The methodological quality was defined as the control of bias in the treatment comparison. The assessment was based on published reports and information provided by the authors of included trials. Based on previous evidence, the randomization methods were classified as the primary measure of bias control.21, 22 The randomization methods were evaluated by the allocation sequence generation (classified as adequate if based on a table of random numbers, computer-generated random numbers, or similar) and allocation concealment (classified as adequate if based on central randomization, identically Vemurafenib appearing coded drug containers, serially numbered opaque sealed

envelopes, or similar). We also extracted blinding (whether the trial was described as double-blind or single-blind, the method of blinding; whether patients, investigators, outcome assessors or other persons involved in the trial were blinded; and whether the adequacy of blinding was assessed),23 the risk of attrition bias (numbers and reasons for dropouts and withdrawals and whether all patients

were accounted for in the report and analysis of the trial), whether the primary outcome measure was defined and reported, whether a crossover design was used, whether sample size calculations were performed, Maraviroc and whether the preset sample size was achieved. For trials terminated prematurely, we registered whether this was based on predefined criteria. The analyses were performed using RevMan version 5.0.5 (Nordic Cochrane Centre, Copenhagen, Denmark). Meta-analyses were performed using random effects models due to expected clinical heterogeneity. Results are presented as the relative risk (RR) for binary and weighted mean differences for continuous outcomes, both with 95% confidence intervals (CIs).

I2 values were calculated as measures of the degree of intertrial heterogeneity. Data on all patients randomized were extracted to allow intention-to-treat analyses. For patients with missing data, carry-forward of the last observed response was used. Only data from the first period of crossover trials were included. For the primary outcome measure, we performed subgroup analyses of trials stratified by the treatment regimen, the type of HRS, and methodological quality. Based on differences in the duration of follow-up in individual Calpain trials, we performed a post hoc analysis to evaluate the relationship between the treatment effect on mortality and the duration of follow-up. Based on discrepancies between the number of patients who survived and the number of patients with reversal of HRS, we performed a post hoc analysis that combined these two outcome measures. We originally planned to perform regression analyses to detect the risk of bias, including publication bias.24 However, we did not perform these analyses, because the power to detect bias was insufficient due to the small number of trials included.