Frailty and type 2 diabetes (T2D) is fast becoming a clinical challenge especially in the elderly population. Despite clear guidance on the need to evaluate individualised HbA1c targets in connection with frailty status with mortality risk, there is a lack of formal evidence on these associations. To guide healthcare professionals in setting appropriate HbA1c targets, this study aimed to investigate the association between mortality in connection with electronic frailty index (eFI) and HbA1c, as well as exploring the clinical utility of eFI as a tool to individualise HbA1c targets in frail patients with T2D.

Methods: We included patients with T2D aged 65+ within UK primary care electronic database (the Clinical Practice Research Datalink) and Hospital Episode Statistics from 2014-19. All-cause mortality (ACM) risk associated with 4 frailty groups (non-frail; mild; moderate and severe) (defined by electronic frailty index scores, eFI) and seven HbA1c groups at 1.0% intervals were assessed using multivariate Cox regression analysis adjusted for demographic and clinical characteristics. The reference group was defined as “non-frail” individuals with HbA1c 6.5-7.5%.

Results: 179,723 individuals (mean±SD; age 74.8±7.6years, diabetes duration 7.3±6.1years, HbA1c 55.7±15.8mmol/mol) were followed for mean 1.7 years. Within all frailty groups, the risk of ACM was greatest at HbA1c<5.5%, with highest risk observed in moderate (HR 3.34; 95%CI 2.80, 3.99) and severe frailty (HR 3.27; 95%CI 2.73, 3.91) groups. In all groups, ACM risk increased with higher HbA1c levels compared to HbA1c 6.5-7.5% (a “U-shaped” curve).

Conclusion: Both high and low HbA1c levels increase the risk of ACM for people 65+ with T2D and frailty in a “U-shaped” association between HbA1c and ACM. This study supports eFI as a useful tool to facilitate HbA1c target setting in older patients with T2D and to manage ACM risk.


T.S.J.Crabtree: Speaker's Bureau; Abbott Diabetes, Novo Nordisk, Lilly, Sanofi, Insulet Corporation. Y.Vinogradova: None. R.Aldafas: None. J.Gordon: Employee; Health Economics and Outcomes Research Ltd. I.R.Idris: None.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at