Article of the Week: Doctors with Less Experience Cost More

by Sonny Tat

This week’s article is from “Health Affairs” and examines potential drivers of cost of health care and how the Affordable Care Act (ACA) might try to bend that cost curve. The ACA is asking health plans to create “physician cost profiles,” which are measures of how many health resources individual physicians use compared to their peers, controlling for severity of illness. This will be used as a proxy of how much excess a physician costs the health system. Health plans could then use this profile to bend the cost curve by identifying high cost physicians and reducing their payments.

The study is a cross-sectional analysis of insurance claims of over 1 million adult patients in Massachusetts from 2004-2005. This includes different specialties, largely internal medicine but did include some emergency medicine. Their models adjusted for patient illness severity. Using linear regression models, they found that the strongest predictor of cost was physician experience, with less experienced physicians having higher cost profiles. There was no association between cost profile and other physician characteristics such as recent malpractice claim, gender, practice group size, academic affiliation, and board certification status. In addition, the experience-to-cost association follows a dose response relationship, as shown in this graph I made:

Why is this case? The authors aren’t sure, since the original hypothesis of the study was to see if malpractice influenced cost profiles (which it didn’t). Assuming no major errors in methodology (like inadequate risk adjustment), two main possibilities exist to me. One is that newer physicians are familiar with newer, more expensive therapies than “more experienced” physicians. The second is that newer physicians are more uncertain and therefore order more tests and are more aggressive with their treatments. The study isn’t able to validate or refute either of these possibilities with the data it has. At the same time, knowing the answer has important cost implications. If it’s about knowing newer, more costly therapies driving the costs then we can expect costs to rise exponentially as this cohort continues in their costly ways. If it’s about uncertainty, we’d expect that to resolve with time and experience. Presumably, costs could then rise more slowly.

As someone in the “inexperienced” cohort, am I at risk for providing costlier care and subsequent targeting by health plans in the future. I do think that many behaviors--including health care spending--are powerfully driven by incentives. Using cost-profiling to encourage costlier doctors to standardize their care could be useful in helping providers think about how much their care costs. This approach would be superior to our current fee-for-service system that incentivizes only more care without looking at quality. However, our methods must be both sensitive and specific. There must be clear data that less expensive care for specific illnesses can result in non-inferior care. In addition to costs, incentives must also address care outcomes. Otherwise, superficial profiling may lead to more harm than good.

What do you think?

Mohetra A, et al. "Physicians with the least experience have higher cost profiles than do physicians with the most experience." Health Affairs. Nov 2012.