Infections in children with cancer: a continued need for the comprehensive physical examination.

Few studies have addressed the influence of profound myelosuppressive therapy in contemporary protocols on infectious morbidity in pediatric oncology patients. This study evaluates the types of infections and the methods used to diagnose infection in patients enrolled in current Children's Cancer Group (CCG) protocols. Data were collected on patients enrolled in CCG protocols from January 1, 1992, through December 31, 1995. Of the 155 protocol patients, 102 were completely evaluated and had data collected through August 1, 1996. Patients were divided into two diagnosis groups: leukemia/lymphoma (N = 51) and solid tumor (N = 51). Eighty-five (83%) patients had documented infections and 17 (17%) did not. Overall, 96 (94%) patients had in-dwelling central venous catheters. Twelve categories of infection were identified. Data were analyzed for age, gender, diagnosis, neutropenia, organism, and disease state (primary active, recurrent active, primary remission, and secondary remission). Statistical comparisons were made only on rates, whereas descriptive comparisons were given for numbers of infections and organisms. The infection rates for patients with active disease were 1.01 and 1.15 per 100 patient days (primary versus recurrent) and 0.59 and 0.38 per 100 patient days for patients with disease in remission (primary versus secondary). Diagnosis-group infection rates were 0.66 and 0.68 per 100 patient days for patients with solid tumors and leukemia/lymphoma, respectively. Three hundred thirty infections, including 19 polymicrobial infections, were recorded. The three most common types of infection were otitis media, septicemia, and urinary tract infection. More infections were associated with an age at diagnosis of less than 3 years, a leukemia/lymphoma diagnosis in remission, and an absolute neutrophil count >500 cells/microL. One hundred ninety-four organisms were isolated from 330 infections. Gram-positive organisms (n = 74) such as coagulase-negative Staphylococci (n = 38) predominated over gram-negative organisms (n = 63) for all infectious categories. Specifically, gram-positive organisms (n = 39) were isolated more often from blood cultures than were gram-negative organisms (n = 27). The overall mortality for patients was 36%. Seven of the 37 (19%) patient deaths were attributed to infection. These patients predominantly were girls with neutropenia and leukemia/ lymphoma in active disease who died of gram-negative sepsis. Infections with gram-positive organisms continue to be major causes of morbidity in pediatric oncology patients receiving contemporary CCG protocols. However, infection-related mortality, especially with gram-negative organisms, occurs less frequently than does malignancy-related mortality. Common childhood infections (such as otitis media) seem equally as prevalent as bacteremia in pediatric oncology patients. Thus, a comprehensive physical examination is as imperative as the microbiologic evaluation in diagnosing infection in this patient population.