House Committee on Ways and Means

Statement of Jeffery Rich, M.D., Chairman, The Society for Thoracic Surgeon’s Task Force on Pay for Performance, Norfolk, Virginia

Testimony Before the Subcommittee on Health
of the House Committee on Ways and Means

March 15, 2005

Thank you Madame Chairman and members of the subcommittee for inviting me to speak with you today about quality measurement, quality improvement, and clinically appropriate and achievable cost containment.   I believe that for the first time since the inception of Medicare, the attainment of “value”, or higher quality for each Medicare dollar spent, is within our reach.  I am here today to demonstrate to you that within our specialty of cardiothoracic surgery and applicable to all of medicine in general, there is a developing body of evidence that links quality improvement to cost containment in healthcare delivery.

We have all witnessed the past and present attempts to contain costs in U.S. healthcare delivery: the poorly designed control of access and resource utilization by HMOs and other payers; the application of the principle of “picking the low hanging fruit” by streamlining purchasing, eliminating easily identifiable excesses, and discharging patients earlier without appropriate safety nets; the attempts to control physician services through the Sustainable Growth Rate Formula.  Although some short-term transient gains have selectively been realized, we remain in a healthcare financing crisis with costs rising at multiples of inflation and an unsustainable physician payment system.  The hearing today focuses on quality based payments to physicians and their impact on cost of care efficiencies for Medicare beneficiaries.  Is Pay for Performance an answer to the problem?

The Society of Thoracic Surgery believes that the answer to this is “yes” if done correctly.  We believe that only through a focus on quality can sustainable reductions in healthcare costs be achieved.  By lowering complications and using quality-guided resource utilization management, savings can be achieved for all of medicine, with these savings accruing immediately.  But to accomplish this we must collect clinically relevant data and allow providers to develop reliable, valid and trusted measures of care that are scientifically credible. They must then be used to guide quality improvement and meaningful, safe cost containment.  This is exactly what we have accomplished through the use of the Society of Thoracic Surgery National Cardiac Database (STS NCD).

My name is Jeffrey Rich, and I am a practicing cardiothoracic surgeon in Norfolk, Virginia.  I am testifying today on behalf of The Society of Thoracic Surgeons (STS), where I serve on the Board of Directors, and chair the STS Pay for Performance Task Force.

I am also a board member of the National Quality Forum, and serve as Chairman of the NQF Research and Quality Improvement Council.  Last, but not least, I am the Chair of the Board of Directors of the Virginia Cardiac Surgery Quality Initiative (VCSQI), a regional STS-based consortium that is in the process of demonstrating the link between quality improvement and cost containment in cardiac surgery.

The Society of Thoracic Surgeons is in a unique position among physician specialties for one reason.  Our cardiac surgeons have been collecting uniform clinical data on their patients for the purpose of quality improvement for fifteen years.   With over 2.7 million patient records in the STS NCD we have been able to learn valuable lessons about what works and what does not in physician quality measurement and improvement.   I would like to share with you how we measure quality of care in cardiac surgery, and how our experience can be used to simultaneously improve care for our beneficiaries and reduce costs to the health care system.  In short, physicians can save lives and improve health while saving money.

I would also like to discuss the process for achieving Medicare savings in pay for performance, and outline what we believe is a road map that should get all physicians to the point where savings can be generated through higher quality, reduced complications, and more efficient care. 

We are well aware that most physician groups are not yet ready to participate in pay for performance.  However, we do not believe that this should be a barrier to moving forward with this important new concept in physician reimbursement.  There are three action items that Congress, CMS, and all physicians along with their respective specialty societies can take to reach the level where quality can be improved and value can be achieved.  They are:

1. Adopt structural measures using Pay for Participation:  In the March 2005 Report to Congress, MedPAC urged the development of clinical IT systems by physicians and that “functions of IT systems that are linked to quality improvements be included as measures in pay-for-performance initiatives.”  Creating incentives for the collection of relevant clinical data by providers – eventually through electronic health records (EHR)- is the cornerstone of quality improvement.  This is best accomplished through the development of relevant measures by providers and collected through participation in a database. 
2. Develop a consensus set of process measures for each specialty or disease area that is linked to quality improvement, and 
3. Develop a consensus set of risk-adjusted outcomes measures that will lead to reductions in death and complications.  Both sets of measures should be subject to the consensus-building process at the National Quality Forum.

Systematic participation in a standardized clinical database should be used to foster a culture of quality and quality improvement utilizing data collection, analysis, the development of evidence-based medicine and Continuous Quality Improvement (CQI) processes of care for performance improvement and cost containment. These steps are the road map to higher quality care and lower costs for all Medicare beneficiaries.  Incentives should be created to help physicians reach each of these levels.

Structural measures - We must start with clinical data

The most important initial ingredient in quality measurement and quality improvement is uniform clinical data.  This is where the rubber meets the road in determining what works in healthcare.  Claims based data – though easy to collect – presently have limited application for quality improvement.  Claims or administrative data simply will not allow physicians to make the crucial links between co-morbidities or disease conditions and the relative outcomes of treatments for their patients.  Additionally, a Virginia study showed that errors in Medicare administrative data varied widely across hospitals, averaging 15% statewide when compared to the STS National Cardiac Database (NCD).

The STS began collecting clinical data on open-heart surgery patients in 1989.  We now have over 2.7 million patient surgeries in our National Cardiac Database, collected from almost 600 heart surgery programs across the country.  This database contains nearly 200 data points on each patient, ranging from demographic factors to clinical risk factors, encompassing the whole spectrum of the complexities of cardiac surgery

Fifty-eight peer reviewed studies have been published using our database;  11 more await publication.  These studies have improved quality of care in areas from racial and gender disparities to efficacy of specific devices and techniques. 

Since the creation of the database in 1989, we have documented the trend that our Medicare patients have become sicker, older, more overweight, with a higher prevalence of previous cardiologic interventions.  In short, the expected mortality rate for bypass patients has significantly increased by approximately 35%.  However, over the same period from 1990 to 1999, both the observed and risk-adjusted mortality in this Medicare CABG population decreased by approximately 30%.  The chart below shows that risk-adjusted mortality rates have dropped markedly despite this increase in preoperative risk.  It is important to note that over this period, Medicare payment rates for CABG surgery decreased approximately 40% as shown in the lower line on the chart.  All of these trends, increasing expected mortality, decreasing observed mortality and decreasing payment rates have continued through 2003.

So, how did we improve survival despite increasing co-morbidities? The answer is that we, as a specialty society have developed the infrastructure through the NCD for the collection, analysis, and feedback of local data compared against regional and national benchmarks of care.  This process of collection and sharing of clinical data led to significant improvements in quality.  Our physicians consider it an important part of their professional responsibility to continually improve the quality of the care they provide.

In essence, we feel that a correctly designed “pay for participation” model as a start to rewarding performance will lead to quality improvement in cardiac surgery and other areas of medicine. 

Process measures - Clinical interventions that improve care must be communicated

The existence of this clinical database, which is warehoused and analyzed at the Duke Clinical Research Institute, has allowed us to make a quantum leap in quality measurement and quality improvement.   With these clinical data,  we have completed the largest randomized trial in medicine of Continuous Quality Improvement in a study sponsored by the Agency for Healthcare Research and Quality.  This national trial studied 267,917 patients undergoing coronary artery bypass graft (CABG) surgery at 400 hospitals across the country.  Identified were two potential best practices, which when communicated to our physicians, altered their behavior significantly in a period of 18 months (Ferguson, JAMA 2003; 290, 49-56).  This proved the ability to rapidly communicate, improve, and measure two care processes in medicine.  Importantly, it was demonstrated at the end of the trial from parallel studies that incorporation of these two measures into clinical care reduced risk-adjusted mortality for CABG; this link to improved mortality provides the scientific basis for incorporating these process measures into evidence-based practice for CABG patients.

The success of this trial highlighted the shortcoming of implementing a bonus payment system that rewards compliance with process only, without linkage to improvement in outcomes.   A pay for performance system designed like this could pay bonuses to doctors to prescribe more medications and order more tests that may have little clinical relevance to the care they provide.   Such a process-oriented system has the potential to increase costs, with little if any knowledge of whether the patient’s condition actually improved or if complications, ER visits, and other problems were reduced.

To ensure that improvements in processes of care actually improve Medicare beneficiary health in the real world, we must measure risk-adjusted outcomes.

Outcome measures – Risk-adjusted patient outcomes must be measured to show health quality improvement

With the use of the STS database, we are able to correlate performance measures to outcomes and judge their relative impact on patient health and survival. As mentioned earlier, in an era of increasingly older and more severely ill patients, the mortality for Coronary Artery Bypass Grafting has fallen.  This alone has validated the concept that participation in a clinical specialty driven database without linkage to payment has worked to improve patient care.  Last year in an unprecedented move by a physician specialty group the STS worked with the National Quality Forum to create the “National Voluntary Consensus Standards for Cardiac Surgery” through their consensus building process.  Out of over 160 proposed measures, the NQF Board approved a set of 21 measures that are most relevant to cardiac surgery. Of the 21, 16 are derived from the STS database.   

Now, armed with consensus measures (six of which are risk-adjusted outcome measures), and a clinical database, all stakeholders can evaluate cardiac surgical care with a level playing field across the nation using clinical data, processes, and outcomes.  Every cardiac surgical program can be measured against the same yardstick.  This allows doctors to see where care in specific areas can be improved, with the ability to analyze the techniques of “Best Practices” and apply these processes of care to improve quality and lower costs in their own practices.

Of course, the integrity of the data is crucial. The STS is developing a three-part approach for validating the data in its database.  First, there are internal checks for data accuracy with rejection of data that are out-of-bounds.  This will be coupled with a newly developed on-site audit.  Secondly, the STS is in discussions with CMS about a partnership involving a chart abstraction audit through their CDAC mechanism.  We are hopeful that this will be approved at CMS shortly.  And lastly, we are pursuing a longer-range validation of mortality data by using the social security National Death Index to validate deaths 18 months after surgery.

Validation and audit mechanisms allow both providers and payers to rely on data for quality and cost implications.  Trust is the foundation that physicians must have to participate in the process and to make changes in their care patterns based on feedback they get from a database. It is important that the STS NCD is a VOLUNTARY effort by the participants performing these cardiac surgical procedures, and an example of what the medical profession can do when agendas are aligned.

In Virginia, we took this quality improvement feedback loop one step further.  Fellow heart surgeons and I established the Virginia Cardiac Surgery Quality Initiative (VCSQI), to systematically improve care while reducing costs.  The VCSQI is a voluntary consortium of 16 hospitals and 10 cardiac surgical practices providing open-heart surgery in Virginia. They are diverse in patient population, geographic location, size and resources.

We used the clinical quality data in the STS database, and using a third-party software solution from ARMUS Corporation, mapped it to the Medicare Part A payment data from the standardized UB-92 files.  This enables us to examine the relationship between quality improvement and cost, and to address the question of whether improved quality can equal reduced costs.  In short we can now evaluate VALUE in health care delivery.

Please allow me to give you a demonstration of how we are in the process of reducing costs by improving quality using real data from actual patients in Virginia.

We allocated all costs into 21 revenue categories to better illustrate where resources were being spent.  These categories are shown below, and include drugs, ICU costs, OR costs, lab, etc.

Now we can compare the hospitals with lower spending to those with higher spending in each category.  We can examine how higher spending in any particular category correlated with outcomes, and can identify interventions or treatment protocols that lead to better and more efficient care.  By maintaining a focus on quality we can begin to examine resource utilization management that is patient-centered, safe and without negative consequences.  In short, we have developed cost savings models that lead to improved quality of care.

We also measure costs by surgeon…….

And when we compared spending to mortality rates, we found that higher spending does not necessarily equal higher quality.  In fact, in Virginia, the lowest spending hospital had the lowest observed to expected mortality ratio.

One of the key questions facing you today is how pay for performance can be implemented in the current difficult budget environment.  The recommendation to repeal the SGR and replace it with incentive payments to physicians enabling them to make the IT investments required for quality improvement will not be easy, and must be done thoughtfully. 

A budget neutral framework for pay for performance must not be the “tournament model” where the funds are taken from the lower performers and given to the best.  We feel strongly that the tournament model will not produce the savings you seek and could hurt access to care by vulnerable populations.   By punishing lower quartile providers, system capacity may be reduced, adversely affecting disadvantaged or minority patients. The primary goal of pay for performance must be quality improvement.  Savings will accrue from improved quality.

Let me demonstrate how this works using an actual example from our Virginia initiative:

One common complication from open-heart surgery is atrial fibrillation.  This is where the heart’s electrical rhythm is out of sync and cannot pump blood efficiently.  While not often deadly, we found that each instance of atrial fibrillation (A-fib) adds $2,366 to the direct cost of care during the hospitalization.  In addition, it can lead to much more serious and expensive consequences such as stroke and hospital readmissions. 

In analyzing our cost and quality data, we found that one hospital had significantly lower rates of A-fib after surgery.  While the rate statewide was 16%, this “best practice” had a rate of 10% A-fib. The treatment protocol to accomplish this was shared with all other programs in the state and implemented within their practices. With an anticipated reduction of A-fib statewide to 10%, estimated cost savings to the healthcare system will be $1.3 million dollars every two years.

When extrapolated nationally using the STS database, cost savings can reach as much as $80 million dollars over 2 years

Again, a key point here is that to achieve savings, you must improve the care of the lower performers until inter-institutional variation is minimized, and all quality is improved. That improvement would not have occurred had there been a budget neutral or “tournament style” P4P system in place. The creation of winners and losers discourages the best performers from sharing their best practices with others.  In short, up-front budget neutrality, that robs Peter to pay Paul, stifles the communication that is essential to quality improvement and cost containment.  The strides that the VCSQI is making in quality improvement would not be occurring without effective and open communications. 

In Virginia, we can also measure the costs of other common complications following cardiac surgery, and can show the incremental cost of each complication, as in the table below: 

Using modest estimates of achievable reductions in the rate of each of these complications, one can estimate the potential savings to Medicare beneficiaries at the national level using the STS database and its CQI processes. 

As you can see, we believe that through achievable improvements in quality, we can save $346 million in the U.S. each year by reducing these 5 complications in cardiac surgery.  That equals a billion dollars every 3 years - imagine the savings you could achieve if all physicians were systematically participating in a clinical database and its associated CQI processes and reducing their specialty-specific complications.  This system is designed to be replicated in other specialties and implemented in outpatient and chronic disease care as long as the process improvements are linked to outcomes measurement and quality improvement.

This is where the funding for pay for performance should be generated.  Incentives to reduce costly complications have immediate savings potential for the healthcare system.  We strongly disagree with the MedPAC recommendation to use an across-the-board reduction from all physician fees to create a bonus pool.

Although that approach would be a windfall for cardiac surgeons who are ready, it would likely have the unintended effect of taking resources from those who need them most to invest in health IT and develop clinical datasets.

Physician practices are very different than hospital systems in terms of their readiness and ability to purchase needed technology.  Reducing physician fees would not be the positive incentive needed for investment in new systems.  STS database participants pay an average of $50,000 per practice to submit and analyze clinical data.  They must purchase software, hire a data manager, and spend their time improving care processes.  These costs are not reimbursed in any way and they are not recognized by Medicare.

This is also why currently mandated SGR cuts threaten our ability to move forward with quality improvement.  Facing larger than 5 percent fee reductions each year, physician practices are not in a position to invest scarce funds in new technology.  We believe that the answer to inappropriate care lies in performance measures based on clinical data created by each specialty.  Compounding the inability to invest is the uncertainty brought by the lack of standards for electronic health records.

Lastly, we appreciate the need for CBO and OMB to have evidence of more tangible, immediate savings in order to score these programs accurately. In this regard, I would recommend a national demonstration project to evaluate the effect of incentives on participation in the STS National Cardiac Database and to document the cost savings that can be achieved through a collaborative quality improvement effort. This can be instituted immediately given the high level of readiness of the STS with the potential for rapid replication in other specialties.

Congress and CMS must recognize that not all physicians have reached the same level of readiness as the STS, but must also recognize that the system of quality improvement and cost containment employed by the STS can be effective for every physician participating in Medicare. With this in mind, we believe that incentives should be established to encourage development and attainment of each component of a meaningful Pay for Performance Program that lends itself to quality measurement and improvement:

1. Structural measures – pay for participation: collect, analyze and share clinical data with providers.
2. Process measures – develop clinically relevant measures through a voluntary consensus process and measure compliance with a link to patient outcomes.
3. Outcome measures – develop consensus risk-adjusted measures of patient outcomes to evaluate better care.
4. Compare costs and link quality improvement to cost savings.

It is crucial to understand that these incentives must be positive updates to the current Medicare rate.  The avoidance of reductions in payment as proposed by MedPAC is not incentive enough for physicians to make the IT investments necessary to participate in these programs. More importantly, a system whereby the best performers are rewarded by reductions in pay to the lowest performers is counterintuitive to the spirit necessary to allow sharing of best practices.  This approach will pit providers in the healthcare system against each other, stifling improvement and ultimately cost containment.  In fact, many argue that the biggest incremental gains in quality improvement will occur by focusing on the lowest performers, and that incentives should be provided there equally.

For these reasons Congress and CMS must do everything in their power to create incentives that promote the inclusive collaboration of physicians and all providers to improve healthcare quality for Medicare beneficiaries AND contain costs.  If a common theme must emerge then let that be one of “Include and Improve” rather than “Divide and Conquer”. 

In conclusion, we believe the answer to many of the questions policy makers have sought in health care is to re-engage the profession in husbanding what is an increasingly scarce resource, the health care dollar.  We believe that incentivizing and supporting the development of condition-specific databases is one step in that process.  Comparative effectiveness of treatments, long term efficacy of drugs and devices, appropriateness criteria for utilization, and racial or gender disparities can all be answered with valid clinical data.

If we have the foresight to prevent cuts mandated by an ineffective formula, it will allow us to take the major steps that will move us from making budget-based health policy, to making clinically appropriate health policy.  And that is what our patients and your beneficiaries deserve.

Thank you for this opportunity and honor to appear before you.