Pediatric AIDS prognosis using somatic growth velocity

Abstract

To describe the natural history of somatic growth in HIV infection by constructing age-specific growth velocity norms and to assess specific prognostic information available using these norms. Observations on 1338 HIV-infected children aged 3 months to 15 years who participated in one of four US clinical trials of pediatric anti-HIV therapies were pooled; baseline growth velocity data were obtained using the first 6 months of observation for each child. Distributions of physical growth velocities in HIV-infected children in the Pediatric AIDS Clinical Trials Group were computed. Statistical smoothing of growth histories was employed to derive velocity estimates, and quantile regression analysis of growth velocities was performed to allow comparisons of growth rates in age- and gender-heterogeneous cohorts in the context of HIV infection. The quantile regressions provide corrected z-scores for growth velocity that appropriately compare HIV-infected children with one another for the purpose of distinguishing more from less favorable prognoses. Consistent deficits in growth velocity amongst HIV-infected children were revealed when compared with the Fels Institute velocity standards. Approximately 33% of height (and 20% of weight) age- and sex-corrected velocity measurements obtained in the first 6 months of clinical trial participation lay beneath the corresponding third percentiles of the Fels reference distributions, which are commonly regarded as critical indicators of growth failure. Proportional hazards regression tests indicated that both weight and height velocity contributed significant information on the risk of death among children with AIDS after adjusting for antiretroviral therapy received, CD4 cell counts, and age at trial enrollment. Comparing subjects who differ in initial weight velocity by one age- and sex-corrected SD, the relative hazard of death was 0.63 (95% confidence interval, 0.55-0.72; P < or = 0.0001) in favor of the child with greater weight velocity, controlling for antiretroviral therapy received, age and CD4 cell count at baseline. The analogous hazard ratio for height velocity was 0.68 (95% confidence interval, 0.57-0.79; P < or = 0.0001). Suitably normalized growth velocities are informative and inexpensive criteria for pediatric AIDS prognosis; the growth velocity distributions presented will be useful for comparing growth effects of new therapeutic strategies to those of single and combination antiretrovirals employed for maintenance of pediatric HIV infection in the mid-1990s.