Abstract
BACKGROUND: Hospitalized patients with diabetes have experienced a disproportionate reduction in mortality over the past decade. OBJECTIVE: To examine whether this differential decrease affected all patients with diabetes, and to identify explanatory factors. DESIGN: Serial, cross-sectional observational study. SETTING: Academic medical center. PATIENTS: All adult, nonobstetric patients with an inpatient discharge between January 1, 2000 and December 31, 2010. MEASUREMENT: We assessed in-hospital mortality; inpatient glycemic control (percentage of hospital days with glucose below 70, above 299, and between 70 and 179 mg/dL, and standard deviation of glucose measurements), and outpatient glycemic control (hemoglobin A1c). RESULTS: We analyzed 322,938 admissions, including 76,758 (23.8%) with diabetes. Among 54,645 intensive care unit (ICU) admissions, there was a 7.8% relative reduction in the odds of mortality in each successive year for patients with diabetes, adjusted for age, race, payer, length of stay, discharge diagnosis, comorbidities, and service (odds ratio [OR]: 0.923, 95% confidence interval [CI]: 0.906-0.940). This was significantly greater than the 2.6% yearly reduction for those without diabetes (OR: 0.974, 95% CI: 0.963-0.985; P < 0.001 for interaction). In contrast, the greater decrease in mortality among non-ICU patients with diabetes did not reach significance. Results were similar among medical and surgical patients. Among ICU patients with diabetes, the significant decline in mortality persisted after adjustment for inpatient and outpatient glucose control (OR: 0.953, 95% CI: 0.914-0.994). CONCLUSIONS: Patients with diabetes in the ICU have experienced a disproportionate reduction in mortality that is not explained by glucose control. Potential explanations include improved cardiovascular risk management or advances in therapies for diseases commonly affecting patients with diabetes.