One hundred and eighteen RCTs (n=15,060) were included in the review.
Nineteen per cent of the trials scored at least 3 on the Jadad quality score. Twenty-five per cent of the trials reported some form of blinding, with 14 of the 30 trials reporting a statement on double-blinding; however, only 4 trials actually described the method of blinding. A sample size calculation was reported in 20% of the trials, and allocation concealment was considered adequate in only 12% of the trials.
The pooled OR for treatment effect varied according to the interventions assessed. There was a statistically significant increase in sustained ALT normalisation with IFN treatment in comparison with placebo or no treatment (32 RCTs, n=2,499); the OR was 9.5 (95% confidence interval, CI: 6.3, 14.2). No significant heterogeneity was observed across this group of trials.
There was a statistically significant effect of treatment with standard IFN compared with alternative IFN schedules (43 RCTs, n=7,454); the OR was 1.6 (95% CI: 1.4, 1.9). Significant heterogeneity was observed across these trials.
There was a statistically significant effect of treatment with IFN in combination with another drug versus IFN alone (30 RCTs, n=4,737); the OR was 2.0 (95% CI: 1.6, 2.6). However, these trials were significantly heterogeneous.
The meta-regression (complete data set, 171 trial arms; n=10,580) showed that sustained response was most likely in experimental arms of IFN in combination with ribavirin or other drugs (OR=2.4); in arms using a yearly dosing schedule (OR=2.0); in studies in which the principal trial author was from Asia, (OR=1.7); in trials with a sample size greater than 200 participants (OR=1.4); and in trial arms enrolling less than 50% of cirrhotics (OR=1.3). The results of the meta-regression also showed a number of significant interaction effects. The effect of trial quality interacted with the recorded funding body, with more benefit observed if the body was not for profit and less if it was for profit. The effect of trial funding also interacted significantly with the location of the first author, with more benefit being observed if the first author was from Asia. Overall, the three main effects of experimental arm, cirrhosis, funding, and the interaction effect of type of funding and the location of the principal author, explained 31% of the between-study variability.
The results of the meta-regression on the sub-group of trials reporting data on HCV genotype (93 arms; n approximately 7,000) showed that enrolling less than 50% of patients with genotype 1 or 4 per arm, and publishing the results of the study in a specialist journal, were significant predictors of either biochemical (transaminases) or virological (HCV-RNA) sustained response in a model including the same main effects identified in the complete data set analysis (reported above).