Variation among 10 pediatric hospitals: Sepsis evaluations for infants with bronchiolitis

Randomly selected inpatients with lower respiratory tract infections were s elected from April 1, 1995, to September 30, 1996, from 10 pediatric hospit als (n = 804). Those I 90 days of age with bronchiolitis (ICD-9 466.1, n = 303) are included. Medical records were abstracted. Pediatric Comprehensive Severity Index was used for severity scoring. Sepsis evaluation eras defin ed as any culture of blood, urine, or cerebrospinal fluid, or parenteral an tibiotic. Growth of any bacterial pathogen defined a serious bacterial infe ction (SBI). Rate of sepsis evaluations among sites (13% to 84%) was signif icantly different; mean age (49 days) and severity were not different. Inte nsive care stay (PICU, 22% to 87%), average length of stay (ALOS, 3-9 days) , and mean total costs ($3,490-$16,147) were significantly different among hospitals. Logistic regression predicting sepsis evaluation showed signific ant predictor variables to be: age, severity, and PICU stay (Odds Ratio [OR ] = 3.3). After controlling for these variables, significant variation due to site (OR by site ranged from 0.1 to 4.6) was observed. Total costs were predicted by severity, PICU stay, and sepsis evaluation. There were four in fants with SBI (1.3%), all positive for Respiratory Syncytial Virus (RSV). Infants mere similar among 10 sites with respect to age and severity; there was a significant difference among sites for sepsis evaluation, ALOS, and costs, after controlling for age, severity, and PICU stay. Risk of SBI was low. Unwarranted Variation should be addressed and reduced.