Criteria for considering studies for this review
Types of studies
All published and unpublished randomized and quasi-randomized controlled trials. There was no language restriction for this review.
Types of participants
All individuals diagnosed with CF based on standard diagnostic criteria (sweat testing or genetics and clinical features or family history) were included. There were no restrictions on pulmonary disease severity, gender, or pancreatic status.
Types of interventions
Trials investigating thiol derivatives given at any dose or frequency, via nebulized or oral administration, for a minimum of four weeks duration were eligible for inclusion. We felt four weeks to be the minimum amount of time needed to see an effect from the treatment. Outcomes for participants who received thiol derivatives were compared to participants receiving no treatment or to control groups including placebo or any other medication. Nebulized and oral interventions were considered separately in this review. Thiol derivatives considered for inclusion were acetylcysteine (or N-acetylcysteine), sodium-2-mercaptoethane sulphonate, carbocysteine, erdosteine, nacystelyn, glutathione, cysteine and mecysteine.
Types of outcome measures
The following outcomes were recorded separately for both nebulized and oral thiol derivatives.
Pulmonary function testing (PFT)
forced expiratory volume in one second (FEV1) per cent predicted (change from baseline and absolute data)
forced vital capacity (FVC) per cent predicted (change from baseline and absolute data)
Other PFT measurements which reflect airflow obstruction or gas trapping or both, e.g. peak expiratory flow (PEF), vital capacity (VC), FEV1/VC, mid-expiratory flow 25-75 (MEF25-75), residual volume/total lung capacity (RV/TLC) (change from baseline and absolute data)
Inflammatory markers (change from baseline)
serum (white blood cell (WBC), C-reactive protein (CRP), erythrocyte sedimentation rate (ESR))
Quality of life (QOL) (change from baseline as measured by a validated QOL scale)
Need for oral antibiotics for pulmonary exacerbation
number of weeks of treatment
number of courses
Need for intravenous antibiotics for pulmonary exacerbation
number of weeks of treatment
number of courses
Number of days in hospital for respiratory exacerbation
Acquisition of new respiratory pathogens (%)
Six-minute walk distance
Sputum characteristics (including measures of viscosity and elasticity)
Search methods for identification of studies
We identified relevant trials using the Cystic Fibrosis Trials Register using the terms: acetylcysteine OR N-acetylcysteine OR carbocysteine OR erdosteine OR mecysteine OR nacystelyn OR glutathione OR cysteine OR sodium-2-mercaptoethane sulphonate.
The Cystic Fibrosis Trials Register is compiled from electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL) (updated each new issue of The Cochrane Library), quarterly searches of MEDLINE, a search of EMBASE to 1995 and the prospective handsearching of two journals - Pediatric Pulmonology and the Journal of Cystic Fibrosis. Unpublished work was identified by searching the abstract books of three major cystic fibrosis conferences: the International Cystic Fibrosis Conference; the European Cystic Fibrosis Conference and the North American Cystic Fibrosis Conference. For full details of all searching activities for the register, please see the relevant sections of the Cystic Fibrosis and Genetic Disorders Group Module.
Date of last search: 13 June 2013.
To ensure completeness of our literature search, we conducted a separate PubMed search on 26 February 2013 (Appendix 1). We manually reviewed abstracts and identified randomized controlled trials published since 2008.
Data collection and analysis
We used RevMan software to conduct the statistical analysis (RevMan 2011).
Selection of studies
Two authors (EN, AS during the initial review; JT, AS during the update) assessed the articles identified against the inclusion criteria for this review in an independent fashion. Where possible, authors contacted primary investigators if any details were unclear. They resolved any disagreements by discussion and advice from the other co-authors.
Data extraction and management
Two authors (EN, AS during the initial review; JT, AS during the update) extracted data from the articles in an independent fashion. Where possible, authors contacted primary investigators if their methodology was unclear. They resolved any disagreements by discussion and advice from the other co-authors.
Where possible we grouped outcome data into those measured at one month, up to three months, up to six months, up to twelve months and then annually thereafter. If outcome data were recorded at other time periods we examined these as well.
One of the included trials is a three-arm trial, and we are including it in two comparisons (oral thiols versus placebo and oral thiols versus other oral agents) (Ratjen 1985). By so doing, we realize that some participants will be counted twice in this review.
One trial’s published results are presented in graph form, or median values and interquartile ranges (Dauletbaev 2009). The authors attempted to contact the investigators to obtain raw data, but did not receive a response in time for this article. The trial’s results are described in this review, but are not included in the statistical analysis.
The authors of the three published abstracts listed as 'Awaiting assessment' were not contacted for this update, but will be contacted and we hope to have data to assess these for inclusion in a future update of the review.
Assessment of risk of bias in included studies
The authors assessed the articles to establish the risk of bias in each trial using a component approach and selected judgements in line with guidance in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Where possible, authors contacted primary investigators if their methodology was unclear.
Measures of treatment effect
When assessing differences between groups, we recorded the mean difference (MD) from baseline for each group as our treatment effect measure for continuous variables. Where applicable, we calculated a pooled estimate of treatment effect by calculating the MD and the 95% confidence intervals (CIs).
With binary outcomes, we planned to use the risk ratio (RR) and the 95% CIs as a measure of treatment effect; however, we were only able to present results using continuous data.
Unit of analysis issues
We included both parallel group trials as well as trials with a cross-over design. Ideally we would have liked to analyse cross-over trials using techniques outlined by Elbourne (Elbourne 2002); however, due to limitations on the data available we were only able to treat these trials as parallel trials and present data for the end of treatment. We realise that this approach is conservative as it ignores within-patient correlation and so does not make use of the advantages of the cross-over design. Also, this approach ignores the fact that patients appear in both arms of the trial and are not independent of each other (Elbourne 2002). We did not combine the data from cross-over trials with parallel trials as we were unable to employ the methods recommended by Curtin (Curtin 2002).
For longitudinal data, we reported the time-points that measurements were taken by the primary investigators and which measures were reported in the papers. In the review we present data from end of treatment. We preferred to report data for the change from baseline at end of treatment, but if the original paper reported only absolute data (means, standard deviations (SD) of groups or raw data), we planned to calculate the mean difference and the variance of the difference imputed using the Follmann method (Follmann 1992). This method allows the use of separate sources of incomplete information to help choose a better variance estimate. This method can be summarised by the formula: Var(change) = Var(pre-test) + Var(post-test) - 2 x SD(pre-test) x SD(post-test) x pre-test-post-test correlation coefficient. We also planned to report these absolute values post-treatment. Where the correlation coefficient was unknown we estimated it to be 0.4 (a moderate correlation) in order to perform the calculation.
Dealing with missing data
When possible, the authors (JT, EN, AS) contacted primary investigators if missing data were required.
Assessment of heterogeneity
We planned to assess the heterogeneity of the trials using the Q test, with heterogeneity being considered to be present if the Q test was statistically significant at the P < 0.10 level. When assessing the magnitude of any heterogeneity present, we planned to use the I2 statistic (Higgins 2003), the value of which depends on the magnitude and direction of effects, and evidence for heterogeneity (Higgins 2011). We interpret values of I2 below 33% as representing a low amount of heterogeneity, greater or equal to 33% and below 66% representing a moderate amount of heterogeneity, and greater or equal to 66% representing a high amount of heterogeneity.
Assessment of reporting biases
We intended on assessing publication bias using a funnel plot; however, there were insufficient trials (minimum of 10 required) to conduct this analysis.
With regards to selective reporting of outcomes, we were unable to compare the original trial protocols with the final published papers; however, there did not appear to be any obvious omissions to the outcomes reported on by the trial investigators.
We did not identify any heterogeneity between trials, and used a fixed-effect model in our analysis; however, if in future we establish moderate or high degrees of heterogeneity, we will utilize a random-effects model.
Subgroup analysis and investigation of heterogeneity
Several thiol derivatives are available, and therefore we planned to perform subgroup analysis for each compound. The dose of the thiol derivative used may alter outcomes, and therefore we planned to perform subgroup analysis according to dose. There were insufficient trials to perform either of these subgroup analyses on this occasion, but we plan to perform these analyses in a future update of the review when we are able to combine a sufficient number of trials to allow this. Clinically important outcomes vary depending on gender, age and severity of lung disease (FEV1 70% to 80% will be considered mild; 60% to 70% moderate; 50% to 60% moderately severe; 34% to 50% severe; and less than 34% very severe (ATS 1991)), and therefore we also planned to perform an analysis using these subgroups. We planned to analyze age and lung function as continuous variables as well as categorical variables in an attempt to identify any high-risk groups. The degree of airway inflammation may also impact on the response to therapy with thiol derivatives, and therefore we also planned to perform subgroup analysis according to whether the trial was performed during an acute pulmonary exacerbation or during a period of disease stability. Again, there were insufficient trials to perform these subgroup analyses in this review, but we plan to perform these analyses in a future update where possible. Lung function will be categorized according to ATS guidelines for disease severity as described above (ATS 1991), and age will be categorized in 10-year blocks where possible.
We intended to conduct a sensitivity analyses based on the degree of risk of bias of the trials in the review; however, there were insufficient trials to proceed with this analysis.