Jackson Williams is vice president, public policy at Dialysis Patient Citizens. Previously he worked at the Centers for Medicare and Medicaid Services (CMS); was a health services researcher in the AARP Public Policy Institute; and a lobbyist on health policy issues for 3 nonprofit associations. His more than 2 decades of experience monitoring, studying, and overseeing federal demonstration projects include time as a project officer at CMMI and serving as the consumer representative on a Medicare pilot site’s governing body.
After a few false starts, the Center for Medicare and Medicaid Innovation (CMMI) is proceeding with new mandatory demonstrations.
Mandatory demonstrations randomly assign providers, by geographic unit, to intervention and comparison sites. Their purpose is to combat the tendency of voluntary demos to attract early-adopter providers whose heightened motivation or skills can confound evaluations of a payment model’s effectiveness.
Over their first 5 years, CMMI mandatory demonstrations have tested episode payment reforms that arguably encourage better care for patients but may also encourage stinting. Their primary goal has been savings for Medicare, not improved outcomes. But with the ESRD Treatment Choices (ETC) Demonstration, CMMI has for the first time deployed a mandatory demo to incentivize particular therapies: home dialysis and transplants for kidney patients. In appropriate patients, these represent unqualifiedly superior treatment options. This distinction radically alters the dynamics of random geographic assignment, leading here to an absurd and amoral result.
Designing Demonstrations to Avoid the “Coals-to-Newcastle” Pattern
A few years ago I looked at participation in 15 then-ongoing HHS programs in which states were the primary applicants or that had geographic service areas. A striking pattern was evident: Of 200 sites across the programs, 107 were located in just 16 states. Twelve of those states scored in the top half of “America’s Health Rankings” as calculated by the United Health Foundation. Eleven were in the top half of states ranked by per capita income. Only 3 of the states, Ohio, North Carolina, and Kentucky, were not in the top half of either list.
In a 2005 paper, Jonathan Skinner and Douglas Staiger noted that the geographic pattern of adoption by hospitals of administering beta-blockers to heart attack patients resembled the pattern of adoption by farmers of hybrid corn. They attributed this to local reservoirs of social capital, which historically have been higher in the northern tier of states. Certain states—which seem to turn up on many lists of “best states for…”—have high levels of social capital and with it, superior levels of educational attainment, less poverty, less corruption, and so on. These states also have healthier populations and high-quality, efficient health care.
Such states, which include Colorado, Oregon, Washington, Wisconsin, Minnesota, Connecticut, Massachusetts, and Vermont, tend to be overrepresented in Medicare demonstrations despite their populations having less need for the extra attention. Ideally we would want to see innovative programs serving states commonly associated with the “stroke belt,” “diabetes belt,” “obesity belt,” and the like, which is to say, in the South and Greater Appalachia.
Random geographic assignment ensures that at least some interventions will be tried, and hopefully proven, in less advantaged areas. In the case of the ETC demo, numerous hospital referral regions (HRRs) have been designated for intervention in Southern states that historically have low transplantation rates. The demonstration assigns randomly selected HRRs for bonus-and-penalty payment adjustments meant to incentivize dialysis facilities and nephrologists to boost education to kidney patients about home dialysis and transplants.
Random Assignment Yields a Perverse Result in the ETC Demonstration
Unfortunately, and perhaps not unexpectedly, random assignment also results in a large number of coals-to-Newcastle interventions. All 8 high-social-capital states I referenced above have HRRs designated for intervention, even though 7 already have above-average rates of transplantation.
What is most concerning about the assignments is that a potential opportunity to address racial disparities in transplantation has been squandered. The ETC demonstration is also unique in involving resources in short supply, including donor organs; mandatory episode demos have heretofore been deployed to reduce the use of resources in plentiful supply, namely postacute care. With kidneys scarce, perhaps the best outcome that could have been hoped for was greater equity in their allocation. ETC might address in this in Southern states, where different races live in relative proximity and HRRs are diverse. But outside the South, where migration patterns have resulted in residential segregation, most HRRs are disproportionately White, leading to perverse results.
In Illinois, the intervention regions are Evanston, Rockford, and Peoria. All of them are overwhelmingly White relative to the national average; the demonstration did not target any of the Illinois regions with diversity reflecting overall racial composition. The Evanston HRR, encompassing Chicago’s North Shore suburbs, is one of the most affluent in America, and 7 of the 14 wealthiest zip codes in Illinois are included in the intervention, while only 5 of the 25 poorest are included. One particularly perverse aspect of this is that other, more diverse hospital regions in the densely populated Chicago area are contiguous to the Evanston region. Transplant experts I’ve consulted fear that provider organizations with broad geographic reach could redirect educational efforts from patients excluded from the demo to patients in adjacent regions, in order to avoid ETC’s stiff penalties.
The assignments also misallocate efforts to increase home dialysis use. The Great Falls, Montana, HRR is included in the intervention, despite having the highest rate of home dialysis use in the United States. Meanwhile, home dialysis “dead zones” in the midst of surrounding areas with high penetration, such as LaCrosse, Wisconsin, and Rapid City, South Dakota, are omitted. We can infer from the uptake in neighboring counties—home dialysis is popular in Western and rural locales—that something is amiss in these areas, but nobody will be drilling down to find out what barriers are at work. This is unfortunate because one outcome we might hope for from Medicare pilots is identifying weaknesses in the delivery system and disseminating solutions.
With the turnover in administrations, incoming officials can put a second set of eyes on this project. It was designed before Kaiser Permanente disseminated its processes for increasing home dialysis use, and before the recent renewed focus on disparate treatment of Black Americans. It’s worth going back to the drawing board on this.