Pediatric barium enema examination: Optimizing patient selection with univariate and multivariate analyses

We prospectively evaluated a total of nineteen symptoms, signs, and laboratory findings in 471 of 557 consecutive pediatric patients (from newborn to age 17) referred for barium enema examinations, to determine predictors of an abnormal study. A univariate analysis was performed, and a logistic regression model was developed. The most frequent indicators for the barium enema examinations were abdominal pain (48%), constipation (27%) and tenderness (25%). Twenty-two percent of the examination were abnormal, and the most common diagnoses were intussusception (n=22), appendicitis (n=17), infectious colitis (n=15), and Hirschsprung disease (n=14). The indicators that were most helpful to predict a barium enema abnormality were abdominal mass, leukocytosis, guaiac-positive stools, diarrhea, anemia, tenderness, and age less than 1 year. If barium enema examinations were performed only when at least one of the predictive indicators was present, 29% of examinations would be eliminated, and 4.8% of patients with detectable disease would be missed. The data indicate that identification of certain clinical variables can provide an effective initial strategy for selecting patients to undergo barium enema examinations.