Mortality under early access to antiretroviral therapy vs . Eswatini’s national standard of care: the MaxART clustered randomized stepped‐wedge trial

Abstract

Objectives Current WHO guidelines recommend the treatment of all HIV-infected individuals with antiretroviral therapy (ART) to improve survival and quality of life, and decrease infection of others. MaxART is the first implementation trial of this strategy embedded within a government-managed health system, and assesses mortality as a secondary outcome. Because primary findings strongly supported scale-up of the 'treat all' strategy (hereafter Treat All), this analysis examines mortality as an additional indicator of its impact. Methods MaxART was conducted in 14 Eswatinian health clinics through a clinic-based stepped-wedge design, by transitioning clinics from then-national standard of care (SoC) to the Treat All intervention. All-cause, disease-related, and HIV-related mortality were analysed using the Cox proportional hazards model, censoring SoC participants at clinic transition. Median follow-up time among study participants was 292 days. There were 36/2034 deaths in SoC (1.77%) and 49/1371 deaths in Treat All (3.57%). Results Between September 2014 and August 2017, 3405 participants were enrolled. In SoC and Treat All interventions, respectively, the multivariable-adjusted 12-month all-cause mortality rates were 1.42% [95% confidence interval (CI): 0.66-2.17] and 1.60% (95% CI: 0.78-2.40), disease-related mortality rates were 1.02% (95% CI: 0.40-1.64) and 1.10% (95% CI: 0.46-1.73), and HIV-related mortality rates were 1.03% (95% CI: 0.40-1.65) and 0.99% (95% CI: 0.40-1.58). Treat All had no impact on all-cause [hazard ratio (HR) = 1.12, 95% CI: 0.58-2.18, P = 0.73], disease-related (HR = 1.04, 95% CI: 0.52-2.11, P = 0.90), or HIV-related mortality (HR = 0.93, 95% CI: 0.46-1.87, P = 0.83). Conclusion There was no immediate benefit of the Treat All strategy on mortality, nor evidence of harm. Longer follow-up of participants is needed to establish long-term consequences.