Tuesday, August 18, 2009

Healthcare Reform: What's Missing

It is the middle of August, the Public Option is on life support, and it won't be getting End of Life Counseling. Comparative Effectiveness Research is alive but may have to operate with one hand tied behind its back. But the healthcare reform effort continues to move forward, bloody and only slightly bowed by a combination of special interests and irrational fears. And there are a good number of important and constructive changes that are likely to be enacted and which will contribute to better quality and control of costs: near universal coverage; more insurance sector competition (likely in the form of not for profit cooperatives); elimination of pre-existing condition exclusions; mandated minimum coverage requirements including preventive services; electronic health records; etc.

The politics of healthcare reform in 2009 have confirmed, not for the first time and certainly not for the last, something I used to stress to my political science students 30 years ago: the U.S. political system isn't designed to do major programmatic reform; we do incremental change, usually in small steps. That is often a good thing, keeping both conflict and risk dampened. It can be a problem in crises that can only be respolved by major and rapid directional change. Time will tell whether we are doing enough fast enough to correct the faults that are increasingly apparent in our healthcare delivery and financing systems.

I can, however, point to at least two possible reforms that are consistent with the expressed goals of the reform effort, that could make a substantial contribution to quality improvement and cost control, but that are not being considered: payment for clinical decision support systems; and capitation to implement pay-for-performance principles. In different ways, each of these measures would address flaws in the dominant fee-for-service healthcare payment system. In diferent ways, their challenge to the fee-for-service model makes them politically difficult. I'll address each separately.

Clinical decision support systems. Every day our store of data linking linking personal physical and/or genetic characteristics and diagnostic test results to the appropriateness and effectiveness of therapeutic options grows. In the near future, with widespread use of electronic health recorda and broadened funding for comparative effectiveness research, the rate of accretion of new dat will increase dramatically. There is more to know, more information to process and incorporate into clinical practice, than any one can relly keep up with. One solution to this information overload is the ongoing development and implementation of sophisticated clinical decision support systems - computerized models that provide practitioners with the data management and processing tools they need to find the right therapy for the right patient at the right time - the fundamental goal of the emerging personalized medicine movement, the necessary mantra for giving evey patient the best care we know how to provide. Well designed and maintained clinical decision support systems will assure that therapy prescription is not one size fits all, that comparative effectiveness findings will not be used to dictate the same treatment for everyone, and that every clinician has effective and efficient access to the most recent research and clinical practice findings that bear on the patient before them.

Broadly implemented, such systems would yield enormous quality improvement and cost reduction. But they will also be quite expensive to develop, distribute and maintain. The investment would yield a very high payback to society ... but the investment needs to be made, and the tragic fact is that our current healthcare financing system provides no obvious mechanism for a payback to a private investor (or group). There is currently no way to get paid for using a decision support system, and therefore no way for a physician or group or hospital to recoup the capital and maintenance costs associated with using it. In the context of a fee-for-service system, the upfront cost of installation for a non-revenue-producing tool would be prohibitive; and that means that the upfront cost of development would also be prohibitive. Maybe some extraordinarily wealthy foundation would step up to the development cost; still no implementation funding. The government could afford to fund development, and perhaps some implementation - but if end of life counseling raised fears of mandatory euthenasia, what would be the response to government-developed treatment protocols imbedded in a computerized decision (support) system? The political hurdle is daunting.

Here is an idea. Any physician, group or institutional provider acquiring, implementing and using a "certified" clinical decision support system would receive 105% (I don't know if that is the right number, but you get the idea) of the established fee-for-service payment amount for every service billed. The end user has a rationale for buying the system; the potential develop has a customer base willing to pay .... The system savings would be fa greater than the 5%, and quality would be greatly enhanced. Do it all in the private sector. It might just work.

Capitation. Sceptics say we tried capitation in the eighties and it failed; the people rejected the capitated HMO model. The unconvinced say that the capitation model was shanghaied by the HMOs' cost accountants, who found it easier to make money by squeezing resources out than by actually managing care. Now we talk about pay-for-performance, and implement it by imposing penalties for failure to achieve quantitative quality targets - a step in the right direction, but a mixed message at best.

Once again, Massachusetts may lead the way on an important healthcare reform dimension. It is unlikely to be by implementing a major systemic change, but Mass. has already developed one feasible model for implementing near-universal capitation without major disruption to the insurance coverage system. The work was done by the Special Commission on the Healthcare Payment System, a body created by the General Court in Section 44 of Chapter 305 of the Acts of 2008, and is described in the Commission's Recommendations issued July 16, 2009. The report was referenced in news articles when issued, and has receded from view as the national reform debate heated up. Whether it will surface again as a serious proposal remains to be seen.

The Special Commission's proposal, based primarily on work done by social policy research firm Mathematica, calls for insurers to make capitated payments, risk-adjusted in order to prevent selection bias, to Accountable Care Organizations (ACOs). ACOs, which could take any number of forms, would then be responsible for providing all of the necessary care enrollees require, either directly or through contracts with providers. The critical aspect of the plan is that there would be a strong incentive to provide preventive care and early intervention in order to avoid the high cost associated with unnecessary acute episodes of care, equally strong incentives to design and implement more cost-effective and less waseful or duplicative, models for the delivery of care. Patients would retain choice, but might have to pay a premium for "out of network" providers.

There are problems with the Special Commission's recommendations. For one thing, the system would need to encompass both public and private payers, and that would require a statewide Medicare waiver. For another, reconciling a major state model change with the substantial national reforms likely to be enacted will be an enormous challenge in cordination. But Massachusetts has at the very least generated a creative and plausible model for changing the perverse incentives of the fee-for-service health payment system without destroying the underlying structure of our health insurance system. It is a viable model for achieving many of the goals of healthcare reform that are well recognized but poorly addressed in the proposals under review by the Congress.

Tuesday, August 4, 2009

Medicare's Inpatient New Technology Add-on: The Trials of InfraReDx, the Triumph of Spiration

The problem of adequate payment for innovative new technologies has from the beginning been a difficult one for the Medicare Inpatient Prospective Payment System (IPPS) for acute care hospitals. Since cases are assigned to DRGs on the basis of diagnosis and/or surgical procedure, and payment for each DRG is based primarily upon historical cost data, the system from the outset had no mechanism to properly recognize the cost of newly introduced technologies that improved outcomes but entailed incremental spending.

Effective FY 2002, Medicare implemented a mechanism - the Inpatient New Technology Add-on - designed to address this issue (see 66 CFR 46917). Under the new program, if a qualifying new technology added substantial cost, Medicare payment to the hospital would be the applicable DRG payment plus 50% of the cost attributable to the new technology. The Program held, over hospital and medtech industry protest, that 50% payment struck a good balance between a financial impediment to technology adoption and a blank check. Besides, the add-on was simply a temporary mechanism, allowing reasonable adoption until the cost of the technology could be reflected in the standard calculation of payment for the DRG. The most notable observations from 9 years of experience with the add-on (including decisions just announced for FY 2010) are the surprisingly small number of technologies that have applied for the extra payment, how few of the applications have been successful, and how minor an impediment the 50% limit on incremental payment has been. We'll use two applicants for FY 2010 - InfraReDx's Lipiscan vulnerable plaque diagnostic and Spiration's IBV intrabronchial valve system - to document the process and the problem.

CMS has clearly articulated 3 standards that must be met for new tech add-on qualification:
  1. Newness - the first commercial distribution of the technology must be sufficiently recent so that cost of the technology is not reflected in the data used to calculate DRG relative weights; effectively, this means first commercial introduction less than 2 years prior to application; commercial introduction is the FDA clearance date, unless there is some reason distribution was delayed;
  2. Cost - the incremental cost attributable to the new technology must bring the charge per case above the lesser of a) 1.75 times the mean standardized charge (MSC) for the DRG or b) the MSC for the DRG + 75% of one standard deviation above the MSC; CMS annually publishes the threshold amount for each DRG; if a technology would be used for patients in multiple DRGs a volume-weighted MSC is calculated; and
  3. Substantial clinical improvement - there must be evidence that the new technology provides real clinical benefits as compared to existing technologies or services.
Only one technology qualified for the new technology add-on for FY 2009 - the Syncardia CardioWest temporary total artificial heart. On its face, the CardioWest device is anything but new: a direct descendant of the Jarvik 7 heart, CardioWest has been used as a bridge to transplant with FDA approval since October 2004 (after use in clinical trials for many years). But because artificial hearts were denied Medicare coverage until a coverage policy revision in May 2008, data on the cost of the CardioWest has never found its way into calculation of DRG payment rates. Consequently, Syncardia's application for an add-on was accepted for FY 2009, and has now been extended to FY 2010. It will likely qualify for a third and final year of add-on payments in 2011. The case for meeting the cost threshold was ironclad, and the CardioWest does inarguably succssfully bridge some patients to transplant who have no other therapeutic option.

Only two new add-on applications were pursued to the end of the process for FY 2010 (four applications referenced in the NPRM for FY 2010 were withdrawn for various reasons - different forms of "inevitable rejection" - before the Final Rule was completed). In the case of the InfraReDx Lipiscan, CMS had in the NPRM questioned whether the device met any of the three qualifying criteria, and requested public comment on all three. The challenge re: the "newness" criterion was most interesting. The Lipiscan was cleared for marketing in 2008 under the 510k route, referencing as predicate an earlier InfraReDx technology first cleared, for a different indication, in 2006. How, CMS asked, could a device be claimed to be "substantially equivalent" to existing technology in an application to one agency but new and innovative in an application to another. Further, CMS quoted FDA's 510k approval letter to make its point:

“The LipiScan Coronary Imaging System utilizes the same basic catheter design as the predicate, the InfraReDx NIR Imaging System (June 23, 2006). These devices have a similar intended use, use the same operating principal, incorporate the same basic catheter design, have the same shelf life, and are packaged using the same materials and processes. The modifications from the lnfraReDx NIR Imaging System to the LipiScan Coronary Imaging System are the improved
catheter design, improved user interface (including PBR and console), and the additional
testing required to support an expanded indication for use.”

Was the new indication for use sufficient to establish Lipiscan as "new" for add-on payment purposes? The answer seemed to be "no", but the question was left open for public comment. And InfraReDx ultimately had a compelling comment to make - the predicate device had never been distributed because it had commercially unacceptable operating problems - i.e. it didn't work. No one had ever bought one, so no cost for it had ever been reflected in DRG weights. An interesting inversion: if the predicate device had been clinically and operationally acceptable, the Lipiscan would have failed the newness criterion.

This one victory was not sufficient to carry the day. Nor was CMS' eventual agreement, after much parsing of numbers, that LipiScan met the cost criterion. The sticking point became substantial clinical improvement. InfraReDx relied on the fact that it had demonstrated a diagnostic capability - the identification of vlulnerable plaque - that was wholly original, and that the use of that diagnostic capability would allow better identification of risks and targeting of therapeutic interventions. It even submitted an opinion from counsel arguing (in essence) that under the applicable statute "better diagnosis" constituted "clinical benefit". This may have been a tactical error. CMS acknowledged the unique diagnostic capability of Lipiscan, but concluded that there was no documented clinical evidence to demonstrate, or any consensus among clinicians, that the diagnostic - whatever its virtues - produced improved outcomes. That, the agency insisted, was their standard: application denied.

On the newness criterion, the Spiration IBV valve system confronted a problem opposite to that of both the Lipiscan and the CardioWest heart. The IBV was approved by FDA under the Humanitarian Device Exemption (HDE) mechanism in 2008. There was now comparable prior product on the market. But use under an HDE requires approval by each user hosital's IRB, and at publication of the NPRM no IRB approvals had yet been received. As a device cannot be new until it receives final regulatory approval, qualification for the add-on depended upon receipt of IRB approval before publication of the Final Rule for FY 2010. Happily, Spiration was able to document initial IRB approval on March 12, 2009, and that was recognized by CMS as the initial commercial distribution date for the IBV System.

Spiration's challenge with regard to substantial clinical benefit was closely related to the IBV's HDE approval route. An HDE requires demonstration of "safety and potential benefit", far less than the demonstration of safety and effectiveness required by the PMA process. HDEs are typically granted based on very limited clinical trials - in Spiration's case, fewer than 10 patients. There was no randomized clinical trial required for HDE approval, and no data from a randomized trial available for review. The NPRM questioned whether there was evidence of cliical benefit and requested public comment.

The comment received was universally positive. Every commenter made essentially the same case from personl experience: postoperative air leaks from pulmonary surgery are extremely dificult to treat, clinically problemmatic, and expensive; the experience with the IBV was positive; the system appeared to close air leaks effectively. CMS, impressed by the consistency of comments from the clinical community, accepted the propositio being argued. The final rule makes it clear that the agency isn't fully and finally convinced, but that the argument is strong enough to carry the day: application approved.

Both the Lipiscan and the IBV System ultimately were judged to qualify under the cost criterion, but in bolth cases the process was laborious. One reason was that both technologies could be used on patient swho fell into a large number of different DRGs, and the method for calculating a weighted standardized cost across mltiple DRGs is convoluted. Furthermore, there is a lot of moving back and forth between costs and charges required in the calculation; every time charge data must be converted to costs, or cost data must be converted to charges, there are statistical manipulations that are clear to CMS but not to many outside the agency. Professional help is essential here. But a further interesting problem was that both InfraReDx and Spiration refused to provide information on the selling price of their devices - even though that information would have simplified the calculations enormously and (in both of these cases) guaranteed meeting the criterion. Spiration went even further to protect proprietary data: it would not give CMS data on the number of valves used in each procedure (the number can vary depending on patient-specific needs), necessitating an enormously convoluted and potentially misleading process of inference.