Predicting death from HIV/AIDS: a case-control study from Florida public HIV/AIDS clinics

After markedly decreasing for 3 years, HIV/AIDS mortality declined only slightly in 1999.

The authors conducted a case-control study in four Florida urban public health HIV clinics to evaluate modifiable factors associated with HIV/AIDS mortality in a non-research setting. Structured chart review was conducted for 120 case-patients who died in 1999 and for 240 randomly selected control-patients. Risk factors associated with death in univariate analysis were entered into three conceptually related, matched logistic regression models.

In the final multivariate model, homelessness (adjusted odds ratio [AOR], 9.98; 95% confidence interval [CI], 2.34-42.5), Medicaid insurance (AOR, 3.10; 95% CI, 1.43-6.74), having a documented adherence problem (AOR, 3.50; 95% CI, 1.64-7.47), injection drug use (AOR, 2.46; 95% CI, 1.11-5.43), non-specific liver failure (AOR, 76.9; 95% CI, 6.79-870.9), interrupted highly active antiretroviral therapy (HAART) secondary to side effects (AOR, 4.00; 95% CI, 1.46-10.9), and not receiving HAART (AOR, 2.62; 95% CI, 1.03-6.68) were independent predictors of mortality.

In addition to medical and clinical indicators, several sociobehavioral-demographic factors remained important throughout the multivariate analysis. Improvement in care should include a focus on social circumstances of infected people. Special attention to the homeless, those with adherence problems, and those with liver disease is clearly indicated.