Prediction rule for etiology of vague abdominal pain in the emergency room: utility for imaging triage

Abstract

Rationale and objectives: To determine the predictive value of clinical parameters in patients with nonspecific abdominal pain undergoing computed tomography (CT) evaluation of the abdomen and pelvis in the emergency room (ER).

Materials and methods: A cross-sectional study of a total of 164 sequential abdominal CT exams of the abdomen and pelvis during a 4 month period for nonspecific abdominal pain in the ER setting identified 100 abnormal scans in 164 patients (61 men, 103 women) of average age 46 years (range 4-97). Patient demographic characteristics (age, sex, temperature, white blood cell (WBC) count, and presence of peritoneal signs) were recorded at the time of CT examination. Results of the CT studies were correlated with the clinical data and discharge diagnosis to assess their positive predictive value using ordinal logistic regression.

Results: There were 17 cases of appendicitis, 9 cases of diverticulitis, 3 neoplasms, 3 abdominal abscesses, 2 pancreatitis, 2 duodenitis, 5 with fluid collections, 1 buttocks abscess, and 1 epiploic appendagitis were diagnosed with CT, 57 patients had unrelated findings on CT (common but not usually associated with vague pain). A diagnosis of appendicitis correlated with; elevated WBC count (>11.5) ( = 0.002), male sex ( = 0.001), and younger age (<25 years old) ( = 0.002). A positive CT correlated with an elevated WBC >11.5 (OR, 7.7; 95% CI, 3.3-18). The presence of peritoneal signs and fever did not correlate with a positive CT finding and diverticulitis had no predictive variables. Alternative diagnoses were correlated with female sex ( = 0.014). The combination of; age, sex, and WBC count allowed for a prediction rule with Area under the receiver operator curve of 0.92 to be generated.

Conclusion: An elevated white blood cell (WBC) count is strong evidence of the presence of an inflammatory process. Alternative diagnostic considerations should be entertained in the context of a normal WBC count without strong clinical suspicion, particularly in women. The use of these factors alone allowed the construction of a prediction rule that can be used for CT protocol optimization.