HIV-associated executive dysfunction in the era of modern antiretroviral therapy: A systematic review and meta-analysis.

OBJECTIVE: While some reports suggest that HIV+ individuals continue to display executive function (EF) impairment in the era of cART, findings have been contradictory and appear to differ based on the aspect of EF being measured. To improve the understanding of how discrete executive abilities may be differentially affected or spared in the context of HIV infection, we conducted a systematic review and meta-analysis to (a) determine whether and to what extent HIV+ adults experience deficits in EFs, and (b) understand how demographic and clinical characteristics may modify the associations between HIV infection and executive abilities.

METHOD: Studies comparing HIV+ and HIV-uninfected groups on measures of working memory, set-shifting, inhibition, decision-making, and apathy between 2000 and 2017 were identified from three databases. Effect sizes (Cohen's d) were calculated using inverse variance weighted random effects models. Meta-regression was used to examine the moderating effect of demographic and clinical variables.

RESULTS: Thirty-seven studies (n = 3935 HIV+; n = 2483 HIV-uninfected) were included in the meta-analysis. Pooled effect sizes for deficits associated with HIV infection were small for domains of set-shifting (d = -0.34, 95% CI [-0.47, -0.20]) and inhibition (d = -0.31, 95% CI [-0.40, -0.21]), somewhat larger for measures of decision-making (d = -0.41, 95% CI [-0.53, -0.28]) and working memory (d = -0.42, 95% CI [-0.59, -0.29]), and largest for apathy (d = -0.87, 95% CI [-1.09, -0.66]). Meta-regression demonstrated that age, sex, education, current CD4 count, and substance dependence differentially moderated the effects of HIV infection on specific EFs. However, lower nadir CD4 count was the only variable associated with greater deficits in nearly all EF domains.

CONCLUSIONS: Our results suggest that discrete domains of EF may be differentially affected by HIV infection and moderating demographic and clinical variables. These findings have implications for the development of targeted cognitive remediation strategies.