Pneumocystis carinii pneumonia in South African children infected with human immunodeficiency virus

Abstract
Pneumocystis carinii pneumonia (PCP) has been regarded as uncommon in HIV-infected patients in Africa, but diagnostic difficulties and geographic variability may partly account for this. There is little information on the incidence of PCP in HIV-infected children in Africa. To investigate (1) the incidence and associated features of PCP in African HIV-infected children and (2) the usefulness of sputum induction and nasopharyngeal aspirates (NPAs) for diagnosis of PCP. HIV-infected children hospitalized with pneumonia were prospectively enrolled in a 1-year study in South Africa. History, examination, chest radiology and blood tests (including HIV testing) were performed. Sputum induction (5% NaCl nebulization) or nondirected bronchoalveolar lavage in intubated patients was performed for P. carinii identification using immunofluorescence and silver stain; immunofluorescence was also done on nasopharyngeal aspirates (NPAs). Of 151 HIV-infected children [47% female; median age, 9 (range, 3 to 23) months], 87 had been previously diagnosed with HIV whereas 64 (42.4%) were found to be HIV-positive at the time of admission. PCP occurred in 15 children (9.9%; 95% confidence interval, 5.9 to 15.5) and was the AIDS-defining infection in 13 of 64 (20.3%; 95% confidence interval, 11.8 to 31.5). Only 1 of 59 children receiving prophylaxis (1.7%) developed PCP compared with 14 of 92 (15.2%) not taking prophylaxis [relative risk, 0.11 (0.02 to 0.82), P = 0.007]. PCP-infected children were younger [3 (range, 3 to 4) vs. 10 (range, 4 to 24) months, P < 0.001] and presented with more severe pulmonary disease as indicated by a higher respiratory rate [63 (range, 60 to 73) vs. 50, (range, 40 to 60) P < 0.001], heart rate [160 (range, 136–180) vs. 140 (range, 120–152) P = 0.025] and a greater incidence of cyanosis (53% vs. 26%, P = 0.025). Clinical signs of HIV infection, occurring in 96% of children, were equally prevalent in both groups. High serum lactate dehydrogenase was the only laboratory investigation that distinguished PCP-infected from uninfected children [626 (range, 450 to 1098) vs. 307 (range, 243 to 465) units/l], P < 0.001. No radiologic features were found to be diagnostic of PCP. P. carinii was identified in 9 sputa and 6 bronchoalveolar lavage specimens, but all corresponding NPAs were negative. Seven of 15 (47%) children with PCP died while hospitalized compared with 24 of 136 (18%) without PCP [relative risk, 1.21 (range, 0.99 to 1.47), P = 0.008]. PCP is an important pathogen in HIV-infected infants in South Africa and is associated with a high mortality. Induced sputum is effective for obtaining lower respiratory tract secretions for diagnosis of PCP but an NPA is not useful.