Introduction:
Diabetes mellitus (DM) is associated with increased mortality in hospitalized adults. However, data regarding the impact of DM on long-term mortality after discharge in very old patients is scarce.

Material and methods:
This prospective study assessed 3-year post-discharge mortality and its predictive factors in older medical patients, focusing on possible differences between patients with and without DM. Medical history, chronic medication use, clinical and laboratory characteristics, Charlson Comorbidity Index (CCI), 5-Fried Frailty score (FFS), Clinical Frailty Scale (CFS), Barthel index (BI), and Katz Index were recorded on admission.

Results:
A sample of 815 older adults (46.0% males) with a median age of 83.0 years (IQR: 77.0–88.0) were included in the study. The three-year mortality rate was 54.9% in patients with DM (n=368) and 60.2% in patients without DM (n=447, p=0.13 between groups). In multivariate logistic analysis, nursing home residency, higher CCI, higher CFS, higher FFS, lower BI, the total number of days of hospitalization the past year, and hospital-acquired infections were independently associated with the three-year mortality in both groups. In individuals with DM, lower body mass index (BMI) and elevated urine albumin-to-creatinine ratio (UACR) were identified as additional independent predictors of mortality.

Conclusions:
A high post-discharge mortality rate was observed in very old medical patients. DM was not identified as an independent factor of post-discharge mortality. Assessment of frailty and disability in very old patients is important for predicting long-term post-discharge mortality. Additionally, in patients with DM, evaluating BMI and UACR may aid in better prediction of three-year mortality.