Updated 8:07 p.m. | For years, many health care experts, medical professionals and politicians have been touting the benefits of electronic health records (EHRs). They say EHRs will not only improve quality and efficiency but also reduce costs by tens of billions of dollars annually to a system that too often orders unnecessary or duplicate tests and procedures.
But a new study out Monday in Health Affairs challenges that premise and says that doctors may actually order more testing in some cases — potentially adding to costs.
The findings come as the Obama administration has included new assistance to speed along the adoption of EHRs, which are a part of the Affordable Care Act reforms. And incentive payments were made available to hospitals and health professionals as part of the stimulus law.
The study, which was done by Cambridge Health Alliance’s Danny McCormick and David Bor and CUNY School of Public Health’s Stephanie Woolhandler and David Himmelstein, found that office-based doctors were more likely to order an imaging test if they had access to EHRs. The authors studied records of more than 28,700 patient visits to nearly 1,200 doctors. Physicians who could look at EHRs ordered imaging 18 percent of the time, compared to about 13 percent of the time with doctors who couldn’t use EHRs.
We spoke with McCormick, the lead author of the study. An edited conversation is below.
Q: What were the most important findings of this study?
McCormick: I think the most important finding is that in contrast to what has sort of become common wisdom, which is that health IT is likely to decrease overall health care costs through the reduction in test ordering, we find in outpatient practice across the county in a nationally representative sample, there was no decrease in outpatient practices that have the capacity to view prior imaging studies or laboratory studies. And in fact, it appears that capacity actually drove up test ordering both for imaging and for laboratory studies.
Q: For years people in health policy have stated that EHRs would not only decrease testing but would save the country money on health care. Why do you think you found opposite results?
McCormick: I think that those conclusions or that idea was based on a few small studies that were done at cutting-edge hospitals years ago that in fact showed that when you have highly skilled health IT departments developing customized health IT systems for their own purposes — understanding the needs of their institutions and carefully tailoring the health IT systems to their needs — that you actually could decrease both redundant test ordering and unnecessary test ordering. Unfortunately, that really is the only data that exists on the question. No one has actually looked into actual practice — whether those same findings were generalizable. This study finds that in fact it drives up costs.
We speculate that there are a couple of things: the two main mechanisms that people point to for decreasing lab testing and image testing ordering. One is that if computers could talk with each other in an effective way, you could decrease test ordering. If you could see a test that was done yesterday at another hospital, for example, you wouldn’t necessarily have to order it again today. So sort of redundant test ordering.
The second is that computers could be devised so that they can help doctors make better decisions about who actually needs a radiology test, for example, in a particular situation. I think what we’re seeing is that both of those conditions don’t really exist in current-day outpatient medical practices in the U.S. That is, we don’t have good interoperability — the ability for computers to communicate — and we may not have adequate decision aids that help doctors make those better decisions about who actually needs a test. Even though in idealized settings, that may work, in practice — where health IT systems are often selected with billing features as the predominant criteria — it may just be that the decisions aids are not adequate. They don’t work well enough in general practice to decrease test ordering.
Q: What more do we understand about the reasons for this?
McCormick: We weren’t able to study the motivation of physicians in this study. What may be happening is that if you make something easier to do, people will do it more often. What we think may be happening is that if physicians have the electronic ability to view test results and imagining results, it may lower the barrier to ordering a test because you know you can retrieve the test results the next day at your computer screen with certainty and without hassle. In contrast, physicians who don’t have that ability often face the onerous task of chasing down the laboratory imaging tests that they order.
Frequently, it’s not uncommon that if you don’t have computerized way of doing the images, you may have to call a radiology facility the next day and try and track that result down. And they fax it over on paper that you can’t really read because there are lines all over it — that kind of thing. And you may have another call back to the facility or you may have to put somebody in your office on the case to retrieve those results. We think that it might make it slightly easier in borderline cases.
We don’t think this phenomenon would influence decisions where clearly a patient needed an MRI or clearly they didn’t. It’s really in those gray-zone patients, where there’s a moment of contemplating whether to get a test or an imaging study.
Q: What will the impact of this study have on those doctors who are wondering whether they should go to the effort to computerize patients medical records?
McCormick: I think it’s important to point out that the conclusions of this study really aren’t about the value of health IT overall. We really address one dimension of the purported benefits of health IT, and that’s costs. It doesn’t address quality or efficiency at all. There may be a lot of reasons from a quality point of view or an efficiency point of view that people would want to get health IT. And in fact, if you asked me as a doctor if I would give up my fairly sophisticated health information technology system, which I use every day and was just on a second ago, I would say no. I see the value of it.
However, what this study addresses is the cost piece. With regard to cost, it also doesn’t say that in no system at no time with no amount of effort, a system couldn’t be devised that would lower imaging (ordering). What it does say is that in current practice in the U.S. — a sort of national snapshot based on more than 28,000 visits — it does not seem to decrease test ordering or cost.
I think that the biggest implication is the national health policy implication. A lot of the discussion that led up to passage of the HITECH Act, the historic federal investment in health information technology that was passed in 2009, was predicated on the assumption that there would be substantial cost savings to the U.S. health care system. I think these results have most implications for people making those claims and should raise the question about whether that’s actually correct and whether future investments in health IT are appropriately framed. That is to say, do they take into consideration the lack of evidence that costs savings will be realized? That may or may not influence whether the federal government makes investments in health IT. Again, it may be appropriate because of improvements in quality.
Q: What do you think this study says about the role IT may play in saving money on overall health care? Does it have the potential to throw cold water on this whole move to get the entire country on board with IT?
McCormick: I think it has implications on several levels, and I think the biggest one is the one I mentioned, which is the U.S. health care system costs. I think it’s become common wisdom that health IT will substantially decrease those costs. I think that the study raises a substantial concern that may not come to pass. I think that if we figure that out, we can better align future federal investments in health IT or in other things with our national health priorities. And presumably they include coming up with a way of saving costs. If this may not be it, than it means that we need to consider alternatives that may be more affective.
I think there are some implications, to go backward, for health IT developers and for physicians. I think that those implications are that we need to make sure that health IT systems that are being developed actually perform well and are driving by clinical needs as opposed to billing needs or both.
And then physicians also will need help in selecting among the wide range of products that are out there, and (they) will need help implementing (the systems). Most physicians now in practice are incredibly busy taking care of patients and have not been trained to implement health IT — how to gear their practice, how to fit it into health IT systems in such a way that they’re efficient, effective and decrease test orders. So there are lots of different steps I think that will need to be focused on, but I think what the study is saying is that, as currently implemented, it doesn’t appear to be working to produce lower costs.
We also asked for a short response from Dr. David Blumenthal, who was the national coordinator for health information technology at the Department of Health and Human Services for the Obama administration from 2009-2011.
Blumenthal: This is one of many, many studies, and the studies are overwhelmingly supportive of cost-reductions by using electronic health records. Anytime you see anything that is complicated like this (the implementation of EHR throughout the system), some studies will be positive, some will be negative. But if you look at the total review of literature that was done a year ago, it puts it into context: 92 percent of studies were positive. I expect that if you study something 100 times, there’s going to be some variation.
Second, this is really not a study of costs. This is a study of test orders. It studies results, and it’s a viewing of those results, not EHR overall. EHRs have many other aspects that are not included here. It doesn’t look at EHR and total health costs. It only looks at particular tests. It’s possible that the increased cost from imaging may reduce costs in other ways. Ordering a particular test that’s being observed may or may not increase costs. You can’t really infer any major conclusions about costs from this study. And the study doesn’t look at the benefits for quality of care at all. It’s possible the use of tests by some of the doctors could have avoided other costs. This study has no way of assessing the overall implications of the behavior that it’s finding.