Macrolide antibiotics for bronchiectasis
Abstract
Background
Bronchiectasis is a chronic respiratory disease characterised by abnormal and irreversible dilatation and distortion of the smaller airways. Bacterial colonisation of the damaged airways leads to chronic cough and sputum production, often with breathlessness and further structural damage to the airways. Long‐term macrolide antibiotic therapy may suppress bacterial infection and reduce inflammation, leading to fewer exacerbations, fewer symptoms, improved lung function, and improved quality of life. Further evidence is required on the efficacy of macrolides in terms of specific bacterial eradication and the extent of antibiotic resistance.
Objectives
To determine the impact of macrolide antibiotics in the treatment of adults and children with bronchiectasis.
Search methods
We identified trials from the Cochrane Airways Trials Register, which contains studies identified through multiple electronic searches and handsearches of other sources. We also searched trial registries and reference lists of primary studies. We conducted all searches on 18 January 2018.
Selection criteria
We included randomised controlled trials (RCTs) of at least four weeks' duration that compared macrolide antibiotics with placebo or no intervention for the long‐term management of stable bronchiectasis in adults or children with a diagnosis of bronchiectasis by bronchography, plain film chest radiograph, or high‐resolution computed tomography. We excluded studies in which participants had received continuous or high‐dose antibiotics immediately before enrolment or before a diagnosis of cystic fibrosis, sarcoidosis, or allergic bronchopulmonary aspergillosis. Our primary outcomes were exacerbation, hospitalisation, and serious adverse events.
Data collection and analysis
Two review authors independently screened the titles and abstracts of 103 records. We independently screened the full text of 40 study reports and included 15 trials from 30 reports. Two review authors independently extracted outcome data and assessed risk of bias for each study. We analysed dichotomous data as odds ratios (ORs) and continuous data as mean differences (MDs) or standardised mean differences (SMDs). We used standard methodological procedures as expected by Cochrane.
Main results
We included 14 parallel‐group RCTs and one cross‐over RCT with interventions lasting from 8 weeks to 24 months. Of 11 adult studies with 690 participants, six used azithromycin, four roxithromycin, and one erythromycin. Four studies with 190 children used either azithromycin, clarithromycin, erythromycin, or roxithromycin.
We included nine adult studies in our comparison between macrolides and placebo and two in our comparison with no intervention. We included one study with children in our comparison between macrolides and placebo and one in our comparison with no intervention.
In adults, macrolides reduced exacerbation frequency to a greater extent than placebo (OR 0.34, 95% confidence interval (CI) 0.22 to 0.54; 341 participants; three studies; I2 = 65%; moderate‐quality evidence). This translates to a number needed to treat for an additional beneficial outcome of 4 (95% CI 3 to 8). Data show no differences in exacerbation frequency between use of macrolides (OR 0.31, 95% CI 0.08 to 1.15; 43 participants; one study; moderate‐quality evidence) and no intervention. Macrolides were also associated with a significantly better quality of life compared with placebo (MD ‐8.90, 95% CI ‐13.13 to ‐4.67; 68 participants; one study; moderate‐quality evidence). We found no evidence of a reduction in hospitalisations (OR 0.56, 95% CI 0.19 to 1.62; 151 participants; two studies; I2 = 0%; low‐quality evidence), in the number of participants with serious adverse events, including pneumonia, respiratory and non‐respiratory infections, haemoptysis, and gastroenteritis (OR 0.49, 95% CI 0.20 to 1.23; 326 participants; three studies; I2 = 0%; low‐quality evidence), or in the number experiencing adverse events (OR 0.83, 95% CI 0.51 to 1.35; 435 participants; five studies; I2 = 28%) in adults with macrolides compared with placebo.
In children, exacerbation frequency was reduced more with macrolides than with placebo (IRR 0.50, 95% CI 0.35 to 0.71; 89 children; one study; low‐quality evidence). However there was no significant difference in this age group with regard to: hospitalisations (OR 0.28, 95% CI 0.07 to 1.11; 89 children; one study; low‐quality evidence), serious adverse events, defined within the study as exacerbations of bronchiectasis or investigations related to bronchiectasis (OR 0.43, 95% CI 0.17 to 1.05; 89 children; one study; low‐quality evidence), or adverse events (OR 0.78, 95% CI 0.33 to 1.83; 89 children; one study), in those receiving macrolides compared to placebo. The same study reported an increase in macrolide‐resistant bacteria (OR 7.13, 95% CI 2.13 to 23.79; 89 children; one study), an increase in resistance to Streptococcus pneumoniae (OR 13.20, 95% CI 1.61 to 108.19; 89 children; one study), and an increase in resistance to Staphylococcus aureus (OR 4.16, 95% CI 1.06 to 16.32; 89 children; one study) with macrolides compared with placebo. Quality of life was not reported in the studies with children.
Authors' conclusions
Long‐term macrolide therapy may reduce the frequency of exacerbations and improve quality of life, although supporting evidence is derived mainly from studies of azithromycin, rather than other macrolides, and predominantly among adults rather than children. However, macrolides should be used with caution, as limited data indicate an associated increase in microbial resistance. Macrolides are associated with increased risk of cardiovascular death and other serious adverse events in other populations, and available data cannot exclude a similar risk among patients with bronchiectasis.
PICOs
Plain language summary
Macrolide antibiotics for bronchiectasis
Background to the question
Bronchiectasis is a long‐term respiratory condition. The airways in the lungs are damaged, and people are prone to infection. Symptoms are chronic cough and the production of sputum (coughed‐up material (phlegm) from the lower airways). Moreover, bronchiectasis is associated with a mortality rate more than twice that of the general population.
Long‐term antibiotic therapy with macrolides (such as azithromycin, roxithromycin, erythromycin, and clarithromycin) may reduce the cycle of reinfection, reduce symptoms, and improve quality of life. We wanted to do this review to look at the evidence on use of macrolides in people with bronchiectasis. This review is intended to help people such as guideline producers, doctors, and patients make decisions about whether to use or recommend macrolides.
Study characteristics
We found 15 studies that compared macrolides with placebo (a substance or treatment with no benefit) or no intervention. Eleven studies involved 690 adults (aged 18 years and older) and four studies involved 190 children. Among adults, six used azithromycin, four roxithromycin, and one erythromycin. The four studies with children used azithromycin, clarithromycin, erythromycin, or roxithromycin. This review is current to January 2018.
Main results
The studies on azithromycin reported improved quality of life in adults. We do not have sufficient evidence from other macrolides to make a robust judgement on their use, and we similarly have insufficient evidence from children to draw clear conclusions.
Although we found only a few trials, they do show a possible increase in antibiotic resistance. Antibiotic resistance is seen when an antibiotic becomes less effective at killing the bacteria causing the chest infection.
We know that macrolides are associated with higher risk of cardiovascular death and other serious adverse events when they are used to treat other conditions. The data in our review suggest it is possible that people with bronchiectasis are at risk for these adverse effects when taking macrolides.
Quality of the evidence
Generally the limited number of studies evaluating macrolides and the variation among them indicate that we cannot be sure of the overall effect of their use in bronchiectasis. Further high‐quality studies are needed to examine the role of long‐term macrolide antibiotics in the treatment of adults and children with bronchiectasis.
Authors' conclusions
Summary of findings
| Macrolides compared with placebo for adults with bronchiectasis | ||||||
| Patient or population: adults with bronchiectasis | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Risk with placebo | Risk with macrolides | |||||
| ≥ 1 exacerbation | 714 per 1000 | 459 per 1000 | OR 0.34 | 341 | ⊕⊕⊕⊝ | 2 studies azithromycin (1750 mg/week for 52 weeks; 1500 mg/week for 6 months) 1 study erythromycin (3500 mg/week for 48 weeks) |
| Hospitalisation: all cause | 133 per 1000 | 79 per 1000 | OR 0.56 | 151 | ⊕⊕⊝⊝ | 2 studies azithromycin (1000 mg/week for 12 weeks; 1750 mg/week for 52 weeks) |
| Serious adverse events | 86 per 1000 | 44 per 1000 | OR 0.49 | 326 | ⊕⊕⊝⊝ | 2 studies azithromycin (1500 mg/week for 6 months; 1000 mg/week for 12 weeks) 1 study erythromycin (3500 mg/week for 48 weeks) |
| All‐cause mortality | 0 per 1000 | 0 per 1000 | not estimable | 540 | ⊕⊕⊝⊝ | 4 studies azithromycin (1000 to 1750 mg/week for 12 to 52 weeks) |
| Quality of life: endpoint | Mean SGRQ score at endpoint in placebo groups was 39.1 points. | MD 8.90 lower (13.13 lower to 4.67 lower) | ‐ | 68 | ⊕⊕⊕⊝ | 1 study azithromycin (1000 mg/week for 12 weeks) |
| Quality of life: change | Mean change in SGRQ score ranged from ‐1.3 to ‐8.9 points. | MD 2.86 lower | ‐ | 305 | ⊕⊕⊝⊝ | 1 study azithromycin (1500 mg/week for 6 months) |
| *The risk in the intervention group (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. | ||||||
| aEffect observed only with azithromycin (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). bUnclear allocation concealment and baseline imbalances on Lourdesamy (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). cTwo small studies and wide confidence interval (one point deducted for imprecision). dWide confidence interval (one point deducted for imprecision). eIn three of the seven studies, study methods were not clearly reported (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). fA total of 28 participants across four studies were lost to follow‐up with no further details available and unclear details of withdrawals in one study (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). gRandomisation, blinding, and other study methods unclear in two studies (Asintam; Juthong) (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). hWide confidence interval and mean difference does not exceed the threshold for clinical significance (one point deducted for imprecision). | ||||||
| Macrolides compared with no intervention for adults with bronchiectasis | ||||||
| Patient or population: adults with bronchiectasis | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Risk with no intervention | Risk with macrolides | |||||
| ≥ 1 exacerbation | Study population | OR 0.31 | 43 | ⊕⊕⊕⊝ | Roxithromycin (1050 mg/week for 6 months) | |
| 762 per 1000 | 498 per 1000 | |||||
| Hospitalisations ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Serious adverse events ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Mortality | No deaths in two trials, although in 1 study (azithromycin), 6 participants were lost to follow‐up | not estimable | 88 | ⊕⊝⊝⊝ | 1 study azithromycin (750 mg/week for 3 months) | |
| QoL SGRQ: endpoint total score | Mean SGRQ: endpoint total score of 51.7 points | MD 8.81 lower (14.33 lower to 3.28 lower) | ‐ | 89 | ⊕⊕⊕⊝ | 1 study roxithromycin (1050 mg/week for 6 months) |
| QoL SGRQ: change in total score | Mean SGRQ: change in total score of 4.1 | MD 12 lower | ‐ | 30 | ⊕⊕⊕⊝ | Azithromycin (750 mg/week for 3 months) |
| *The risk in the intervention group (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. | ||||||
| aOpen‐label study (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). bUnclear randomisation and study methods (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). c6 participants in one study lost to follow‐up and no further details reported (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). | ||||||
| Macrolides compared with placebo for children with bronchiectasis | ||||||
| Patient or population: children with bronchiectasis | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Risk with placebo | Risk with macrolides | |||||
| Exacerbation frequency | Number of exacerbations 195 (median: 4 range 0‐14) | Number of exacerbations 104 (median: 2 range 0‐9) | IRR 0.50 95% CI 0.35 to 0.71 | 89 | ⊕⊕⊝⊝ | Azithromycin (30 mg/kg/week for up to 24 months) |
| Hospitalisation: all‐cause | 205 per 1000 | 67 per 1000 | OR 0.28 | 89 | ⊕⊕⊝⊝ | Azithromycin (30 mg/kg/week for 24 months) |
| Serious adverse events | 432 per 1000 | 246 per 1000 | OR 0.43 | 89 | ⊕⊕⊝⊝ | Azithromycin (30 mg/kg/week for 24 months) |
| Mortality | 1 child died but study group was not stated. | ‐ | 42 | ⊕⊕⊝⊝ | Erythromycin (875 to 1750 mg/kg/week for 52 weeks) | |
| Quality of life not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| *The risk in the intervention group (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. | ||||||
| aWide confidence interval that includes 1 (no difference) (one point deducted for imprecision). bLow event rates and low numbers (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). cUnclear information on randomisation, blinding, and other study methods (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). dNo information on participants lost to follow‐up (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). | ||||||
| Macrolides compared with no intervention for children with bronchiectasis | ||||||
| Patient or population: children with bronchiectasis | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Risk with no intervention | Risk with macrolides | |||||
| Exacerbations ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Hospitalisation ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Serious adverse events ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Mortality | 0 per 1000 | 0 per 1000 | not estimable | 34 | ⊕⊕⊝⊝ | Clarithromycin (105 mg/kg/week for 3 months) |
| Quality of life ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| *The risk in the intervention group (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. | ||||||
| aInsufficient information on study methods and procedures (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). bNot blinded (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). | ||||||
Background
Description of the condition
Bronchiectasis is a chronic respiratory disease characterised by abnormal and irreversible dilatation and distortion of the airways (Pasteur 2010). Bacterial colonisation of the damaged airways leads to chronic cough and sputum production, often with breathlessness and further structural damage to the airways. Diagnosis is made by computed tomography (CT) scanning of the chest when appropriate clinical symptoms are identified (Chang 2010), but asymptomatic radiological evidence of bronchiectasis may be noted (Kwak 2010).
Bronchiectasis has many causes, generally involving major or repeated insults to the lungs. Severe infections including pneumonia, tuberculosis, and pertussis may cause bronchiectasis, particularly if they occur during childhood while the lungs are still developing. Connective tissue disorders and defects in the immune system are other common causes of bronchiectasis, but many cases are idiopathic. Cystic fibrosis leads to severe, progressive bronchiectasis and usually is considered a separate entity from 'non‐cystic fibrosis' bronchiectasis. This review will exclude bronchiectasis secondary to cystic fibrosis.
Prevalence estimates are unclear owing to variable diagnostic strategies (Weycker 2005), a well as higher prevalence rates in developing countries (Habesoglu 2011), but the global disease burden is increasing, with mortality rising by 3% per year between 2001 and 2007, in England and in Wales (Roberts 2010), and hospitalisations by 3% per year in the United States (Seitz 2010). Prevalence is higher in women and those over 60 years of age (Roberts 2010; Seitz 2012). However, prevalence rates may be increasing more rapidly than was previously estimated, with 67 cases per 100,000 general population reported in Germany (Ringshausen 2015), and with UK prevalence rising from 350.5 to 566.1 in women and from 301.2 to 485.5 in men, affecting around 262,900 adults (Quint 2016). Similarly, UK incidence rates increased by 63% over nine years to 2013, rising from 21.2 to 35.2 in women and 8.2 to 26.9 in men, per 100,000 person‐years (Quint 2016). In paediatric populations, younger children and more frequent exacerbations are associated with worse quality of life (Kapur 2012). A higher prevalence of bronchiectasis has been reported among indigenous children in Australia (15:1000) and Alaska (16:1000) (Chang 2002). Incidence rates of 3.7 per 10,000 per year in children under 15 years of age have been reported in New Zealand (Twiss 2005). This equates to a prevalence of 1:3000 children overall and 1:625 children of Pacific Island descent (Twiss 2005). However, these increases may be due to improved diagnosis resulting from easier access to high‐quality CT scanners, rather than reflecting a true rise in prevalence (Goeminne 2016).
Estimated European mortality rates are 0.3 per 100,000 general population in EU countries (ranging from 0.01 in Germany to 1.18 in the UK) and 0.2 per 100,000 in non‐EU countries (ranging from 0.01 in Azerbaijan to 0.67 in Kyrgyzstan), based on data to 2009 (European Lung White Book 2013). Recent age‐adjusted mortality rates in the UK are estimated to be 2.26 times higher in women and 2.14 times higher in men compared with the wider population (Quint 2016). This information is based on estimated mortality rates for bronchiectasis of 1437.7 per 100,000 and for the general population of 635.9 per 100,000 (Quint 2016).
Description of the intervention
Chronic airway infection with pathogens such as Pseudomonas aeruginosa and Haemophilus influenzae and neutrophil‐mediated airway inflammation are key drivers of disease progression and poor outcomes in bronchiectasis (Chalmers 2012; Chalmers 2014; Finch 2015). Long‐term antibiotic therapy therefore is often prescribed with the intention of suppressing bacterial load and reducing airway inflammation (Chalmers 2012). This in turn aims to reduce exacerbations, improve symptoms, and improve quality of life (Haworth 2014). Prolonged antibiotic treatment can be administered in the form of oral or inhaled antibiotics. Inhaled antibiotics offer the advantage of delivering a higher dose of the drug directly to the site of bronchiectasis infection, with less potential for collateral damage and resistance; however, they are often time consuming to administer (Brodt 2014). Oral antibiotics by contrast are typically cheaper and easier to administer than inhaled antibiotics.
Oral antibiotics may be given at lower doses than those used to treat acute infection, with the aims of reducing adverse effects and promoting compliance (Haworth 2014). Macrolide antibiotics are antibacterial agents with anti‐inflammatory and immunomodulatory properties (Haworth 2014). Long‐acting macrolide antibiotics such as azithromycin can be given intermittently rather than requiring daily dosing. Penicillins, tetracyclines, and macrolides have all been tested as prolonged therapy in bronchiectasis (Pasteur 2010). National guidelines for bronchiectasis, such as those provided by the British Thoracic Society, suggest that use of long‐term antibiotic treatment should be considered for patients with three or more exacerbations per year (Pasteur 2010).
Long‐term use of macrolides in bronchiectasis is supported by their ease of administration, their effectiveness in cystic fibrosis and other neutrophilic lung diseases, and their reported anti‐inflammatory properties (Saiman 2003). Balanced against these traits is the potential for macrolides to induce antibiotic resistance and produce antibiotic‐related adverse effects, hearing impairment, and cardiotoxicity (Serisier 2013a).
How the intervention might work
Exacerbations, symptoms, and quality of life are directly linked to bacterial infection and airway inflammation in bronchiectasis (Chalmers 2012; Chalmers 2014). Macrolides are given as both antibacterial and anti‐inflammatory drugs, although it is unclear which of these properties is primarily responsible for the clinical effect observed in cystic fibrosis or bronchiectasis. Macrolides bind reversibly to the 50s ribosomal subunit, preventing bacterial protein synthesis (Haworth 2014). They therefore have broad activity against Gram‐positive organisms such as staphylococci and streptococci and exhibit a degree of activity against Gram‐negative organisms such as Haemophilus bacteria. It is interesting to note that macrolides show no bacteriocidal activity against P aeruginosa but may modify virulence by interfering with quorum sensing and virulence factors (Kohler 2010).
The anti‐inflammatory effects of macrolides have been known for decades and are classically demonstrated in their effectiveness against diffuse panbronchiolitis (Amsden 2005). Macrolides contain a macrocytic lactone ring that is thought to be responsible for most anti‐inflammatory effects (Haworth 2014). Macrolides are classified according to the number of lactone rings as 14‐, 15‐, and 16‐member ring macrolides. Macrolides confer potentially beneficial effects at every level of the 'vicious cycle' of bronchiectasis. They reduce the secretion of pro‐inflammatory cytokines from epithelial cells, inhibit leucocyte recruitment to the airway, inhibit neutrophil activation, and reduce oxidative stress (Zarogoulidis 2012).
Thus potential benefits of macrolides include suppression of bacterial infection, leading to reduced exacerbations, reduced cough and sputum production, and improved lung function and quality of life.
Why it is important to do this review
Bronchiectasis is associated with a mortality rate more than twice that of the general population ‐ 2.26 times higher in women and 2.14 times higher in men (Quint 2016). Data on the economic burden of bronchiectasis are few; however a 2001 US study reported 2.0 more days in hospital, 6.1 more outpatient appointments, and 27.2 more days of antibiotic treatment (Weycker 2005). It is estimated that additional annual costs associated with bronchiectasis ranged from USD 5681 to USD 7827 during the period between 2001 and 2009 (Joish 2013; Seitz 2010; Weycker 2005).
Frequent exacerbations lead to impaired quality of life and progressive lung damage with permanent loss of lung function (Martínez‐García 2007). Therefore, drug interventions that are effective in reducing the frequency of exacerbations should offer both short‐term and long‐term benefit for patients with bronchiectasis. A Cochrane Review of short‐term antibiotics provided little evidence on which to base a recommendation, with one small trial showing evidence of global improvement and pathogen eradication in sputum (Wurzel 2011). Another Cochrane Review of long‐term antibiotic therapy included 18 trials of moderate quality and provided evidence of reduced exacerbation frequency and hospitalisation but increased drug resistance (Hnin 2015). Neither of these Cochrane Reviews examined effects by class of antibiotics, and neither specifically created subgroups by macrolide therapy. A Cochrane Overview concluded that further evidence is required on the efficacy of antibiotics in terms of eradication of specific bacterial colonisation and the extent of antibiotic resistance (Welsh 2015). Recent recommendations from the European Task Force on Bronchiectasis further reinforced the importance of this question by identifying research on macrolide therapy as one of the key priorities in bronchiectasis (Aliberti 2016). Macrolides may potentially reduce exacerbations of bronchiectasis. Given their drawbacks, particularly cardiac toxicity as described by Ray 2012 and the potential for selecting antibiotic‐resistant organisms as discussed by Leclercq 2002, robust evidence on the effectiveness of macrolides is needed if they are to be used with confidence for this indication.
This review is being conducted alongside two other, closely related reviews: "Dual antibiotics for bronchiectasis" and"Head to head trials of antibiotics for bronchiectasis".
Objectives
To determine the impact of macrolide antibiotics in the treatment of adults and children with bronchiectasis.
Methods
Criteria for considering studies for this review
Types of studies
We included randomised controlled trials (RCTs) of at least four weeks' duration. We included cross‐over studies but used only data from the first pre‐cross‐over phase to eliminate potentially irreversible carry‐over effects (e.g. antibiotic resistance). We included studies reported as full text, those published as abstract only, and unpublished data.
Types of participants
We included adults and children with a diagnosis of bronchiectasis by bronchography, plain film chest radiography, or high‐resolution computed tomography who reported daily sputum expectoration for at least three months. We did not exclude participants whose condition was diagnosed by radiography alone. When a study included participants with different respiratory conditions, we included the study only if investigators performed a separate subgroup analysis for participants with bronchiectasis. We excluded studies if participants had been receiving continuous or high‐dose antibiotics immediately before enrolment, or if they had received a diagnosis of cystic fibrosis, sarcoidosis, or allergic bronchopulmonary aspergillosis. We defined children as individuals from six months to 18 years of age.
Types of interventions
We included studies comparing macrolide antibiotics with placebo, no intervention, or non‐macrolide antibiotics for long‐term management of stable bronchiectasis and reporting these different comparisons separately. We excluded studies looking at short‐term macrolides for management (as opposed to prevention) of exacerbations of bronchiectasis.
Types of outcome measures
We used exacerbation and hospitalisation rates as reported by study authors. We collected outcome data at a range of follow‐up points that best reflected available evidence from included studies (e.g. end of study, end of follow‐up, change from baseline).
Primary outcomes
-
Exacerbations (defined by study authors' criteria)
-
Hospitalisation (defined by study authors' criteria)
-
Serious adverse events defined by Hansen 2015, as follows: adverse events that result in death or life‐threatening events, requirement for hospitalisation or prolongation of existing hospitalisation, persistent or significant disability, or congenital anomalies; or events that are considered medically important
Secondary outcomes
-
Sputum volume and purulence
-
Measures of lung function (e.g. forced expiratory volume in one second (FEV1))
-
Systemic markers of infection (C‐reactive protein (CRP))
-
Adverse events (e.g. cardiac arrhythmias, gastrointestinal symptoms, hearing impairment)
-
Mortality (with this review indicating whether defined as all‐cause or bronchiectasis‐related in individual studies)
-
Emergence of resistance to antibiotics
-
Exercise capacity (e.g. the Six‐Minute Walk Distance test (6MWD))
-
Health‐related quality of life (e.g. St. George's Respiratory Questionnaire (SGRQ))
Reporting in the study one or more of the outcomes listed here was not an inclusion criterion for this review.
Search methods for identification of studies
Electronic searches
We identified studies from the Cochrane Airways Trials Register, which is maintained by the Information Specialist for the Group. The Cochrane Airways Trials Register contains studies identified from the following sources.
-
Monthly searches of the Cochrane Central Register of Controlled Trials (CENTRAL), through the Cochrane Register of Studies Online (crso.cochrane.org).
-
Weekly searches of MEDLINE Ovid SP 1946 to date.
-
Weekly searches of Embase Ovid SP 1974 to date.
-
Monthly searches of PsycINFO Ovid SP 1967 to date.
-
Monthly searches of Cumulative Index to Nursing and Allied Health Literature (CINAHL) EBSCO 1937 to date.
-
Monthly searches of Allied and Complementary Medicine (AMED) EBSCO.
-
Handsearches of the proceedings of major respiratory conferences.
Studies contained in the Cochrane Airways Trials Register are identified through search strategies based on the scope of Cochrane Airways. We have provided details of these strategies, as well as a list of handsearched conference proceedings, in Appendix 1. See Appendix 2 for search terms used to identify studies for this review.
We also conducted a search of the US National Institutes of Health Ongoing Trials Register, ClinicalTrials.gov (www.clinicaltrials.gov), and the World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch).
We searched all databases from their inception to January 2018 and imposed no restrictions on language of publication.
Searching other resources
We checked the reference lists of all primary studies and review articles for additional references and searched relevant manufacturers' websites for study information. We searched PubMed (www.ncbi.nlm.nih.gov/pubmed) for errata or retractions from included studies published in full text and reported within the review the date this was done.
Data collection and analysis
Selection of studies
Two review authors (DE and LF) independently screened titles and abstracts for inclusion of all potential studies identified as a result of the search and coded them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. We retrieved full‐text study reports/publications, and two review authors (CK and LF) independently screened the full texts, identified studies for inclusion, and identified and recorded reasons for exclusion of ineligible studies. We encountered no disagreements, so the need to consult a third review author (SS or SJM) did not arise. We identified and excluded duplicates and collated multiple reports of the same trial, so that each trial rather than each report was the unit of interest in the review. We recorded the selection process in detail in the PRISMA flow diagram and the Characteristics of excluded studies table (Moher 2009).
Data extraction and management
We used a data collection form that was piloted on at least one study in the review to extract study characteristics and outcome data. One review author (LF) extracted the following characteristics from included studies.
-
Methods: study design, total duration, details of 'run‐in' period, number of centres and their locations, settings, withdrawals, and dates the study was carried out.
-
Participants: number, mean age and range, gender, bronchiectasis severity, diagnostic criteria, baseline lung function, smoking history, inclusion and exclusion criteria.
-
Interventions and comparisons: intervention, comparison, concomitant medications, and excluded medications.
-
Outcomes: primary and secondary outcomes reported and follow‐up time points.
-
Notes: funding source and notable conflicts of interest of study authors.
Two review authors (LF and NR) independently extracted outcome data from the included studies. When investigators did not report outcome data in a usable way, we noted this in the Characteristics of included studies table. We resolved disagreements by reaching consensus or by involving a third review author (SS or SJM). One review author (LF) transferred data into the Review Manager 5 file (RevMan 2014), and a second review author (SS) verified and validated the information. One review author (CK) spot‐checked study characteristics for accuracy against the trial reports.
Assessment of risk of bias in included studies
Two review authors (NR and LF) independently assessed the risk of bias for each study, using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), according to the domains below. We resolved disagreements by discussion or by consultation with another review author (SS or SJM).
-
Random sequence generation.
-
Allocation concealment.
-
Blinding of participants and personnel.
-
Blinding of outcome assessment.
-
Incomplete outcome data.
-
Selective outcome reporting.
-
Other bias.
We graded each potential source of bias as bringing high, low, or unclear risk, provided a quote from the study report, and recorded our judgement in the 'Risk of bias' table. We summarised risk of bias judgements across different studies for each of the domains listed and reported these in a 'Risk of bias summary table' and a 'Risk of bias graph'. We considered blinding separately for different key outcomes when necessary (e.g. for unblinded outcome assessment, risk of bias for all‐cause mortality may be very different than for a patient‐reported pain scale). When information on risk of bias was related to unpublished data or correspondence with a study author, we noted this in the 'Risk of bias' table.
When considering treatment effects, we took into account the risk of bias for studies that contributed to that outcome.
Assessment of bias in conducting the systematic review
We conducted the review according to the previously published protocol and have reported deviations from it in the Differences between protocol and review section.
Measures of treatment effect
We analysed dichotomous data as odds ratios (ORs) and continuous data as mean differences (MDs) or standardised mean differences (SMDs). We analysed hospitalisation and exacerbation rates as rate ratios when possible. We entered data as a scale with a consistent direction of effect. We undertook meta‐analyses only when these were meaningful (i.e. when treatments, participants, and the underlying clinical question were similar enough for pooling to make sense). We narratively described skewed data reported as medians and interquartile ranges, as well as data not suitable for meta‐analysis (e.g. data from mixed methods regression). Our review did not include trials with multiple intervention arms, but if future updates of the review should identify this type of trial, we will include only the intervention arms relevant to this review. When we combined two comparisons (e.g. drug A vs placebo and drug B vs placebo) in the same meta‐analysis, we halved the control group to avoid double‐counting.
Unit of analysis issues
The study participant was the unit of analysis in all included studies. For exacerbation and admission rates, we focused on the number of events experienced by the participant during the trial. For cross‐over trials, we used only data from the first pre‐cross‐over phase to minimise potential bias from carry‐over effects.
Dealing with missing data
We contacted investigators or study sponsors to verify key study characteristics and to obtain missing numerical outcome data when possible (e.g. only abstract available). When this was not possible and we thought that missing data might introduce serious bias, we explored the impact of including such studies in the overall assessment of results by performing a sensitivity analysis.
Assessment of heterogeneity
We used the I2 statistic to measure heterogeneity among the studies in each analysis. When we identified substantial heterogeneity (> 50%), we reported this in the text and explored possible causes by conducting prespecified subgroup analyses (e.g. adults vs children).
Assessment of reporting biases
We were not able to pool more than 10 studies for any comparison; therefore, we were unable to explore small‐study effects and publication biases by using a funnel plot.
Data synthesis
We included outcomes in meta‐analyses when we considered study designs, interventions, and outcomes as sufficiently similar. When we identified substantial heterogeneity (> 50%), we reported outcomes in the text, revealing the direction and size of the effect, along with the strength of the evidence (risk of bias). Antibiotic studies varied by population, design, and outcomes. However, we identified few studies for each comparison, and estimates from a random‐effects model therefore may have been unreliable, we used a fixed‐effect model, reported data with 95% confidence intervals (CIs), and evaluated the impact of model choice by performing a sensitivity analysis, when appropriate. We synthesised and reported dichotomous and continuous data separately for each outcome (e.g. exacerbation/no exacerbation or exacerbation duration), and when study authors reported both end‐of‐study point estimates and change from baseline scores, we analysed these separately.
'Summary of findings' tables
We created 'Summary of findings' tables by using the following primary and secondary outcomes: exacerbations, hospitalisations, serious adverse events, deaths, and quality of life. We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of evidence from studies contributing data to meta‐analyses for these outcomes. We used methods and recommendations as described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), and we used GRADEproGDT software (GRADEproGDT). We justified all decisions to downgrade or upgrade the quality of evidence provided by studies by using footnotes and adding comments to aid understanding when necessary.
Subgroup analysis and investigation of heterogeneity
We planned to carry out the following subgroup analyses, although data were insufficient for comparisons of all subgroups. However, we chose to present the data for different macrolides as subgroups for all outcomes.
-
Macrolides versus other classes of long‐term antibiotics.
-
Types of macrolides.
-
Dose and frequency.
-
Duration.
We planned to use the following outcomes in conducting subgroup analyses.
-
Exacerbations.
-
Hospitalisations.
-
Serious adverse events.
We used the formal test for subgroup interactions provided in Review Manager 5 (RevMan 2014).
Sensitivity analysis
We evaluated effects of methodological study quality by removing studies at high or unclear risk of bias for the domains of random sequence generation and allocation concealment.
Results
Description of studies
Results of the search
A systematic search, conducted on 18 January 2018, identified 103 unique records of potentially relevant trials. Of these, we considered 63 records irrelevant following inspection of their titles and abstracts. We obtained and read full texts for the remaining 40 records and formally excluded eight records (documented in Excluded studies). We contacted the authors of one study (two records) awaiting classification (see Studies awaiting classification). A total of 15 trials, with 30 records, met our inclusion criteria for studies of macrolides for bronchiectasis. We have summarised the selection process in the study flow diagram (Figure 1).
Study flow diagram.
Included studies
In 11 trials (Altenburg 2013; Asintam 2012; Cymbala 2005; Diego 2013; Juthong 2011; Liu 2012; Liu 2014; Lourdesamy 2014; Sadigov 2013; Serisier 2013; Wong 2012), study participants were adults, and in the remaining four trials (Koh 1997; Masekela 2013; Valery 2013; Yalcin 2006), participants were children. See Characteristics of included studies for further details. See Table 1 for an overview of study characteristics.
| Study | Adults/ Children | No. of participants | Type of macrolide | Macrolide dose | Frequency | Delivery mode | Combined weekly dose | Comparison | Duration (months unless stated) |
| Altenburg 2013 | Adults | 83 | Azithromycin | 250 mg | Once daily | Oral | 1750 mg | Placebo | 12 |
| Asintam 2012 | Adults | 30 | Roxithromycin | 300 mg | Once daily | Oral | 2100 mg | Placebo | 12 weeks |
| Cymbala 2005 | Adults | 12 | Azithromycin | 500 mg | 3 days per week | Oral | 1000 mg | No intervention | 6 |
| Diego 2013 | Adults | 36 | Azithromycin | 250 mg | 3 days per week | Oral | 750 mg | No intervention | 3 |
| Juthong 2011 | Adults | 26 | Roxithromycin | 300 mg | Once daily | Oral | 2100 mg | Placebo | 8 weeks |
| Koh 1997 | Children | 25 | Roxithromycin | 4 mg/kg | Twice daily | Oral | 56 mg/kg | Placebo | 12 weeks |
| Liu 2012 | Adults | 50 | Roxithromycin, ambroxol hydrochloride | 150 mg | Once daily | Oral | 1050 mg | Ambroxol hydrochloride (no intervention) | 6 |
| Liu 2014 | Adults | 52 | Roxithromycin | 150 mg | Once daily | Oral | 1050 mg | No intervention | 6 |
| Lourdesamy 2014 | Adults | 78 | Azithromycin | 1000 mg | Weekly | 1000 mg | Placebo | 3 | |
| Masekela 2013 | Children | 42 | Erythromycin | 125 mg for children weighing < 15 kg and 250 mg ≥ 15 kg | Daily | Oral | 875 mg for children weighing < 15 kg and 1750 mg ≥ 15 kg | Placebo | 12 |
| Sadigov 2013 | Adults | 65 | Azithromycin | 500 mg | 3 days per week | Oral | 1500 mg | Placebo | 6 |
| Serisier 2013 | Adults | 117 | Erythromycin | 250 mg | Twice daily | Oral | 3500 mg | Placebo | 11 |
| Valery 2013 | Children | 89 | Azithromycin | 30 mg/kg up to a maximum of 600 mg | Once a week | Oral | 30 mg/kg up to a maximum of 600 mg | Placebo | 24 |
| Wong 2012 | Adults | 141 | Azithromycin | 500 mg | 3 days per week | Oral | 1500 mg | Placebo | 6 |
| Yalcin 2006 | Children | 34 | Clarithromycin, supportive therapies | 15 mg/kg | Daily | Oral | 105 mg/kg | Supportive therapies (no intervention) | 3 |
Methods
Fourteen of the 15 included studies were parallel‐group RCTs, and the remaining study was an RCT with a cross‐over design (Cymbala 2005). Nine trials were double‐blind, five were open‐label, and one did not report information on study blinding. The intervention duration ranged from eight weeks in Juthong 2011 to 24 months in Valery 2013. The percentage of participants who withdrew after randomisation ranged from 0 in Juthong 2011and Yalcin 2006 to 27% in Masekela 2013, with an average withdrawal proportion of 8.7% across all included studies.
Seven trials were conducted in Asia (Asintam 2012; Juthong 2011; Koh 1997; Liu 2012; Liu 2014; Lourdesamy 2014; Sadigov 2013); three in Australia/New Zealand (Serisier 2013; Valery 2013; Wong 2012); three in Europe (Altenburg 2013; Diego 2013; Yalcin 2006); one in South Africa (Masekela 2013); and one in the USA (Cymbala 2005). Please see Figure 2 for the global distribution of trials. The oldest study concluded in 1996 (Koh 1997), and the most recent in 2013 (Lourdesamy 2014). Three studies recruited participants through multiple centres (Altenburg 2013; Valery 2013; Wong 2012); the remainder were conducted at single centres (Figure 2).
Global distribution of studies.
Six trials used intention‐to‐treat analyses (Altenburg 2013; Juthong 2011; Serisier 2013; Valery 2013; Wong 2012; Yalcin 2006), and seven trials included in analyses only participants who completed the study (Cymbala 2005; Diego 2013; Koh 1997; Liu 2012; Liu 2014; Lourdesamy 2014; Masekela 2013); the analyses performed in two studies were unclear (Asintam 2012; Sadigov 2013). Nine studies reported power calculations for sample size estimation (Altenburg 2013; Asintam 2012; Cymbala 2005; Diego 2013; Lourdesamy 2014; Masekela 2013; Serisier 2013; Valery 2013; Wong 2012), and all six remaining studies reported statistically significant results (Juthong 2011; Koh 1997; Liu 2012; Liu 2014; Sadigov 2013; Yalcin 2006).
Note:We could not include results fromCymbala 2005in the review, as data from the pre‐cross‐over phase alone were not available owing to ineffective randomisation procedures. SeeCharacteristics of included studiesand the associated risk of bias table for additional details.
Participants
We chose to present separately data from adults and children and data on different macrolides.
Adults
Eleven studies included a total of 690 adults aged 18 years and older, with a diagnosis of bronchiectasis confirmed by high‐resolution computed tomography (HRCT). Three studies specified the following numbers of exacerbations in the preceding year as screening criteria: at least three (Altenburg 2013); two (Serisier 2013); and one (Wong 2012). The number of randomised participants in each study ranged from 12 in Cymbala 2005 to 141 in Wong 2012, with a mean age range of 48 years in Liu 2012 to 70.8 years in Cymbala 2005, although one study did not report this information (Sadigov 2013). Data on gender were missing for 81 randomised participants: Three trials reported gender distribution only for those who completed the study (Cymbala 2005; Diego 2013; Liu 2014), and one did not report gender (Sadigov 2013). Of 601 participants for whom data were available, 373 were female and 236 were male, with the percentage of male participants ranging from 23% in Asintam 2012 to 54% in Juthong 2011, across individual studies.
Three studies reported baseline disease severity in terms of Bhalla score: 9.5 (Liu 2014), 12.5 (Asintam 2012), and 26.5 (Diego 2013). Seven studies reported baseline FEV1 % predicted ranging from moderate impairment at 57% of predicted (Diego 2013), to mild impairment at 80.7% of predicted (Altenburg 2013), and two further studies reported baseline FEV1 as 1.08 L in Lourdesamy 2014 and 1.42 L in Juthong 2011. The remaining two studies did not report baseline lung function (Liu 2012; Sadigov 2013). Seven studies reported smoking status, with the proportion of current smokers ranging from none in Asintam 2012 and Serisier 2013 to 28% in Lourdesamy 2014 one study reported 66% current or ex‐smokers (Cymbala 2005), and four studies did not report this information (Diego 2013; Liu 2012; Sadigov 2013; Wong 2012).
Children
Four studies included a total of 190 randomised children, consisting of 81 girls and 98 boys (gender of participants lost to follow‐up was not reported in Masekela 2013), younger than 18 years of age (Koh 1997; Masekela 2013; Valery 2013; Yalcin 2006). Four studies reported a diagnosis of bronchiectasis by HRCT, and Valery 2013 included children with chronic suppurative lung disease, thus meeting clinical criteria when HRCT was not available. Sample sizes ranged from 25 in Koh 1997 to 89 children in Valery 2013, with mean age ranging from four years in Valery 2013 to 13 years in Koh 1997. Participants in Valery 2013 were indigenous children from Australia and New Zealand. Children in Masekela 2013 had confirmed HIV infection, were receiving highly active antiretroviral therapy (HAART), and had undergone a sweat test to rule out cystic fibrosis. Children in Koh 1997 had increased airway responsiveness, confirmed by a metacholine challenge test. Valery 2013 specified at least one exacerbation in the preceding year as one of its inclusion criteria.
Three studies reported baseline FEV1 % predicted as follows: 54.8% (Masekela 2013), 76.5% (Yalcin 2006), and 83% (Koh 1997). Valery 2013 did not report lung function.
Interventions
Adults
The 11 adult studies evaluated three types of oral macrolides. Six studies used azithromycin with doses ranging from 750 to 1750 mg per week for a period of 12 to 52 weeks, four studies used roxithromycin with doses ranging from 1050 to 2100 mg per week for 8 to 24 weeks, and one study used erythromycin at a dose of 3500 mg per week for 48 weeks. Seven studies compared the intervention with placebo (Altenburg 2013; Asintam 2012; Juthong 2011; Lourdesamy 2014; Sadigov 2013; Serisier 2013; Wong 2012), three studies compared the intervention with no intervention (Cymbala 2005; Diego 2013; Liu 2014), and one study compared the intervention plus an antimucolytic with the antimucolytic alone (Liu 2012). We have summarised in Table 1 further details of interventions provided in individual studies.
Note: For all outcomes from Diego 2013, we have included only the mean change score, pending clarification by study authors of reported standard deviations. Therefore, we have included these data in the text narratively.
Outcomes
One study reported all of our prespecified outcomes (Altenburg 2013).
Seven adult studies reported the frequency of exacerbations (Altenburg 2013; Asintam 2012; Diego 2013; Liu 2014; Sadigov 2013; Serisier 2013; Wong 2012), and three reported the time to first exacerbation (Altenburg 2013; Sadigov 2013; Wong 2012). Two children's studies reported the frequency of exacerbations (Masekela 2013; Valery 2013), and one also reported the time to first exacerbation and the duration of the exacerbation (Valery 2013).
Two adult studies reported hospitalisations (Altenburg 2013; Lourdesamy 2014), as did one paediatric study (Valery 2013).
All three adult studies reported serious adverse events (Lourdesamy 2014; Serisier 2013; Wong 2012), as did one study in which the participants were children (Valery 2013).
Five of the adult studies reported sputum volume (Asintam 2012; Cymbala 2005; Diego 2013; Lourdesamy 2014; Serisier 2013), as did two paediatric studies (Koh 1997; Yalcin 2006). Data from Cymbala 2005 were not usable (see note above).
Nine adult studies reported lung function, measured as FEV1 or forced vital capacity (FVC), or both (Altenburg 2013; Asintam 2012; Cymbala 2005; Diego 2013; Juthong 2011; Lourdesamy 2014; Sadigov 2013; Serisier 2013; Wong 2012), as did all four paediatric studies (Koh 1997; Masekela 2013; Valery 2013; Yalcin 2006). Data from Cymbala 2005 were not usable (see note above).
Two adult studies reported FEV1 before and after bronchodilation (Diego 2013; Wong 2012), and one also reported FVC before and after bronchodilation (Wong 2012). The remaining studies did not specify whether lung function was measured before or after bronchodilation.
Three adult studies reported systemic markers such as C‐reactive protein (Altenburg 2013; Masekela 2013; Serisier 2013).
Five adult studies reported adverse events (Altenburg 2013; Juthong 2011; Lourdesamy 2014; Serisier 2013; Wong 2012), as did two studies with children (Liu 2014; Valery 2013).
All 15 studies directly reported or inferred all‐cause mortality due to completion of the study period by all participants (Altenburg 2013; Asintam 2012; Cymbala 2005; Diego 2013; Juthong 2011; Koh 1997; Liu 2012; Liu 2014; Lourdesamy 2014; Masekela 2013; Sadigov 2013; Serisier 2013; Valery 2013; Wong 2012; Yalcin 2006).
Four adult studies reported the emergence of resistance to antibiotics (Altenburg 2013; Juthong 2011; Serisier 2013; Wong 2012), as did one study that included children (Valery 2013).
Two adult studies reported exercise capacity as measured by the 6MWD test (Serisier 2013; Wong 2012).
Nine adult studies reported health‐related quality of life using SGRQ (Altenburg 2013; Asintam 2012; Diego 2013; Juthong 2011; Liu 2012; Liu 2014; Lourdesamy 2014; Serisier 2013; Wong 2012).
Note: Eight studies reported a formal sample size calculation (Altenburg 2013; Asintam 2012; Cymbala 2005; Diego 2013; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012), but two of these studies did not recruit the target number of participants (Asintam 2012; Cymbala 2005). Six studies provided details of online trial registration (Altenburg 2013; Diego 2013; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012). Eight studies included conflict of interest statements (Altenburg 2013; Cymbala 2005; Juthong 2011; Liu 2014; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012). Nine studies explicitly stated funding sources for the study (Altenburg 2013; Cymbala 2005; Diego 2013; Juthong 2011; Liu 2014; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012), but only six studies reported the role of funding sources in the trial (Altenburg 2013; Cymbala 2005; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012).
Subgroup analysis
One study with children conducted several post hoc subgroup analyses based on intervention compliance, intervention duration, bronchiectasis diagnosis, frequency of exacerbations at baseline, and positive bacterial infection at the beginning of the trial (Valery 2013).
Excluded studies
We excluded eight studies from this review. Six of these were not RCTs (Kudo 1988; Min 1988; Ming 2005; Rikitomi 1988; Saito 1988; Unoura 1986), one study was of less than four weeks' duration and therefore did not meet our inclusion criteria (Tagaya 2002), and one study served as the protocol for a trial (Chang 2013). Please see Characteristics of excluded studies for additional details.
Risk of bias in included studies
Full details of the risk of bias judgements can be found in the 'Risk of bias' section at the end of each Characteristics of included studies table. Figure 3 and Figure 4 also provide a summary of the risk of bias in all included studies. Two independent review authors (LF and NR) independently assessed the risk of bias for each of the included studies and reached agreement.
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
Review authors considered the methods used to generate randomisation sequences as low risk in six studies (Altenburg 2013; Liu 2012; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012). Lourdesamy 2014 and Valery 2013 randomised participants using computer‐generated random numbers in a 1:1 ratio, and Valery 2013 also reported using a block design. Altenburg 2013 described an independently performed computer‐generated random allocation sequence that used a permuted block size of 10. Serisier 2013 also used a computer‐generated random allocation sequence but with block sizes of 2, 4, and 8, and stratified patients by baseline sputum Pseudomonas. Wong 2012 used a similar sequence generation with block size of 6 and stratified participants by centre. Liu 2012 randomised participants by using random number tables. The remaining nine studies provided unclear details regarding generation of the randomisation sequence (Asintam 2012; Cymbala 2005; Diego 2013; Juthong 2011; Koh 1997; Liu 2014; Masekela 2013; Sadigov 2013; Yalcin 2006).
We judged allocation concealment as having low risk of bias in four studies (Altenburg 2013; Serisier 2013; Valery 2013; Wong 2012), and we assigned unclear risk in 11 studies (Asintam 2012; Cymbala 2005; Diego 2013; Juthong 2011; Koh 1997; Liu 2012; Liu 2014; Lourdesamy 2014; Masekela 2013; Sadigov 2013; Yalcin 2006). Altenburg 2013 assigned identification codes with double‐blind allocation to treatment groups. Valery 2013 used sequentially numbered, double‐sealed, opaque envelopes to conceal group allocation. Serisier 2013 used an independent trial pharmacist to dispense blinded study drug according to the randomisation sequence. Wong 2012 randomly assigned participants to groups using a study‐independent statistician. Studies considered at unclear risk of allocation concealment bias did not provide adequate details of study methods to inform a clear judgement.
Blinding
We judged performance of the trial to be at low risk of bias in six studies (Altenburg 2013; Juthong 2011; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012). Juthong 2011 and Altenburg 2013 reported identical tablets in both groups. Juthong 2011Lourdesamy 2014, Serisier 2013Valery 2013, and Wong 2012 stated that study personnel (patients, supervisors, staff, researchers, investigators) were blinded to treatment allocation at all times. We judged three studies as having high risk of performance bias, as they were open‐label trials (Diego 2013; Liu 2012; Liu 2014). Investigators reported methods in the remaining studies in insufficient detail to permit a clear judgement of performance bias (Asintam 2012; Cymbala 2005; Koh 1997; Masekela 2013; Sadigov 2013; Yalcin 2006).
Six studies clearly stated blinding of outcome assessments (detection bias); we therefore judged these studies to be low risk of bias (Altenburg 2013; Liu 2012; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012). However, the remaining nine studies did not report methods in sufficient detail to inform a clear judgement of the risk of detection bias (Asintam 2012; Cymbala 2005; Diego 2013; Juthong 2011; Koh 1997; Liu 2014; Masekela 2013; Sadigov 2013; Yalcin 2006).
Incomplete outcome data
We judged incomplete outcome data (attrition bias) to introduce low risk of bias in nine studies (Altenburg 2013; Juthong 2011; Koh 1997; Liu 2014; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012; Yalcin 2006). Four studies reported no dropouts (Altenburg 2013; Juthong 2011; Serisier 2013; Yalcin 2006). Five studies clearly reported attrition rates and reasons for withdrawal (Koh 1997; Liu 2014; Lourdesamy 2014; Valery 2013; Wong 2012). We judged the remaining six studies to have unclear risk of attrition bias owing to insufficient reporting (Asintam 2012; Cymbala 2005; Diego 2013; Liu 2012; Masekela 2013; Sadigov 2013).
Selective reporting
We judged six of the included studies to have low risk of reporting bias (selective reporting) (Altenburg 2013; Diego 2013; Lourdesamy 2014; Serisier 2013; Valery 2013; Wong 2012), as the study protocols were available, and all outcomes of interest had been reported in the prespecified way. We judged the risk of reporting bias as unclear in nine studies (Asintam 2012; Cymbala 2005; Juthong 2011; Koh 1997; Liu 2012; Liu 2014; Masekela 2013; Sadigov 2013; Yalcin 2006), as a full trial protocol was not available.
Other potential sources of bias
We did not identify any other potential sources of bias in five studies (Altenburg 2013; Diego 2013; Serisier 2013; Valery 2013; Wong 2012), but we could not adequately assess this in seven other included studies (Asintam 2012; Juthong 2011; Koh 1997; Liu 2012; Liu 2014; Masekela 2013; Sadigov 2013). We judged three studies to have high risk of other potential sources of bias (Cymbala 2005Lourdesamy 2014; Yalcin 2006). Group allocation was ineffective in the pre‐cross‐over phase of Cymbala 2005, with eight of 11 participants receiving the intervention. In Lourdesamy 2014, baseline sputum volume (primary outcome) was significantly higher in the intervention arm compared with the placebo group. Similarly, in Yalcin 2006, baseline cytokine assay levels were again significantly higher in the intervention group compared with the placebo group.
Effects of interventions
See: Summary of findings for the main comparison Macrolides compared with placebo for adults with bronchiectasis; Summary of findings 2 Macrolides compared with no intervention for adults with bronchiectasis; Summary of findings 3 Macrolides compared with placebo for children with bronchiectasis; Summary of findings 4 Macrolides compared with no intervention for children with bronchiectasis
Macrolide versus placebo: adults
Primary outcomes
Exacerbations
Two adult studies of azithromycin ‐ Altenburg 2013 and Wong 2012 ‐ and one adult study of erythromycin ‐ Serisier 2013 ‐ with a total of 341 participants were included in a meta‐analysis. Results show that macrolides reduced the frequency of exacerbations to a greater extent than placebo (OR 0.34, 95% CI 0.22 to 0.54; I2 = 65%; Analysis 1.1; moderate‐quality evidence). This translates to 714 per 1000 in the placebo group experiencing one or more exacerbation compared with 459 per 1000 in the macrolide group (95% CI 355 to 574) or a number needed to treat for an additional beneficial outcome (NNTB) of 4 (95% CI 3 to 8) (Figure 5).
Analysis 1.1. Cates plot showing the absolute reduction in numbers of participants experiencing one or more exacerbations in adults treated with macrolides compared with placebo (OR 0.34, 95% CI 0.22 to 0.54). 714 people per 1000 in the placebo group experienced one or more exacerbations compared with 459 (95% CI 355 to 574) per 1000 in the macrolide group.
As heterogeneity was substantial, we tested the impact of a random‐effects model on the pooled effect size, which remained unchanged (OR 0.34, 95% CI 0.15 to 0.75). However, we noted significant differences between azithromycin and erythromycin subgroups (test for subgroup differences: Chi2 = 5.63, df = 1 (P = 0.02), I2 = 82.2%) and beneficial effects related to the two azithromycin studies (OR 0.23, 95% CI 0.13 to 0.40; I2 = 0%) (Altenburg 2013; Wong 2012). Data show no differences between groups in the erythromycin study (OR 0.74, 95% CI 0.34 to 1.63). Two further studies did not report exacerbations in sufficient detail for inclusion in meta‐analyses. In one study of azithromycin (1500 mg/week for six months) involving 65 adults, trial authors reported that the intervention "significantly decreased the rate of event‐based exacerbations and significantly increased the time to the first event‐based exacerbation compared to placebo", but no further details were available (Sadigov 2013). In another study of roxithromycin involving 30 adults, two participants in the intervention group and one participant in the control group developed an exacerbation but researchers reported no further details (Asintam 2012).
Three adult studies reported significantly reduced incidence rate ratios in the intervention group as follows: 0.48 fewer exacerbations per year (95% CI 0.65 to 0.26) (Altenburg 2013); 0.57 fewer exacerbations per year (95% CI 0.77 to 0.42) (Serisier 2013); and 0.38 fewer exacerbations per year (95% CI 0.54 to 0.25) (Wong 2012).
One adult study reported time to first exacerbation following a post hoc analysis, with a hazard ratio of 0.29 (95% CI 0.16 to 0.51) favouring azithromycin (Altenburg 2013).
Hospitalisations
We included in a meta‐analysis two studies of azithromycin involving 151 adults (Altenburg 2013; Lourdesamy 2014); results show no evidence of a reduction in hospitalisations in the azithromycin group compared with the placebo group (OR 0.56, 95% CI 0.19 to 1.62; I2 = 0%; Analysis 1.2; low‐quality evidence), although these results should be interpreted with caution owing to the low event rate.
Serious adverse events
Meta‐analysis included two studies of azithromycin involving 209 adults (Lourdesamy 2014; Wong 2012), along with one study of erythromycin with 117 adults (Serisier 2013). Serious adverse events included pneumonia, respiratory and non‐respiratory infections, haemoptysis, gastroenteritis, hernia, congestive heart failure, stroke, and skin carcinoma. Results show no difference in the numbers of participants with serious adverse events between study groups (OR 0.49, 95% CI 0.20 to 1.23; I2 = 0%; Analysis 1.3; low‐quality evidence) and no evidence of subgroup differences between azithromycin and erythromycin (test for subgroup differences: Chi2 = 0.48, df = 2 (P = 0.79), I2 = 0%), although results should be interpreted with caution owing to low event rates. Removing the study with unclear risk of bias for allocation concealment ‐ Lourdesamy 2014 ‐ from the meta‐analysis had little impact on the pooled treatment effect (OR 0.39, 95% CI 0.12 to 1.23; I2 = 0%).
Secondary outcomes
Sputum volume and purulence
One study of azithromycin (1000 mg/week for 12 weeks) with 78 adults reported no difference in sputum volume between study groups (MD 3.70, 95% CI ‐5.78 to 13.18; Analysis 1.4) (Lourdesamy 2014). One study of erythromycin (3500 mg/week for 48 weeks) with 117 adults reported a significant reduction in the change from baseline in 24‐hour sputum weight, favouring the intervention (median change ‐4.4 grams, interquartile ratio (IQR) ‐7.8 to ‐1; P = 0.01) (Serisier 2013). One study of roxithromycin (2100 mg/week for 12 weeks) with 30 adults reported no improvement in sputum volume in either study group but provided no further details (Asintam 2012).
Measures of lung function
Forced expiratory volume in one second (FEV1)
Seven adult studies reported FEV1 as litres, percent of predicted, or both (Altenburg 2013; Asintam 2012; Juthong 2011; Lourdesamy 2014; Sadigov 2013; Serisier 2013; Wong 2012). One trial of azithromycin (1000 mg/week for 12 weeks) with 78 participants showed no evidence of benefit in FEV1 % predicted from the intervention at the end of the study (MD 2.98, 95% CI ‐6.15 to 12.11; Analysis 1.5) (Lourdesamy 2014). One trial of azithromycin (1750 mg/week for 52 weeks) with 83 participants reported an increase of 1.03% in FEV1 % predicted in the intervention group every three months compared with a decrease of 0.10% in the placebo group (P = 0.047) (Altenburg 2013). One trial of erythromycin (3500 mg/week for 48 weeks) with 117 participants reported a significant difference in FEV1 %predicted change from baseline between groups, favouring macrolides (MD 2.40, 95% CI 0.34 to 4.46; Analysis 1.6) (Serisier 2013). One study of azithromycin (1500 mg/week for 6 months) with 65 participants reported significant improvements in prebronchodilator and postbronchodilator FEV1 but provided no further details (Sadigov 2013). One study of roxithromycin (2100 mg/week for 12 weeks) with 30 participants reported no improvement in either study group but provided no further details (Asintam 2012).
A meta‐analysis of data from two studies showed no benefit from azithromycin or roxithromycin in FEV1 at the end of the study (MD 0.02 L, 95% CI ‐0.17 to 0.22; Analysis 1.7) (Juthong 2011; Lourdesamy 2014). Results show were no significant differences between the two macrolides (test for subgroup differences: Chi2 = 0.43, df = 1 (P = 0.51), I2 = 0%). Another study of azithromycin (1500 mg/week for 6 months) with 141 participants also showed no benefit from the intervention in change in FEV1 during the study (MD 0.04 L, 95% CI ‐0.03 to 0.11; Analysis 1.8) (Wong 2012).
Forced vital capacity (FVC)
Four adult studies reported FVC as percent of predicted, in litres, or both ways (Altenburg 2013; Juthong 2011; Lourdesamy 2014; Wong 2012). One trial of azithromycin (1000 mg/week for 12 weeks) with 78 participants showed no benefit from the intervention at the end of the study in terms of FVC % predicted (MD 1.07, 95% CI ‐9.27 to 11.41; Analysis 1.9) (Lourdesamy 2014). Another study of azithromycin (1750 mg/week for 52 weeks) with 83 participants reported an increase in FVC of 1.33% predicted in the intervention group and a decrease of 0.30% predicted in the placebo group every three months (Altenburg 2013).
A meta‐analysis of data from two studies with 94 participants showed no benefit at the end of the study from azithromycin or roxithromycin in terms of FVC (MD 0.08 L, 95% CI ‐0.19 to 0.36; Analysis 1.10) (Juthong 2011; Lourdesamy 2014). Results show no significant differences between the two macrolides (test for subgroup differences: Chi2 = 1.57, df = 1 (P = 0.21), I2 = 36.3%). One study of azithromycin (1500 mg/week for six months) with 141 participants showed no benefit from the intervention in changes in FVC (MD 0.08 L, 95% CI ‐0.53 to 0.69; Analysis 1.11) (Wong 2012).
FEV1/FVC ratio
One study of azithromycin (1000 mg/week for 12 weeks) with 78 participants reported the FEV1/FVC ratio showing no evidence of benefit from the intervention at the end of the study (MD 3.57, 95% CI ‐3.89 to 11.03; Analysis 1.12) (Lourdesamy 2014).
Systemic markers of infection
One trial of azithromycin (1750 mg/week for 52 weeks) with 83 participants reported no significant differences between median CRP values at the end of the study (azithromycin 2.6 mg/dL, IQR 1.5 ‐ 7; control 3.9 mg/dL, IQR 2 ‐ 6.15) and no changes in serum levels, although P values were not reported (Altenburg 2013). Similarly, one trial of erythromycin (3500 mg/week for 48 weeks) with 117 participants reported no differences between groups in CRP levels (median change difference ‐0.2 mg/L, IQR ‐1.5 to 1.2), although again significance values were not reported (Serisier 2013).
Adverse events
Five studies of three different macrolides (azithromycin, erythromycin, and roxithromycin) with 435 adult participants were included in a meta‐analysis (Altenburg 2013; Juthong 2011; Lourdesamy 2014; Serisier 2013; Wong 2012), showing no differences between study groups in the numbers of people experiencing adverse events (OR 0.83, 95% CI 0.51 to 1.35; I2 = 28%; Analysis 1.13). Trials provided no evidence of differences between the three different macrolides (test for subgroup differences: Chi2 = 2.07, df = 2 (P = 0.36), I2 = 3.3%). Removing two studies from the analysis with unclear risk of bias for sequence generation or allocation concealment had little impact on the pooled treatment effect (OR 0.83, 95% CI 0.50 to 1.39; I2 = 27%) (Juthong 2011; Lourdesamy 2014).
All‐cause mortality
Data show no deaths during the intervention period in six of the adult studies (Altenburg 2013; Asintam 2012; Juthong 2011; Sadigov 2013; Serisier 2013; Wong 2012). One study of azithromycin (1000 mg/week for 12 weeks) with 78 participants reported no deaths in the placebo group and two deaths in the intervention group attributed to bronchopneumonia and not considered treatment‐related (Lourdesamy 2014). In performing our GRADE assessment, we judged this outcome to be of low quality (summary of findings Table for the main comparison).
Emergence of resistance to antibiotics
One study of azithromycin with 83 adults reported no differences between groups in the emergence of resistance to antibiotics (OR 0.71, 95% CI 0.30 to 1.69; Analysis 1.14) (Altenburg 2013). In another study of roxithromycin (2100 mg/week for 8 weeks) with 26 adults, none of the participants experienced antibiotic resistance to any bacterial strain (Juthong 2011). However, a study of erythromycin (3500 mg/week for 48 weeks) with 117 participants reported a higher proportion of macrolide‐resistant oropharyngeal streptococci in the intervention group compared with the placebo group (median change difference 25.5%, IQR 15% to 33.7%; P = 0.001) (Serisier 2013).
Exercise capacity
One study of erythromycin (3500 mg/week for 48 weeks) with 117 adults reported no differences in the change in 6MWD between study groups (MD ‐6.30, 95% CI ‐28.86 to 16.26; Analysis 1.15) (Serisier 2013). Similarly, a study of azithromycin (1500 mg/week for 6 months) with 141 adults reported no significance difference in change in 6MWD between study groups (mean change difference 6.48 m, 95% CI ‐11.28 to 24.22; P = 0.4) (Wong 2012).
Health‐related quality of life
One study with 68 adults showed a significantly lower (better) SGRQ total score at the end of the study in the intervention group compared with the placebo group (MD ‐8.90, 95% CI ‐13.13 to ‐4.67; Analysis 1.16; moderate‐quality evidence) and an MD that exceeded the 4‐unit threshold of clinical significance (Lourdesamy 2014). We included in a meta‐analysis four adult studies with 305 participants showing greater improvements from baseline to study endpoint in quality of life with macrolides (MD ‐2.86, 95% CI ‐5.67 to ‐0.04; Analysis 1.17; low‐quality evidence) (Asintam 2012; Juthong 2011; Serisier 2013; Wong 2012). Although data show no significant differences between azithromycin and roxithromycin (test for subgroup differences: Chi2 = 0.01, df = 1 (P = 0.92), I2 = 0%), the beneficial effect was largely observed in the azithromycin group. Differences between groups in all four studies were below the threshold of clinical significance. Removing from the meta‐analysis the two studies with unclear risk of bias for sequence generation and allocation concealment had no impact on the pooled treatment effect (MD ‐2.97, 95% CI ‐5.94 to ‐0.00; I2 = 0%) (Asintam 2012; Juthong 2011).
One study of azithromycin (1750 mg/week for 52 weeks) with 83 participants reported a significant improvement in quality of life (SGRQ total) with the intervention compared with placebo (intervention group mean change ‐6.09, control group mean change ‐2.06; P = 0.05) (Altenburg 2013).
Macrolide versus no intervention: adults
Primary outcomes
Exacerbations
One study of roxithromycin with 43 adults (1050 mg/week 6 months) did not find a clear difference in the frequency of exacerbations between groups (OR 0.31, 95% CI 0.08 to 1.15; Analysis 2.1; moderate‐quality evidence; summary of findings Table 2) (Liu 2014).
None of the included studies reported our other primary outcomes: hospitalisation and adverse or serious adverse events.
Secondary outcomes
Sputum volume and purulence
One study of azithromycin with 30 adults (750 mg/week 3 months) reported a decrease in sputum volume with macrolides (MD ‐11.00 mL/d; P < 0.05) (Diego 2013). Following the intervention, sputum volume decreased by 8.9 mL/d in the azithromycin group and increased by 2.1 mL/d in the control group by three months. Data show no differences in changes in sputum purulence scores between groups (mean change score: azithromycin 0.8, control 0.7) by three months.
Measures of lung function
One study of azithromycin with 30 adults (750 mg/week 3 months) found no differences between groups in FEV1 or FVC (Diego 2013). Relative to baseline, FEV1 increased by 0.06 L and 0.04 L in azithromycin and control groups, respectively, and FVC decreased by 0.07 L and 0.08 L in azithromycin and control groups, respectively, by three months.
Systemic markers of infection
One study of azithromycin with 30 adults (750 mg/week 3 months) found no differences between groups in CRP levels (Diego 2013). This study reported a reduction in CRP levels by three months compared with baseline in both groups: mean reduction: ‐17 in the azithromycin group, ‐11 in the control group.
Adverse events
One adult study reported adverse events as event rates but did not report the number of participants experiencing at least one adverse event (Liu 2014).
Mortality
Two adult studies reported no deaths during the study (Diego 2013; Liu 2014). Through GRADE assessment, we judged this outcome to be of very low quality (summary of findings Table 2).
Quality of life
Two roxithromycin studies with 89 adults (1050 mg/week 6 months) reported significantly better quality of life with macrolides compared with no intervention at the end of six months (MD ‐8.81, 95% CI ‐14.33 to ‐3.28; Analysis 2.2; moderate‐quality evidence) (Liu 2012; Liu 2014). One study of azithromycin with 30 adults (750 mg/week for 3 months) reported a significant improvement in quality of life, measured by the SGRQ total score, after three months compared with no intervention (MD ‐12.00; P < 0.05; low‐quality evidence) (Diego 2013). The total score decreased by 7.9 units in the azithromycin group and increased by 4.1 units in the control group. It was not considered appropriate to combine these outcomes in a meta‐analysis owing to differences in macrolides, doses, and study duration.
The included studies did not report our other secondary outcomes ‐ exercise capacity and resistance to antibiotics.
Macrolide versus placebo: children
Primary outcomes
Exacerbations
Two studies reported exacerbation frequency in children. One study of azithromycin (30 kg/week for up to 24 months) reported that in children, exacerbation frequency was reduced more with macrolides than with placebo (IRR 0.50, 95% CI 0.35 to 0.71; 89 children; one study; low‐quality evidence). However, we feel this should be interpreted with a degree of caution as the same study reported no significant benefit from the intervention in the number of children with at least one exacerbation (OR 0.40, 95% CI 0.11 to 1.41; low‐quality evidence) nor in the time to first exacerbation (hazard ratio 0·63, 95% CI 0·40–1·00; log‐rank P = 0.12) (Valery 2013). Another study of erythromycin (875 < 15 kg > 1750 mg/week for 52 weeks) in 42 children reported that three children in the intervention group remained exacerbation‐free during the study, and all children in the placebo group had at least one exacerbation (Masekela 2013).
Hospitalisations
One study of azithromycin with 89 children showed no evidence of a reduction in numbers of children hospitalised for exacerbations between study groups (OR 0.28, 95% CI 0.07 to 1.11; Analysis 3.1; low‐quality evidence), although again these results should be interpreted with caution owing to the low event rate (Valery 2013).
Serious adverse events
One study of azithromycin with 89 children reported serious adverse events (Valery 2013), showing no differences between groups in the number of children experiencing at least one event (OR 0.43, 95% CI 0.17 to 1.05; Analysis 3.2; low‐quality evidence). The majority of the serious adverse events related either to an exacerbation or an investigation related to bronchiectasis (e.g., bronchoscopy).
Secondary outcomes
Sputum volume and purulence
One study of roxithromycin with 25 children reported a reduction in sputum purulence score with the intervention (MD ‐0.78, 95% CI ‐1.32 to ‐0.24; Analysis 3.3) (Koh 1997).
Measures of lung function
Forced expiratory volume in one second (FEV1)
A meta‐analysis of two studies with 65 participants showed no evidence of benefit from azithromycin or roxithromycin in FEV1 expressed as percent of predicted by the end of the study (MD 1.73, 95% CI ‐3.32 to 6.78; Analysis 3.4) (Koh 1997; Valery 2013). Removing the study, which had unclear risk of bias for sequence generation and allocation concealment from the meta‐analysis, had little impact on the treatment effect (MD 3.70, 95% CI ‐5.99 to 13.39) (Koh 1997). Another study of erythromycin (875 < 15 kg > 1750 mg/week for 52 weeks) with 42 children who had HIV and were all receiving highly active antiretroviral therapy (HAART) reported no difference in FEV1 % predicted between groups at the end of the study (MD 5.50, 95% CI ‐7.26 to 18.26) (Masekela 2013).
Forced vital capacity (FVC)
Masekela 2013 reported no significant difference in FVC % predicted between groups at the end of the study (MD 5.00, 95% CI ‐5.61 to 15.61).
Systemic markers of infection
One erythromycin study (875 mg < 15 kg > 1750 mg/week for 52 weeks) of 42 children with HIV who were receiving HAART reported no differences in CRP levels between groups (MD 1.60, 95% CI ‐38.38 to 41.58) (Masekela 2013).
Adverse events
One study of azithromycin (30 mg/kg/week for 24 months) in 89 participants reported no differences between study groups in the numbers of children experiencing adverse events (OR 0.78, 95% CI 0.33 to 1.83; Analysis 3.5) (Valery 2013).
All‐cause mortality
Masekela 2013 reported the death of one randomised participant but did not state the study group to which that participant had been assigned (low‐quality evidence).
Emergence of resistance to antibiotics
One study of azithromycin in 89 children reported an increase in macrolide‐resistant bacteria (OR 7.13, 95% CI 2.13 to 23.79; Analysis 3.6), an increase in resistance to Streptococcus pneumoniae (OR 13.20, 95% CI 1.61 to 108.19; Analysis 3.7), and an increase in resistance to Staphylococcus aureus (OR 4.16, 95% CI 1.06 to 16.32; Analysis 3.8) with macrolides compared with placebo (Valery 2013).
The included studies did not report our other secondary outcomes: exercise capacity and health‐related quality of life.
Macrolide versus no intervention: children
Primary outcomes
The included study did not report our primary outcomes: exacerbations, hospitalisations, and adverse and serious adverse events.
Secondary outcomes
Sputum volume and purulence
One study of clarithromycin with 34 children (105 mg/week 3 months) reported a significantly greater reduction in sputum volume with macrolides compared with placebo (P = 0.0001) but did not report exact values for each group (Yalcin 2006).
Measures of lung function
Yalcin 2006 also reported no differences in FEV1 between groups but did not report exact values and significance levels.
Mortality
Yalcin 2006 reported no deaths (low‐quality evidence).
The included study did not report our other secondary outcomes: systemic markers of infection, adverse events, resistance to antibiotics, exercise capacity, and quality of life.
Discussion
Summary of main results
Fifteen randomised trials met the inclusion criteria for this systematic review; 11 studies included adult participants only (Altenburg 2013; Asintam 2012; Cymbala 2005; Diego 2013; Juthong 2011; Liu 2012; Liu 2014; Lourdesamy 2014; Sadigov 2013; Serisier 2013; Wong 2012), and in four studies (Koh 1997; Masekela 2013; Valery 2013; Yalcin 2006), the participants were children. Considerable clinical heterogeneity was evident on a range of other factors, including four different types of macrolides, doses ranging from 750 mg/week to 3500 mg/week with regimens varying from twice daily to once a week, intervention duration ranging from eight weeks to 24 months, and background therapies administered to all groups in two studies. None of the included studies compared one type of macrolide versus another or versus a non‐macrolide antibiotic.
Evidence shows a reduction in exacerbations seen in our aggregation of data from four adult studies, including Altenburg 2013, Sadigov 2013Serisier 2013 and Wong 2012, and from one study of children ‐ Valery 2013 ‐ comparing macrolides with placebo, and we used GRADE criteria to assess the quality of this evidence as moderate. Most of these studies (with the exception of Serisier 2013) used azithromycin. Masekela 2013 reported no reduction in the number of exacerbations over 52 weeks with erythromycin compared with placebo. This study was carried out in South African children with HIV who were receiving antiretrovirals and showed varying degrees of HIV virological suppression. The specifics of this population make it difficult to generalise the findings of Masekela 2013 to individuals with bronchiectasis in less specialised circumstances. Studies comparing macrolides with no intervention were insufficient to establish clear effects. For hospitalisations, data show no evidence of benefit with azithromycin based on aggregation of data from two adult studies (Altenburg 2013; Lourdesamy 2014), along with one study of children (Valery 2013), and on evidence of low quality. We are unable to draw any conclusions on which macrolides may be most beneficial, as data were not available for all of our planned subgroups. Available low‐quality evidence from four adult studies, including Altenburg 2013Lourdesamy 2014Serisier 2013 and Wong 2012, and from one study of children ‐ Valery 2013 ‐ suggests that participants receiving macrolides experienced more adverse events. We are again unable to draw clear conclusions regarding the effectiveness of different macrolides, as four of the five studies used azithromycin. Studies comparing macrolides with no intervention did not report this outcome.
Overall our review provides promising, but inconclusive, results for our three predefined primary outcomes, but on the basis of currently available evidence, we are unable to present robust conclusions.
For our secondary outcomes, aggregated data from six studies comparing macrolides against placebo and yielding moderate‐ to low‐quality evidence (Altenburg 2013; Asintam 2012; Juthong 2011; Lourdesamy 2014; Serisier 2013; Wong 2012) indicate that macrolides have a positive impact on health‐related quality of life, as measured by St. George's Respiratory Questionnaire (SGRQ). Similarly, three studies comparing macrolides with no treatment showed improved quality of life with the intervention (Diego 2013; Liu 2012; Liu 2014). Data on sputum volume and purulence, measures of lung function, markers of infection, and demonstrated exercise capacity provided no indication of benefit from macrolides in adults. One of the largest adult studies of erythromycin provided limited evidence of increased resistance to macrolides (Serisier 2013).
None of the four children's studies measured quality of life. Macrolides were associated with improved sputum characteristics in two children's studies (Koh 1997; Yalcin 2006). Studies with children provided no evidence of benefit from macrolides in terms of measures of lung function, markers of infection, or demonstrated exercise capacity. One study of azithromycin with children provided limited evidence of increased resistance to macrolides (Valery 2013).
Evidence of moderate to very low quality from the 15 included studies provided no indication of a higher mortality rate with macrolides.
In relation to our predefined secondary outcomes, health‐related quality of life data further strengthen the impression noted in our primary outcomes that this intervention merits further exploration in high‐quality clinical trials.
Overall completeness and applicability of evidence
We have identified studies of macrolides in bronchiectasis reporting exacerbation and hospitalisation rates. Data for planned secondary outcomes, particularly adverse effects, are lacking. Our findings are based on studies totalling 690 adults and 190 children. Investigators used several different macrolide antibiotics (azithromycin, erythromycin, roxithromycin, and clarithromycin) in these populations in a variety of international settings. This breadth enhances the generalisability of findings but may conceal an advantage of, for example, one macrolide over another or use in adults over use in children, as none of the included studies reported direct comparisons between different macrolides or between adults and children. Small and short‐term studies mean that we may not detect small but clinically important increases in absolute risk for serious adverse events. Such adverse events, including mortality as reported in Ray 2012 and hearing loss as described in Albert 2011, have been reported in larger studies of macrolides for indications other than bronchiectasis. Although this review provides limited evidence of benefit associated with macrolide antibiotics, their relative benefit compared with benefit derived from other types of antibiotics remains unknown, as we did not identify any studies that included these comparisons.
Apart from Altenburg 2013, Juthong 2011, Serisier 2013, Wong 2012, and Valery 2013, we found a lack of information on microbial resistance associated with the macrolides used in the reports of included studies. No studies were designed to evaluate changes in resistance patterns in the wider community.
Quality of the evidence
The overall quality of studies included in this review ranges from very low to moderate for outcomes included in the GRADE assessment. From adult studies comparing macrolides versus placebo, the evidence for frequency of exacerbations and mortality from all causes is of moderate quality owing to imprecision of the effect (limited to azithromycin). Evidence for hospitalisations, serious adverse events, and quality of life was of low quality owing to study design limitations (unclear study methods, open‐label approach) and imprecision of the effect (few studies and wide confidence intervals), which resulted in downgrading of the quality of outcomes. Evidence for all‐cause mortality is of poor quality owing to unclear reporting and losses to follow‐up. From adult studies that compared macrolides versus no intervention, the evidence for frequency of exacerbations and quality of life as assessed by the SGRQ is of moderate quality owing to limitations in study design (open‐label study) and imprecision of the effect (few studies and no confidence intervals). The quality of evidence for mortality from all causes is very low owing to serious design limitations (open‐label study, unclear study methods) and inadequate reporting of participants lost to follow‐up.
Studies that compared macrolides versus placebo in children have provided low‐quality evidence on frequency of exacerbations, hospitalisations, serious adverse events, and mortality owing to design limitations (unclear methods), imprecision (wide confidence intervals and low event rates), and no information on participants lost to follow‐up. The single small study that compared macrolides versus no intervention in children provided low‐quality evidence on mortality from all causes owing to insufficient information on trial methods and imprecision.
We judged only four of our 15 included studies ‐ three with adults and one with children ‐ as having low risk of bias across all domains. Selection bias was unclear in nearly half of the included studies owing to lack of detailed reporting on random sequence generation and allocation concealment. Most studies blinded participants to group allocation, but several studies described investigator blinding in a way that was unclear, and most trials reported blinding of outcome assessment that was also unclear. None of the included studies had high risk of attrition or reporting bias, although some studies provided no information on participants who were lost to follow‐up. Furthermore, only six studies explicitly reported the role of funders in the trial; six studies registered their protocol on trial registries; and eight studies reported that investigators performed a formal sample size calculation before the start of the trial.
Potential biases in the review process
We used a comprehensive systematic search, conducted by a highly experienced information specialist, to identify potentially eligible studies. We searched multiple resources, including electronic databases, journals, conference proceedings, reference lists of included studies, citations of included studies, and trial registries. Nevertheless, we recognise the possibility of publication bias in this review that could either overestimate or underestimate effects of the intervention in terms of the different outcomes included in the review. Trials showing no, or negative, effects are less likely to be offered for publication, and if offered are less likely to be accepted, resulting in a biased set of data available for review. As we included only a few studies for each outcome, we were unable to assess the presence of publication bias through formal testing.
Furthermore, some papers may have been misclassified as not eligible for inclusion in this review. Two review authors independently assessed all studies, and a third review author verified the data; we are confident that we assessed studies excluded from analyses on the basis of consistent and appropriate criteria. For some full‐text reports, it is possible that we could have entered some data into analyses incorrectly, although we double‐checked all data to attempt to avoid extraction errors.
We contacted the investigators of three included studies based on conference proceedings that were available as an abstract, to obtain study characteristics and other numerical outcome data. Although we received responses from all of these investigators, we found that data from only two studies were provided. The same investigator was involved in both of these studies, which were conducted in similar settings and used similar interventions. Owing to the small number of included studies, we did not explore the impact of excluding studies with missing outcome data in the overall assessment of results by performing a sensitivity analysis. Finally, data were insufficient to permit all planned subgroup analyses, so we included only types of antibiotics, which we considered most clinically important.
Agreements and disagreements with other studies or reviews
Major findings of the present review are largely in agreement with the results of previously published meta‐analyses of the impact of macrolides on outcomes in bronchiectasis (Gao 2014), which have shown a reduced frequency of exacerbations, improved lung function, and improved quality of life with prolonged macrolide treatment. Differences in effect size between our present review and these previously published meta‐analyses reflect differences in the inclusion and exclusion criteria of studies, as well as the inclusion of some recent studies. In general, the strict methods applied in the present review have resulted in pooling of fewer studies and therefore more precise effect estimates.
The three largest macrolide trials have the longest duration of follow‐up and are concordant with overall results of the meta‐analysis, with each demonstrating benefit in terms of frequency of and/or time to first exacerbation (Altenburg 2013; Serisier 2013; Wong 2012). It is striking that all three studies are of high quality (Figure 4).
The benefit in terms of exacerbations was largely attributable to azithromycin. This should not be taken to indicate that azithromycin is superior to other macrolides because the characteristics of participants in the BLESS study of erythromycin were different from those of participants enrolled in azithromycin studies. Our analysis is not designed to determine whether the drug or the patient cohort is responsible for the apparent differential effect.
Global distribution of studies.
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.
Analysis 1.1. Cates plot showing the absolute reduction in numbers of participants experiencing one or more exacerbations in adults treated with macrolides compared with placebo (OR 0.34, 95% CI 0.22 to 0.54). 714 people per 1000 in the placebo group experienced one or more exacerbations compared with 459 (95% CI 355 to 574) per 1000 in the macrolide group.
Comparison 1 Macrolide versus placebo: adults, Outcome 1 ≥ 1 exacerbation.
Comparison 1 Macrolide versus placebo: adults, Outcome 2 Hospitalisation: all‐cause.
Comparison 1 Macrolide versus placebo: adults, Outcome 3 Serious adverse events.
Comparison 1 Macrolide versus placebo: adults, Outcome 4 Sputum weight (g): endpoint.
Comparison 1 Macrolide versus placebo: adults, Outcome 5 FEV1 (% predicted): endpoint.
Comparison 1 Macrolide versus placebo: adults, Outcome 6 FEV1 (% predicted): change (post bronchodilator).
Comparison 1 Macrolide versus placebo: adults, Outcome 7 FEV1 (L): endpoint.
Comparison 1 Macrolide versus placebo: adults, Outcome 8 FEV1 (L): change.
Comparison 1 Macrolide versus placebo: adults, Outcome 9 FVC (% predicted): endpoint.
Comparison 1 Macrolide versus placebo: adults, Outcome 10 FVC (L): endpoint.
Comparison 1 Macrolide versus placebo: adults, Outcome 11 FVC (L): change.
Comparison 1 Macrolide versus placebo: adults, Outcome 12 FEV1/FVC: endpoint.
Comparison 1 Macrolide versus placebo: adults, Outcome 13 Adverse events.
Comparison 1 Macrolide versus placebo: adults, Outcome 14 Azithromycin‐resistant bacteria (any).
Comparison 1 Macrolide versus placebo: adults, Outcome 15 6‐Minute walk test: change.
Comparison 1 Macrolide versus placebo: adults, Outcome 16 Quality of life: endpoint.
Comparison 1 Macrolide versus placebo: adults, Outcome 17 Quality of life: change.
Comparison 2 Macrolide versus no intervention: adults, Outcome 1 ≥ 1 exacerbation.
Comparison 2 Macrolide versus no intervention: adults, Outcome 2 QoL SGRQ: endpoint total score.
Comparison 3 Macrolide versus placebo: children, Outcome 1 Hospitalisation: all‐cause.
Comparison 3 Macrolide versus placebo: children, Outcome 2 Serious adverse events.
Comparison 3 Macrolide versus placebo: children, Outcome 3 Sputum purulence score: endpoint.
Comparison 3 Macrolide versus placebo: children, Outcome 4 FEV1 (% predicted): endpoint.
Comparison 3 Macrolide versus placebo: children, Outcome 5 Adverse events.
Comparison 3 Macrolide versus placebo: children, Outcome 6 Azithromycin‐resistant bacteria (any).
Comparison 3 Macrolide versus placebo: children, Outcome 7 Azithromycin‐resistant Streptococcus pneumoniae.
Comparison 3 Macrolide versus placebo: children, Outcome 8 Azithromycin‐resistant Staphylococcus aureus.
| Macrolides compared with placebo for adults with bronchiectasis | ||||||
| Patient or population: adults with bronchiectasis | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Risk with placebo | Risk with macrolides | |||||
| ≥ 1 exacerbation | 714 per 1000 | 459 per 1000 | OR 0.34 | 341 | ⊕⊕⊕⊝ | 2 studies azithromycin (1750 mg/week for 52 weeks; 1500 mg/week for 6 months) 1 study erythromycin (3500 mg/week for 48 weeks) |
| Hospitalisation: all cause | 133 per 1000 | 79 per 1000 | OR 0.56 | 151 | ⊕⊕⊝⊝ | 2 studies azithromycin (1000 mg/week for 12 weeks; 1750 mg/week for 52 weeks) |
| Serious adverse events | 86 per 1000 | 44 per 1000 | OR 0.49 | 326 | ⊕⊕⊝⊝ | 2 studies azithromycin (1500 mg/week for 6 months; 1000 mg/week for 12 weeks) 1 study erythromycin (3500 mg/week for 48 weeks) |
| All‐cause mortality | 0 per 1000 | 0 per 1000 | not estimable | 540 | ⊕⊕⊝⊝ | 4 studies azithromycin (1000 to 1750 mg/week for 12 to 52 weeks) |
| Quality of life: endpoint | Mean SGRQ score at endpoint in placebo groups was 39.1 points. | MD 8.90 lower (13.13 lower to 4.67 lower) | ‐ | 68 | ⊕⊕⊕⊝ | 1 study azithromycin (1000 mg/week for 12 weeks) |
| Quality of life: change | Mean change in SGRQ score ranged from ‐1.3 to ‐8.9 points. | MD 2.86 lower | ‐ | 305 | ⊕⊕⊝⊝ | 1 study azithromycin (1500 mg/week for 6 months) |
| *The risk in the intervention group (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. | ||||||
| aEffect observed only with azithromycin (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). bUnclear allocation concealment and baseline imbalances on Lourdesamy (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). cTwo small studies and wide confidence interval (one point deducted for imprecision). dWide confidence interval (one point deducted for imprecision). eIn three of the seven studies, study methods were not clearly reported (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). fA total of 28 participants across four studies were lost to follow‐up with no further details available and unclear details of withdrawals in one study (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). gRandomisation, blinding, and other study methods unclear in two studies (Asintam; Juthong) (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). hWide confidence interval and mean difference does not exceed the threshold for clinical significance (one point deducted for imprecision). | ||||||
| Macrolides compared with no intervention for adults with bronchiectasis | ||||||
| Patient or population: adults with bronchiectasis | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Risk with no intervention | Risk with macrolides | |||||
| ≥ 1 exacerbation | Study population | OR 0.31 | 43 | ⊕⊕⊕⊝ | Roxithromycin (1050 mg/week for 6 months) | |
| 762 per 1000 | 498 per 1000 | |||||
| Hospitalisations ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Serious adverse events ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Mortality | No deaths in two trials, although in 1 study (azithromycin), 6 participants were lost to follow‐up | not estimable | 88 | ⊕⊝⊝⊝ | 1 study azithromycin (750 mg/week for 3 months) | |
| QoL SGRQ: endpoint total score | Mean SGRQ: endpoint total score of 51.7 points | MD 8.81 lower (14.33 lower to 3.28 lower) | ‐ | 89 | ⊕⊕⊕⊝ | 1 study roxithromycin (1050 mg/week for 6 months) |
| QoL SGRQ: change in total score | Mean SGRQ: change in total score of 4.1 | MD 12 lower | ‐ | 30 | ⊕⊕⊕⊝ | Azithromycin (750 mg/week for 3 months) |
| *The risk in the intervention group (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. | ||||||
| aOpen‐label study (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). bUnclear randomisation and study methods (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). c6 participants in one study lost to follow‐up and no further details reported (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). | ||||||
| Macrolides compared with placebo for children with bronchiectasis | ||||||
| Patient or population: children with bronchiectasis | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Risk with placebo | Risk with macrolides | |||||
| Exacerbation frequency | Number of exacerbations 195 (median: 4 range 0‐14) | Number of exacerbations 104 (median: 2 range 0‐9) | IRR 0.50 95% CI 0.35 to 0.71 | 89 | ⊕⊕⊝⊝ | Azithromycin (30 mg/kg/week for up to 24 months) |
| Hospitalisation: all‐cause | 205 per 1000 | 67 per 1000 | OR 0.28 | 89 | ⊕⊕⊝⊝ | Azithromycin (30 mg/kg/week for 24 months) |
| Serious adverse events | 432 per 1000 | 246 per 1000 | OR 0.43 | 89 | ⊕⊕⊝⊝ | Azithromycin (30 mg/kg/week for 24 months) |
| Mortality | 1 child died but study group was not stated. | ‐ | 42 | ⊕⊕⊝⊝ | Erythromycin (875 to 1750 mg/kg/week for 52 weeks) | |
| Quality of life not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| *The risk in the intervention group (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. | ||||||
| aWide confidence interval that includes 1 (no difference) (one point deducted for imprecision). bLow event rates and low numbers (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). cUnclear information on randomisation, blinding, and other study methods (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). dNo information on participants lost to follow‐up (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). | ||||||
| Macrolides compared with no intervention for children with bronchiectasis | ||||||
| Patient or population: children with bronchiectasis | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Risk with no intervention | Risk with macrolides | |||||
| Exacerbations ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Hospitalisation ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Serious adverse events ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| Mortality | 0 per 1000 | 0 per 1000 | not estimable | 34 | ⊕⊕⊝⊝ | Clarithromycin (105 mg/kg/week for 3 months) |
| Quality of life ‐ not reported | ‐ | ‐ | ‐ | ‐ | ‐ | |
| *The risk in the intervention group (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. | ||||||
| aInsufficient information on study methods and procedures (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). bNot blinded (one point deducted in relation to design and implementation of available studies suggesting likelihood of bias). | ||||||
| Study | Adults/ Children | No. of participants | Type of macrolide | Macrolide dose | Frequency | Delivery mode | Combined weekly dose | Comparison | Duration (months unless stated) |
| Altenburg 2013 | Adults | 83 | Azithromycin | 250 mg | Once daily | Oral | 1750 mg | Placebo | 12 |
| Asintam 2012 | Adults | 30 | Roxithromycin | 300 mg | Once daily | Oral | 2100 mg | Placebo | 12 weeks |
| Cymbala 2005 | Adults | 12 | Azithromycin | 500 mg | 3 days per week | Oral | 1000 mg | No intervention | 6 |
| Diego 2013 | Adults | 36 | Azithromycin | 250 mg | 3 days per week | Oral | 750 mg | No intervention | 3 |
| Juthong 2011 | Adults | 26 | Roxithromycin | 300 mg | Once daily | Oral | 2100 mg | Placebo | 8 weeks |
| Koh 1997 | Children | 25 | Roxithromycin | 4 mg/kg | Twice daily | Oral | 56 mg/kg | Placebo | 12 weeks |
| Liu 2012 | Adults | 50 | Roxithromycin, ambroxol hydrochloride | 150 mg | Once daily | Oral | 1050 mg | Ambroxol hydrochloride (no intervention) | 6 |
| Liu 2014 | Adults | 52 | Roxithromycin | 150 mg | Once daily | Oral | 1050 mg | No intervention | 6 |
| Lourdesamy 2014 | Adults | 78 | Azithromycin | 1000 mg | Weekly | 1000 mg | Placebo | 3 | |
| Masekela 2013 | Children | 42 | Erythromycin | 125 mg for children weighing < 15 kg and 250 mg ≥ 15 kg | Daily | Oral | 875 mg for children weighing < 15 kg and 1750 mg ≥ 15 kg | Placebo | 12 |
| Sadigov 2013 | Adults | 65 | Azithromycin | 500 mg | 3 days per week | Oral | 1500 mg | Placebo | 6 |
| Serisier 2013 | Adults | 117 | Erythromycin | 250 mg | Twice daily | Oral | 3500 mg | Placebo | 11 |
| Valery 2013 | Children | 89 | Azithromycin | 30 mg/kg up to a maximum of 600 mg | Once a week | Oral | 30 mg/kg up to a maximum of 600 mg | Placebo | 24 |
| Wong 2012 | Adults | 141 | Azithromycin | 500 mg | 3 days per week | Oral | 1500 mg | Placebo | 6 |
| Yalcin 2006 | Children | 34 | Clarithromycin, supportive therapies | 15 mg/kg | Daily | Oral | 105 mg/kg | Supportive therapies (no intervention) | 3 |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 ≥ 1 exacerbation Show forest plot | 3 | 341 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.34 [0.22, 0.54] |
| 1.1 Azithromycin | 2 | 224 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.23 [0.13, 0.40] |
| 1.2 Erythromycin | 1 | 117 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.74 [0.34, 1.63] |
| 2 Hospitalisation: all‐cause Show forest plot | 2 | 151 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.56 [0.19, 1.62] |
| 2.1 Azithromycin | 2 | 151 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.56 [0.19, 1.62] |
| 3 Serious adverse events Show forest plot | 3 | 326 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.49 [0.20, 1.23] |
| 3.1 Azithromycin | 2 | 209 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.51 [0.20, 1.34] |
| 3.2 Erythromycin | 1 | 117 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.32 [0.01, 8.07] |
| 4 Sputum weight (g): endpoint Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 4.1 Azithromycin | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 5 FEV1 (% predicted): endpoint Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 5.1 Azithromycin | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 6 FEV1 (% predicted): change (post bronchodilator) Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 6.1 Erythromycin | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 7 FEV1 (L): endpoint Show forest plot | 2 | 94 | Mean Difference (IV, Fixed, 95% CI) | 0.02 [‐0.17, 0.22] |
| 7.1 Azithromycin | 1 | 68 | Mean Difference (IV, Fixed, 95% CI) | ‐0.01 [‐0.23, 0.21] |
| 7.2 Roxithromycin | 1 | 26 | Mean Difference (IV, Fixed, 95% CI) | 0.15 [‐0.27, 0.57] |
| 8 FEV1 (L): change Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 8.1 Azithromycin | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 9 FVC (% predicted): endpoint Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 9.1 Azithromycin | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 10 FVC (L): endpoint Show forest plot | 2 | 94 | Mean Difference (IV, Fixed, 95% CI) | 0.08 [‐0.19, 0.36] |
| 10.1 Azithromycin | 1 | 68 | Mean Difference (IV, Fixed, 95% CI) | ‐0.02 [‐0.34, 0.30] |
| 10.2 Roxithromycin | 1 | 26 | Mean Difference (IV, Fixed, 95% CI) | 0.38 [‐0.16, 0.92] |
| 11 FVC (L): change Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 11.1 Azithromycin | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 12 FEV1/FVC: endpoint Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 12.1 Azithromycin | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 13 Adverse events Show forest plot | 5 | 435 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.83 [0.51, 1.35] |
| 13.1 Azithromycin | 3 | 292 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.77 [0.41, 1.45] |
| 13.2 Erythromycin | 1 | 117 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.16 [0.51, 2.62] |
| 13.3 Roxithromycin | 1 | 26 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.13 [0.01, 2.83] |
| 14 Azithromycin‐resistant bacteria (any) Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 14.1 Azithromycin | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 15 6‐Minute walk test: change Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 15.1 Erythromycin | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 16 Quality of life: endpoint Show forest plot | 1 | 68 | Mean Difference (IV, Fixed, 95% CI) | ‐8.90 [‐13.13, ‐4.67] |
| 16.1 Azithromycin | 1 | 68 | Mean Difference (IV, Fixed, 95% CI) | ‐8.90 [‐13.13, ‐4.67] |
| 17 Quality of life: change Show forest plot | 4 | 305 | Mean Difference (IV, Fixed, 95% CI) | ‐2.86 [‐5.67, ‐0.04] |
| 17.1 Azithromycin | 1 | 141 | Mean Difference (IV, Fixed, 95% CI) | ‐3.25 [‐7.19, 0.69] |
| 17.2 Erythromycin | 1 | 117 | Mean Difference (IV, Fixed, 95% CI) | ‐2.60 [‐7.12, 1.92] |
| 17.3 Roxithromycin | 2 | 47 | Mean Difference (IV, Fixed, 95% CI) | ‐1.86 [‐10.63, 6.91] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 ≥ 1 exacerbation Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 1.1 Roxithromycin | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2 QoL SGRQ: endpoint total score Show forest plot | 2 | 89 | Mean Difference (IV, Fixed, 95% CI) | ‐8.81 [‐14.33, ‐3.28] |
| 2.1 Roxithromycin | 2 | 89 | Mean Difference (IV, Fixed, 95% CI) | ‐8.81 [‐14.33, ‐3.28] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Hospitalisation: all‐cause Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 1.1 Azithromycin | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2 Serious adverse events Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 2.1 Azithromycin | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 3 Sputum purulence score: endpoint Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3.1 Roxithromycin | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 4 FEV1 (% predicted): endpoint Show forest plot | 2 | 65 | Mean Difference (IV, Fixed, 95% CI) | 1.73 [‐3.32, 6.78] |
| 4.1 Azithromycin | 1 | 40 | Mean Difference (IV, Fixed, 95% CI) | 3.70 [‐5.99, 13.39] |
| 4.2 Roxithromycin | 1 | 25 | Mean Difference (IV, Fixed, 95% CI) | 1.0 [‐4.91, 6.91] |
| 5 Adverse events Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 5.1 Azithromycin | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 6 Azithromycin‐resistant bacteria (any) Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 6.1 Azithromycin | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 7 Azithromycin‐resistant Streptococcus pneumoniae Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 7.1 Azithromycin | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 8 Azithromycin‐resistant Staphylococcus aureus Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 8.1 Azithromycin | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
