Types of studies

Randomised trials were included irrespective of blinding, publication status, or language. Quasi‐randomised trials and observational studies were excluded.

Types of participants

We included patients with chronic hepatitis C. The diagnosis was based on presence of hepatitis C virus RNA plus elevated transaminases for more than six months or chronic hepatitis documented on liver biopsy. We also included patients diagnosed with 'non‐A, non‐B' chronic hepatitis as some trials may have been conducted before HCV‐RNA analyses were widely available. Based on previous antiviral treatment, the included patients were classified as naive (not previously treated), relapsers (patients with a transient response to previous treatment), or non‐responders (patients without response to previous treatment). We excluded patients who had human immune deficiency virus infection or had undergone liver transplantation.

Types of interventions

This review includes randomised comparisons of any dose or duration of

• ribavirin versus no intervention or placebo;

• ribavirin versus interferon.

Types of outcome measures

Primary outcome measures

• Failure of serum (or plasma) sustained virological response: number of patients with detectable hepatitis C virus RNA at least six months after treatment;

• Liver‐related morbidity plus all‐cause mortality: number of patients who developed cirrhosis, ascites, variceal haemorrhage, hepatic encephalopathy, hepatocellular carcinoma, or died;

• All adverse events: number of patients with adverse events: number of patients with different types of adverse events, number of necessary dose reductions, and treatment discontinuations.

Secondary outcome measures

• Failure of end of treatment virological response: number of patients with detectable hepatitis C virus RNA at the end of treatment;

• Failure of sustained biochemical response (number of patients without normalisation of alanine aminotransferase (ALT) or aspartate transaminase (AST) six months or more after treatment);

• Failure of end of treatment biochemical response (number of patients without normalisation of ALT or AST at the end of treatment);

• Failure of histological response: number of patients without improvement of histology (inflammation score (grading) or fibrosis score (staging) as defined by the individual trials);

• Quality of life.

Electronic searches

As described in Appendix 1, we searched The Cochrane Hepato‐Biliary Group Controlled Trials Register (Gluud 2009), the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library, MEDLINE, EMBASE, and Science Citation index Expandeed (Royle 2003). We searched for ongoing clinical trials and unpublished trials on http://controlledtrials.gov, http://clinicaltrials.gov, and http://centerwatch.com. Searches were performed in March 2009.

Searching other resources

Trials were also identified through manual searches of bibliographies in relevant articles and through hand searches of conference proceedings. We also wrote to authors of included trials and pharmaceutical companies.

Statistical methods

We performed the review and meta‐analyses following the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008) and The Cochrane Hepato‐Biliary Group Module (Gluud 2009). The analyses were performed using Review Manager 5.0 and (RevMan 2008) and Trial Sequential Analysis version 0.8 (TSA 2008).

Selection of studies

We listed the identified trials, and JB and LLG evaluated whether the trials fulfilled the inclusion criteria. Disagreements were resolved by discussion together with CG. Excluded trials were listed with the reason for exclusion.

Data extraction and management

Data from all included trials were extracted by JB and LLG. Disagreements were resolved by discussion with CG. We wrote to the principal investigator of included trials to ask for data that were not presented in the published reports. From each trial we extracted the virological response to previous antiviral therapy, mean age, proportion of men, proportion of patients with cirrhosis, proportion of patients with hepatitis C virus genotype 1, limit of detectable HCV‐RNA, dose and duration of ribavirin and interferon therapy, type of interferon, bias risk components, funding, publication status, duration of follow‐up, and all outcome measures.

Assessment of risk of bias in included studies

We assessed the components below to estimate the risk of bias (Als‐Nielsen 2004; Higgins 2008; Wood 2008; Gluud 2009).

Sequence generation 
‐ Adequate, sequence generation was achieved using computer random number generation or a random number table. Drawing of lots, tossing a coin, shuffling cards and throwing dice were considered as adequate if performed by an independent adjudicator.

‐ Unclear, the trial was described as randomised but the method of sequence generation was not specified.

‐ Inadequate allocation sequence generation using dates, names, or admittance numbers (quasi‐randomised trials) were excluded from the present review.

Allocation concealment
‐ Adequate, allocation concealment through a central and independent unit, sealed envelopes, or similar.

‐ Unclear, the trial was described as randomised but the method used to conceal the allocation was not described.

‐ Inadequate allocation sequence was known to the investigators who assigned participants through use of an open table of random numbers or similar or the study was quasi‐randomised as explained above.

Blinding
 ‐ Adequate, double blinding was achieved using identical placebo or active treatments. We classified trials that used the word 'placebo' as adequate even though it was not specified that placebo had same smell, appearance, taste, etc.

‐ Unclear, the trial was described as double blind, but the method of blinding was not described.

‐ Inadequate, the trial was not double blind.

Baseline imbalance
‐ Adequate, if there was no baseline imbalance in important characteristics.

‐ Unclear, if the baseline characteristics were not reported.

‐ Inadequate if there was a baseline imbalance due to chance or due to imbalanced exclusion after randomisation.

Early stopping
‐ Adequate, if sample size calculation was reported and the trial was not stopped, or the trial was stopped early by formal stopping rules at a point where the likelihood of observing an extreme intervention effect due to chance was low.

‐ Unclear, if sample size calculation was not reported and it is not clear whether the trial was stopped early or not.

‐ Inadequate, if the trial was stopped early due to informal stopping rules.

Incomplete reporting of outcome data
‐ Adequate the numbers and reasons for dropouts and withdrawals in all intervention groups were described or if it was specified that there were no dropouts or withdrawals. 

‐ Unclear, the report gave the impression that there had been no dropouts or withdrawals, but this was not specifically stated. 

‐ Inadequate, the number or reasons for dropouts and withdrawals were not described.

Selective outcome reporting
‐ Adequate, if the trial protocol is available and all of the trial's pre‐specified outcomes that are of interest in the review have been reported or similar.

‐ Unclear, there is insufficient information to assess whether the magnitude and direction of the observed effect is related to selective outcome reporting.

‐ Inadequate, not all of the trial's pre‐specified primary outcomes have been reported or relevant outcomes are omitted.

Source of funding bias
‐ Adequate, the trial's source(s) of funding did not come from any parties that might had conflicting interest (eg, a drug or a device manufacturer).

‐ Unclear, uncertain risk of bias if the source of funding was not clear.

‐ Inadequate, high risk of bias if the trial was funded by a drug or a device manufacturer.

Funnel plot were also applied to assess bias and publication bias (Egger 1997)

Measures of treatment effect

Data were analysed by both a random‐effects model (DerSimonian 1986) and a fixed‐effect model (DeMets 1987). We only present the results of the fixed‐effect model if both models provide the same result regarding statistical significance. We present data with risk difference (RD) and 95% confidence intervals (CI) due to the large number of trials with zero events. Such analyses may, however, provide too narrow CI (Brok 2003; Keus 2009).

Dealing with missing data

We analysed data by intention to treat including all patients irrespective of compliance or follow‐up. Because many patients relapse after the end of therapy, patients lost to follow‐up were counted as non‐responders. A large number of patients did not have post treatment liver biopsies. An additional per protocol sensitivity analysis of histological response was, therefore, performed.

Assessment of heterogeneity

Heterogeneity was explored by chi‐squared test with significance set at P‐value less than 0.10, and the quantity of heterogeneity was measured by I² (Higgins 2008). We assessed sources of heterogeneity in random‐effects meta‐regression analyses for our primary outcome if the meta‐analysis included at least 10 trials. These analyses included all the extracted patient, intervention, and trial characteristics.

Sensitivity analysis

We performed sensitivity analyses excluding trials with patients diagnosed with 'non‐A, non‐B' chronic hepatitis.