Excess mortality during hospital stays among patients with recorded diabetes compared with those without diabetes.

AIM: To assess the additional mortality during hospital admissions among patients with recorded diabetes and identify the extent of variation in English provider trusts. METHODS: Inpatient admissions to all English hospitals between April 2010 and March 2012 were extracted from Hospital Episode Statistics. Binary logistic regression was used to standardize for age, sex, deprivation, method and reason for admission, co-morbidities and type of trust. Trust level standardized mortality ratios for inpatients with recorded diabetes were compared with those without recorded diabetes and with published measures of hospital mortality. RESULTS: Of the 10 169 003 hospital admissions analysed, 11.2% had recorded diabetes, but 21.5% of inpatient deaths occurred in this group. After adjustment for case mix, hospital admissions in patients with recorded diabetes had a 6.4% greater risk of dying (2052 more deaths over 2 years) than would be expected compared with similar patients without recorded diabetes. The additional risk of death was significantly greater in smaller trusts. The highest additional risk of death was found in hospital admissions of younger female patients admitted electively. At provider trust level, 37.4% of variation in adjusted mortality in patients with recorded diabetes was explained by the mortality in those without recorded diabetes. CONCLUSION: A diagnosis of diabetes has an adverse impact on hospital mortality that cannot be explained by usual case-mix adjustments, and the additional risk of dying is greatest among hospital admissions that would normally have a low risk of death. This implies that diabetes may override the usual risk factors for hospital mortality.