OBJECTIVES : To estimate the prevalence of cancer among Medicare beneficiaries, and examine the impact of Medicare Part D on total prescription drug expenditures among these patients.

METHODS : The study was a retrospective cross-sectional analysis, conducted using the Medical Expenditure Panel Survey (MEPS) data. Regression discontinuity model was adopted where the age 65 was the cutoff point. Patients (n=1796) aged 55 years old and above with cancer diagnosis in the MEPS data in 2014 were identified. Treatment group (n=733) was Medicare beneficiaries (65 years old and above) with cancer diagnose, excluding individuals who were dual eligible for Medicaid or other public health insurance than Medicare. Control group (n=440) was near-elderly (55-64 years old) individuals with cancer diagnose, excluding those covered by Medicare, Medicaid, or other public health insurance than Medicare.

RESULTS : The average age for Medicare beneficiaries was 73.9 and the nearly-old group was 59.7. Medicare beneficiaries tend to have more comorbidities than the control group (4.6 vs 3.4), less likely to be married (56.6% vs 71.1%), and lower household income (8.6% vs 5.3% in the low-income category). The ethnicity and region composition were significantly different between the two groups. Treatment group has significantly more public insurance coverage than the control group (45.5% vs 1.4%). Medicare beneficiaries had higher mean and median total prescription drug expenditure than the nearly-old group (p<0.05). Around age 65, there was no sudden change of total prescription drug expenditure; however, the trends before and after age 65 were significantly different. Medicare beneficiaries claimed $14.3 (SE=7.2) less than near-elderly individuals for total prescription drug expenditure when holding other factors constant. Other indicators that significantly associated with the outcome measure are insurance type and BMI category.

CONCLUSIONS : Eligible Medicare beneficiaries were associated with lower total prescription drug expenditure than near-elderly individuals when controlled for demographics and health status.