Abstract
We assessed the effectiveness and safety of standardised, shorter multidrug-resistant tuberculosis (MDR-TB) regimens by pooling data from observational studies.Published studies were identified from medical databases; unpublished studies were identified from expert consultation. We conducted aggregate data meta-analyses to estimate pooled proportions of treatment outcomes and individual patient data (IPD) meta-regression to identify risk factors for unsuccessful treatment in patients treated with 9- to 12-month MDR-TB regimens composed of a second-line injectable, gatifloxacin/moxifloxacin, prothionamide, clofazimine, isoniazid, pyrazinamide and ethambutol.We included five studies in which 796 out of 1279 (62.2%) individuals with confirmed MDR-TB (98.4%) or rifampin-resistant TB (1.6%), and not previously exposed to second-line drugs, were eligible for shorter regimens. 669 out of 796 participants were successfully treated (83.0%, 95% CI 71.9-90.3%). In IPD meta-regression (three studies, n=497), failure/relapse was associated with fluoroquinolone resistance (crude OR 46, 95% CI 8-273), pyrazinamide resistance (OR 8, 95% CI 2-38) and no culture conversion by month 2 of treatment (OR 7, 95% CI 3-202). Two participants acquired extensive drug resistance. Four studies reported grade 3 or 4 adverse events in 55 out of 304 (18.1%) participants.Shorter regimens were effective in treating MDR-TB; however, there is uncertainty surrounding the generalisability of the high rate of treatment success to less selected populations, to programmatic settings and in the absence of drug susceptibility tests to key component drugs.