Diagnostic performance of Baveno IV criteria in cirrhotic patients with upper gastrointestinal bleeding: analysis of the F7 liver-1288 study population.

ABSTRACT

BACKGROUND & AIMS: The definition of failure to control bleeding agreed upon at the Baveno IV consensus meeting, included the Adjusted Blood Requirement Index [ABRI number of blood units/(final-initial hematocrit+0.01)]. ABRI ≥0.75 denotes failure. However, timing for hematocrit measurements was not defined. The aims of this study were (1) to assess the Baveno IV criteria performance to classify treatment success or failure to control bleeding at 5 days, (2) to determine the appropriate timing for hematocrit. METHODS: Two hundred and forty-two cirrhotic patients with gastrointestinal bleeding were independently classified by three clinical experts according to the Baveno IV criteria, by analysis of the database of a randomized trial. ABRI was calculated by using the closest hematocrit to the 5 day time point from the first trial product administration (ABRI-1) or after the latest transfusion within the 5-day period (ABRI-2). The gold standard for success/failure for 5-day control of bleeding was the clinical judgment of the three independent observers based on all the clinical and follow-up data. RESULTS: Inter-observer agreement for the final outcome assessment was 0.82 and a final consensus was obtained in 236/242 patients. Inter-observer agreement on patient classification with Baveno IV criteria was 0.70 with ABRI-1 and 0.84 with ABRI-2. c-statistics for correct patients classification were 0.86 for ABRI-1, 0.84 for ABRI-2, and 0.88 for Baveno IV criteria without ABRI. ABRI-1 caused misclassification of 27 patients and ABRI-2 of 39. CONCLUSIONS: Baveno IV criteria are accurate to assess outcome of patients with variceal bleeding. There is a substantial observer variability linked to timing of hematocrits for ABRI calculation. With the current definition ABRI does not add to the performance of the other criteria.