Early mortality (<48 wks) after ART initiation in resource-limited settings is well recognised, but less well understood are the causes of later mortality, which are widely assumed to be due to virological failure (VF) or non-adherence. We investigated HIV viral load (VL) in patients who died after 48 weeks of continuous first-line ART in the DART study.
The DART trial randomised 3,316 Ugandan and Zimbabwean patients to laboratory monitoring (LCM; CD4 cell count every 12 weeks) or clinically driven monitoring (CDM). Prospective VL testing was not undertaken. Previous analyses found that low pre-ART CD4 cell count was strongly associated with higher mortality during the first year of ART, which was predominantly from infectious causes. All late mortalities had stored plasma samples from the closest visit to the date of death retrospectively tested for VL. Logistic regression models were used to determine predictors of mortality with virological suppression (VS) status (VL<200 copies/mL) in patients who died. Fractional polynomials were used for continuous variables, but non-linear risk was not found.
210/382 (55%) deaths during the DART study occurred after week 48; 112 were on continuous first-line and 78 on second-line ART. The late mortality rate was low (10.7/1,000 PY). VL data were available for 105/112 (94%) mortalities at a median (IQR) of 10 (6-14) weeks before death. 43/105 (41%) patients were virologically suppressed (VS) at the time of death. VF deaths were more often due to opportunistic infections (26% vs 12%;p=0.09). CD4 cell count was significantly lower at the time of death in patients with VF than VS (Median: 62 vs 238 cells/mm3; p<0.001) and a greater proportion had CD4<100 cells/mm3 (66% vs 19%). In multivariate logistic regression analyses (Table), patients in the CDM arm had reduced odds of 0.28 (95% CI: 0.11 – 0.68) death with VS, with no evidence of a change over time. The odds of death with VS were almost 4 times higher for each additional 100 cells/mm3 increase in baseline CD4 count. Gender, age, initial ART regimen, CD4 cell count at week 48, baseline VL and opportunistic infections were not associated with VS at death in multivariate analyses.
40% of late deaths on ART occurred without VL criteria for treatment switch being fulfilled. There were significantly more deaths with VS among patients who received CD4 cell count monitoring. Further research is required to elucidate the cause of deaths in those without VF.