2011 HSR&D National Meeting Abstract
1025 — Increased Risk of Hypoglycemia in Older Patients with Dementia and Cognitive Impairment: Do HEDIS Performance Measures Risk Safety?
Feil DG (West Los Angeles VA Healthcare Center), Rajan M
(East Orange REAP), Soroka O
(East Orange REAP), Miller D
(Bedford VA COE), Pogach L
(East Orange REAP)
Current HEDIS performance measures for glycemic control ( < 8 percent) in the elderly (65-74) do not exclude dementia and cognitive impairment (CI). Our objective was to evaluate the prevalence of these conditions among older veterans and to evaluate rates of coded hypoglycemia.
We conducted a retrospective cross-sectional data analysis of the VA Diabetes Epidemiology Cohort (DEpiC) for fiscal years 2002 and 2003. We identified 497,900 veterans aged 65 years and older with diabetes. We extracted ICD-9-CM codes for all-cause dementia and cognitive impairment (combined years 2002 and 2003); hemoglobin A1c lab test results (2002 and 2003); medical comorbidities and anti-glycemic medications (2002); and diagnoses of hypoglycemia from outpatient, emergency, and admission visits (2003). Associations between dementia/CI and coded hypoglycemia were evaluated using multivariate logistic regression.
The prevalence rates of combined dementia and CI were 13.2 and 24.2 percent for veterans aged 65-74 and 75 years and older, respectively. Rates of insulin use were higher in veterans with dementia (28 percent) and CI (26 percent) than without (20 percent). Coded hypoglycemia rates were 2.5 times higher in dementia (15 percent) and 83 percent higher in CI (11 percent). In both, A1c levels were lower. Rates of hypoglycemia were highest for insulin therapy and/or lower A1c levels. Odds ratios (OR) for coded hypoglycemia were 2.42 (95 percent OR = 2.36-2.48) for dementia and 1.72 (1.65-1.79) for CI; adjusted odds ratios were 1.84 (1.79-1.89) for dementia and 1.22 (1.17-1.27) for CI.
The prevalence of dementia and CI is high even among veterans 65-74 years old who would be included in the denominator of the current HEDIS measures. Their diabetes is managed more intensively and rates of coded hypoglycemia are clinically significant. We suggest cognitive screening of older patients with diabetes, and excluding those with dementia/CI from the HEDIS measure. Healthcare providers may accordingly individualize glycemic goals, factoring in safety and quality of life.
By assessing risks associated with diabetes care in veterans with dementia/CI, VHA may better define the healthcare needs of this vulnerable population and align clinical management, health services, and policies with VHA-DOD Guidelines.