Knowledge of Medicare beneficiaries with high medical expenditures can help identify target populations for intervention and strategies to reduce overall Medicare expenditures. Using 100% Medicare fee-for-service claims, we examined characteristics and spending patterns of high spenders among Medicare beneficiaries with diabetes (DM) . We defined high spenders as those whose expenditures in a year were in the top 10% of beneficiaries with DM. We identified 7,321,228 high spenders in 2015 and followed them through 2017. We grouped high spenders in 2015 into three groups: those who died during 2015-2017 (group 1) , were alive and a high spender in all 3 years (group 2) , and alive but a high spender for 1/2 years (group 3) . We used regression models to identify factors associated with each group. In 2015 high spenders accounted for 48% of all medical expenditures in Medicare beneficiaries with DM. Of the high spenders, 34% were in group 1, 20% in group 2, and 46% in group 3. High spenders in group 1 had a mean annual expenditure of $110,429 and were more likely to be male, be older, have dual eligibility for Medicaid, and have a higher number of comorbid conditions compared with groups 2 and 3. Over 60% of the expenditures in group 1 were spent on either acute inpatient care (39%) or on skilled nursing facility (23%) . Groups 2 and 3 had mean annual expenditures of$83,2 and \$42,527, respectively. High spenders in group 2 were more likely to be male, be older, have dual eligibility for Medicaid, and have a higher number of comorbid conditions compared with group 3. Compared to those in group 3, high spenders in group 2 spent a higher proportion on hospice (13.3% vs. 7.9%) and inpatient care (16.7% vs. 9.9%) , but less on acute inpatient care (19.5% vs. 31.3%) . High-spender Medicare beneficiaries with DM had different characteristics and spending patterns. Future studies could identify modifiable factors and develop strategies to lower Medicare expenditures among high spenders with DM.

Disclosure

Y.Wang: None. P.Zhang: None.