OR22-4 Recurrent Hypoglycemia, Fragmentation of Care, and Mortality in Chicago

Abstract

Hypoglycemia is an acute complication of diabetes management that has been linked to increased morbidity and mortality. In this study, we analyzed whether recurrent hypoglycemia is related to fragmentation of care. The Chicago HealthLNK Data Repository (HDR) consists of merged, de-duplicated and de-identified electronic health records from 6 institutions and employs a hashing and matching algorithm to create a unique ID for each patient (pt) so that pt care can be tracked across different institutions without sharing protected health information. The HDR contains pt demographic and clinical information. Hypoglycemia was identified using a validated algorithm using ICD-9 codes. Fragmentation was defined as an emergency department or inpatient hypoglycemia encounter at >1 institution over the 7-year study period 2006-2012. Of 187,644 patients (pts) with a diabetes diagnosis (250.x), 9,741 (5.2%) pts were identified as having hypoglycemia, representing 18,443 unique encounters, with a mortality rate of 27.7%. 1,035 pts (10.6% of all pts with hypoglycemia) had ≥ 4 hypoglycemic encounters and accounted for 40.3% (n=7,434) of all hypoglycemia encounters. Having any hypoglycemia was associated with race (p<0.001), younger age (p <0.001), and insurance status (p<0.001). Of the 9,741 pts with hypoglycemia, 304 (3.1%) had fragmented care. In 1,035 pts with ≥4 hypoglycemia encounters, 96 pts (9.3%) had fragmented care, and in 1822 pts with 2-3 hypoglycemia encounters, 208 (11.4%) had fragmented care (p=0.074). Hypoglycemic pts with fragmented care relative to those with non-fragmented care had a higher percentage of Medicare (57.9% vs 46% p < 0.001), Medicaid (16.8% vs 9.3% p < 0.001), and self-pay (11.5% vs 4.6% p <0.001). Pts with any hypoglycemia (N=9,741) had greater mortality (2,696 deaths, 27.7%) than those without any hypoglycemia (20,188 deaths, 11.4%, p < 0.00001). In comparing those with ≥ 4 episodes vs 2-3, those with ≥ 4 episodes had fewer complications (39.7% vs. 54.3% p < 0.001) and less mortality (16.0% vs. 30.6% p <0.001). Limitations of this study include use of EHR data and a previously validated algorithm that may misclassify diagnoses/encounters. Overall, hypoglycemia was associated with increased mortality. Pts who had ≥ 4 hypoglycemic encounters represented only 10.6% of patients with hypoglycemic encounters, but accounted for 40.3% of all hypoglycemic encounters, therefore likely contributing to higher healthcare costs. Increased frequency of hypoglycemia encounters was not associated with fragmentation of care. Our findings need to be validated in other large data sets.