2019 HSR&D/QUERI National Conference

4078 — After prandial insulin initiation: evaluating adverse outcomes in older veterans

Lead/Presenter: Chin-Lin Tseng,  Veterans Affairs-New Jersey Health Care System
All Authors: Tseng C (Veterans Affairs-New Jersey Health Care System), SOROKA O (Veterans Affairs-New Jersey Health Care System), POGACH L (Veterans Affairs-New Jersey Health Care System)

Objectives:
Prandial insulin increases risks of hypoglycemia, especially in high-risk population. We aimed to evaluate rates of adverse outcomes: serious hypoglycemia (HYPO; emergency department visits or hospitalizations for hypoglycemia), death, and potential overtreatment (A1c < 6.5%) within one-year after initiation of prandial insulin in older Veterans.

Methods:
This was a retrospective cohort study of veterans with diabetes and dually enrolled in the Veterans Health Administration and fee-for-service Medicare in 2009-2013, with prandial insulin initiated in 2010-2012, receiving basal insulin (without other diabetes medications) in the prior year, and aged > = 65 years. We compared event rates by race/ethnicity (Hispanics, non-Hispanic Blacks, non-Hispanic Whites, and others) using Chi-squared tests. We also assessed subsequent refills of prandial insulin after the first (index) HYPO in the one-year follow-up.

Results:
Of the 4,885 studied patients: 69.6% aged 65-79 years, 79.3% were non-Hispanic Whites, 9.2% non-Hispanic Blacks. Most (84.5%) patients had comorbidities including advanced diabetes complications, cardio vascular conditions, etc. Overall, 637 (13%) patients died or incurred HYPO. Of the 318 (6.5%) that incurred HYPO, 14.5%, 20.1%, and 26.7% had the event within 1, 2-3, and 4-6 months, respectively. Of the 339 (6.9%) deaths, 9.1%, 11.5%, and 26.8% occurred in the same time intervals. Among the 3,472 (71.1%) patients having some A1c values within a year after prandial initiation, 10.3% had an A1c < 6.5%. After the index HYPO, the prandial refill rate was 28% within 30 days, an additional 27% rate was observed within 31-90 days. HYPO rates by race/ethnicity were 8.9% (non-Hispanic Blacks), 8.6% (Hispanics), 6.9% (others), and 6.1% (non-Hispanics Whites) [p = 0.08].

Implications:
Within one-year after prandial initiation among basal insulin users, HYPO rates were higher in non-Whites and one in ten had an A1c < 6.5%. There was a significant refill rate of prandial insulin after a HYPO event.

Impacts:
Stratification by race is necessary to compare magnitude of serious hypoglycemic events and death for patients receiving insulin. Professional societies should make efforts to address potential overtreatment with prandial insulin in older adults with complex comorbidities. Benefits and risks of prandial insulin may be re-evaluated after HYPO, and reduction in dose or discontinuation be considered.