Background: A severity staging system predictive of mortality for perinatally HIV-infected children is needed for clinical and research purposes.
Methods: A pediatric AIDS severity score (PASS) was developed using baseline sociodemographic, clinical, immunologic, and functional measures obtained from 786 perinatally HIV-infected children enrolled into a prospective cohort study (PACTG 219) in the pre-highly active antiretroviral therapy (HAART) era (pre-1996). PASS was then validated among 392 perinatally HIV-infected children randomly sampled from the original source population (n = 1178). Survival estimates and hazard ratios (HRs) were obtained using the Kaplan-Meier method and proportional hazards models, respectively. The most predictive models were determined using Harrell's "C" statistic.
Results: Overall survival was 95% and 90% at 1 and 2 years of follow-up, respectively. The most comprehensive model for predicting mortality, termed the "Full" PASS, included CD4% <15 (HR = 3.9), CDC category C (HR = 2.6), BMI <10% (HR = 2.4), a low (<70) neuropsychological score (HR = 2.6), a general health rating <5 (HR = 2.4), and an elevated symptoms score (HR = 1.9). These determinants were highly predictive of mortality (C statistic = 0.841).
Conclusions: PASS will be helpful in assessing the effectiveness of ART among children with HIV infection, particularly when randomized clinical trials are not possible due to ethical and feasibility concerns.