A recent policy perspective was published in the New England Journal of Medicine March 5, 2015 by the Secretary of Health And Human Services, Sylvia Burwell, on setting value-based payment goals for Medicare [1]. These goals include having at least 30 % of the Medicare payments provided through such mechanisms as accountable care organizations (ACOs) and bundled episodes of care by 2016, rising to 50 % by 2018. ACOs are commonly full risk or risk shared arrangements where the provider is “on the hook” for the cost of care provided to a set group of Medicare enrollees. In other words, if at the end of a period of time (e.g. one year) the cost for care is less than the “yearly budget”, then the provider keeps the difference (e.g. profit). If the cost for care is more than this, the provider is “on the hook” for the loss. As well, bundled episodes for care are a derivative of the diagnostic related group (DRG), which has been in place by Medicare since 1983. DRGs are essentially bundled payments for the hospital portion of a patient’s stay and are acute in nature. Under bundled episodes of care, commonly all provider services (hospital plus physician) are extended past this period of hospital stay in caring for a patient and can include: physician care, hospital care, nursing home care post-hospital discharge, and home care. These same goals were expressed by a representative from Center for Medicare and Medicaid Services (CMS) at a recent (April 10, 2015) Samuel Martin Memorial lecture held by the Leonard Davis Institute (LDI) at the University of Pennsylvania.

Is it very interesting to note how aggressive CMS has been in setting and announcing these payment reform goals. At a recent Academy Health meeting held in early February 2015 in Washington, DC, payment reform was discussed by Professors Michael Chernew and Michael McWilliams from Harvard and Peter Hussey from the RAND Corp. All three are nationally regarded experts on payment reform. This panel’s conclusions on how well these types of payment reform initiatives were progressing and the resultant cost saving were not encouraging. First, a bundled payment demonstration that took place from 2010–2013 in California (by the Integrated Health Association and funded by the government) was not successful. The hospitals that were involved in this demonstration only enrolled 35 patients over a 3-year period. The issues stated by those involved in the demonstration were that it was very complex—it was too much work, there was a lack of technical infrastructure, and as well a lack of trust between parties. The conclusion made was that: “Despite great initial support, enthusiasm and effort, episode-of-care payment does not offer an easy fix to the nation’s health-care financing problems [2]”. As it related to ACOs, again the way in which they were configured created issues—with an unsuccessful end result. The recommendations made during this session were that patients needed to be kept in the ACO for the entire year (and cannot be allowed to float in and out of this arrangement) and that the risk needed to be two sided (e.g. both Medicare and the providers being at risk). The take-home messages from their findings w as follows: better data are required from these demonstrations, and the payment models will take time before they are fully baked.

The above Academy Health presentation by Professor Chernew was further elucidated upon in a New England Journal of Medicine analysis of the Pioneer Accountable Care Organizations [3]. While this analysis demonstrated a small savings of 1.2 % for the ACOs versus a control group (Medicare enrollees living in the same hospital referral regions served by the Pioneer ACO), there were notable limitations to this analysis. The first limitation related to the distinct possibility that the likely desire/efforts of the Pioneer ACOs to constrain spending and to possibly work extremely hard to accomplish this (e.g. the Hawthorne effect) and to have the pieces in place ahead of time. The fact that significant savings were not realized, may speak to the difficulty in doing so with an ACO model. Secondly, and as mentioned in the analysis, the costs that CMS has incurred in attempting to make ACOs work were not included in the analysis. These additional costs may have negated the 1.2 % savings realized. Further, back in 2012, Burns and Pauly commented on the difficulty that ACOs may have in reducing costs and improving quality—based on the failures of the integrated network experience of the 1990s [4]. Today, the main issue that remains is care coordination, which is an integral part of reducing fragmented and duplicitous care, and which is a significant cost driver. It has been noted that patients with chronic conditions refer themselves to many specialists [4] and integrating information technology (IT) on these patients could provide for more seamless care (there are hundreds of different systems capturing and reporting on patients—how do we get IT to “talk together” in such a short period of time as proposed by Burwell above?). As a matter of fact, care coordination (with substantial interaction with the patient) when evaluated recently was found at best to be cost neutral [5].

CMS recently presented the findings as reported in an article by McWilliams et al. [3], but in a much different light [6]. The CMS findings demonstrated close to a US$400 million savings in Medicare enrollees in the ACO model versus fee for service over a 2-year period with the differences in spending being statistically different for each year. However, the spending difference in year 2 versus year 1 was much smaller. This may portend difficulties in the sustainability of spending reductions by ACOs over time. These difficulties in maintaining sustainable reductions may also relate to some of the issues identified by Burns and Pauly [4].

CMS has also reported on the results of bundled payments with early findings that there were significantly lower episode payments in orthopedic surgery over a 90-day bundled episode of care (included initial hospitalization, all professional services, and all other services delivered within that 90-day period such as rehabilitation, follow-up physician visits, etc.) (Note: this was the only episode of care where there were enough numbers to report/comment on) [7]. On examining this study in more detail however, there are a number of issues that should be mentioned which might be driving these findings. First, providers whose outcomes were improving without involvement in the 90-day bundle model may have been those who signed up for it in the first place—leading one to a false conclusion that this might work for all [8]. Second, one-quarter of a calendar year of cost decreases does not make for sustainable cost decreases. Third, as the Lewin report states, these episodes of care were in hospitals located in areas where more affluent populations existed versus the universe of Medicare-participating hospitals [8]. This may mean that the overall health of these patients might be better (e.g. needing less care over time and with better social support systems), and that these institutions had more resources to engage in care redesign [7]. Not evaluated in the Lewin report, but likely should be, are the social determinants of health resulting in better health and lower costs per capita [9].

So where does that leave us? When the above CMS representative was asked at the LDI meeting by the author of this article, if a contingency plan was in place in case these models do not work, the question was answered in this manner: “CMS will continue to tinker with the ACO and bundled payment models until they are corrected”. So how much will this cost and when is the appropriate point if ACOs or bundled payments do not work for CMS to say “we need to try something different?” Does CMS even have a fallback position/plan? No mention of one was made by the CMS representative. One was left with the impression that it is full steam ahead no matter the consequences.

For now, the rollout by CMS of these types of payment models continues unabated in a very aggressive manner (as per Burwell above) which will likely cost the tax payer millions of dollars. It appears that CMS is absolutely certain that these types of payment and delivery models will ultimately work. The question that has been left unanswered is: What if these reforms do not work? When Professor Michael Chernew was asked this same question during the Academy Health meeting his response was: “It may be back to fee-for-service payment but with a ‘governor’ included to limit growth”. Burns and Pauly supported this position as espoused by Chernew in a similar manner—that Medicare is likely to move toward providing what is effectively a budget-determined capitation payment (a governor) as an end–of-year adjustment to accumulated fee-for-service payments [4].

As noted above, health economists at some of the leading academic teaching institutions (Harvard and University of Pennsylvania, The Wharton School) have weighed in on these findings with a healthy dose of skepticism. It is also fine for the rest of us to examine these findings in the same way and to question them. It is not fine however, to accept reports of success and not allow healthy dialogue on these issues to take place. This dialogue is important for us to solve the problem of continually rising health-care costs (viewed as a percent change in health spending versus GDP). As an industry we need to ask good questions and continue to challenge executives and policy makers at institutions such as the CMS—as it controls over US$600 billion in spending [10].

Stay tuned—Medicare may need a different plan for controlling spending or this may take longer than Medicare is envisioning based on its pronouncements and presentations—resulting in the goals outlined in the New England Journal of Medicine March 5, 2015 article being pushed back.