Most politicians in Washington agree that the health care system in this country needs to be reformed. Their ideas of what needs to be done are quite different, however. I spend most of my time praising the left for its ideas. Today, I’m going to take a closer look at one of the major issues on the right: The need to reform the medical malpractice system. I’ve talked about this issue before once or twice, but this time around I’ll be making use of a recently published paper on the issue written by Dr. Brandon Roberts and Dr. Irving Hoch, which appears in the December 2009 issue of Health Economics (subscription required).
The basic idea from proponents of tort reform is that malpractice drives up costs both because physicians are forced to pay large sums–which increase with the size of jury awards–for malpractice insurance premiums and because physicians pull out all the stops in their diagnosis and treatment of patients in hopes of staving off a lawsuit–a practice aptly labeled “defensive medicine.” Limit the size of jury awards and malpractice premiums will fall. Create no-fault remedies for physicians and patients and the practice of defensive medicine will taper off. As a result, health care costs will grow at a much slower rate. That’s the argument anyway.
The real question, however, is how much potential savings are represented by medical malpractice? In a fancy econometric analysis, Drs. Roberts and Hoch examine the association between the number of malpractice cases per capita and the per enrollee Medicare Part B (physician visit) expenditures across a nationally-representative sample of metro and non-metro areas. Controlling for a variety of demographic characteristics, medical services, differences in state tort law, and general times–as well as site-level fixed effects (if you don’t know what those are, ask me later)–they conclude that medical malpractice accounts for between 2 and 10% of national health care expenditures. Prior studies placed this figure between 1 and 2%. So what gives?
First of all, the authors’ model captures the total effect of malpractice on health care costs. That is, it will include both the direct costs (e.g., malpractice insurance premiums) and the indirect costs (e.g., overuse motivated by defensive medicine). Whether or not other studies were treated in this way, I can’t say. Anyway, I like what they’ve done here, because I think the total effect is of more concern than either of its separate components.
However, I do have one major problem with the study, and that is that the authors’ do not control for health care utilization. If you’re looking at the relationship between the number of malpractice suits per capita and the average Medicare spending per enrollee you need to control for how much care is being used. Why? Simple: Imagine two cities, both with a population of 100,000 people. In one city (City A), the average Medicare enrollee goes to the doctor five times a year. In the other city (City B), the average Medicare enrollee goes to the doctor ten times a year. It’s safe to assume that average Medicare expenditures will be higher (roughly double) in City B than in City A. It’s also likely that, given the higher number of total visits in City B (1 million) than City A (500,000), there will be more opportunities for an “event” to occur that leads to a malpractice suit being brought by a patient. But, the way the study treats malpractice suits (per capita), City B would seem to have a higher per capita rate of lawsuits than City A (because they have the same population size). Therefore, health care utilization is a “left-out common cause” that would, in this scenario, cause both Medicare expenditures and number of lawsuits per capita to be higher in City B than in City A. Consequently, the relationship the authors’ find is likely to be highly biased, calling their results into question.
I wondered about this issue, so I contacted Dr. Roberts (the lead author of the study) and asked him for his thoughts on the matter. Here’s what he had to say:
“Very important point, but what you find is that utilization can in fact be controlled for in large part by the demographic factors and the area dummies. In a sense, our dependent variable IS a measure of utilization. Utilization and cost is higher in certain areas, but these differences do tend to hold over time. This is borne out in the literature. This is one of the fascinating things about the study to me – even controlling for this variation across geographies, the cost- litigation association still holds.”
To the extent that utilization is constant over time, he’s right: the area dummies will control for it. So, too, for the demographic factors. But that’s not controlling for everything, and thus, some bias will likely remain. As to the comment that the “dependent variable IS a measure of utilization” — that doesn’t hold up well, because different services cost different amounts. Anyway, it’s an interesting study, and its limitations are not necessarily enough to discard its contribution to the important question of just how much medical malpractice drives up health care costs. I’d turn my attention to the unsettling fact that the authors’ generalize from their study of the Medicare population to make statements about health care spending and litigation in the U.S. population as a whole, but something tells me you’ve had enough for today.