On October 31, 2013 the House Ways & Means and Senate Finance Committees unveiled a bicameral and bipartisan press release and Discussion Draft of a proposal to repeal the SGR (sustainable growth rate) update to professional fees and replace it with: (1) reforms for future fee-for-service (FFS); (2) incentives for alternative payment models (APMs); and (3) provisions regarding care coordination, relative valuation, service use, and data availability. After a decade of simply kicking the SGR down the road with last-minute and temporary authorizations, a “permanent fix” to this “fundamentally broken” program will be universally welcome. The task at hand: analyzing the elements of what is proposed to replace it.
We all know that the future of Medicare is at a critical state, and it is therefore imperative that we use this rare opportunity of Congressional harmony to do it right. I am delighted to contribute to this discussion, especially as feedback was solicited by the Committees and due to them by November 12. My overall conclusion: the objectives should be applauded, the legislation should be supported, but some changes to the details need to be encouraged before the final legislative text is reported.
Medicare Fee For Service
With FFS, individual physicians and other health care professionals (referred to in the Draft as “professionals”) are paid for each and every specific item or service they furnish, regardless of how many – hence the disparaging moniker “paying for volume not value”. I have long believed that FFS as a method of compensation is highly appropriate because the items and services (CPT codes) vary so dramatically. Instead, the problem with traditional FFS for Medicare and private health insurance has been the historical practice of fixed payments: paying the same amount to all practitioners regardless of the varying necessity, efficiency, or effectiveness of their care, especially when viewed over time.
Think about compensation by time: hourly rates for services in our economy are common and appropriate, but paying everyone in the same field at the same hourly rate is not. The same goes for salaries. The solution, then, is variable FFS payments, with the extent of variation derived from measured performance. Performance-based compensation would offer practitioners both the incentive to distinguish their patient care and the equity of being paid commensurately. Believe it or not, there is evidence this can be done and that it works – for example, the “Selections” program of King County Medical Blue Shield (now Regence) during the late 1990s: individual physicians received varying supplemental payments to their FFS, determined by their differential patient care performance, from an aggregate performance pool funded by demonstrated plan savings.
“Value-Based Performance” (VBP) Payment Program
The Discussion Draft impressively and progressively proposes transitioning Medicare’s current fixed FFS system to one of variable FFS. By the end of 2016 the penalties in two existing programs (quality reporting and meaningful use) and the payment adjustments in the yet-to-be implemented value modifier program “would sunset”. Beginning in 2017, a new VBP payment program would begin. At its core is an annual performance funding pool, created as a percentage of estimated spending for eligible professionals. This pool would allow for supplemental incentive payments that would vary based on assessed performance, to be added to reduced fixed FFS payments. The performance pool is proposed to be 8 percent in 2017, 9 percent in 2018, 10 percent in 2019, and at least 10 percent thereafter.
“The entire funding pool for a year would be paid out to eligible professionals based on their VBP composite score for a specified performance period, with those achieving the highest scores receiving the greatest incentive payment.” This suggests that the assessment of “performance” would be on the basis of the relative distribution of scores rather than any absolute levels; this would be a highly desirable and essential advance.
From an analytical perspective, assessing “performance” when there is variation (e.g., in education, athletics, and yes, professions) has always been defined in relative rather than absolute terms. For example, although scoring “par” in golf might be the “standard”, this score will almost never win a tournament among professionals; payments (i.e., prize money) are not based on attaining a par score (or any other), and are not the same for each participating professional. Instead, performance and payments are based on the distribution of the scores.
Because “payment increases … would be offset by payment reductions” so to keep the program “budget neutral” overall, one must presume that the ranges for payment adjustments, assuming a “normal” distribution of composite performance scores, would be twice these amounts. During 2017, for example, the incentive payments from the performance pool would be 0 to 16 percent, resulting in final payments (when added to fixed FFS payments reduced to 92 percent) ranging from 92 percent to 108 percent of allowances, i.e. ±8 percent. If the distribution of scores turns out to be “skewed”, then the range could either be (a) somewhat more than twice the pool amounts, to keep the average score at 100 percent, or (b) “forced” to twice the pool amounts, and allow the average score to be somewhat greater or less than 100 percent.
If the legislative text permits ranges for the incentive payment amounts to be any less than twice the pool amount, then the actual size of the performance pools would then either be less than the funding amounts specified or not be budget neutral. These payment ranges represent amounts that many analysts, including myself, believe are the minimum necessary to incent value in FFS and to equitably recognize differences in performance. In fact, when the DC Blues (now CareFirst) asked me to design the performance-based payment program for specialists in their CapitalCare plan during the mid-1990s, our range was ±10 percent.
Four categories of assessment are proposed for the VBP composite score: (1) quality measures; (2) resource use measures; (3) adoption of clinical practice improvement activities; and (4) compliance with EHR meaningful use requirements.
The first category begins with “quality measures used in current law PQRS and other incentive programs” (I would assume for ACOs), but these would be expanded over time based on funding to develop more measures and annual solicited recommendations. Importantly, the Draft directs that “higher overall weight [be] given to outcomes measures”, which have been woefully few to date. From my experience, mining patient claims data for such outcomes measures would be both a treasure trove and a procedure less burdensome to practitioners. In addition, we must define quality performance as more than merely reporting quality data.
The Draft also directs that there would be “credit for attainment and achievement”. If this means measurement on both absolute and relative scales, this could be desirable. If, instead, this phrase was meant to say attainment (or achievement) “and improvement”, this would be undesirable. Performance scoring should be common and applicable for all. A separate scale beneficial only for those performing lower on the common scale would make little analytic sense.
Measuring resource use is the most important category for assessing professional performance, as this is where the greatest variation and opportunities for improvement exist. As opposed to “snap-shots” of professional activities or patient services at points in time as with the other categories, resource use during episodes of care and extended periods of time gives a robust “moving picture” of patient care. Only with extended-time measurement can we avoid evaluating individual services and focus instead on the cumulative impact. Therefore, doing too few or the wrong services at a point in time will be reflected later in the excessive use of other services, while doing too many or more costly (among substitutable) services will be revealed as unnecessary over time.
Based on my twenty-five years’ experience with performance “profiling”, the keys to assessing professionals’ patient care performance are (a) to attribute individual patients to those professionals who are responsible for their care (in whole or in part), and (b) to measure and compare the overall resource use (in total and by types of services) of their patients to all “like” patients within distinct groups distinguished by their meaningfully-different acute and chronic clinical conditions and personal characteristics, i.e., “risk stratification”. “Overall” resource use means the items and services furnished and ordered by a practitioner as well as all other items and services received by their patients, whether suggested by the practitioner or not. Thus, evaluating professionals is not properly accomplished by measuring what they do, but rather what happens to their patients.
In addition, specific resource use, or the cessation of further use, can be powerful proxy measures of quality. Finally, and contrary to the oft-repeated refrain that individual professionals cannot be evaluated because of their “small ‘n’” (i.e., numbers), analytic and statistical techniques are, and have been, available to improve and assure significance and sustain even the most critical scrutiny.
Unfortunately, because neither the specific relationship of each professional to each patient nor the distinct care episode and patient condition groups are designated by professionals on their claim forms, analysts including me have instead been forced to rely on complicated and mysterious computer algorithms and theoretical formulas to impute this critical information. Developing these algorithms and formulas have been a lengthy and expensive task, both for the private sector and, more recently, for the government, and efforts to convince professionals that they are valid, and that the measurements derived with them are reliable, have been controversial to say the least.
The Congressional proposal “would involve the health care professional community in furthering the measurement of resource use” by establishing a process to determine such episode and condition groups and patient relationships and to have practitioners indicate them on their claim forms. This would be a fundamental improvement that would significantly and swiftly advance the state of the art of both resource use and quality measurement, and it should be welcome news to everyone affected and involved, particularly eligible professionals. Indication by professionals has to be more accurate than imputation by computers. I would therefore recommend that the proposed weights for the quality and resource use categories be equal when the VBP begins in 2017, not 2019 as in the Draft.
The third category for performance assessment would be new for Medicare. The specific clinical practice improvement activities would fall under five listed sub-categories “from which professionals can select”. These activities, however, would seem to be less about “performance” — how well patient care is furnished — and more about organizational changes that might then improve performance. That said, because the activities “would be established through a collaborative process with professionals and other stakeholders”, most should view this positively. What should not be cast in statute, however, is the “automatic” receipt of the full or one-half “highest possible score” simply for practicing in a certified medical home or participating in any Medicare APM — let the collaborative process work its wisdom.
The fourth category is an as-is carry-over from the existing EHR Meaningful Use program, with compliance proposed to have an initial weight of 25 percent of the composite and to “drop to 15 percent once adoption reaches 75 percent”. I take serious issue with attributing so much of “performance” to such a clearly procedural initiative. An EHR is neither an outcome nor a goal itself – it is technology to facilitate the information and care enhancement processes. Although I believe that the use of EHRs will ultimately assist professionals to improve and distinguish their patient care performance on the quality and resource use measures, and that is why they should be adopted and used, we should not give “extra credit” just for having one, and current “meaningful use” is not sufficient.
Also, as compliance is binary (yes-no), such high weights would seem unfair to those without the access or resources to acquire such technology, and to those with high patient care performance without it. Furthermore, there continues to be Federal stimulus money available to practitioners to encourage its adoption. Since the use of EHRs is really just another clinical practice improvement activity, having 40 percent of the VBP composite score based on these activities would seem way too high. However, in order to be “politically-correct” and keep attention on EHR adoption, I would recommend reducing the initial weight to 10 percent, and then dropping it to zero, and adding the amount of reduced weight half to quality and half to resource use.
Encouraging Alternative Payment Model (APM) Participation
These policies would seem not to apply to all APMs, but only to a subset of APMs referred to as “advanced” APMs. This subset requires (a) two-sided financial risk, i.e., both gains and losses, (b) an unspecified “quality measurement component”, and (c) an APM “revenue threshold” of either a rising share (25 percent 2016-17; 50 percent 2018-19; 75 percent 2020-21) of their total Medicare revenues [APM$ ÷ (APM$ + FFS$)], or a rising share (50 percent 2018-20; 75 percent 2020-21) of their total, all-payer revenues along with a 25 percent threshold for their Medicare revenues. Professionals participating in such APMs would be (1) rewarded with “a five percent bonus [on their total Medicare revenues] each year from 2016-2021”, (2) exempted from the VBP Payment Program, and (3) exempted from “the EHR meaningful use information exchange” and from “quality reporting requirements”, needing instead only “to use a certified EHR product”.
The Discussion Draft indicates that only this financial model should be singled out for encouragement. Unfortunately, without any documented evidence of actual performance or advantage, I do not yet see the basis for this conclusion. Such proposed rewards and exemptions would be “costly” both in terms of added Medicare expenditures and reduced Medicare information, except for the likelihood that few APMs might ever qualify as an “advanced” one as proposed. Also, we must consider the future consequences of reorganizing professionals into a model that ultimately could fail to meet expectations. However, rather than disparage the specifics of this model, I will instead offer recommendations to achieve the objective in a more balanced and cautious way.
First and foremost, professionals in all APMs should be included in the VBP payment program. This would allow their performance to be compared to all others in Medicare – a “level playing field”. If it comes to pass that their care coordination activities, etc., and risk-based financing achieve improvements to patient quality and resource use, then they will qualify for high levels of VBP incentive payments. Of course, these payments would not count in the computation of an APM’s savings or losses. Let’s “pay for performance” rather than “pay for participation”, just like the proposal would pay for quality rather than pay for reporting and would focus on outcome measures rather than process measures. Also, by having APM patient data in the VBP program, we could compare professionals among APMs and between APMs and FFS, as well as avoid the undesirable consequences of comparing FFS professionals only to themselves.
Second, I believe that encouragement of professionals to participate in APMs would be better achieved by adjusting the specific terms of the APM financial models to be more favorable to the plans, especially during their initial performance periods. An aggregate appropriated amount could be established for this purpose. Finally, if encouraging APM start-up is still considered desirable, a provision could be written to create, for example, a one or two-year “signing bonus” (e.g., five percent of Medicare revenues) for 2015 and/or 2016, the years before the VBP program would begin.
The bipartisan and bicameral Discussion Draft for SGR Repeal and Medicare Physician Payment Reform is an impressive example of Congressional cooperation that reflects a tremendous amount of effort, insight, and leadership in moving this troubled payment system in the right direction. I believe it offers an outline of reforms that, with a few but critical changes, will be highly effective in advancing the future viability of the Medicare program. Obviously, we will have to await the legislative text before drawing definitive conclusions.