Low haemoglobin predicts early mortality among adults starting antiretroviral therapy in an HIV care programme in South Africa: a cohort study
1 London School of Hygiene and Tropical Medicine, London, UK
2 Aurum Institute, Johannesburg, South Africa
BMC Public Health 2010, 10:433 doi:10.1186/1471-2458-10-433Published: 23 July 2010
Antiretroviral therapy (ART) has dramatically reduced morbidity and mortality among people with HIV infection; however, mortality after the start of ART is high in resource-limited settings. We investigated risk factors for mortality among adults starting ART in a multi-clinic community programme in South Africa.
Cohort of adults starting ART at 27 clinics between February 2005 and June 2006, followed to 31st March 2007. Kaplan-Meier survival estimates were used to describe overall mortality. Shared frailty Cox regression was used to identify baseline risk factors for early mortality.
Among 1350 participants (median age 35.5 years, 60% female, median CD4 count 83/μL [interquartile range (27 - 147)], median follow-up 13.4 months), there were 185 deaths, overall mortality rate 13/100 pyrs; for 0-3, 3-9 and >9 months from ART start mortality rates were 24, 13 and 6/100 pyrs respectively. 43% of the deaths were in the first 3 months of treatment. Risk factors for mortality in univariable analysis were baseline CD4 count, viral load, haemoglobin and body mass index, in multivariable analysis adjusting for age and gender, only CD4 count and haemoglobin remained independently associated with proportional hazards not being satisfied for haemoglobin. Adjusted hazard ratios (aHR) for participants with haemoglobin <8, 8.1-9.9, >11.9(f)/12.9 (m) g/mL were 4.99, 3.05 and 0.12 respectively comparing to 10-11.9 (f)/12.9 (m)g/mL in the first 3 months of ART. aHRs for CD4 counts were 0.40, 0.38 and 0.34 for 50-99, 100-200 and >200/μL comparing to <50/μL.
The high mortality rate in the first 3 months underlines the need for earlier HIV diagnosis so that ART can be initiated earlier. Low haemoglobin and low CD4 count are both strong predictors of mortality, and could be used to identify individuals at high risk who might benefit from intensive case management.