The interplay between <scp>CD</scp>4 cell count, viral load suppression and duration of antiretroviral therapy on mortality in a resource‐limited setting

<jats:title>Abstract</jats:title><jats:sec><jats:title>Objective</jats:title><jats:p>To examine the interaction between <jats:styled-content style="fixed-case">CD</jats:styled-content>4 cell count, viral load suppression and duration of antiretroviral therapy (<jats:styled-content style="fixed-case">ART</jats:styled-content>) on mortality.</jats:p></jats:sec><jats:sec><jats:title>Methods</jats:title><jats:p>Cohort analysis of <jats:styled-content style="fixed-case">HIV</jats:styled-content>‐infected patients initiating <jats:styled-content style="fixed-case">ART</jats:styled-content> between April 2004 and June 2011 at a large public sector clinic in Johannesburg, <jats:styled-content style="fixed-case">S</jats:styled-content>outh <jats:styled-content style="fixed-case">A</jats:styled-content>frica. A log‐linear model with Poisson distribution was used to estimate risk of death as a function of the interaction between current <jats:styled-content style="fixed-case">CD</jats:styled-content>4 count, current viral load suppression and duration on <jats:styled-content style="fixed-case">ART</jats:styled-content> in 12‐month intervals. We calculated predicted mortality using estimated coefficients within combinations of predictors.</jats:p></jats:sec><jats:sec><jats:title>Results</jats:title><jats:p>Amongst 14 932 <jats:styled-content style="fixed-case">ART</jats:styled-content> patients, 1985 (13.3%) died. Current <jats:styled-content style="fixed-case">CD</jats:styled-content>4 was the strongest predictor of death (<50 <jats:italic>vs</jats:italic>. ≥550 cells/mm<jats:sup>3</jats:sup> – <jats:styled-content style="fixed-case">RR</jats:styled-content>: 46.3; 95% <jats:styled-content style="fixed-case">CI</jats:styled-content>: 26.8–80), while unsuppressed current viral load <jats:italic>vs</jats:italic>. suppressed (<jats:styled-content style="fixed-case">RR</jats:styled-content>: 1.8; 95% <jats:styled-content style="fixed-case">CI</jats:styled-content>: 1.5–2.1) and short duration of <jats:styled-content style="fixed-case">ART</jats:styled-content> (0–11.9 <jats:italic>vs</jats:italic>. 66–71.9 months <jats:styled-content style="fixed-case">RR</jats:styled-content>: 1.7; 95% <jats:styled-content style="fixed-case">CI</jats:styled-content>: 1.2–2.3) also predicted death. Our interaction model showed that mortality was highest in the first 12 months on treatment across all <jats:styled-content style="fixed-case">CD</jats:styled-content>4 and viral load strata. As current <jats:styled-content style="fixed-case">CD</jats:styled-content>4 and duration on <jats:styled-content style="fixed-case">ART</jats:styled-content> increased and viral load suppression occurred, mortality dropped. <jats:styled-content style="fixed-case">CD</jats:styled-content>4 count was the strongest predictor of death. The relative effect of current <jats:styled-content style="fixed-case">CD</jats:styled-content>4 count varied strongly by viral load and duration of <jats:styled-content style="fixed-case">ART</jats:styled-content> (from 1.3 to 55). Lack of suppression increased the risk of mortality upwards of six‐fold depending on time on <jats:styled-content style="fixed-case">ART</jats:styled-content> and current <jats:styled-content style="fixed-case">CD</jats:styled-content>4.</jats:p></jats:sec><jats:sec><jats:title>Conclusions</jats:title><jats:p>Our findings show that while <jats:styled-content style="fixed-case">CD</jats:styled-content>4 count is the strongest predictor of death, the effect is modified by viral load and the duration of <jats:styled-content style="fixed-case">ART</jats:styled-content>. Assessment of risk should take into account all three factors.</jats:p></jats:sec>