Background: Tuberculosis is a major cause of morbidity and mortality in HIV-infected children, but is difficult to diagnose. We studied mortality and its determinants in antiretroviral treatment (ART)-naive HIV-infected children presenting with suspected tuberculosis. Methods: In this observational cohort study, HIV-infected children aged 13 years or younger with suspected tuberculosis were followed up for 6 months as part of the ANRS 12229 PAANTHER 01 cohort in eight hospitals in four countries (Burkina Faso, Cambodia, Cameroon, and Vietnam). Children started ART and antituberculosis treatment at the clinician's discretion and were retrospectively classified into one of three groups by tuberculosis documentation: confirmed by culture or Xpert MTB/RIF, unconfirmed, and unlikely. We assessed mortality and associated factors using Kaplan-Meier methods and Cox proportional hazard models. The ANRS 12229 PAANTHER 01 study is registered at ClinicalTrials.gov, number NCT01331811. Findings: 266 (61%) of 438 children enrolled in the study between April 27, 2011, and May 31, 2014, were ART-naive and included in the analysis (40 had confirmed tuberculosis, 119 unconfirmed tuberculosis, and 107 unlikely tuberculosis). 112·5 person-years of follow-up were available. 154 children (58%) started antituberculosis treatment and 212 (80%) started ART. 50 children (19%) died. Mortality by 6 months was higher in children with confirmed tuberculosis (14 deaths; 2 month survival probability 65·0% [95% CI 50·2–79·8]) compared with unconfirmed tuberculosis (19 deaths; 83·5% [76·8–90·3]) and unlikely tuberculosis (17 deaths; 83·5% [76·3–90·7]; log-rank p=0·0141) and was lower in children with confirmed or unconfirmed tuberculosis who started antituberculosis treatment (p<0·0001 for both). In a multivariate analysis, ART started during the first month of follow-up (hazard ratio 0·08; 95% CI 0·01–0·67), confirmed tuberculosis (6·33; 2·15–18·64), young age (5·90; 2·02–17·19), CD4 less than 10% (2·63; 1·25–5·53), miliary features (4·08; 1·56–10·66), and elevated serum transaminases (4·40; 1·82–10·65) were all independently associated with mortality. Interpretation: In our cohort, mortality was high in the first 6 months after suspicion of tuberculosis in ART-naive children. ART should be started early, particularly in children with factors associated with high mortality. Documented or empirical tuberculosis treatment decision should be accelerated to reduce mortality and allow early ART initiation. Funding: ANRS and Fondation Total.
