Physicians Trained in High-Cost Regions Spend More

Researchers find that physicians whose residencies were in higher-spending regions spent 29% more on average than their peers who had trained in lower-spending areas of the country.

A study this month in the Journal of the American Medical Association  found that physicians from residency programs with higher healthcare spending brought more-costly habits with them when they began their practices.

The study by researchers at the Milken Institute SPH at George Washington University found that physicians whose residencies were in higher-spending regions spent 29% more on average than their peers who had trained in lower-spending areas of the country.

When adjusted for patient demographics, regional cost of living, and other factors, the study showed that mean spending for physicians trained in higher-cost residency programs averaged $522 more per Medicare patient each year.

Those cost traits persisted even if the new physicians went to another region of the country with lower healthcare spending. Those higher-cost practice traits appeared to level off at mid-career as the research found no significant variance related to training location 16 to 19 years after residency.

Fitzhugh Mullan, MD, the Murdock Head Professor of Medicine and Health Policy at Milken Institute SPH and senior author of the paper, discussed his team's findings in an interview last week. The following is an edited transcript.

HLM: Why do training costs vary so greatly among teaching hospitals?

Mullan: I don't know the answer precisely. All teaching hospitals aren't the same. There are some that are well more expensive than others, and there is high variability, even within the more expensive hospitals. It ranges from simple calls about office visits to complicated decisions about expensive surgeries. It would seem in some communities there is a propensity to do more interventions and in others to practice more frugally. It's a complicated set of factors that we call the culture of practice.

HLM: Did the higher cost hospitals produce better results?

Mullan: We did not look at outcomes, but others have. The general findings are that there is not a correlation between the costs of medical practice in a higher-spending hospital referral region than in a lower spending one. These costs adjust on race, age. You adjust those communities so you are looking at the same age cohort and you find there is not an association with outcomes based on costs.

HLM: What are the lower-cost teaching hospitals doing differently?

Mullan: Many of them are in less-urban areas. The urban areas tend to be more expensive, and even when you adjust for cost of living they remain very expensive. There are many reasons why that might be.

When you get into rural areas, hospitals tends to be smaller and tend to have more limited training programs largely in primary care, perhaps general surgery and OB/GYN. They are not into neurosurgery, plastic surgery, advanced nanotechnology, etc. They are training more of the basic physicians and probably as such their use of technology and the frequency of the use is less.

One might say we need the expensive teaching hospitals in the cities. That proposition needs to be challenged, because our national problem isn't an access of technology, it is an absence of coverage in some areas and it is an absence of prudent coverage in many areas. Those are major issues that tax payers in all locations should be concerned with.

Yes, it is good to have burn units in certain areas, but people burning to death is way less of a national problem than having people not get services and die prematurely from things that could have been prevented with good primary care. We are not focused on producing those doctors with the same vigor that we tend to be for producing the more specialized and rarified physicians.

HLM: Why do physician costs fall in line with community averages as their practice matures?

Mullan: It was a surprise finding for us. We found what we call "imprinting." The nature of the practice in the area where the person trains imprints on them a standard and intensity of practice that they carry into their practices. In the first seven years that is very powerful, with a 29% difference for those trained in low-expense areas versus high-expense areas.

That effect diminishes in the 8−15 year period and we see no difference in the 16−19 year period. That suggests that whatever the imprinting they carried with them out of residency diminishes over time as they become more a part of the culture of the community in which they are practicing.

That means that their decisions on diagnostic activities and therapeutic activities become more in common with the practice standards of that community. Although it works in reverse, too.

HLM: How can high-cost teaching hospitals act on your findings?

Mullan: First of all, it's knowing the hospital referral region. Some leaders of teaching hospitals may know what their attributable area is. Others may not. What role do they play?

They second thing, and I expect many would be defensive, is to ask 'if we are more expensive, is it because we are practicing good medicine?' As the costs of medicine in general and the variability of cost become better known and the nation becomes more aware of what a huge chunk of our national treasure is becoming invested in healthcare these topics won't be as easily avoided.

This is a preamble period where we are gathering data and perspective for what could be a fairly major change in our notions of training prudent physicians. We are way out of line with the rest of the world in terms of our expenditures without the presumed benefits. Our outcomes are less good than many countries in the world that spend a good deal less.

HLM: What sort of challenges do teaching hospitals face with these self-assessments?

Mullan: The question of a CEO or CMO looking at their culture is what I would like to see happen. It's a little trickier when you consider that you have in place 50 or 100 or 1,000 academic physicians, many of them experts in their field, and many physicians in the community probably trained at that hospital, practicing in similar fashions.

The nature of the decisions being made about clinical interventions is diffuse. It's not like you get the chief of surgery to say to his faculty and residents 'be more frugal.' It's a question of looking at a given service or diagnoses and analyzing the approach and costs at the institution versus others versus standards of practice.

HLM: What role do you see for value-based payments in this transformation?

Mullan: For an institution to take on a serious intervention to be more cost conscious is a challenge. It's also a challenge because right now, by and large, they get rewarded for volume. It is going to be hard acting on this teaching hospital by teaching hospital.

As the system sends more signals back to large institutions that we are going to manage more care in the community—or if it's an ACO where there is a bundled payment or any kind of organization that is paying—there will be pressure to be more prudent providers.

My guess is [that] over time that will be more effective; the invisible hand as opposed to a principled and intellection approach of trying to look more like national standards.

HLM: Given their training, specialty, and research mission, should not we expect that care in teaching hospitals would be more expensive?

Mullan: The perception is right. What is the norm? Residency is a wonderful part of a physician's life, in which you move from training wheels to no training wheels. It is a revered and ancient way of learning things in the midst of our modern techno-medical center.

But you do what you see as you rotate from service to service to within a specialty. There is more cost consciousness and efforts are underway in institutions to embrace value thinking, but it is still predominantly [based on] what you see your elders doing, whether that elder is a chief resident or a world-renown specialist.

That is why there is imprinting. We think there is documented variability in hospital costs, presumably because of decisions around therapeutic and diagnostic services. You get the clinical leaders doing it X way. As a learner you see it X way, you adopt it X way in your practice. The opportunity and narrative for imprinting is very clear.

HLM: Do you want a more rigorous role for value as a metric when dispersing GME funding?

Mullan: The short answer is yes, but I should amplify that. We have many parts of the country that are way under-doctored and other areas that are generously doctored. We have specialties that are clearly undersupplied, such as primary care, and others that are arguable oversupplied.

The federal government, despite billions of dollars awarded to these institutions, has no ability to discuss, let alone effect changes with that money. That is the basic problem with Medicare funding.

The answer is yes, there should be a way that not just government, but public policy around the training of doctors can be gently redirected based on the huge investment that the taxpayers of America make in graduate medical education.

Right now that doesn't exist. Teaching hospitals by and large like the system as it is now. They don't want the topic discussed because they feel they won't be able to defend and maintain it in a way they like.

The playing field is tilted toward larger more research intensive institutions and that is just fine with them. Those people are the most vocal in defending it, largely in the Northeastern part of the country. That is the standoff we have right now.

Can we discuss a better use, modernizing and bringing into the 21st century a generous reward system for hospitals from the 20th century when our concerns about costs and quality were not what they are today?