ABSTRACT
BACKGROUND: The prognostic value of an elevated international normalised ratio (INR) as part of initial risk stratification in nonvariceal upper gastrointestinal bleeding (NVUGIB) remains poorly characterised. AIM: To assess the usefulness of the initial INR in patients with NVUGIB. METHOD: After a systematic review, we included the presenting INR and other validated prognosticators in multivariable models predicting rebleeding and mortality. Data are reported as odd ratios and 95% confidence intervals. RESULTS: Only two of 769 candidate studies were useful, but reported disparate, highly selected NVUGIB patients with varying threshold initial INR values, yielding conflicting results on predictive ability. The RUGBE cohort included 1869 patients (mean age 66.3 ± 16.8, 38.1% female) with a mean presenting INR of 1.5 ± 1.7. 462 (24.7%) patients had an elevated INR (INR > 2.5 in 7.6%). INR did not add to the prediction of rebleeding. An INR >1.5, in contradistinction, significantly predicted mortality (OR: 1.96; 95% CI: 1.13-3.41). CONCLUSIONS: An elevated INR at initial presentation does not predict rebleeding in NVUGIB. A value 1.5 or greater, however, is associated with increased patient mortality after adjustment for validated prognosticators. The INR appears most useful as proxy of co-morbid burden at the time of initial assessment in NVUGIB.
