Venue: The Fuqua School of Business, Duke University, 1 Towerview Drive, Durham, NC 27708-0120
Presentation
Canadian Border Crossing For Prescription Drugs: Evidence From Medicare Drug Discount Card Data
Introduction It has been widely reported that seniors on fixed incomes with inadequate insurance coverage are often forced to forego medication, split pills to extend supplies, or acquire prescription drugs in Canada. Although reference pricing policies lead to cheaper medications at Canadian pharmacies, American consumers must incur time and travel costs to conduct these transactions. Thus, an individual's net economic benefit of border crossing for prescription drugs should depend on his proximity to Canada. The present study is likely the first to examine spatial correlation to quantify the extent to which seniors were filling prescriptions in Canada prior to the implementation of Medicare Part D.
Data: Eligibility and pharmacy claims data from 15 Medicare Drug Discount Card (MDDC) programs managed by Caremark were analyzed. A total of 15,341 cardholders who had at least 6 months of enrollment and 1 or more prescriptions during the period June 1, 2004 through November 12, 2005 were included. Non-seniors and those who received the Transitional Assistance Program subsidy were excluded. Using the same criteria, a second dataset of 52,162 employer-insured individuals aged 65 and older was also studied for comparison.
Methods: The analytical datasets were geo-coded by individual zip code and street address. X-Y coordinates for 83 intersections of the national road network and the U.S.-Canadian border were manually recorded. For each person, the estimated driving time in hours to the nearest border crossing location was derived using travel speeds by U.S. highway type and a cost-weighted distance function. Seniors who resided within 15 hours of Canada (the midpoint of the 0 to 31-hour range) were analyzed; more distant individuals were excluded given their closer proximity to Mexico, another potential source of lower-cost medications. For each of the two samples, gamma-log link generalized linear models of annualized generic and brand utilization were estimated as functions of age, gender and driving time to Canada (including a squared term).
Results: In general, the closer seniors lived to Canada, the fewer prescriptions they filled in the U.S. This was true for both samples up to distances of about 9 hours, where utilization differentials narrowed substantially. The magnitude of the marginal effect of driving time, however, was much greater among MDDC (uninsured) enrollees than employer-insured individuals. Moreover, proximity to the Canadian border had a larger association with brand utilization than generic utilization. For instance, in the MDDC sample, brand use was 30% lower (and generic use 20% lower) for cardholders living 2 hours from the border versus those residing 11 hours away. The comparable effects in the employer-insured group were significantly smaller, 10% lower for brands and 3% lower for generics. Additional specifications, which controlled for income, state, and program/insurer effects, yielded similar findings.
Conclusions: The results suggest that seniors obtained prescriptions, particularly brands, from Canadian pharmacies when they lived within 9 hours of the border. As expected, border crossing does appear to be inversely related to its cost, as measured by driving time. Insurance significantly reduces, but may not eliminate this activity.