Clinical Question: What diagnostic strategy best identifies patients with nonspecific abdominal pain who will require urgent intervention?

Setting: Emergency department

Study Design: Decision rule (development only)

Synopsis: This study was designed to determine which test or combination of tests, performed in conjunction with a history and physical examination, can identify patients with nonspecific abdominal pain who will not require urgent intervention. Nonspecific abdominal pain was defined as abdominal or pelvic pain of no more than seven days’ duration that was unaccompanied by signs of peritonitis, instability, or other clinically obvious need for intervention (e.g., puncture wound). The investigators excluded patients with a primary urogenital complaint. The primary outcome for this study was the need for urgent intervention, defined as the need for surgical, endoscopic, or nonoperative intervention without which the patient might die or experience severe morbidity. Because some patients with nonspecific abdominal pain may have low-grade continuous conditions, the investigators evaluated and followed up with 165 adult patients for six months after initial presentation to make certain that none of these diagnoses subsequently developed.

The authors studied a convenience sample representing approximately 94 percent of the potentially eligible patients. They compared four decision models: (1) unaided history and physical examination; (2) history and physical examination plus laboratory tests; (3) history and physical examination plus laboratory tests and radiography; and (4) history and physical examination plus laboratory tests, radiography, and nonenhanced helical computed tomography (CT). The first three models were fairly insensitive (25 to 56 percent) but reasonably specific (81 to 92 percent). The positive and negative likelihood ratios (2.9 to 3.2 and 0.5 to 0.8, respectively) for these models were not strong enough to significantly alter diagnostic probabilities. The last model (history and physical examination plus laboratory tests, radiography, and nonenhanced helical CT), however, was much stronger. That model correctly identified 92 percent of patients needing urgent intervention and 90 percent of those not needing it (positive likelihood ratio = 9.2; negative likelihood ratio = .09).

Bottom Line: A structured clinical evaluation combined with laboratory analysis, radiography, and noncontrasted helical CT best identifies patients with nonspecific abdominal pain who will need urgent intervention. (Level of Evidence: 3b)