Clinical frailty and long-term mortality in elderly subjects with diabetes.

Elderly subjects are characterized by a high prevalence of diabetes and clinical frailty. This study aimed to examine the predictive role of clinical frailty on long-term mortality in elderly subjects with and without diabetes. The study evaluated mortality after 12-year follow-up in 188 subjects with diabetes and 1,100 subjects without diabetes selected in 1992. Clinical frailty was assessed according to the "Frailty Staging System" and stratified in tertiles. After 12-year follow-up, mortality was 50.5 % in subjects without and 66.5 % in subjects with diabetes (p < 0.001). With increasing frailty, mortality increases from 57.9 to 79.0 % (p for trend <0.01) in subjects without and from 75.9 to 87.0 % in subjects with diabetes (p for trend <0.001). Multivariate analysis shows that both diabetes (hazard ratio = 1.38; 95 % confidence interval = 1.12-1.95; p = 0.02) and frailty score (hazard ratio = 1.58 for each unit of increase; 95 % confidence interval = 1.41-2.35; p = 0.04) are predictive of long-term mortality. Moreover, when Cox regression analysis was performed by selecting sex, frailty increases the risk of long-term mortality for each unit of increase by 14 % (hazard ratio = 1.14; 95 % confidence interval = 1.10-1.18; p < 0.01) in women and by 60 % in men (hazard ratio = 1.60; 95 % confidence interval = 1.21-2.12; p < 0.001) in the absence and by 31 % (Hazard ratio = 1.31, 95 % confidence interval = 1.03-1.85, p = 0.03) in women and by 60 % in men (hazard ratio = 1.99, 95 % confidence interval = 1.75-3.05, p < 0.001) in the presence of diabetes, respectively. We concluded that diabetes predicts long-term mortality in elderly subjects. Moreover, clinical frailty significantly predicts mortality in subjects without and even more in those with diabetes. This phenomenon is particularly evident in men. Thus, clinical frailty may be considered a new prognostic factor to identify subjects with diabetes at high risk of mortality.