Background: To analyze the assistance requirements generated in a hospital emergency unit by the upper gastrointestinal bleeding (UGIB) syndrome, and to assess some strategies intended to improve its syndromic diagnosis prior to admission.
Patients and methods: Prospective study of a cohort including 1,029 consecutive cases assisted because of presumed hematemesis or melenas. At the emergency room UGIB was diagnosed or ruled out according to criteria based on examination of vomits and stools. Those cases not classified were observed before establishing the preadmission diagnosis. Sensitivity (Se), specificity (Sp) and likelihood ratio for a positive or negative test were estimated in each step of the process. The efficacy of the proposed criteria for diagnosis was compared with those accepted by the World Organization of Gastroenterology (WOG). The diagnosis at hospital discharge was used a reference pattern.
Results: A final diagnosis of UGIB was established in 718 cases (59%). The emergency room criteria were applied to 848 out of the 1,029 patients assisted, bleeding being discarded in 216 cases (Se = 0.43; Sp = 0.99) and estimated as present in 632 (Se = 0.79%; Sp = 0.87%). By combining these results with those obtained after observation of the remaining 361 patients, the preadmission diagnosis showed a sensitivity of 0.99% and a specificity of 0.86%. Should the WOG criteria were adopted, the same process was equally sensitive but less specific (Sp = 0.55%).
Conclusions: In a high proportion of patients assisted in a hospital for a suspected UGIB, this diagnosis is not confirmed. The examination of vomits and stools by nursing or medical staff, followed if needed by an observation period, improves the accuracy of the syndromic diagnosis prior to admission and may avoid an elevated number of unnecessary hospital staying.