Data, use and misuse, regional variation, Data-Driven Health Care: An Interview with Jerry Reeves, MD

This interview appeared on June 8 in The Health Carre Blog. In conducting interviews, my policy it to let the interviewee speak his/her mind at length. The resulting interview does not represent my opinion.


An under-the-radar debate is occurring in health care between those who say data shows that practice variations across the land are “unwarranted” and those who maintain that such variation is inevitable given socioeconomic population differences and cost of practice differences in major metropolitan and rural areas.

• Data transparency proponents say costs of medical practice, to achieve the same outcomes, should not vary much.

• Data transparency opponents say data can be shaped to fit a premise that variation is unwarranted, while ignoring the human and economic realities and inherent variability driven by different regional cultures.

That Medicare law forbids doctors to compare fee schedules to avoid monopolitistic behavior, that costs of episodes of care vary greatly with points of patient entry into the system, that third parties generally set physician payments, and that reformers and physicians have fundamentally different economic points of view confounds and complicates the argument.

What follows is an interview with Jerry Reeves, MD, an articulate spokesperson for using data to reduce practice variation, promoting value-based purchasing by payers, and achieving higher levels of physician performance.

In God we trust, all others bring data.


W. Edwards Deming. 1900-1993.American statistician who taught top management how to improve design, product quality, and sales in global markets.

Researchers have estimated nearly 30 percent of Medicine’s costs could be saved without negatively affecting health outcomes if spending in high- and medium-cost areas could be reduced to the level of low-cost areas – and those estimates could probably be extrapolated to the health system as a whole.

Peter Orszag, Director of the Congressional Budget Office, “Opportunities to Increase Efficiencies in Health Care," statement before Congress, June 16, 2008

It is no surprise that the I.T. candidate, Barack Obama, is intent on being the I.T. president. To succeed, he will have to remind his administration early and often, that he is committed to transparency – and that the threat of embarrassment is no justification for secrecy.

“Data.gov,” New York Times editorial, May 26, 2009
*****
Prelude:

Dr. Reeves is Chief Medical Officer of Hotel Employees and Restaurant Employees International Union (H.E.R.E.I.U.) Welfare Fund. The Fund offers multi-employer health insurance coverage for 90,000 eligible employees and their family members. He is also Principal of Health Innovations LLC which provides health benefits, wellness, and health management consulting services for health plan sponsors and coalitions. He is a Director and Chairman of the Board of Health Insight, the Quality Improvement Organization for Nevada. And he is Medical Director of the Nevada Business Coalition for Health Improvement. Dr. Reeves previously served as Chairman of WorldDoc Inc., Chief Medical Officer of Humana Inc. and Sierra Health Services Inc., and as Chief of Clinical Medicine at USAF Headquarters in Europe. He served two terms on the Board of Health of the State of Nevada and has served on the faculty of three medical schools.
*****

Q: What does Health Innovations do?

Reeves: We work with self-insured employers, Taft-Hartley Trusts, and business coalitions on health to promote programs of transparency and accountability that engage hospitals, doctors and patients in improving their health with a focus on primary and secondary prevention and wellness. We engage people in more rational lifestyle choices.

Q: And what about your work at HEREIU Welfare Fund, the health plan for Hotel Employees and Restaurant Employees International Union members and their families?

Reeves: We align incentives in the benefit designs of our health insurance coverage to engage beneficiaries in taking their medications, getting their tests, and seeing the same doctor regularly. We collaborate with doctors, hospitals and utilization management firms to modify behaviors that generate excessive costs and to minimize payments for low value services to achieve better control of costs and outcomes. Much of my work involves analyzing medical and pharmacy claims data and other self reported data reflecting patient risks and provider practice patterns. The findings help us prioritize interventions. We then use ongoing measures to monitor and improve the impacts of our chosen interventions.

Q: You have been doing a study and giving a slide presentation with the title “Variations of Care, Comparisons of more than 450 Episode Treatment Groups - Evaluating Physicians in 4 States.” Tell us about that study, and the 4 states you are talking about.

Reeves: The 4 states are Nevada, Illinois, New York, and Pennsylvania. The study is based on health plan data. We pay medical and pharmacy claims for our members who seek medical services under our health plan. We collect information from those claims and from health risk questionnaires and other biometric measures to track patterns of care.

We can compare plans geographically because we have plans spread across several states. I can see significant patterns between providers in the same specialty not only in multiple States but also in the same town. They are treating hotel and restaurant workers with many socio-demographic similarities and are paid using similar fee schedules.

The variations in physician preferences are substantial For high volume episodes in primary care – otitis media, bronchitis, urinary tract infections, or chest pain – the most expensive high volume adult primary care doctor is about 7 to 8 times more expensive per episode than the least expensive doctor in the same specialty within the same town paid on the same fee schedule. The outcomes – no more ear pain, no more cough, no more dysuria, and no more chest pain – are the same.

These variations are not just in the primary care arena. They exist in the specialty arenas as well – cardiology, ob/gyn, orthopedics. The most expensive cardiologist who takes care of an episode of angina is 5 times as expensive as the least expensive cardiologist; the most expensive orthopedic knee surgeon is on average 2 ½ times more expensive than the least expensive. We see no differences in outcomes.

So there are dramatic differences. The question is: with health care as expensive as it is already, can we as a country afford to sustain overpayment to outliers who are so much more expensive than their peers in the same town when they are each achieving the same results?

Admittedly there may be variation in the severity of the cases. But we attempt to minimize the impacts of varied illness burden by comparing multiple cases, typically more than 30 cases per doctor. That probably washes out most of the severity related variations. To make things fair, we also take the top 3% and the bottom 3% outliers out of the comparisons to make our estimates more reliable.
Are these variations sustainable in today’s economic environment and in today’s era of global competition?

Q: Are doctors you are comparing aware of what their peers are doing? Do you send them de-identified information on these variations?

Reeves: We do. We also sit down with them and share comparative information when we believe we’re using fair measures. In some cases, there’s a reasonable explanation for the variations. For instance, we learned that an outlier orthopedist was known throughout the town as the doctor who did the best job with redoing surgery for patients with failed back surgeries. That likely explained why episode costs were higher for him. We want to learn from them what a rational explanation might be. Sometimes we get logical valid explanations. What we notice, though, is that after we’ve had these discussions, the trends move more towards the middle instead of staying at the extremes.

Q: So there’s a swing towards the mean?

Reeves: Yes, there’s a regression towards the mean. The overall system becomes more efficient, and cost trends decrease.

Q: I notice you indicate variations in total costs of certain episodes of care differ greatly depending on the site of patient entry into the system - the hospital, the ER, an urgent care clinic, or the office.

Reeves: People have known for years that when the first visit is at the hospital, expenses soar. The typical hospital admission costs 12.7 times as much as an ER visit; an ER visit costs 10.7 times as much as an office visit; a hospitalization costs 136 times as much as an office visit.

When you think about it, you could get a lot more office visits for the cost of one hospital admission or one emergency room visit. We would rather pay more for patients regularly visiting their continuing care physician than going into the hospital or emergency room.

From our point of view, a patient showing up at the hospital or ER represents a failure of outpatient management. The great majority of all care should be going on between doctor and patient in less restrictive settings and safer environments than hospitals and ERs.

Q: You must offer some educational process informing your members of these cost considerations.

Reeves: We do. We expend a lot of effort on developing systems to engage our members – posters and brochures, newsletters and explanation of payment (EOP) stuffers, reminders, making available telephonic nurses and health coaches, making it easy for people to call in, making it convenient to reserve a next day appointment. We show them the comparative out of pocket cost of going to the ER and hospital and explain alternatives that can get their problems solved faster and at less cost.

There is value derived from this approach. For instance, Microsoft has found they can save a lot of money for both the company and the beneficiary if they pay for a doctor to spend an hour at a patient’s home rather than having that patient go to the ER. Also it’s more personal and more likely to lead to a continuous care process that identifies needs earlier.

Also, to align incentives with desired behaviors, we’ve made it much more costly for our members to go the ER rather than seeing their doctor. We also work with our doctors to make sure they have slots available so people can be seen within one business day when they are worried they are getting worse.

It takes two to tango. We need to engage both the doctor and the patient in improving the availability and affordability of care that improves health. We combine a number of incentives for members and information campaigns about choices resulting in less out of pocket costs, with incentives and interventions with doctors to provide more efficient care.

Q: Do you have data indicating significant cost reductions.

Reeves: We do. In Las Vegas, we ran three campaigns: one, to have more patients adhering to their chronic medications; second, a community wide campaign to champion the use of generic drugs; and third, free pharmacies where our people could come to fill prescriptions for generic drugs for chronic conditions with no out-of-pocket expense. Simultaneously, we profiled and gave performance feedback to primary care physicians displaying their apparent adherence to quality of care guidelines and their comparative efficiency expressed in terms of average costs per episode. We used things like ratios of HbA1C testing, microalbuminuria testing, hypertension management, compared to their peers – 22 quality indicators in all. The episode comparisons were for those episodes most common to their specialty with ratios expressed in comparison with the median values among their peers.

We rewarded 155 of those primary care physicians with bonuses and displayed them as Gold Star physician in our provider directory. At the same time, there were 50 doctors who persistently underperformed despite our sitting down with them and showing them their patterns. We discontinued our contracts with them for lack of a business reason to continue their contracts. At the same time, we intervened regarding the use of Oxycontin, which was being hugely overused at the time. It turned out that the physicians discontinued from the network had been prescribing more than 50% of the Oxycontin used for our whole population in that town.
The result of our suite of interventions was that out of the $268 million spent the baseline year, we saved $69 million over the next two years according to actuarial projections; our medication adherence for chronic conditions went up 8% even while our drug costs went down dramatically; and our adherence to mammography, Pap smears, and lipid management guidelines improved.

Q: Just to give us perspective, how many people live in Las Vegas and how many doctors are in your network?

Reeves: The total population is about 1.6 million and we had about 1900 doctors in our network then. Our patient base was about 120,000 lives including children then.
Q: Is inpatient cost control a different animal?

Reeves: Inpatient cost control has similar patterns. We looked at inpatient costs obligated by physicians who sometimes admit patients to hospital in 4 different states – Nevada, Illinois, New York, and Pennsylvania. In this group, we focused on the inpatient facility costs as part of the overall costs of episodes of care that might result in hospitalization.

When we compared the variance from the expected median inpatient facility costs for various episodes managed by internists, the most expensive internist was $71,000 more expensive than the least expensive internist. The most expensive cardiologist was $203,000 more expensive than the least expensive cardiologist for episodes managed by cardiologists. The discrepancy between the most expensive and least expensive general surgeons was $284,000; and for obstetricians the discrepancy was $305,000.

It appeared that a primary driver for excess cost among obstetricians related to wide variations in the prevalence of primary cesarean section deliveries (among women in their first pregnancy). Among obstetricians delivering more than 200 births per year, we had obstetricians with 54% primary c-section rates and others with 9% primary c-section rates.

It is not defensible or believable that all of these obstetricians can be right. There are some variations that go on that are frankly just plain unsafe. Malpractice insurer underwriters told me they spend much more malpractice insurance payouts for major surgery complications of c-sections than payments for babies who might have fared better from cesarean delivery.

Q: What incentives do you use to encourage doctors to perform better? What are your techniques? You’ve mentioned sitting down with them, showing them data, rewarding them with bonuses. Anything else?

Reeves: We’ve used a suite of multiple interventions. Doctors deserve multiple opportunities to correct these variations. It should be three strikes before you’re out.

The most common comment I get back is: “Nobody ever told me this.” “How come nobody has ever said this before?” It’s a little bit like patients who have previously seen urgent care doctors for quick symptom relief when you tell them they have hypertension or diabetes. They often say, “Nobody ever told me that.” The doctors are right. Few payers give comparative performance feedback to rank and file physicians.

Essentially once doctors finish their residency program, they are on their own unless they work in a large multispecialty group with internal peer review. By and large in private practice offices, there is not much performance feedback.
Q: So many private practice physicians, if you will, function in a data-vacuum?
Reeves: Exactly. Once you share the information, they will often point out deficits in the data, even though the data ultimately comes from them. We are open to challenges and want to continuously improve the quality, accuracy, and reliability of the data.

The first strike is defined by a doctor’s or hospital’s response to reviewing the data so they can internalize it and take action to address the root causes that lead to these variations.

The second strike we may take toward corrective action is sharing the information more widely so that doctors and hospitals understand payers are reluctant to keep paying extra for something that doesn’t result in a superior outcome. The people paying these bills are hurting, trying to pay their employees and to stay afloat and to compete with other companies. On a larger scale, they are trying to compete with companies in other countries making the same product but not bearing the same costs. This kind of understanding can sometimes bring accountability and behavior change. We have doctors who call us and ask when they can see their next report. Of course, most of those are performing well and want to disseminate that news.

The third strike is discontinuing contracts with physicians or hospitals that do not alter excessive charging and wasteful practices. Sometimes it may involve discontinuing payment for particular services being overused. For doctors this might mean payers would discontinue payments to a primary care physician who owns a machine for nerve conduction velocity testing or ultrasound imaging who orders dramatically more of these tests per 100 patients compared to their peers who do not own and profit from such equipment. For hospitals, CMS and some other payers have discontinued paying for “Never Events” – high costs incurred as a result of certain hospital acquired conditions like venous thromboembolic events occurring after knee or hip surgery.

Q: As you know, as a nation, we are in the hot heat of the health reform debate, and the Obama administration, particularly Peter Orszag, the budget director, has put a lot of stock in John Wennberg’s work at Dartmouth. Wennberg and his colleagues, using Medicare data, keep emphasizing that most practice variation is “unwarranted,” and the nation could save 30% on total costs by bringing down costs in high spending regions, like some large cities, to those in low spending regions, like the upper Midwest and the South.

But some critics of the Dartmouth studies, like Dr. Richard Cooper, a professor of medicine at Penn and a principal at the Leonard Davis Institute of Health Economics at Penn, have challenged the Dartmouth interpretation of the data in Health Affairs and his blog, www. Buzcooper.com by saying you can’t compare spending, say in Los Angeles with a 70% Hispanic population, many of whom are poor and sick, with Rochester, Minnesota, with a 90% white population, most of whom are well.

Is your approach similar to the Dartmouth studies on Medicare studies, except that it’s done on a more local level and includes data from commercial insurers on the under 65 population?

Reeves: The Dartmouth and Obama Administration approaches are similar except their focus is mostly on Medicare hospital data. The end-of-life years are the most expensive by far. Our data is on hotel and restaurant workers and their families, perhaps more similar to Medicaid patients, but covering a wider spectrum of diseases and conditions. Most of our costs are for prescription drugs and professional services and new technologies.

Even so, our experience is that the overall patterns of practice variation in the commercial world and the Medicare world are parallel. I believe we need to merge the doctor and hospital data from Medicare and Medicaid payments with data available from other commercial payers serving working age populations. We need to consolidate all that data into common data warehouses.

The data should include not only the professional services of doctors and other clinicians but also the laboratory and diagnostic data, the inpatient and outpatient facility data, and data on high technology hardware (imaging procedures, implants,) and drugs (biologics, cancer therapies) at least by regions so we can compare these regions against each another. Then adequate sample sizes could become generally available to analyze and display care patterns, technologies, and drugs offering the most value (best outcomes) for the dollars invested.

Q: I just read in the May 22 New York Times a piece entitled “I.B.M. Unveils Software to Process Vast Amounts of Data “for quick analyses of massive chunks of combined data. Is that what it will take to carry out your vision? Would that be a breakthrough?

Reeves: It would be breakthrough, but I do not see technology as our major challenge. The challenge is political will. We have a competitive risk issue that leads to carriers not wanting to share their data in a data warehouse for fear that proprietary rates or payments might be revealed or confidentiality agreements that they have negotiated with various providers might be breeched. Some also worry that personal health information might be revealed.

The banking industry has been able to deal with the issues relating to money for many years and has brought dramatic improvements in efficiency and choice. I remember waiting in long lines to deposit a check on payday. Now I am irritated if it takes an ATM window more than 30 seconds to complete my transaction at a drive through window at midnight. Much of the fear of disclosure of health information is already addressed by HIPAA law and protected by reliable systems of maintaining confidentiality and security.

I cannot emphasize enough how important it is to merge the various data sources into master data files like the Dartmouth Atlas to include physician patterns of care and physician groups’ patterns to enable purchasers, consumers, and patients to get fairer representation of the choices they have, much like they do for buying cars and dishwashers.

Q: Are you talking of public disclosure?

Reeves: I would start out with feedback to the providers for their internal quality improvement initiatives and root cause analyses. We do not need to start with public disclosure. But we will need to move down the track of accountability and transparency in order to keep our country afloat because we simply can’t maintain competitive advantage globally with current health care cost trends.

Q: The Dartmouth Group did a study of five major academic centers – Mayo, UCLA, the Cleveland Clinic, NYU, and Hopkins – and it showed a significant variation in costs.
Reeves: That’s absolutely true, and I see the same thing happening in our domains. Take our data in Chicago and Pittsburgh. We can rank order costs in hospitals in those cities by diagnostic group, and the most expensive hospital may often be 5 to 8times as expensive as the least expensive hospital with the same outcomes for patients with apparently similar risk and case mix.

Q: Is it realistic to believe we can homogenize these cost differences across the country, given the different institutional, regional, and cultural differences? After all, there are different expenses and profits required in New York City and rural Alabama.

Reeves: I don’t know we can do that, but I think we can compare rural Alabama to rural Georgia. And we can compare Chicago to New York. What’s right is right.
The right way to practice family practice is the right way the world around. The right way to practice internal medicine is the right way the world around. It is both feasible and advisable to decrease the incredible discrepancies between good and bad practices of medicine in our country and elsewhere.

Yes, there are culture differences, and there are habits and preferences that vary from location to location. For example, we have more problems with back surgeries and re-dos and excessive narcotic use in places like Las Vegas than in New York. And we have a lot bigger problem with obesity and bariatric surgery in West Virginia than in Chicago.

There are variations driven by demographics, socioeconomics, and patterns of living. But when it comes to delivering effective, efficient care, 800% differences are not defensible. If you have two cars that drive the same speed, look similar, and last the same thousands of miles, and one is eight times more expensive than the other, how many people would buy the more expensive car at the 700% higher cost?

If you apply that same principle to health care, how long do we really believe we can sustain this kind of variation and turn a blind eye to it? It doesn’t seem like a reasonable proposition to me.

Q: Do you think the Obama administration’s proposal to create a National Comparative Effectiveness Institute would address these issues?

Reeves: It would simply be an extension of what’s going on already. A number of organizations have been doing comparisons of relative therapeutic effectiveness and cost efficiency of new technologies and drugs for years. They have graded the scientific evidence for level of proof of what works best, cost effectiveness, and safety. Managed care plans and carriers have been using these rank order grades to help them decide what their insurance plans are going to cover. We already have a long history of comparative effectiveness studies. For instance, the Medical Letter does this kind of thing for drugs, and looks in a nonbiased way at outcomes and costs. Do we really need more “me too” drugs and expensive images that don’t change care effectiveness and outcomes commensurate with their costs? I sometimes think new technologies are developed mostly because it’s possible, then the developers go looking for problems the technology might help. Can we really afford that? Who should pay for all of that? Under what special circumstances should society as a whole pay for that?

I think in the future we will see the rank ordering of ratings of various health care services much like in Consumer Reports. The day of secrecy and behind the scenes behaviors hidden to the public will eventually be coming to an end. There are multiple initiatives going on in state legislatures, business coalitions, and other organizations that are collecting comparative effectiveness data and displaying them to the public. For instance, you have the Leapfrog measures of safety, and the Institute for Healthcare Improvement’s 5 Million Lives campaign and the publicly displayed CMS core measures of hospital performance. You have publicly available data bases of Medicare claims payments in most States and all payer data bases in 17 states comparing hospitals’ data to that of competitors in the same market. And the National Business Coalition on Health and some States are collecting data comparing health plans to each other and displaying performance metrics on public websites. It’s all about transparency and accountability. Congress has a track record of strongly favoring this approach.

Q: And yet, despites all these rankings and initiatives and talk of transparency and accountability and nearly 40 years after Wennberg’s original paper on Medicare practice variation, the variations remain high.

Reeves: They do, but progress is occurring. Over a 3 year period in Las Vegas, we’ve been giving quarterly reports to hospitals comparing their Leapfrog results, patient satisfaction results, and CMS core measure national percentile rankings to those of their Las Vegas hospital peers. At the beginning, the rankings ranged from the 88th percentile for one hospital to 2nd percentile rank for another.
After regular meetings with senior executives of these hospitals, discussing their quality improvement initiatives, the hospital at the 2nd percentile moved up to the 38th percentile nationally at the end of the 3rd year. So substantial improvement is possible through transparency. Even more improvement is possible through incentives such as bonuses, as demonstrated by the Premier Project with CMS, in which hospitals received 2% more in payments for meeting quality standards. Those hospitals participating in the Pay for Performance cohort showed substantially more improvements than those subjected only to public reporting.

Value-based design of CMS and private health plan coverage works. They decrease out of pocket costs for high value services and treatments and may raise the out of pocket costs for interventions with marginal effectiveness and value. Cost trends bend downward, and value and quality go up. They have to, if this country is going to survive in a competitive world economy.

Along with 4 large company CEOs, a large company benefits manager, and a State health officer, I met recently with President Obama in the White House. This was the day after his historic meeting with national health leaders – the AHA, the AMA, America’s health Insurance Plans, PharMa, the Service Employees International Union, and others - who pledged to reduce national health care spending by $2 trillion over the next decade. Our roles were to explain our interventions that have lowered cost trends and improved health outcomes so they could be adopted for federal employees and other Americans. President Obama and his administration are determined to reform the system to achieve lower cost trends and better health status for Americans. It will take insurance plan designs that align incentives with desired behaviors, and data based reporting of impacts of positive and negative incentives that engage physicians and patients more actively for this effort to succeed. Working together, we can do a lot better than we have the past several years.

*****
The interviewer is Richard L. Reece, MD, author of Innovation-Driven Health Care: 34 Keys to Transformation (Jones and Bartlett, 2007), Obama, Doctors, and Health Reform; A Doctor Assesses Odds for Success (Universe, 2009), and blogger, www.medinnovationblog.blogspot.com.

June 8, 2009 in Dartmouth Atlas, Economics, evidenced-based medicine, Health Plans, Obama administration, Patient Safety, Physicians, Policy/Politics, primary care, Quality, The Industry, Transparency | Permalink
Comments
Dr. Reeves says, "We collect information from those claims and from health risk questionnaires and other biometric measures to track patterns of care."
Until the clinical information from the documents of clinical activity (DCAs), composed by the clinicians attending the patient, is collected and analyzed, all we can do is guess at what happened. Claim's data is never appropriate for analysis of any clinical activity.
This is why the quality assurance and utilization review efforts require the on-site examination of the patient's chart--the compendium of DCAs--by a nurse or doctor who is used to operating in the clinical domain for accurate analysis and reporting.
Claims creation is an activity for the business realm and rarely requires clinician input. DCA creation is the unique activity of the clinician in the clinical domain. Claims are supposed to reflect the activity of the DCA, but rarely do because of the limitations of coding.