OBJECTIVE: To assess the importance of baseline characteristics including medical history, indicators of current disease status, therapeutic drug use, in vitro drug susceptibility, immune status and mycobacterial load on bacteriologic response and survival in HIV-positive patients with Mycobacterium avium complex (MAC) bacteremia.
DESIGN: An observational substudy of an open-label randomized controlled trial of two alternative therapeutic regimens for MAC.
SETTING: Twenty-four hospital-based HIV clinics in 16 Canadian cities.
MAIN OUTCOME MEASURES: The main outcome measures were survival and bacteriologic response, defined by consecutive negative blood cultures for MAC at least 2 weeks apart within 16 weeks of study entry.
RESULTS: Prior AIDS diagnosis, low Karnofsky score, active unstable AIDS-related conditions, absence of antiretroviral therapy and absence of Pneumocystis carinii pneumonia prophylaxis were associated with shorter survival by univariate regression using the proportional hazards model. On multivariate analysis, antiretroviral therapy was not an independent predictor of mortality, and previous rifabutin prophylaxis was independently associated with poor survival outcomes, a result consistent across study treatment. Using a logistic regression model, baseline quantitative mycobacterial load [relative odds of clearing, 1.97 for a decrease of 1 log10 colony forming count; 95% confidence interval (CI), 1.36-2.87; P < 0.001] and Karnofsky score were the only statistically significant univariate predictors of clearance, although previous prophylaxis with rifabutin was also a significant predictor in a multivariate model (relative odds of clearing, 0.39; 95% CI, 0.17-0.88; P < 0.05).
CONCLUSIONS: This study indicates that although the level of MAC bacteremia is an important predictor of clearance, it is not associated with survival.