87, 2 RCTs), although data are heterogeneous (I2=65%). Medium‐term data also did not differ significantly (n=1040, 8 RCTs), but again, data are heterogeneous (I2=90%). There are no obviously outlying studies as all confidence intervals overlap. Removing the possibly outlying studies, either Prien 1968 or Hall 1955, does not restore homogeneity, nor does their removal change the results.
1.6.2 Unchanged
Both short‐term (Schiele 1961, n=40) and medium‐term (n=68, 2 RCTs) dichotomous data did not reveal any significant differences between chlorpromazine and placebo when assessing change in participants behaviour (Analysis 1.12).
1.6.3 Rosenthal Rating Scale
Tetreault 1969 provided data from the Rosenthal Rating Scale (Analysis 1.13) and we found short‐term data were not significantly different (n=30, 1 RCT) between chlorpromazine and the placebo group. Medium‐term data were also not significantly different (n=30, 1 RCT).
1.6.4 Parkside Behaviour Rating Scale
Baker 1959 used the Parkside Behaviour Rating Scale (Analysis 1.14) to assess behaviour and we found that those given chlorpromazine had a significantly better rating in their behaviour compared with the placebo group (n=14, MD 6.00 CI 1.97 to 10.03).
1.6.5 Fergus Falls Behavioural Rating Scale
Ramu 1999a used the Fergus Falls Behavioural Rating Scale (Analysis 1.15) to assess change in behaviour of participants and we found that behaviour was not significantly different (n=42, 1 RCT) between chlorpromazine and the placebo group.
1.7 Adverse effects
1.7.1 Extrapyramidal symptoms
There is evidence that chlorpromazine increases a person's chances of experiencing acute movement disorders (dystonia) (n=942, 5 RCTs, RR 3.47 CI 1.50 to 8.03), parkinsonism (n=1468, 15 RCTs, RR 2.11 CI 1.59 to 2.80), tremor (n=392, 7 RCTs, RR 1.66 CI 1.01 to 2.73) and rigidity (n=412, 7 RCTs, RR 2.24 CI 1.42 to 3.54). Akathisia (subjective feeling of restlessness that may lead to agitation) was dominated by one trial (Prien 1968) and did not occur more frequently in the chlorpromazine group than placebo (n=1164, 9 RCTs), nor did tardive dyskinesia (Clark 1977, n=18) nor ataxia (Hankoff 1962, n=97). We found extrapyramidal adverse effects were equivocal at both short‐ and medium‐term assessments from one small scale study (n=30) by Tetreault 1969. (See Analysis 1.16; Analysis 1.17).
1.7.2 Central nervous system
Chlorpromazine is clearly sedating (n=1627, 23 RCTs, RR 2.79 CI 2.25 to 3.45). There is also evidence that chlorpromazine increases a person's chances of experiencing fits (n=695, 3 RCTs, RR 3.11 CI 1.05 to 9.18) and weakness (n=92, 3 RCTs, RR 3.33 CI 1.02 to 10.88). Convulsions did not occur more frequently in the chlorpromazine group than placebo (Gwynne 1962, n=52; Analysis 1.18).
1.7.3 Blood, skin, liver and eyes
We found no significant differences in blood problems such as agranulocytosis and leucopenia (n=394, 7 RCTs), or rashes and itching (n=1313, 13 RCTs). Liver problems, mainly jaundice were also not significant (n=249, 4 RCTs). Further data from early trials suggest that chlorpromazine may well cause photosensitivity (n=799, 6 RCTs, RR 6.04 CI 3.22 to 11.32), and eye opacities or pigment problems (n=657, 2 RCTs, RR 3.09 CI 1.87 to 5.11) when large dosages of chlorpromazine are used. (See Analysis 1.19).
1.7.4 Other
Chlorpromazine clearly causes a lowering of blood pressure with accompanying dizziness (n=1488, 18 RCTs, RR 2.38 CI 1.74 to 3.25). Chlorpromazine is constipating, when compared with placebo (n=1117, 10 RCTs, RR 2.05 CI 1.33 to 3.15). We found data that urinary problems (n=926, 5 RCTs), and also blurred vision were not significantly different between chlorpromazine and placebo. We found that chlorpromazine does cause dry mouth (n=1015, 7 RCTs, RR 4.56 CI 2.35 to 8.85). Chlorpromazine increases participants’ weight (n=165, 5 RCTs, RR 4.92 CI 2.32 to 10.43). We found significantly more participants given chlorpromazine experienced nausea (n=1024, 5 RCTs, RR 2.07 CI 1.14 to 3.73). Salivation occurred significantly more frequently in the chlorpromazine group (n=830, 3TCTs, RR 3.37 CI 1.07 to 10.57). We found no clear evidence that chlorpromazine precipitates the frequency of amenorrhoea, menorrhagia or lactation problems. (See Analysis 1.20).
2. Subgroup analyses
2.1 Men versus women
Few studies reported outcomes for only men or women. The only primary outcome for which data were available for comparison is 'Behaviour deteriorated/disturbed/uncooperative' (Analysis 1.21). Schiele 1961 included only men (n=40) and three studies report the same outcome for women alone (Clark 1970b; Fleming 1959; Somerville 1960, total n=158). Results of randomised trials were equally significant for the subgroups.
2.2 Under 18 years of age versus 18‐65 years old versus older than 65
We could not perform this subgroup analysis as data were only available for people between the ages of 18 and 64 years.
2.3 Acutely ill people (< one month in duration) versus people who have been ill for longer
Limited data were available for a few primary outcomes (Analysis 1.22). We found that people who were chronically ill were more likely to have improved for short‐ and medium‐term global improvement, and disturbed behaviour compared with those whose illnesses were acute. No difference was found for rates of relapse between acute and chronic participants. Results of randomised trials were equally significant for all subgroups. However, these analyses are severely limited by the lack of studies in the acutely ill groups and no firm conclusion can be made.
2.4 High dose (> 501 mg/day) versus low doses (1‐500 mg/day)
For the outcome of relapse between nine weeks and six months, the high‐dose arm of Prien 1968 reports statistically significantly more favourable results for the chlorpromazine group, compared with the low‐dose group of studies (P < 0.005). However, for the low‐dose subgroup there remains significant heterogeneity that is not explained by the exclusion of the high‐dose arm of Prien 1968. For the short‐ and medium‐term outcome 'no global improvement', there is no clear difference between studies using high‐dose chlorpromazine to low dose. Higher dosages of chlorpromazine did not confer an advantage in reducing behavioural disturbances, compared with the low‐dose group (Analysis 1.23).
2.5 People diagnosed according to any operational criteria versus those who have not been diagnosed using operational criteria
For relapse in short‐term studies, the any operational criteria group had a better outcome but the sample size is too small to enable conclusions to be made. Relapse in medium‐term studies favoured participants diagnosed with operational criteria (P < 0.04). Global impression 'not improved' revealed no differences in the short and medium term. There were no apparent differences for severity of illness and behaviour (Analysis 1.24).
2.6 Studies published before 1990 versus studies published between 1990 and the present
Data were available for the outcomes of 'no overall improvement' in the short term and medium term, and 'behaviour deteriorated/disturbed/uncooperative' (Analysis 1.25). Few studies were available in the 1990 to 2007 group limiting the analysis. Results of randomised trials were equally significant for all subgroups.
3. Sensitivity analyses
3.1 Implication of randomisation
For the outcome ‘no overall improvement’ in the medium term, there were no differences in the results for one study (Ramu 1999a) that only implied randomisation and studies that explicitly stated that they were randomised (Analysis 1.26).
3.2 Assumptions for lost binary data
Two studies made assumptions regarding people lost to follow‐up for the outcome ‘no overall improvement’ in the short term; no differences were found (Analysis 1.27).
3.3 Risk of bias
None of the studies had a high risk of bias for allocation concealment, or blinding of participants and outcome assessors (see Figure 2). Those that had a high risk of bias for randomisation were included in the sensitivity analysis above (Analysis 1.26).
3.4 Imputed values
We did not undertake a sensitivity analysis to assess the effects of including data from trials where we used imputed values for ICC as there were no cluster‐randomised trials included in the review.
3.5 Dishonest researchers
It has come to our attention that Dr Richard Borison and Dr Bruce Diamond have been convicted of theft, making false statements and violations of state racketeering law in the USA. At this point, it seems that the crimes were to do with criminal diversion of funds, rather than falsifying study data (http://www.the‐scientist.com/yr1998/oct/notebook_981026.html). Nevertheless, we temporarily removed studies with either of these authors from the analyses to see if this made a substantive difference to the findings. Borison 1991 presented usable data on mental state (average endpoint score on BPRS), the removal of this study did not result in a substantive change in the findings. Borison 1991 is the reports one unique outcome for global state (average endpoint score on Clinical Global Impression (CGI)), so removal of the study results in the deletion of the complete outcome.
Discussion
Summary of main results
The summary below reflects the outcomes chosen for the summary of findings Table for the main comparison, and considered the main findings of this review that can support evidence‐based decision making.
1. Death
It may be surprising that there were no more deaths reported among over 5000 people with schizophrenia who were randomised to chlorpromazine or placebo. The lifetime incidence of suicide for people suffering from schizophrenia is 10% to 13% (Caldwell 1992). Furthermore, the use of large doses of neuroleptic has been associated with sudden death (Jusic 1994), but there are no records of such events within this review. The fact that there are none may reflect the fact that either trial‐care is more vigilant than routine care or that death is an under‐reported outcome.
2. Relapse
Chlorpromazine reduced the number of participants experiencing a relapse compared with placebo during six months to two years follow‐up, but not in the short, medium or long term over two years, although data were heterogeneous. The removal of Prien 1968 and Spohn 1977 did restore homogeneity in the short‐ and medium‐term studies. The Prien 1968 study was different because of the very large dosages of chlorpromazine that were employed, although it had two arms, one with lower doses and another with higher doses, and these two arms were pooled in the analysis. An 86% efficacy was found for the high‐dose arm in preventing relapse and a 68% efficacy for the low‐dose arm; the confidence intervals only minimally overlap, so there is a potential impact of Prien 1968 on this outcome.
3. Global state
The best‐quality data would need to be reported from six months onwards, and trials only reported on this outcome at short‐ and medium‐term follow‐ups. The efficacy of chlorpromazine for improving global state is 39% for short‐term data and 29% for medium term, but the data are heterogenous. Considering that there was very little antipsychotic treatment that preceded the advent of chlorpromazine, such an efficacy can be considered nothing less than revolutionary for those with very serious mental illnesses.
4. Leaving the study early
The finding that using chlorpromazine results in more people staying in the study could be seen as heartening. Perhaps a genuine decrease in the distressing symptoms of schizophrenia leads to an increased concordance with medication despite the unpleasant side effects of this drug. On the other hand, this apparent willingness to comply may be due partly to sedation and hypotension. These effects (the former being linked to emotive terms such as the 'chemical straitjacket') may decrease a person's ability to make his/her own decisions.
5. Mental state
In spite of 45 years of research on this benchmark anti‐psychotic treatment, very little can be said from trials regarding its direct effect on mental state in general or specific symptoms of schizophrenia.
6. Satisfaction with care
No studies reported on this outcome, so it is not possible to make any conclusions as to participants’ satisfaction with chlorpromazine treatment.
7. Behaviour
There are more data regarding behaviour. No difference in the occurrence of behaviour judged to be disturbed or deteriorated was found in both short‐ and medium‐term analyses, but the medium‐term result is based on heterogeneous data (I2=90%). Other measures of behaviour 'unchanged' and the modified Rosenthal scale and the Fergus Falls scale did not result in any significant differences. Continuous endpoint data (Baker 1959) derived from the Parkside Behaviour Rating Scale did favour chlorpromazine but there were only seven participants in each group and we can have no real confidence in this finding.
8. Adverse effects
Clinicians will not be surprised that chlorpromazine produces acute movement disorders, parkinsonism, fits, tremor, rigidity, weakness and sleepiness. This Cochrane review, however, is a rare report of the best available and quantitative data on this compound that is now over half a century old. Estimates of the incidence of movement disorders such as tardive dyskinesia, however, are not available from this review, as these necessitate a long follow‐up period that was only attempted in a few trials. Evidence supporting a link between chlorpromazine and akathisia is much less convincing than that for acute movement disorders, such as oculogyric crisis, and parkinsonism. This suggests that chlorpromazine may be less potent a cause of this unpleasant adverse effect than other compounds.
Taking chlorpromazine commonly causes people to become sleepy. This is an effect that, at times, may be welcomed by clinicians, but not necessarily by those with schizophrenia. Short‐term sedation can be advantageous for clinicians trying to manage people with very disturbed behaviour. Sedation often helps to bring a difficult and dangerous situation under control and gives time for antipsychotic measures to be effective.
In addition, chlorpromazine has a tendency to cause other adverse effects such as jaundice, photosensitivity, eye opacities, low blood pressure, constipation, urinary retention, blurred vision and dry mouth. The worrying data regarding eye opacities is all derived from one trial (Prien 1968). This large trial, however, used up to two grams of chlorpromazine a day and it is likely that the lower doses more usual in current practice would result in less risk of this adverse effect. Chlorpromazine frequently causes weight increase.
9. Cost of care
Again, no studies reported on this outcome, so it is not possible to make any conclusions about the cost of chlorpromazine treatment.
10. Subgroup analyses
As was likely from the start, the power to detect a real difference between studies in any one of the subgroup analyses was very low. Only subsets of already limited lists of trials were available. The wide confidence intervals could be hiding true differences in effect between the groups. The only suggestions of statistically significant differences were for acutely ill versus chronically ill for the outcomes of global improvement and behaviour, and for high‐dose versus low‐dose studies (relapse between nine weeks and six months). It is important to remember that this is now a non‐randomised comparison between studies, rather than within a study, and that this is one of many statistical tests that were undertaken on this dataset. Further complicating matters is the fact that the other outcomes within this particular subgroup analysis did not clearly support or refute this difference between high and low doses. Prien 1968 is an unusual study. Using two grams of chlorpromazine per day would be unacceptable in most situations today, a view supported by some of the findings of this review (n=657, 2 RCTs, RR eye opacities from two grams chlorpromazine per day, 3.09 CI 1.9 to 5.1).
Overall completeness and applicability of evidence
Applicability
The 55 included studies in this version of this review include many people who would be recognisable in everyday practice. There are those with strictly diagnosed illnesses, very likely to suffer from schizophrenia, and people whose illness was diagnosed using less rigorous criteria. The results of the subgroup analyses on diagnostic rigour (see Effects of interventions section 2.5) also support the assertion that the results are widely applicable.
The dose of chlorpromazine in the studies included in this review could be considered high (mean 574.1 mg/day SD 445) but, again, these levels are probably common for people with persistent schizophrenia across the globe.
Although the outcomes that have been used in this review are accessible to both clinicians and recipients of care, generalising to treatment in community settings, could be problematic. Most studies were undertaken in hospital, whereas the great majority of people with schizophrenia are in the community.
Homogeneity
Some results are difficult, or impossible, to interpret because of heterogeneity. The test for homogeneity is based on I2 analysis, and is often fairly weak, as the number of studies is small. However, the results of such tests, when statistically significant, suggest caution when adding trial data together.
Quality of the evidence
The quality of the current evidence is very low based on GRADE (Schünemann 2008). The majority of studies did not report the method of randomisation, and only two trials described method of allocation concealment. Although most studies were reported to be double blind, it was not clear in many whether or not the assessors were blinded to treatment group of participants. Forty‐eight out of the 55 included studies were rated as high risk of bias for selective reporting. Studies frequently presented both dichotomous and continuous data in graphs, or just reported statistical measures of probability (P values). This often made it impossible to acquire raw data for synthesis. It was also common to use P values as a measure of association between intervention and outcomes instead of showing the strength of the association. Although P values are influenced by the strength of the association, they also depend on the sample size of the groups. It is sometimes possible to extract raw data from P values, but their exact values are needed. In the reviewed studies this was not possible, because they were reported as 'P < 0.05' or 'P > 0.05'. Frequently, continuous data were presented without providing standard deviations or standard errors (33/55 trials) or no data were presented at all (11/55 trials). In this way a lot of potentially informative data were lost. In some studies it seemed that attempts had been made to use the original trials as vehicles for answering a host of other questions about schizophrenia. As a consequence, data from the randomised parts of the studies became buried beneath copious subgroup analyses.
Potential biases in the review process
1. Adding the old to the new
This work includes studies that span nearly five decades of evaluative studies within psychiatry. It is possible that the rigour of these experiments has changed over time, as have the participants and even the formulation of the drug; it was thought that introduction of impurities in early formulations of chlorpromazine led to jaundice. There is some empirical evidence that the quality of schizophrenia trial reporting has not changed much over time (Thornley 1998) or, if it has changed, it may even have declined (Ahmed 1998). We have found no time‐related differences in reporting of studies within this review and no suggestion of a change of the effect size over time. Synthesis of the results of studies seems justified.
2. Failing to identify old trials
We identified trials by meticulous searching, including handsearching old files (2002 update) that covered the drug's development over its first two decades. Nevertheless, for this compound formulated so long ago, publication biases may be difficult to avoid. We did not detect any overt asymmetry of the funnel plots.
The strength of this review is that it presents up‐to‐date quantitative data for a benchmark treatment for schizophrenia that is used throughout the world.
3. Sensitivity analyses on dishonest researchers
We felt that it would be harsh to immediately delete all trial data associated with Drs Borison and Diamond without empirical data. That removal of their data makes no discernable difference to any outcome is reassuring.
Agreements and disagreements with other studies or reviews
We do not know of any other systematic reviews on this topic.
'Risk of bias' graph: review authors' judgements about each 'Risk of bias' item presented as percentages across all included studies.
'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 1 Death.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 2 Relapse.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 3 Global state: 1a. No overall improvement (psychiatrist‐rated).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 4 Global state: 1b. No overall improvement (nurse‐rated).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 5 Global state: 2. Average endpoint score ‐ short term (CGI, high score=worse).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 6 Global state: 3a. Severity of illness, severely ill or worse (CGI 5+ points, psychiatrist‐rated).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 7 Global state: 3b. Severity of illness, severely ill or worse ‐ medium term (CGI 5+ points, nurse‐rated).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 8 Leaving the study early.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 9 Mental state: 1. Improved ‐ short term (BPRS, 50% change).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 10 Mental state: 2. Average endpoint scores (BPRS, high score=worse).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 11 Behaviour: 1. Deteriorated/ disturbed/un‐cooperative.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 12 Behaviour: 2. Unchanged.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 13 Behaviour: 3. Average endpoint scores (RRS, high score=worse).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 14 Behaviour: 4. Average endpoint score ‐ short term (PBRS, high score=good).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 15 Behaviour: 5. Average endpoint scores ‐ medium term (Fergus Falls Behavioural rating scale).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 16 Adverse effects: 1. Movement disorders.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 17 Adverse effects: 2. Movement disorders: Average endpoint scores (Extrapyramidal Bilan, high score=worse).
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 18 Adverse effects: 1. Central nervous system.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 19 Adverse effects: 3. Blood, skin, liver, eyes.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 20 Adverse effects: 4. Other.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 21 SUBGROUP ANALYSIS: 1. MEN vs WOMEN: Behaviour: Deteriorated/disturbed/un‐cooperative.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 22 SUBGROUP ANALYSIS: 2. ACUTE vs CHRONIC.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 23 SUBGROUP ANALYSIS: 3. HIGH vs LOW DOSE.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 24 SUBGROUP ANALYSIS: 4. DIAGNOSTIC CRITERIA vs NO DIAGNOSTIC CRITERIA.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 25 SUBGROUP ANALYSIS: 5. STUDIES PRE‐1990 vs STUDIES 1990‐2007.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 26 SENSITIVITY ANALYSIS: 1. RANDOMISATION.
Comparison 1 CHLORPROMAZINE versus PLACEBO, Outcome 27 SENSITIVITY ANALYSIS: 2. ASSUMPTIONS FOR LOST BINARY DATA.
| CHLORPROMAZINE versus PLACEBO for schizophrenia | ||||||
| Patient or population: patients with schizophrenia | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control | CHLORPROMAZINE versus PLACEBO | |||||
| Death | See comment | See comment | Not estimable | 14 | ⊕⊝⊝⊝ | One study specifically reported mortality and there were no deaths in either the chlorpromazine or placebo group. There were no reports of death in any other study. |
| Relapse | 710 per 1000 | 461 per 1000 | RR 0.65 | 512 | ⊕⊝⊝⊝ | 2 trials report this outcome at 0‐8 weeks follow‐up, 3 trials at 6 months to 2 years, and 2 trials 2‐5 years, none showed a significant difference. |
| Global state: no overall improvement | 897 per 1000 | 637 per 1000 | RR 0.71 (0.58 to 0.90) | 1164 | ⊕⊝⊝⊝ | 13 trials also reported on this outcome at 0‐8 weeks follow‐up and showed a significant result in favour of chlorpromazine. |
| Leaving the study early | 200 per 1000 | 128 per 1000 | RR 0.64 | 1831 | ⊕⊕⊝⊝ | 17 trials reported on this outcome at 0‐8 weeks follow‐up and showed significant results in favour of chlorpromazine. 2 trials reported on this outcome at 6 months to 2 years follow‐up, and 1 trial at 2‐5 years, and there was no significant difference. |
| Satisfaction with treatment ‐ not reported | See comment | See comment | Not estimable | ‐ | See comment | No studies reported on this outcome. |
| Behaviour: deteriorated/ disturbed/un‐cooperative | 471 per 1000 | 231 per 1000 | RR 0.49 | 1040 | ⊕⊝⊝⊝ | 2 trials reported on this outcome at 0‐8 weeks follow‐up and found no significant difference. |
| Cost of care ‐ not reported | See comment | See comment | Not estimable | ‐ | See comment | No studies reported on this outcome. |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. 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 | ||||||
| 1 Serious risk of bias: the study had an unclear risk of bias for random sequence generation, allocation concealment and other bias as the drugs were provided by a pharmaceutical company. | ||||||
| Study | Chlorpromazine mean | Chlorpromazine SD | Chlorpromazine N | Placebo mean | Placebo SD | Placebo N |
| ‐4.3 | 19.1 | 53 (LOCF) | ‐2.9 | 16.3 | 53 (LOCF) | |
| LOCF ‐ last observation carried forward | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Death Show forest plot | 1 | 14 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 2 Relapse Show forest plot | 7 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 2.1 short term (0‐8 weeks) | 2 | 74 | Risk Ratio (M‐H, Random, 95% CI) | 0.24 [0.01, 5.49] |
| 2.2 medium term (9 weeks ‐ 6 months) | 4 | 809 | Risk Ratio (M‐H, Random, 95% CI) | 0.61 [0.21, 1.72] |
| 2.3 long term (6 months ‐ 2 years) | 3 | 512 | Risk Ratio (M‐H, Random, 95% CI) | 0.65 [0.47, 0.90] |
| 2.4 2 ‐ 5 years | 2 | 394 | Risk Ratio (M‐H, Random, 95% CI) | 0.74 [0.51, 1.09] |
| 3 Global state: 1a. No overall improvement (psychiatrist‐rated) Show forest plot | 27 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 3.1 short term (0‐8 weeks) | 13 | 728 | Risk Ratio (M‐H, Random, 95% CI) | 0.61 [0.46, 0.82] |
| 3.2 medium term (9 weeks ‐ 6 months) | 14 | 1164 | Risk Ratio (M‐H, Random, 95% CI) | 0.71 [0.58, 0.86] |
| 4 Global state: 1b. No overall improvement (nurse‐rated) Show forest plot | 4 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 4.1 short term (0‐8 weeks) | 1 | 29 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.91 [0.65, 1.27] |
| 4.2 medium term (9 weeks‐6 months) | 3 | 84 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.48 [0.35, 0.66] |
| 5 Global state: 2. Average endpoint score ‐ short term (CGI, high score=worse) Show forest plot | 1 | 19 | Mean Difference (IV, Fixed, 95% CI) | 0.10 [‐0.35, 0.55] |
| 6 Global state: 3a. Severity of illness, severely ill or worse (CGI 5+ points, psychiatrist‐rated) Show forest plot | 4 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 6.1 short term (0‐8 weeks) | 1 | 44 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.46 [0.16, 1.30] |
| 6.2 medium term (9 weeks ‐ 6 months) | 3 | 694 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.80 [0.74, 0.86] |
| 7 Global state: 3b. Severity of illness, severely ill or worse ‐ medium term (CGI 5+ points, nurse‐rated) Show forest plot | 2 | 66 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.63 [0.45, 0.90] |
| 8 Leaving the study early Show forest plot | 45 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 8.1 short term (0‐8 weeks) | 17 | 1065 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.76 [0.63, 0.92] |
| 8.2 medium term (9 weeks ‐ 6 months) | 27 | 1831 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.64 [0.53, 0.78] |
| 8.3 long term (>6 months ‐ 2 years) | 2 | 492 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.06 [0.71, 1.59] |
| 8.4 >2 ‐ 5 years | 1 | 374 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.68 [0.35, 1.36] |
| 9 Mental state: 1. Improved ‐ short term (BPRS, 50% change) Show forest plot | 1 | 106 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.25 [0.36, 4.40] |
| 10 Mental state: 2. Average endpoint scores (BPRS, high score=worse) Show forest plot | 2 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | |
| 10.1 short term (0‐8 weeks) | 2 | 49 | Mean Difference (IV, Fixed, 95% CI) | ‐4.82 [‐8.48, ‐1.15] |
| 10.2 medium term (9 weeks ‐ 6 months) | 1 | 30 | Mean Difference (IV, Fixed, 95% CI) | ‐7.70 [‐14.77, ‐0.63] |
| 11 Behaviour: 1. Deteriorated/ disturbed/un‐cooperative Show forest plot | 10 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 11.1 short term ‐ (0‐8 weeks) | 2 | 87 | Risk Ratio (M‐H, Random, 95% CI) | 0.67 [0.34, 1.35] |
| 11.2 medium term (9 weeks ‐ 6 months) | 8 | 1040 | Risk Ratio (M‐H, Random, 95% CI) | 0.49 [0.24, 1.00] |
| 12 Behaviour: 2. Unchanged Show forest plot | 2 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 12.1 short term (0‐8 weeks) | 1 | 40 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.63 [0.25, 1.58] |
| 12.2 medium term (9 weeks ‐ 6 months) | 2 | 68 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.15 [0.64, 2.07] |
| 13 Behaviour: 3. Average endpoint scores (RRS, high score=worse) Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | |
| 13.1 short term (0‐8 weeks) | 1 | 30 | Mean Difference (IV, Fixed, 95% CI) | ‐2.90 [‐6.74, 0.94] |
| 13.2 medium term (9 weeks ‐ 6 months) | 1 | 30 | Mean Difference (IV, Fixed, 95% CI) | ‐4.60 [‐9.47, 0.27] |
| 14 Behaviour: 4. Average endpoint score ‐ short term (PBRS, high score=good) Show forest plot | 1 | 14 | Mean Difference (IV, Fixed, 95% CI) | 6.0 [1.97, 10.03] |
| 15 Behaviour: 5. Average endpoint scores ‐ medium term (Fergus Falls Behavioural rating scale) Show forest plot | 1 | 42 | Mean Difference (IV, Fixed, 95% CI) | ‐2.36 [‐6.11, 1.39] |
| 16 Adverse effects: 1. Movement disorders Show forest plot | 25 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 16.1 acute movement disorders (dystonia) | 5 | 942 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.47 [1.50, 8.03] |
| 16.2 parkinsonism (includes EPS) | 15 | 1468 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.11 [1.59, 2.80] |
| 16.3 tremor | 7 | 392 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.66 [1.01, 2.73] |
| 16.4 rigidity | 7 | 412 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.24 [1.42, 3.54] |
| 16.5 akathisia | 9 | 1164 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.78 [0.54, 1.11] |
| 16.6 chronic movement disorders (tardive dyskinesia) | 1 | 18 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.5 [0.32, 6.94] |
| 16.7 ataxia | 1 | 97 | Risk Ratio (M‐H, Fixed, 95% CI) | 8.64 [0.94, 79.31] |
| 17 Adverse effects: 2. Movement disorders: Average endpoint scores (Extrapyramidal Bilan, high score=worse) Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | |
| 17.1 short term (0‐8 weeks) | 1 | 30 | Mean Difference (IV, Fixed, 95% CI) | ‐1.40 [‐3.22, 0.42] |
| 17.2 medium term (9 weeks ‐ 6 months) | 1 | 30 | Mean Difference (IV, Fixed, 95% CI) | ‐1.80 [‐4.19, 0.59] |
| 18 Adverse effects: 1. Central nervous system Show forest plot | 25 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 18.1 sleepiness | 23 | 1627 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.79 [2.25, 3.45] |
| 18.2 fits / loss of consciousness | 3 | 695 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.11 [1.05, 9.18] |
| 18.3 weakness | 3 | 92 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.33 [1.02, 10.88] |
| 18.4 convulsions | 1 | 52 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.33 [0.01, 7.82] |
| 19 Adverse effects: 3. Blood, skin, liver, eyes Show forest plot | 20 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 19.1 blood problems (agranulocytosis, leukopenia) | 7 | 394 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.08 [0.74, 5.83] |
| 19.2 rashes/itching/skin disorders | 13 | 1313 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.45 [0.92, 2.29] |
| 19.3 liver problems | 4 | 249 | Risk Ratio (M‐H, Fixed, 95% CI) | 4.31 [0.98, 18.95] |
| 19.4 photosensitivity | 6 | 799 | Risk Ratio (M‐H, Fixed, 95% CI) | 6.04 [3.22, 11.32] |
| 19.5 eye opacity / eye pigment problems | 2 | 657 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.09 [1.87, 5.11] |
| 20 Adverse effects: 4. Other Show forest plot | 21 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 20.1 blood pressure ‐ low +/‐ dizziness/syncope | 18 | 1488 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.38 [1.74, 3.25] |
| 20.2 constipation | 10 | 1117 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.05 [1.33, 3.15] |
| 20.3 urinary problems | 5 | 926 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.73 [0.70, 4.30] |
| 20.4 blurred vision | 7 | 962 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.16 [0.51, 2.65] |
| 20.5 dry mouth | 7 | 1015 | Risk Ratio (M‐H, Fixed, 95% CI) | 4.56 [2.35, 8.85] |
| 20.6 weight increase | 5 | 165 | Risk Ratio (M‐H, Fixed, 95% CI) | 4.92 [2.32, 10.43] |
| 20.7 weight decrease | 5 | 165 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.38 [0.22, 0.66] |
| 20.8 nausea/vomiting | 5 | 1024 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.07 [1.14, 3.73] |
| 20.9 salivation | 3 | 830 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.37 [1.07, 10.57] |
| 20.10 menorrhagia / abnormal menstruation | 2 | 46 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.39 [0.62, 3.13] |
| 20.11 amenorrhea | 1 | 161 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.83 [0.17, 3.99] |
| 20.12 lactation | 2 | 192 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.48 [0.57, 3.81] |
| 21 SUBGROUP ANALYSIS: 1. MEN vs WOMEN: Behaviour: Deteriorated/disturbed/un‐cooperative Show forest plot | 4 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 21.1 only men | 1 | 40 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.04 [0.00, 0.63] |
| 21.2 only women | 3 | 158 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.46 [0.29, 0.73] |
| 22 SUBGROUP ANALYSIS: 2. ACUTE vs CHRONIC Show forest plot | 31 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 22.1 acute ‐ Relapse (9 wks ‐ 6 months) | 1 | 127 | Risk Ratio (M‐H, Random, 95% CI) | 1.21 [0.94, 1.55] |
| 22.2 chronic ‐ Relapse (9 wks ‐ 6 months) | 3 | 682 | Risk Ratio (M‐H, Random, 95% CI) | 0.46 [0.18, 1.15] |
| 22.3 acute ‐ Global state 1. No overall improvement (0‐8 wks) | 2 | 149 | Risk Ratio (M‐H, Random, 95% CI) | 0.68 [0.44, 1.05] |
| 22.4 chronic ‐ Global state 1. No overall improvement (0‐8 wks) | 9 | 459 | Risk Ratio (M‐H, Random, 95% CI) | 0.57 [0.39, 0.84] |
| 22.5 acute ‐ Global state 2. No overall improvement (9 wks‐6 months) | 1 | 23 | Risk Ratio (M‐H, Random, 95% CI) | 0.18 [0.03, 1.28] |
| 22.6 chronic ‐ Global state 2. No overall improvement (9 wks‐6 months) | 12 | 1121 | Risk Ratio (M‐H, Random, 95% CI) | 0.74 [0.64, 0.86] |
| 22.7 acute ‐ Behaviour deteriorated/disturbed/unco‐operative | 1 | 42 | Risk Ratio (M‐H, Random, 95% CI) | 0.88 [0.60, 1.29] |
| 22.8 chronic ‐ Behaviour deteriorated/disturbed/unco‐operative | 9 | 1085 | Risk Ratio (M‐H, Random, 95% CI) | 0.49 [0.26, 0.92] |
| 23 SUBGROUP ANALYSIS: 3. HIGH vs LOW DOSE Show forest plot | 22 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 23.1 high ‐ Relapse (9 wks‐6 Months) | 1 | 420 | Risk Ratio (M‐H, Random, 95% CI) | 0.14 [0.08, 0.26] |
| 23.2 low ‐ Relapse (9 wks‐6 Months) | 3 | 474 | Risk Ratio (M‐H, Random, 95% CI) | 0.51 [0.26, 0.98] |
| 23.3 high ‐ Global state 1. No overall improvement (0‐8 wks) | 2 | 201 | Risk Ratio (M‐H, Random, 95% CI) | 0.48 [0.18, 1.25] |
| 23.4 low ‐ Global state 1. No overall improvement (0‐8 wks) | 9 | 425 | Risk Ratio (M‐H, Random, 95% CI) | 0.72 [0.52, 1.01] |
| 23.5 high ‐ Global state 1. No overall improvement (9 wks‐6 months) | 6 | 576 | Risk Ratio (M‐H, Random, 95% CI) | 0.69 [0.51, 0.94] |
| 23.6 low ‐ Global state 1. No overall improvement (9 wks‐6 months) | 3 | 493 | Risk Ratio (M‐H, Random, 95% CI) | 0.83 [0.60, 1.16] |
| 23.7 high ‐ Behaviour 1. Deteriorated/disturbed/unco‐operative | 2 | 447 | Risk Ratio (M‐H, Random, 95% CI) | 0.50 [0.04, 6.21] |
| 23.8 low ‐ Behaviour 1. Deteriorated/disturbed/unco‐operative | 2 | 462 | Risk Ratio (M‐H, Random, 95% CI) | 0.39 [0.27, 0.56] |
| 24 SUBGROUP ANALYSIS: 4. DIAGNOSTIC CRITERIA vs NO DIAGNOSTIC CRITERIA Show forest plot | 36 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 24.1 diagnostic criteria ‐ Relapse (0‐8 wks) | 1 | 34 | Risk Ratio (M‐H, Random, 95% CI) | 0.75 [0.26, 2.19] |
| 24.2 no diagnostic criteria ‐ Relapse (0‐8 wks) | 1 | 40 | Risk Ratio (M‐H, Random, 95% CI) | 0.05 [0.00, 0.76] |
| 24.3 diagnostic criteria ‐ Relapse (9 wks‐6 months) | 2 | 662 | Risk Ratio (M‐H, Random, 95% CI) | 0.34 [0.13, 0.90] |
| 24.4 no diagnostic criteria ‐ Relapse (9 wks‐6 months) | 2 | 147 | Risk Ratio (M‐H, Random, 95% CI) | 1.03 [0.66, 1.60] |
| 24.5 diagnostic criteria ‐ Global state 1. No overall improvement (0‐8 wks) | 3 | 192 | Risk Ratio (M‐H, Random, 95% CI) | 0.63 [0.40, 0.99] |
| 24.6 no diagnostic criteria ‐ Global state 1. No overall improvement (0‐8 wks) | 11 | 565 | Risk Ratio (M‐H, Random, 95% CI) | 0.64 [0.46, 0.90] |
| 24.7 diagnostic criteria ‐ Global state 1. No overall improvement (9 wks‐6 months) | 3 | 689 | Risk Ratio (M‐H, Random, 95% CI) | 0.77 [0.57, 1.05] |
| 24.8 no diagnostic criteria ‐ Global state 1. No overall improvement (9 wks‐6 months) | 11 | 475 | Risk Ratio (M‐H, Random, 95% CI) | 0.71 [0.59, 0.85] |
| 24.9 diagnostic criteria ‐ Global state 3. Severity of illness (9 wks‐6 months) | 1 | 628 | Risk Ratio (M‐H, Random, 95% CI) | 0.82 [0.75, 0.88] |
| 24.10 no diagnostic criteria ‐ Global state 3. Severity of illness (9 wks‐6 months) | 2 | 66 | Risk Ratio (M‐H, Random, 95% CI) | 0.58 [0.36, 0.95] |
| 24.11 diagnostic criteria ‐ Behaviour 1. deteriorated/disturbed/unco‐operative | 2 | 670 | Risk Ratio (M‐H, Random, 95% CI) | 0.48 [0.14, 1.72] |
| 24.12 no diagnostic criteria ‐ Behaviour 1. deteriorated/disturbed/unco‐operative | 8 | 457 | Risk Ratio (M‐H, Random, 95% CI) | 0.56 [0.33, 0.97] |
| 25 SUBGROUP ANALYSIS: 5. STUDIES PRE‐1990 vs STUDIES 1990‐2007 Show forest plot | 32 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 25.1 Pre‐1990 ‐ Global state 1. No overall improvement (0‐8 wks) | 11 | 555 | Risk Ratio (M‐H, Random, 95% CI) | 0.64 [0.46, 0.91] |
| 25.2 1990 to 2002 ‐ Global state 1. No overall improvement (0‐8 wks) | 3 | 202 | Risk Ratio (M‐H, Random, 95% CI) | 0.61 [0.40, 0.94] |
| 25.3 Pre‐1990 ‐ Global state 1. No overall improvement (9 wks‐6 months) | 13 | 1121 | Risk Ratio (M‐H, Random, 95% CI) | 0.77 [0.67, 0.88] |
| 25.4 1990 to 2002 ‐ Global state 1. No overall improvement (9 wks‐6 months) | 1 | 43 | Risk Ratio (M‐H, Random, 95% CI) | 0.48 [0.30, 0.79] |
| 25.5 Pre‐1990 ‐ Behaviour 1. deteriorated/disturbed/unco‐operative | 9 | 1085 | Risk Ratio (M‐H, Random, 95% CI) | 0.49 [0.26, 0.92] |
| 25.6 1990 to 2002 ‐ Behaviour 1. deteriorated/disturbed/unco‐operative | 1 | 42 | Risk Ratio (M‐H, Random, 95% CI) | 0.88 [0.60, 1.29] |
| 26 SENSITIVITY ANALYSIS: 1. RANDOMISATION Show forest plot | 14 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 26.1 not stated ‐ Global state: 1. No overall improvement (9 weeks ‐ 6 months) | 1 | 43 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.48 [0.30, 0.79] |
| 26.2 stated ‐ Global state: 1. No overall improvement (9 weeks ‐ 6 months) | 13 | 1121 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.80 [0.75, 0.85] |
| 27 SENSITIVITY ANALYSIS: 2. ASSUMPTIONS FOR LOST BINARY DATA Show forest plot | 14 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 27.1 LOCF ‐ Global state: 1. No overall improvement (0‐8 weeks) | 2 | 86 | Risk Ratio (M‐H, Random, 95% CI) | 0.46 [0.17, 1.29] |
| 27.2 not LOCF ‐ Global state: 1. No overall improvement (0‐8 weeks) | 12 | 671 | Risk Ratio (M‐H, Random, 95% CI) | 0.66 [0.49, 0.88] |
