THE article by Millwee, Quinn, and Goldfield is a worthwhile read-providing interesting and useful information for both policy-makers and practitioners. The authors are specific regarding the types of outcomes that should be considered in any "pay for outcomes" (which they shorten to P4O) reimbursement system. The idea of paying for outcomes has been raised frequently over the last decade, but deciding on the metrics to use has remained elusive. Getting specific recommendations about what could or should be included from people who know the relevant information systems currently or soon to be available is therefore very helpful.
They also indicate several of the states, including Florida, Maryland, Minnesota, New York, and Ohio, that can provide precedent on structuring the program using the soon-to-be available Transformed Medicaid Statistical Information System (T-MSIS). I am not familiar with the "transformed" information system, which they indicate will help produce the potential for a tighter tie between payment and use, and between quality and outcomes, but getting comparable information for use across different states has been a challenge for Medicaid analyses in the past. If the new T-MSIS data set resolves the issue, it will be a major step forward.
The authors focus on the importance of using outcome measures that are relevant for the population under consideration. Because the Medicaid population, with its focus on maternal and child health, mental health, and on specific diseases such as HIV/AIDS, is so different from the Medicare population, most of the work that the Centers for Medicare & Medicaid Services (CMS), which has focused on use of quality and outcome metrics for the Medicare population, is not relevant for Medicaid. Too often, the CMS and others have seemed to assume that the same metrics could or should be used for the populations in both programs. Parsimonious use of the metrics required for reporting is important but not if the relevant populations and their illnesses and use differ as much as they do between the Medicare and Medicaid populations. Even the "dual-eligibles"-individuals who are on both Medicare and Medicaid-are more likely to be like the Medicaid population than the Medicare-only population although will probably need some specific metrics for individuals who are in long-term care facilities and other institutionalized settings.
Like several others, Millwee et al criticize the large number of metrics that are currently being required by some parts of the Medicare program such as the 62 different metrics that are used across weighted groups to provide a composite performance measure for the Medicare Stars program.
There seems to be an increasing interest to consider the use of a limited number of uniform metrics across the different payers in an effort to reduce the burden currently being placed on clinicians and institutions but no clear path forward about exactly how to get there and who should lead the effort.
The timetable they think will be necessary for states to adopt outcome metrics that could be used in a P4O program is a little discouraging. They suggest 3 to 5 years from concept to full implementation although they believe that some benefits will occur early on. They cite the experience of Texas as an example of the relevance of this type of this timetable. While I can grudgingly acknowledge the experience reported, it is hard to imagine that slowness of pace generally being tolerated in the private sector although the adoption of some safety equipment or other changes that impose costs on the corporations can also require a long phase-in-mainly for budgetary reasons.
The only disconnect I find is with their generally very positive predictions about the ability of states to adopt a manageable set of metrics for P4O that actually improves care and the discouraging experience the Center for Medicare and Medicaid Innovation (CMMI) has reported on the outcomes of their various programs incorporating pay-for-performance-type programs. This was recently reiterated in a study by Bonfrer and colleagues from the Harvard School of Public Health reported in the BMJ (https://doi.org/10.1136/bmj.j5622) that showed that care for older patients was no better, defined in terms of better process scores or lower mortality rates, at hospitals that had been operating under pay for performance for more than a decade compared with those who were using such incentives for less than 3 years. They concluded that the programs as currently implemented were not likely to be successful even if their timeframes were extended.
Even with this caveat in mind, the movement toward focusing on outcomes rather than only on inputs remains promising-even with the stumbles that have occurred along the way. Extending this focus to Medicaid is an important step in the right direction.