Predicting poor outcome from acute upper gastrointestinal hemorrhage

Uncertainty about the outcome of acute upper gastrointestinal bleeding often results in a longer-than-necessary hospital stay.

We derived and internally validated clinical prediction rules (CPRs) to predict outcome from upper gastrointestinal bleeding. This multisite, prospective cohort study involved consecutive patients admitted for acute upper gastrointestinal bleeding. Multivariate logistic regression was used to derive CPRs on two thirds of the cohort (derivation set) that predicted bleeding-specific outcomes (rebleeding, need for urgent surgery, or hospital death [poor outcome 1]) and bleeding-specific outcomes plus new or worsening comorbidity (poor outcome 2). Both CPRs were then tested on the remaining third of the cohort (validation set).

A total of 391 individuals (99% men; mean age, 63.4 years) were enrolled, of which 4.6% rebled and 3.1% died. Independent predictors of poor outcome 1 were APACHE (Acute Physiology and Chronic Health Evaluation) II score of 11 or greater, esophageal varices, and stigmata of recent hemorrhage. Predictors of poor outcome 2 were these 3 factors plus unstable comorbidity on admission. Of patients with no risk factors, only 1 (1.1%) of 92 experienced poor outcome 1 and only 6 (6.2%) of 97 experienced poor outcome 2. Risks in the validation set were comparable. The CPRs identified 37.8% and 32.2% of patients in the derivation and validation sets, respectively, who were eligible for a shorter hospital stay.

Patients admitted with acute upper gastrointestinal bleeding were unlikely to have a poor outcome if these risk factors were absent. These CPRs might make hospital management more efficient by identifying low-risk patients for whom early hospital discharge is possible.