BENEFIT OF PRIMARY PROPHYLAXIS BEFORE 18 MONTHS OF AGE IN REDUCING THE INCIDENCE OF PNEUMOCYSTIS-CARINII PNEUMONIA AND EARLY DEATH IN A COHORT OF 112 HUMAN IMMUNODEFICIENCY VIRUS-INFECTED INFANTS

Objective. To determine the effectiveness of primary prophylaxis in pr eventing Pneumocystis carinii pneumonia (PCP) in children with perinat ally acquired human immunodeficiency virus 1 (HIV-1) infection. Method s. We conducted a retrospective analysis of a cohort of infants follow ed from birth at six metropolitan hospitals and one outpatient clinic for pregnant, drug-using women in New York City. Outcomes measured wer e histologically confirmed PCP and/or death. The potential confounding effect of the infant's stage of illness, as determined by CD4 count w as controlled by including all CD4 determinations as time-dependant co variates in a Cox proportional hazards analysis. Cases were censored a t PCP onset, death, loss to follow-up, and 18 months of age. Results. One hundred twelve HIV-infected children were enrolled at birth betwee n 1986 and 1993. Sixty of these were tracked beyond 18 months of age; of the others, 21 died before this age, 4 were considered lost to foll ow-up, and 27 had not reached 18 months of age at the last visit. Only 3 cases (4%) of confirmed PCP occurred among the 70 children who rece ived primary PCF prophylaxis before 18 months of age, compared with 12 cases (28%) among 42 children not receiving PCP prophylaxis at any po int before 18 months of age. The Kaplan-Meier estimated incidence of P CP in the first year among children not receiving prophylaxis was 25% (95% confidence interval [CI], 12 to 39). Using Cox methods, the unadj usted risk of PCP among infants not receiving prophylaxis, relative to those receiving it, was 4.1 (95% CI, 1.1 to 15); the relative risk wa s 4.4 (95% CI, 1.2 to 17) adjusting for the percentage of CD4-positive lymphocytes and 5.1 (95% CI, 1.3 to 20) adjusting for the absolute nu mber of CD4-positive cells. Eight of 26 deaths were caused by PCP, and the likelihood of early death was significantly diminished if PCP pro phylaxis was given (relative risk controlling for absolute CD4 cells, 2.57; 95% CI, 1.1 to 6.1). Conclusions. We report evidence that primar y antimicrobial PCP prophylaxis is highly effective in decreasing the frequency of PCP and early death in infants with perinatal HIV infecti on. These findings support the revised National Pediatric HIV Resource Center and Centers for Disease Control and Prevention guidelines for PCP prophylaxis in children.