Background: Mental illness (MI) and substance abuse (SA) are common in HIV-positive patients. MI/SA consistently predict poorer antiretroviral adherence, suggesting that affected patients should be at higher risk of poor virologic and immunologic response to highly active antiretroviral therapy (HAART).
Participants: 198 HAART-naive patients initiated HAART at an academic medical center serving a heterogeneous population.
Methods: Participants were assigned a predicted probability from 0 to 1 of having each of the following: (1) any mood, anxiety, or substance use disorder; (2) clinically relevant depression; (3) alcohol abuse/dependence; and (4) drug abuse/dependence. Probabilities were based on responses to questions on an MI/SA screening instrument (Substance Abuse and Mental Illness Symptoms Screener [SAMISS]) and other clinical and sociodemographic characteristics and were derived using predictive logistic regression modeling from a separate validation study of the SAMISS compared with Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition diagnoses. Using survival analysis techniques, we assessed baseline predicted probability of psychiatric illness as a predictor of time from HAART initiation to virologic suppression (first viral load [VL] <400 copies/mL), from HAART initiation to overall virologic failure (first VL >or=400 copies/mL after suppression, time set to 0 for patients never achieving suppression), from virologic suppression to virologic rebound (first VL >or=400 copies/mL), and from HAART initiation to immunologic failure (first CD4 cell count lower than baseline).
Results: A higher predicted probability of any psychiatric disorder was associated with a slower rate of virologic suppression (adjusted hazard ratio [aHR] = 0.86 per 25% increment, 95% confidence interval [CI]: 0.75 to 0.98) and a faster rate of overall virologic failure (aHR = 1.22, 95% CI: 1.06 to 1.40). Associations with other outcomes were consistent in direction but not statistically significant. Predicted probability of depression was associated with slower virologic suppression (aHR = 0.79, 95% CI: 0.63 to 0.98), and predicted probabilities of alcohol and drug abuse/dependence was associated with faster overall virologic failure (aHR = 1.37, 95% CI: 1.08 to 1.74 and aHR = 1.18, 95% CI: 1.00 to 1.39, respectively).
Conclusions: These results are consistent with an inferior virologic response to first HAART among patients with concurrent mood, anxiety, and substance use disorders, suggesting a clinical benefit to identification and treatment of psychiatric illness among patients initiating antiretroviral therapy.