Objective: To estimate the effect of delaying antiretroviral treatment (ART) for 15, 30, or 60 days after tuberculosis (TB) treatment initiation on mortality and virological suppression.
Design: Cohort of 573 ART-naive HIV-infected children initiated on TB treatment at an outpatient clinic in South Africa between April 2004 and March 2008.
Methods: Hazard ratios for mortality and viral suppression were estimated using marginal structural models and multivariate Cox models, respectively.
Results: During follow-up (median 9.64 months), 37 HIV-infected children died after a median of 62 days of TB treatment. ART was initiated in 461 children at a median of 17 days after TB treatment initiation, 415 (90%) achieved viral suppression. The hazard ratios of death for initiating ART more than 15, more than 30, or more than 60 days of TB treatment compared with initiating within 15, 30 and 60 days, respectively, were 0.82 (95% CI: 0.48, 1.41), 0.86 (95% CI: 0.46, 1.60), and 1.32 (95% CI: 0.55, 3.16). Hazard ratios for analysis restricted to severely immunosuppressed children were: 0.92 (95% CI: 0.51, 1.63), 1.08 (95% CI: 0.56, 2.08), and 2.23 (95% CI: 0.85, 5.80), respectively. Hazard ratios for viral suppression were 0.98 (95% CI: 0.76, 1.26), 0.95, (95% CI: 0.73, 1.23), 0.84 (95% CI: 0.61, 1.15), respectively and did not change with restriction to children severely immunosuppressed.
Conclusion: In this observational study, we found that delaying ART for 2 months or more in children diagnosed with TB may be associated with poorer virological response and increased mortality, particularly in children with severe immunosuppression. These findings should be confirmed in a randomized controlled trial.