Urgent versus post-stabilisation antiretroviral treatment in hospitalised HIV-infected children in Kenya (PUSH): a randomised controlled trial

Abstract

SUMMARYBackgroundUrgent antiretroviral therapy (ART) among hospitalized HIV-infected children may accelerate recovery or worsen outcomes due to immune reconstitution. In an unblinded randomized controlled trial, we compared urgent versus post-stabilization ART among hospitalized HIV-infected children.MethodsWe randomized HIV-infected, ART-naïve children age 0 – 12 years who were eligible for ART in a 1:1 ratio to receive ART within 48 hours (urgent arm) or seven to 14 days (post-stabilization arm). We excluded children with suspected or confirmed central nervous system (CNS) infection. Block randomization, with variable block sizes, was generated by a statistician not involved in study procedures. We followed children for six months for primary outcomes: mortality, drug toxicity, and immune reconstitution inflammatory syndrome (IRIS). Registered in Clinical Trials.gov (NCT02063880). Trial status: complete.FindingsWe began enrollment on 24th April, 2013 and completed follow-up on 17th November, 2015. Of 250 HIV infected children, we enrolled 191/250 (76%) and randomized 183/191 (96%). We included 181/183 (99%) of randomized hospitalized HIV infected children age 0–12 who had no CNS infection in a modified intent-to-treat analysis. Median age was 1·9 years (IQR 0·8–4·8). Baseline sociodemographic, clinical, and virologic characteristics did not differ between arms except median CD4%, which was lower in the urgent arm (13% [IQR 9, 18] versus 17% [IQR 9, 24], p=0·05). Pneumonia, malnutrition, and suspected tuberculosis (TB) contributed to 118/181 (65%), 58/181 (32%), and 27/181 (15%) of admission diagnoses, respectively. Median time to ART was one day (IQR 1, 1) and eight days (IQR 7, 11) in the urgent and post-stabilization arms, respectively. Overall, mortality risk was 61 per 100 person-years. Mortality risk did not differ by arm (70 versus 54 per 100 person-years in urgent versus post-stabilization arms, respectively [HR 1.26 95% CI 0.67, 2.37 p=0.47]), even after adjusting for baseline CD4% (aHR 1·30 [95% CI 0·69, 2·45, p=0·41]). There was no statistical difference in incidence of IRIS or drug toxicity between trial arms. We discontinued randomization at interim review when the futility boundary was crossed.InterpretationEarly mortality risk was extremely high among hospitalized HIV-infected children. Urgent ART did not improve survival.