Representativeness and adverse event reporting in late-phase clinical trials for rifampin-susceptible tuberculosis: a systematic review

Abstract

We did a systematic review and meta-analysis of trials of treatment for rifampicin-susceptible tuberculosis to evaluate the representativeness of participants compared with characteristics of the global population of people with tuberculosis, and the adequacy of adverse event reporting. We searched MEDLINE, Embase, the Cochrane Central Register of Controlled Trials, and the Cochrane Database of Systematic Reviews from Jan 1, 2000, to Dec 10, 2023, for trials that had greater than or equal to 50 participants per arm and had follow-up to at least treatment completion. Studies were excluded if they compared different formulations of standard drugs (eg, fixed-dose combination tablets); aimed to primarily enrol participants with isoniazid-resistant or rifampicin-resistant tuberculosis; evaluated treatment to prevent tuberculosis infection; tested dietary or vitamin supplementation; tested vaccines or other immune-based interventions; tested adherence support or system-related mechanisms; or enrolled participants with tuberculosis, but tuberculosis treatment itself was not randomised (ie, trials of the timing of antiretroviral therapy initiation). Trial protocols and trials not available in English were also excluded. The outcomes were inclusion and exclusion criteria, characteristics of participants, and adverse event reporting. This systematic review was prospectively registered (PROSPERO ID CRD42022373954). We identified 7328 articles, of which 40 were eligible for analysis. Demographic characteristics, including sex, were reported for 20 420 participants, of which 6663 (33%) were female and 13 757 (67%) were male. We found that people who were greatly affected by the global tuberculosis pandemic were frequently excluded from participation: of the 40 trials, 25 (62·5%) excluded people younger than 18 years, 12 (30·0%) excluded people aged 65 years or older, 34 (85·0%) excluded pregnant or lactating people, 12 (30·0%) excluded people with diabetes, and 11 (27·5%) excluded people with excessive alcohol use, drug use, or both. In the nine trials that reported enrolment of people with diabetes, the pooled proportion of participants with diabetes (9%) was lower than global estimates for the proportion of people with tuberculosis who have diabetes (16%). There were important gaps in adverse event ascertainment, analysis, and interpretation. Of the 40 trials, a minority reported measures of regimen acceptability: 14 (35·0%) reported study withdrawal, eight (20·0%) reported temporary and 16 (40·0%) reported permanent discontinuation of assigned therapy, and 11 (27·5%) reported adherence. Participants in trials were not representative of the global tuberculosis pandemic in demographic and clinical characteristics, restricting the generalisability of trial outcomes. Adverse event reporting could be improved through the use of patient-reported outcomes, standardised definitions of key outcomes, and uniform reporting of measures of regimen acceptability. There was no funding for this systematic review.