Background: A multidimensional pediatric AIDS severity score (PASS) has been developed for severity adjustment and as a predictive model for mortality in a pediatric HIV-infected population. While the prognostic value of PASS is relevant in the US setting, there is a need to develop a simpler model of PASS for use in resource-limited settings where CD4% values and HIV RNA levels may not be available to assess prognosis and guide treatment decisions.
Methods: A Simple PASS model was developed including baseline weight percentile, WHO stage, symptoms, a general health rating, total lymphocyte count, packed-cell volume, and albumin measures from 1178 perinatally HIV-infected children enrolled into a prospective cohort study (PACTG 219). This prognostic model was then validated among 952 perinatally HIV-infected children enrolled in other PACTG research studies at the same sites. Survival estimates and Hazard Ratios (HR) were obtained using the Kaplan-Meier method and proportional hazards models, respectively. The predictive ability of the models was determined using Harrell's "C" statistic.
Results: Of the clinical measures and assays considered in this study, weight percentile, WHO stage, symptomatology, general health rating, total lymphocyte count, packed-cell volume, and albumin were found to be predictive of mortality. The simple PASS model including only the simple clinical measures and assays was found to be predictive of mortality (C statistic = 0.852). Its discriminative ability for mortality was comparable to a model consisting of the Simple PASS plus CD4% (C statistic = 0.871).
Conclusion: The Simple PASS scoring system provides a reasonable alternative to CD4% values and HIV viral-load levels to assess prognosis and guide decisions about antiretroviral therapy initiation in resource-limited settings.