Association of Cumulative Multimorbidity, Glycemic Control, and Medication Use With Hypoglycemia-Related Emergency Department Visits and Hospitalizations Among Adults With Diabetes

This cohort study examines the associations of age, cumulative multimorbidity, glycated hemoglobin level, and use of glucose level–lowering medication with hypoglycemia-related emergency department visits and hospitalizations among adults with diabetes.

Analytic Note: All code sets were reviewed for relevance and validity by study investigators experienced in diabetes management and diabetes-related health care delivery research. In circumstances where no validated code sets existed (specifically: falls, urinary incontinence), relevant diagnosis codes were identified by study investigators based on clinical experience and review of pertinent literature.
The Ginde ICD-9 algorithm has been used extensively to identify hypoglycemia-related utilization events, including hospitalizations and emergency department visits, despite its original development in the ED setting. 1 It is used by the Centers for Disease Control and Prevention (CDC) to ascertain hospitalizations and ED visits for hypoglycemia (https://gis.cdc.gov/grasp/diabetes/DiabetesAtlas.html#) using data from the National Inpatient Sample (NIS) and the National Emergency Department Sample (NEDS), respectively.
All ICD9 and ICD10 codes for diabetes complications of retinopathy and neuropathy were adapted from the validated Diabetes Complications Severity Index (DCSI) 6 and also the National Quality Forum (NQF)-endorsed CMS "Risk-Standardized Acute Admission Rates for Patients with Diabetes" measure (ACO-36) 7 . For retinopathy, the DCSI and ACO-36 definitions include both proliferative and non-proliferative retinopathy codes, and proliferative retinopathy codes were extracted by the study investigators. Rationale for focusing on proliferative rather than all retinopathy is that American Diabetes Association (ADA) and Department of Veterans Affairs/ Department of Defense (VA/DoD) specify that patients with severe non-proliferative diabetic retinopathy (NPDR) and proliferative diabetic retinopathy (PDR) should be treated to less stringent glycemic targets.
Because we relied on diagnosis codes to identify patients with PDR, we sought to capture the likely ways that PDR is billed by providers, which means looking not only for codes associated with PDR but also the ways that PDR presents in any given patient. Specifically, PDR (ICD9 362.02) is defined by the presence of neovascularization arising from the disc or retinal vessels, which can lead to pre-retinal and vitreous hemorrhage (ICD9 379.23), fibrosis, and traction retinal detachment (ICD9 361.xx). 8 We included cystoid macular degeneration (ICD9 362.53) because clinically significant macular edema can lead to vision loss. We also included blindness and low vision (ICD9 369.x) because this reflects end-stage or clinically significant eye disease. We did not include diabetic ophthalmologic disease (ICD9 250.5x); background diabetic retinopathy (ICD9 362.01); other background retinopathy and retinal vascular changes (ICD9 362.1x); or retinal hemorrhage, retinal exudates and deposits, and retinal edema (ICD9 362.81-362.83) because these diagnoses are not necessarily reflective of severe NPDR or PDR.

eTable 2. Adjusted Risk Factors for Hypoglycemia-Related ED Visits/
Hospitalizations-Type 1 vs Type 2 Diabetes. These Poisson regression analyses examine the associations between patient characteristics and potential hypoglycemia risk factors with hypoglycemia-related ED visits and hospitalizations among adults with type 1 and type 2 diabetes, with all factors adjusted for simultaneously. The total number of chronic conditions was calculated from among dementia, end-stage renal disease, stage 3-4 chronic kidney disease, myocardial infarction, heart failure, cerebrovascular disease (transient ischemic attack/stroke), chronic obstructive pulmonary disease, cancer (except for non-melanoma skin cancer), cirrhosis, proliferative retinopathy, peripheral neuropathy, hypertension, arthritis, urinary incontinence, depression, and falls. Prior history of a hypoglycemia-related ED visit/hospitalization ("prior hypoglycemia") was considered separately and therefore was not included in the total count of guideline-specified chronic conditions. *With or without other glucose-lowering medications (not insulin, sulfonylurea).

eTable 4. Adjusted Risk Factors for Hypoglycemia-Related ED Visits and Hospitalizations-Secondary Analysis Examining Individual Comorbidities in the Overall Population.
In this Poisson regression analysis, the association between each comorbid health condition, rather than the total count of comorbidities, and hypoglycemia-related ED visits and hospitalizations was assessed. All variables were adjusted for simultaneously, allowing for the examination of the independent associations between each of these risk factors and the risk of hypoglycemia-related ED visits/hospitalizations. *With or without other glucose-lowering medications (not insulin, sulfonylurea).

eTable 5. Adjusted Risk Factors for Hypoglycemia-Related ED Visits/ Hospitalizations-Secondary Analysis Examining Individual Comorbidities Among Patients With Type 1 and Type 2 Diabetes.
In this Poisson regression analysis, the association between each comorbid health condition, rather than the total count of comorbidities, and hypoglycemia-related ED visits/hospitalizations was assessed separately for patients with type 1 and type 2 diabetes. All variables were adjusted for simultaneously, allowing for the examination of the independent contribution of each of these risk factors to the risk of hypoglycemia-related ED visits/hospitalizations. *With or without other glucose-lowering medications (not insulin, sulfonylurea).