349-P: Hospital Utilization for Hypoglycemia among Patients with Type 2 Diabetes Using Pooled Data from Six Health Systems

Abstract

Hypoglycemia is the most common serious adverse effect of diabetes treatment and a major cause of medication-related hospitalization. This study aimed to identify trends and predictors of hospital utilization for hypoglycemia among adults with type 2 diabetes. Using electronic health record data from six academic health systems, we performed a retrospective open cohort study including 532,323 patients aged ≥18 years with type 2 diabetes receiving regular care between 2009 and 2019. The primary outcome was the yearly event rate for hypoglycemia hospital utilization: emergency department visits, observation visits, or inpatient admissions for hypoglycemia identified using a validated ICD-9 algorithm. After the transition to ICD-10 in 2015, we used two ICD-10 code sets for hypoglycemia (limited and expanded) that were used in prior studies. We identified independent predictors of hypoglycemia hospital utilization using multivariable logistic regression analysis with data from 2014. We found that yearly rates of hypoglycemia hospital utilization decreased non-significantly from 3.4 to 2.6 events per 1,000 patients from 2009 to 2014. From 2016 to 2019, yearly event rates ranged from 15.5 to 16.5, or 21.1 to 21.8, using the limited and expanded ICD-10 code sets, respectively. The strongest independent risk factors for hypoglycemia hospital utilization were chronic kidney disease (OR 3.13, 95% CI 2.63-3.70), age 18-39 years (OR 3.00 vs. age 40-64 years, 95% CI 2.42-3.73), and insulin use (OR 2.61 vs. no diabetes medications, 95% CI 2.16-3.15). In this multicenter open cohort study, we found that rates of hypoglycemia hospital utilization were stable from 2009 to 2014 and varied considerably by clinical risk factors such that younger adults, insulin users, and those with chronic kidney disease were at especially high risk. There is a need to validate hypoglycemia ascertainment using ICD-10 codes, which detect a substantially higher number of events compared to ICD-9. Disclosure S. J. Pilla: None. J. L. Kraschnewski: None. E. Lehman: None. L. Kong: None. E. Francis: None. J. M. Poger: None. C. L. Bryce: None. N. M. Maruthur: Other Relationship; Self; Johns Hopkins HealthCare Solutions. H. Yeh: None. Funding National Institutes of Health (KL2TR003099); Patient-Centered Outcomes Research Institute (1306-04912)