Why do people care so much about unraveling the reasons for geographic spending variation? Because of a perception that in higher spending reasons much of the extra care is unnecessary. Because if that is true, understanding variation’s causes may reveal steps that could be taken to reduce spending. So the search continues, with a new paper at the National Bureau of Economic Research being the latest volley. (NBER Paper) These researchers used a test bed of people who have been uninsured but are aging into Medicare, so will have insurance coverage for the first time in a while. Spending and utilization for this group once they hit Medicare is compared across hospital referral regions, to see what happens in high spending regions versus low spending ones. Prior research has indicated that the difference in per capita Medicare spending varies more than 100% across all US counties. For Medicare, as opposed to commercially insured persons, variation should be largely due to utilization, not to price, since Medicare prices are relatively equivalent across the country and not subject to bargaining leverage of providers. Data from the Health & Retirement Survey was used to identify those persons who were uninsured prior to becoming eligible for Medicare.
Once these people become eligible, they had a much greater increase in health use and spending in the high-spending HRRs than in the low spending ones. Simply looking at above and below median-spending HRRs, being in an above median one was associated with a 40% greater probability of a hospital stay in the past two years, 26% greater likelihood of having five or more doctor visits in that time period and 23% greater chance of ten or more physician visits. The authors perform a variety of adjusted analyses to rule out the effect of differences in health status or other factors and find that the results persist. This suggests that it is differences in practice styles across regions that is more responsible for variation in spending. While this is plausible, it still seems reasonable to think differences in patient medical status and needs, and differences in patient preferences for care or care-seeking behavior can account for a fair amount of variation as well. Remember that Medicare has significant cost-sharing, so people’s ability and willingness to pay can affect how much health care they consume, even when covered by Medicare. And given that people rarely choose to go into the hospital these days, the increase in hospital use seems hard to explain as driven by the hospital itself or physicians who might admit the patient. And items like outpatient surgeries and prescription use did not seem to follow the pattern of other categories of care, which the authors illogically suggest are not as susceptible to physician decision-making. Last time I checked, doctors prescribe drugs and recommend and perform outpatient surgeries. All this adds to my skepticism about the results. But I do think practice style is a major, perhaps predominant factor in spending variation.