Utility of the white blood cell count to predict the need for admission in adults presenting to the emergency department with undifferentiated abdominal pain

Abstract

Study objectives: We determine the predictive value of the WBC count in adults presenting to the emergency department (ED) with undifferentiated abdominal pain (UAP). Methods: We conducted a prospective cohort study of adults presenting to an urban ED with UAP during 28 consecutive days. Patients underwent scripted interview and focused physical examination at the ED visit. UAP was defined as nontraumatic abdominal pain of at least an hour's duration. Exclusions included pain primarily in the flank, pregnancy, vaginal bleeding or discharge, frequency, dysuria or foul-smelling urine, and upper or lower gastrointestinal bleed. Physicians were asked for their clinical impression about whether the patient disposition solely according to medical history and physical examination should be (1) treat and release; (2) observe tests and probably discharge; (3) observe tests and probably admit; or (4) admit to the hospital. Patients were followed up for 3 months to determine the most reasonable final diagnosis. WBC count was categorized as low ( 15,000 cells/mL). Neutrophil count was categorized as low ( 80%). Statistics are presented as simple percentages, medians, interquartile ranges, or χ 2 . Ninety-five percent odds ratios (ORs) are given for an α equal to 0.05. Results: Two hundred twenty-four patients presented during the study period with UAP. Ninety-two (41%) had in-person interviews, with physician real-time clinical impression of presumptive diagnosis and need for admission. All patients had detailed study forms filled out. Using the final diagnosis at follow-up as the criterion standard for need to admit, high WBC count was not significantly better than low WBC count (OR 1.83 [95% confidence interval (CI) 0.77 to 4.36]), but very high WBC did significantly predict need for admission (OR 6.85 [95% CI 1.32 to 35.65]). High neutrophil count was significant when high (OR 4.03 [95% CI 1.04 to 15.7]) but not significant when very high (OR 3.10 [95% CI 0.75 to 12.8])in predicting need for admission. Low WBC count with high neutrophil count did not predict need for admission (OR 0.75 [95% CI 0.23 to 2.41]). Neutrophil count, but not WBC, did appear to affect decision to admit (OR 3.02 [95% CI 1.20 to 7.60] for high and OR 2.90 [95% CI 1.10 to 7.61] for very high neutrophil count). Conclusion: Our data suggest that WBC count and neutrophil count may be helpful in addition to clinical impression based on history and physical examination alone in determining need for admission in adults presenting to the ED with UAP. Further investigations need to determine whether these results hold up in other patient cohorts.