1651-P: Diabetes Incidence among Medicare Beneficiaries, 2001-2014

Abstract

Diabetes mellitus affects more than 12% of the U.S. adult population and has doubled in prevalence between 1990 and 2010. Older adults are particularly affected by the increasing prevalence, which doubles roughly every decade of life and then peaks in the mid-60s at more than 25%. However, there was no change in prevalence or incidence nationally between 2009 and 2012 based on data from the National Health Interview Survey. Increasing prevalence rates are due in part to a higher incidence among adults aged 65 and older. However, these estimates are based on survey research with relatively small numbers of geriatric adults. Administrative data sets built from Medicare claims contain millions of cases, affording an opportunity to explore both trends over time and heterogeneity within the geriatric population. Although most older adults in the U.S. are enrolled in Medicare, claims are only available for those beneficiaries who are enrolled in Fee-for-Service (FFS or Parts A and B). Because beneficiaries can switch between FFS and Medicare Advantage plans, the criteria for identifying prevalent and incident cases of diabetes in FFS claims must be defined carefully. To ensure we had all claims for our study population, we included the 50.4 million Medicare FFS beneficiaries (mean age 68.9 y) who were enrolled for all 60 months of each five-year window centered on index years 2001-2014. We identified prevalent cases as those with either one inpatient or two outpatient diabetes claims within two years. Incident cases were required to have a two year “clean” period with no diabetes diagnostic code. The overall national prevalence of diabetes among Medicare FFS beneficiaries increased from 23.3% in 2001 to a high of 32.2% in 2012, and then remained level through 2014. The overall incidence rate increased from 3.4% in 2001 to a high of 4.1% in 2006-7, then decreased to 3.2% in 2014. Diabetes prevalence and incidence varied considerably across states and were higher among blacks, Hispanics and Asians than among whites. Disclosure L.J. Andes: None. Y. Li: None. M. Srinivasan: None. D.B. Rolka: None. E. Gregg: None.