• Why do docs overorder tests?

    There’s an awesome study that was just published in JAMA Internal Medicine, “Appropriate Use of Myocardial Perfusion Imaging in a Veteran Population: Profit Motives and Professional Liability Concerns“:

    Myocardial perfusion imaging (MPI) is performed millions of times annually in the United States to assess patients for coronary ischemia. Some have expressed concern that MPI is being used inappropriately, possibly because of self-referral profit motives and professional liability fears. To inform clinicians about situations in which patients are likely to benefit from MPI testing, appropriate use criteria (AUCs) for MPI were developed, last revised in 2009. Prior investigations have applied AUCs to describe the magnitude and patterns of inappropriate testing. Rates of inappropriate testing have ranged from 7% to 24%. We hypothesized that the single-payer environment of the Veterans Affairs (VA) health system, which eliminates self-referral profit motive and limits liability concern, will result in less inappropriate use of MPI.

    Anyone who visits this blog regularly knows that one of my foci of interest in the fact that doctors do too much. Many of you believe that’s because they fear being sued, and they are practicing defensive medicine. I’ve often argued that I think the profit motive can’t be ignored – the more docs do, the more they generally get paid.

    The genius of this study is that it was conducted in the VA setting. Because it’s the VA, and the docs are salaried government employees, there no profit motive involved. If they do more, they don’t get paid more. Moreover, because they work for the federal government, malpractice is handled differently. Almost all cases are handled administratively. In fact, only three judgments were recorded against the US for malpractice tort cases in 2010. Therefore, there is really no impetus to practice defensive medicine. Logic would tell us, therefore, that there is no reason to order unnecessary tests.

    This study chose to look at myocardial perfusion imaging, which is often used inappropriately. In a four and a half month time period, they identified 332 patients who got the study in the VA system. Only 78% were clearly appropriate. 13% were inappropriate, and 8% were uncertain.

    Why are docs doing this? My guess is that it’s just incredibly hard to change physician behavior. But what’s clear is that it’s not simply greed, and it’s not simply fear. Ironically, those things are easier to fix. We’ve got our work cut out for us.


    • This study is fascinating, but I think it will be tough to draw meaningful conclusions from it. First, stress testing is all over the map. Some emergency docs stress test everyone with chest pain, others stress no one.
      Now that I’ve said we can’t draw conclusions, let me:
      I agree with the main conclusion that it’s likely mostly about “culture,” but I think it’s also that there are multiple incentives all aligned in the same direction. Most emergency docs don’t get paid when they order a stress test (the cardiologists, radiologists, hospital, and/or department do) but they might lower their malpractice risk.
      I also think that many docs are more concerned about a miss in general, and not the malpractice associated with it, both because we care and because we don’t want to be embarrassed in front of their peers and colleagues.
      What we end up with is a whole bunch of incentives all aligned in the same direction: ordering tests of minimal (if any!) benefit.

    • I think a lot of the over-testing is simply a matter of moral hazard, since most doctors don’t really think about the fact that their patients have to pay for every thing they order. Given that the doctor views additional tests as cost-less, there are two reasons, at least that I’ve noticed at my hospital, why doctors order extra tests: 1) scientific curiosity and 2) bad computer interface design.

      For an example of the former, I’ve known doctors who, purely out of scientific curiosity, order genotype testing for conditions where the genotype shouldn’t affect the decision about treatment at all. I secretly think that they make these decisions because subconsciously, they want that data for their research. But, of course, a research interest shouldn’t affect how you treat patients, and such a decision should get IRB review and patient approval.

      As for the bad interface design, it has been the case here, especially with younger doctors who can’t remember the pre-computer days, that the placement of options in the computer interface influences the tests doctors order. The old paper systems didn’t really have a way to automatically order regular daily or weekly tests, the new computer systems do, and the placement of that suggestion has caused many doctors to think that it is more acceptable than it really should be. You can correlate the increases in certain kinds of tests with how prominently those options were made in the software. The bottom line is that it’s clear that doctors need training and guidelines for computerized systems, and the software makers need to start hiring more graphic designers with the medical training needed to know what features are appropriate and where.

    • 7-24 percent is a pretty big range. So would the VA only be performing better if inappropriate testing was 6% or less? In my opinion defensive medicine is a culture learned trough medical school, residency, and the first few years of practice. Being at the VA in the present doesn’t eliminate that cultural phenomenon

      • I was thinking the same thing about that.

        About 8 years (give or take) I personally heard of at least 1 doctor (specialist doing procedures), who was a mutual acquaintance of me & a friend of veteran friend of mine, who told me this doctor started working at the VA.
        This guy was no spring chicken when he decided to start working at the VA, I know he went to medical school in the early 70s, he had to be in his mid to late 50s at the time. And I remember him moaning about the price of malpractice insurance 15 years ago, so I would tend to think that attitude informed his practice either consciously or unconsciously. If it was unconsciously especially, why would that change when he started treating veterans at the VA?

        So upon seeing those percentages on this post… I immediately wondered… okay but of those 13% inappropriate tests… how many were ordered by doctors who never worked anywhere but the VA?

        Some kind of break-down of the 13% by the ordering doctor’s background would’ve been more informative, IMHO.

    • Always illuminating, and reminds me of the NHS and the difficulties they have with care variations. Unlike the US, theoretically, no profit motive, med mal less burdensome, and more solidarity in keeping the system functioning well, recent missteps aside.

      Yet, look at the maps:

      Unexplained variation–where little or none should exist–continually rears its head. I struggle with every Dartmouth release, and then the counterpunch from the other side…

      A short commentary from a UK official speaks to problem in UK hospitals:

      • I’m glad you brought up international comparisons. I was thinking about the Australian situation, which is similar to the UK in that there are fewer incentives for doctors to order tests. I think there are important explanations other than incentives to explore:

        1) It is difficult to overcome the bias to “do something” even if reason tells you that doing that thing wont be effective (or even harmful). This is a bias facing all occupations, but is particularly severe for doctors because of agency problems.

        2) It is often cognitively easier to perform tests than to not perform tests. WIth the related point of fear of missing results.

        You would need disincentives to overcome these factors rather than just the lack of incentives.

    • In addition to moral hazard one could argue that since the physician does not incur the cost of the tests it is optimal for them to order test that have a positive marginal product that > than the marginal costs of their time. Therefore, they should be ordering tests in excess of what society would imply as optimal because they face a lower cost.

    • I think on net, these results:

      1) raise my confidence that profit from procedures (outside of a few specialties like orthopedics) is not driving utilization patterns significantly. This fits with what we see from other fee-for-service countries like France despite the counter-intuitive nature of the results.

      2) lower my confidence that defensive medicine is the central factor driving test ordering. It does not eliminate it though as the VA is staffed by doctors trained elsewhere in the defensive medicine mindset

      3) worry me that this level of testing is indicative of a larger failure of medical education

    • I agree with all of the above. You order tests with no idea what they cost. They also get ordered out of habit, because during residency an attending told you that test A should always be ordered for condition B, because the patient didn’t think to tell you they had the same test with another doctor 6 weeks ago, because you recently missed an exotic diagnosis and this case is kinda, sorta like that one, or because the patient is going to stomp off with knickers in a bunch if they don’t get a certain test or at least some test.

      I did a Pap on a 3 pack per day smoker (right there you know something bad is about to happen). Four years later — years, not months — she was diagnosed with lung cancer and sued the seven primary care docs she had seen in the previous five years for “failure to refer to a specialist in a timely manner”. I’m willing to bet that there are more than a few CYA chest x-rays being ordered by the seven of us. We had a bad case in our group involving an elderly man with cancer and narcotic induced constipation. A few days after he saw a PA, he perfed and the docs who were responsible for the PA were eventually sued for “elder abuse” which is not covered by malpractice insurance and so were sued for their personal assets. I know that the people involved in that one are much faster now to send constipated patients to the ED for the obligatory CT. In the US we have the idea firmly fixed in our belief that anything can be detected adequately in advance to avoid all bad outcomes. If that doesn’t happen, it’s “malpractice”. I asked a new patient about the cause of her parents’ deaths and she laughed and said, “I don’t know, it was in Europe, they just got old.”

    • “Rates of inappropriate testing have ranged from 7% to 24%. ”

      Hmm. As a customer of the medical system, that sounds quite reasonable. I’d be _really_ unhappy if doctors were _undertesting_ by even 1%, so I don’t see how you are going to get it below 10% or so. This is medicine, a game where you don’t know what’s wrong a priori. (Despite the following rant, I do respect MDs and recognize that things ain’t easy.)

      By the way, on malpractice. It’s a real problem in the US: it kills more people than car and workplace accidents combined. And the vast majority of malpractice victims receive no compensation. And the number of malpractice _suits_ is a tiny fraction of the number of malpractice events. So until you figure out a way to compensate more of the victims, you should be thinking yourself lucky there are as few* cases as there are.

      *: It’s pretty much always** been the case, that malpractice cases only occur when the victim or family is hit by large bills. So I was not surprised that “the best hospital in Boston” didn’t bill me for the “care” it “provided” my father. (It did reasonably well for mother, though.)

      **: My father worked at Mass General on medical engineering and safety issues back in the 1970s, so I remember following this even back then.

      • By the way, on malpractice. It’s a real problem in the US: it kills more people than car and workplace accidents combined. And the vast majority of malpractice victims receive no compensation. And the number of malpractice _suits_ is a tiny fraction of the number of malpractice events. So until you figure out a way to compensate more of the victims, you should be thinking yourself lucky there are as few* cases as there are.”

        You are confusing all medical errors with malpractice. The infamous IOM report cited 100k deaths from ALL MEDICAL ERRORS COMMITTED BY ALL PARTIES, not just malpractice committed by doctors.

        In fact, if you actually read the report, most of the errors were done by non-physicians (i.e. pharmacists, nurses, PAs, NPs, etc)

        Furthermore, many of those errors were “systems errors” that are impossible to blame on one person as a mistake.

        You need to actually read the IOM report before you try to cite it.

    • The big questions in my mind are why do the payers not refuse to pay for such tests and if the patient had to pay directly would there be fewer?

      Maybe the if the patient had to pay directly would there would be fewer of these tests ordered and that is why high deductible insurance is such a bargain compared to low deductible insurance.

    • Working at a VA, one important point to consider is how malpractice is handled in VA. It is true that, in malpractice claims against the VA, the Government is the Defendant, and generally speaking individuals are not sued.

      However, because the Government is the defendant and not the physician, Counsel represents the Government, and not the physician. So far, so good; except that Counsel may choose to settle at an amount that is advantageous to the Government but not advantageous to the provider. If the dollar amount of the settlement exceeds a certain amount, it gets reported to the national register as a settlement against a provider.

      As a result, VA physicians still factor in a “fear of being sued” in their mental calculus when determining their plan of care for a Veteran. It may be a smaller part than their private sector colleagues, but it is still there to be sure.

      If anything, I would expect VA to be *higher* in over-utilization simply because there is no cost-containment to consider. There is no utilization review to work through. There is no gatekeeper to disapprove a diagnostic and/or treatment recommendation. Why rates of excessive utilization are not actually higher in VA is the more interesting question.