166-LB: Targeting Deprescribing in Type 2 Diabetes Mellitus Using a Pragmatic Electronic Health Record Measure of Frailty

Abstract

Background: Guidelines recommend less stringent glycemic goals for older adults with type 2 diabetes mellitus (T2DM) and frailty or comorbidity. However, pragmatic, scalable approaches to identify candidates for de-intensification of T2DM regimens are lacking. Methods: Analysis of electronic health record (EHR) data for patients ≥65 years with T2DM from an accountable care organization as of 11/1/2020. Frailty was determined based on a 54-item electronic Frailty Index (eFI) derived from the EHR. Other data included the level of glycemic control, use of higher-risk medications regimens (active prescription of insulin, sulfonylurea, or combinations of the two), the incidence of emergency department (ED) visits and hospitalizations, and all-cause mortality. Results: Amongst 16973 patients, 53.9% were female, 77.8% white, with a mean age of 75.5 (SD=6.9) years. Based on the eFI, 6218 (36.6%) patients were classified as frail (eFI>0.21). During short-term follow-up (median=116 days), compared to fit patients (eFI≤0.10), patients classified as frail exhibited a higher incidence of ED visits and hospitalizations (hazard ratio = 3.05, 95% CI: 2.35 to 3.95) and all-cause mortality (hazard ratio = 7.33, 95% CI: 3.61 to 14.88). A large number of patients classified as frail based on the eFI had HbA1c levels <7.5% based on their most recent measure (N=4544, 73.1%). In this population, 1408 (31%) were prescribed no T2DM medication, 1013 (22.3%) were prescribed metformin alone, and 1755 (38.6%) were on a higher-risk T2DM medication (sulfonylurea or insulin). In frail patients with HbA1c<7.5%, patients taking metformin only exhibited the lowest rate of ED visits and hospitalizations (hazard ratio = 0.58, 95% CI: 0.44 to 0.77 compared to all other groups). Conclusions: The eFI is a pragmatic and scalable tool to identify vulnerable older adults with T2DM that may benefit from de-prescribing consistent with guideline recommendations. Disclosure C. Usoh: None. K. M. Lenoir: None. N. M. Pajewski: None. K. E. Callahan: None.