A test of a linear discriminant for identifying low-risk abdominal pain

This study tests a previously published decision rule for identifying nonspecific abdominal pain (NSAP). The rule, developed for ambulatory male patients in two Veterans Administration (VA) facilities and a prepaid group practice, was studied in an additional 110 VA patients and in 77 patients (predominantly female) from a solo private practice. The group of 58 patients (33%) classified as "low-risk" rarely had abnormal laboratory tests or radiographs, except for upper gastrointestinal series; 15 of these patients had potentially serious disease. Peptic ulcer was the specific diagnosis most often misclassified as NSAP. The accuracy of the rule in our population is similar to the accuracy of the judgment of experienced clinicians.