Multivariate analysis showing prognostic factors with significant and independent influence on mortality at three years following hospital admission with upper GI bleeding.

Notes*Odds ratios for sex are adjusted for age group, odds ratios for social deprivation, inpatient bleed, aetiology/main diagnoses, and major c-morbidities are adjusted for age group and sex†Ref = Reference category‡When applying the Bonferroni correction to adjust for the multiple comparisons (p = 0.0031), all terms in this model remained statistically significant with the exception of gastric varices (which was based on only 30 cases). None of the two-way interaction effects between these terms were significant when applying the Bonferroni correction (p = 0.0004) or at the p = 0.001 level.