ObjectiveIntensive care resources are both scarce and costly in South Africa. We set out to ascertain whether human immunodeficiency virus (HIV) status affects the spectrum of disease and the prognosis in children treated in pediatric intensive care units (ICUs).DesignA prospective, case-controlled study.SettingThe pediatric ICU at King Edward VIII Hospital, Durban, South Africa.PatientsHIV-positive cases were classified as ``acquired immunodeficiency syndrome,'' ``symptomatic HIV,'' and ``HIV-unrelated diseases.'' Controls were matched for age, race, gender, severity of disease, and admitting diagnosis.Measurements and Main ResultsThe clinician was blinded to the serostatus of patients. Informed consent and Ethics Committee approval were obtained for HIV testing and the study. Outcome measures were the duration of intermittent positive-pressure ventilation and ICU stay, maximum ventilatory requirements, infectious complications, and mortality. There were 11 cases of acquired immune deficiency syndrome (AIDS), 24 cases of symptomatic HIV, and 13 cases of HIV-unrelated diseases. Mortality rates for the three groups were as follows: 100% in patients with AIDS compared with 55% in their controls (p less than .01); 38% in patients with symptomatic HIV and 46% in controls; and 54% in patients with HIV-unrelated disease as against 46% in controls. Cytomegalovirus and Pneumocystis carinii infections were significantly increased (p less than .01, p less than .002, respectively) in patients with AIDS as compared with controls. Outcome measures were similar in those patients with symptomatic HIV and those patients with HIV-unrelated diseases, as compared with their respective controls.ConclusionsOutcome in children with AIDS admitted to the pediatric ICU was significantly worse than the outcome in HIV-uninfected children. However, the number of AIDS patients studied was small. HIV antibody-positive children without AIDS do as well as uninfected controls. Ethics committees in resourceconstrained countries may be assisted by such data when making difficult decisions on the admissions of HIV-infected patients to pediatric ICU facilities.(Crit Care Med 1996; 24:963-967)