Fifty-seven RCTs with 21,472 patients in 312 trial arms were included in the review. Twenty-nine RCTs had losses to treatment of under 10%; randomisation was considered appropriate in 40 of the 41 trials describing it. Median follow-up post-treatment was 24 months (interquartile range 18 to 30 months).
Head-to-head comparisons found that failure rates were significantly higher in patients receiving only one to two months of rifampin than in those treated for three to four months (RD 0.3%, 95% CI 0.9 to 1.4; two RCTs), with no heterogeneity. Shorter durations of rifampin were associated with significantly higher rates of relapse compared with longer durations in all analyses. There was no difference between shorter and longer treatment durations for acquired drug resistance. Pooling of comparisons of intermittent schedules was not possible due to different schedule comparisons between trials.
Indirect comparisons confirmed the finding that rifampin durations of one to two months had higher failure and relapse rates than longer durations, and that rates of relapse were higher for shorter durations of rifampin, up to a treatment length of eight months. There were no significant differences between schedules using daily treatment, daily then thrice weekly, daily then twice weekly, or thrice weekly throughout treatment. Only one trial evaluated treatment twice weekly throughout therapy.
Meta-regression confirmed that one to two months of rifampin treatment resulted in higher incidence rates of failure, relapse and acquired drug resistance than longer durations. A number of other factors were found to have statistically significant relationships with treatment outcomes and full details were reported in the paper.