Interview with Dr. Michael Ong: A response to the Dartmouth Atlas of Health Care
2 March 2010
Dr. Michael Ong is the lead author of Circulation: Cardiovascular Quality and Outcomes, a study by California teaching hospitals, carried out in response to the Dartmouth Atlas of Health Care.
JL: Could you explain the origins of the California study?
MO: The first study of variation in hospital resource use by the Dartmouth Atlas was published in 2006. They published a California-focused version in Health Affairs in 2005. One of the variations they cited was the variation in the California hospitals. In the recent Dartmouth report, UCLA has been identified as an institution of high spending on Medicare beneficiaries in the last two years of life. We thought it would be good to bring together all the University of California hospitals in order to try to understand what is unnecessary variation, what variation might actually be warranted and important.
The Dartmouth Atlas use administrative data from Medicare, which only has information about a person’s age, a person’s gender, a person’s ethnicity and some illnesses these patients may have.
The Dartmouth report focused on individuals at the end of life. But when you look at the individuals at the end of life, that doesn’t tell you what might have happened to those people who were facing the same situation but did not die. What happens when you look at everyone—both those people who died as well as those who survived?
We focused on one condition—heart failure. The Dartmouth Atlas report looked at people with many other types of chronic illnesses. We basically focused on heart failure, in part because we know that that’s one of the most common causes of hospitalization. Our study is on a much smaller scale than the Dartmouth reports.
JL: If Medicare is not proxy for the whole population, what conclusions can one draw from the Dartmouth 2008 study?
MO: The Dartmouth Atlas has done a lot of study on variations over the years. And not all of them use this methodology of looking at Medicare patients in their last two years of life. It’s an outgrowth of previous work.
They’ve certainly done a lot of work on geographic variation. This approach that they’ve used to look at individuals in their last years of life is much more recent. The one thing about it is if you were really just studying individuals at the end of life, I think that’s fine. There are still problems using this approach. But if you were making claims about what’s happening at the end of life, it is a little bit more reasonable.
That being said, I think the work they’ve done on hospitals presents a particular problem. What we do for end-of-life care could be improved and there certainly are variations that happen across institutions. But it’s not clear how much is waste. Neither the Dartmouth Atlas researchers nor any large group that’s been looking at this issue can very clearly state what are those things in the medical system that we need to retain and at what cost.
The Dartmouth Atlas has been good at highlighting that there are variations among hospitals. But then reducing the overall spending in our health care system based on geographic variation is not justified by their work. I do think there are some concerns as to how well the findings in the Medicare population translate into all of the individuals that we care for in our health care system.
When you’re combining many different types of patients together, different conditions are treated in different ways. How a cancer patient is treated is going to be very different than how a heart patient is treated. In order to try and understand how to improve overall care, we need to go condition by condition. Some prospective work, like our approach, has been done. But the bottom line is that there is a lot more work to be done for all of us to understand how to advise on improving the health care delivery system.
If we are going to try and improve how we deliver care in our hospitals, we can’t just look at costs alone. We need to balance out what those outcomes are, as well as the costs.
Further, I think it would be unwise to use only the Dartmouth study to determine reimbursements to providers. Reducing our costs by 30 percent would certainly not be a viable solution.
JL: Your study seems to indicate that the more medical intervention, the better was the outcome.
MO: I don’t think it’s entirely that. When we look at the 6 institutions in our study, what we see when we put them side by side, ranking them 1 through 6 in resource use, whether it’s cost or hospital days, and compare them on how they did in terms of mortality rates, there seems to be a fairly good correlation between those sites that spent the most with those with the lowest mortality rates.
Conversely, the lowest amount of resource use had some of the highest mortality rates among the 6 sites. How well does this conclusion translate out to all hospitals in the US? It’s hard to know for sure what all this means; whether this is something we would see nationwide, I can’t really comment on this. But hospital reports looking only at the end of life, like Dartmouth, would not arrive at our conclusions.
JL: Dr. Richard Cooper says income inequality plays a significant role in variation. Could you comment?
MO: The Dartmouth hospital studies don’t adjust for economic status. Most people in health care would agree that your socio-economic status plays a large role in terms of what happens to you in the health care system. I’m an internist, and when I take care of patients there are certainly huge differences in terms of what their backgrounds may be. Patients being discharged from the hospital who are able to recruit all sorts of resources to help them are dissimilar from those coming from an impoverished background.
Dr. Cooper and the Dartmouth Atlas have had a long-standing debate about what we should do about the health care work force (internist/specialist). The hallmark of a good primary care provider is that we know when we reach our limits and when we need to refer to a specialist. I think that most people would agree that the way our health care is delivered is not very well coordinated.
We do need to recognize some kind of constraint in overall care, otherwise the sky will always be the limit in terms of what we can provide. That’s certainly not easy for a physician. When I’m in the room with a patient, I’m just trying to do what is best for the patient rather than trying to think of any incentive in terms of how a payment system might affect me. I’m also cognizant that it does not make sense to order a lot of tests or request a lot of things that in the end will not help the patient.
In our system, we should be looking at costs, but only in a way that will benefit patients.