UCSF

BACKGROUND: 3 months of weekly rifapentine plus isoniazid (3HP) and 4 months of daily rifampicin (4R) are recommended for tuberculosis preventive treatment. As these regimens have not been compared directly, we used individual patient data and network meta-analysis methods to compare completion, safety, and efficacy between 3HP and 4R. METHODS: We conducted a network meta-analysis of individual patient data by searching PubMed for randomised controlled trials (RCTs) published between Jan 1, 2000, and Mar 1, 2019. Eligible studies compared 3HP or 4R to 6 months or 9 months of isoniazid and reported treatment completion, adverse events, or incidence of tuberculosis disease. Deidentified individual patient data from eligible studies were provided by study investigators and outcomes were harmonised. Methods for network meta-analysis were used to generate indirect adjusted risk ratios (aRRs) and risk differences (aRDs) with their 95% CIs. FINDINGS: We included 17 572 participants from 14 countries in six trials. In the network meta-analysis, treatment completion was higher for people on 3HP than for those on 4R (aRR 1·06 [95% CI 1·02-1·10]; aRD 0·05 [95% CI 0·02-0·07]). For treatment-related adverse events leading to drug discontinuation, risks were higher for 3HP than for 4R for adverse events of any severity (aRR 2·86 [2·12-4·21]; aRD 0·03 [0·02-0·05]) and for grade 3-4 adverse events (aRR 3·46 [2·09-6·17]; aRD 0·02 [0·01-0·03]). Similar increased risks with 3HP were observed with other definitions of adverse events and were consistent across age groups. No difference in the incidence of tuberculosis disease between 3HP and 4R was found. INTERPRETATION: In the absence of RCTs, our individual patient data network meta-analysis indicates that 3HP provided an increase in treatment completion over 4R, but was associated with a higher risk of adverse events. Although findings should be confirmed, the trade-off between completion and safety must be considered when selecting a regimen for tuberculosis preventive treatment. FUNDING: None. TRANSLATIONS: For the French and Spanish translations of the abstract see Supplementary Materials section.