Prognostic markers of short-term mortality in AIDS-associated Pneumocystis carinii pneumonia

Background: Since 1990, corticosteroids have been recommended as adjunctive therapy for patients with AIDS-associated Pneumocystis carinii pneumonia (PCP) and respiratory failure. We hypothesized that the natural course of AIDS-associated PCP has changed in the era of adjunctive corticosteroid therapy. Objective: To study variables obtained on hospital admission for possible prognostic value of short-term (3-month) outcome of PCP. Design and patients: Prospective observational study of 176 consecutive HIV-l-infected individuals with PCP between 1990 and 1999. Method: Cox proportional-hazards regression models. Results: Univariate analysis showed that age, one or more prior episodes of PCP, use of antimicrobial therapy other than trimethoprim-sulfamethoxazole (TMP-SMZ), use of PCP prophylaxis at diagnosis, and culture of cytomegalovirus (CMV) in BAL predicted progression to death within 3 months. After adjustment, age (relative risk [RR], 4.1; 95% confidence interval [CI], 1.8 to 9.3), initial antimicrobial therapy other than TMP-SMZ (RR, 3.1; 95% CI, 1.2 to 8.5), use of PCP prophylaxis (RR, 5.6; 95% CI, 2.2 to 14.4), and culture of CMV in BAL fluid (RR, 2.7; 95% CI, 1.3 to 5.9) remained independent predictors of a poor outcome. In contrast, neither Po, nor serum lactate dehydrogenase, which in earlier studies were identified as prognostic markers, were predictors of mortality. Conclusion: Age, initial anti-PCP therapy, use of PCP prophylaxis, and BAL CMV status may be useful predictors of outcome of PCP in patients treated in the era of adjunctive corticosteroid therapy.