Health economists continue to try to gain a better understanding of the nature of geographic variation in health spending. The Bureau of Economic Analysis has issued a paper which takes a disease based approach and tries to evaluate the role of price and of utilization differences in a commercial population. (BEA Paper) The geographic unit used in this study was the Metropolitan Statistical Area. Because markets may not be homogenous in high or low price or utilization status across all diseases, the authors used a disease-based, episode of care expenditure unit to study the relative effects of unit price and utilization for each disease category across the 85 markets studied. This contrasts with a pure per capita measure which aggregates all spending for a person across all diseases. As an example, the Birmingham MSA is 4th highest in per capita spending, but 17th in spending on an episode of care basis, and there is likely even more variability when the episodes of care are looked at on a disease by disease basis. And all of these methods should be adjusted for demographic and socio-economic characteristics in the MSA and for health status and disease severity, although episode of care measures inherently pick some of that up. While variation across MSAs and within MSAs can be measured, the authors were most interested in how much episode costs varied in an MSA from the national average, and then ascertaining how much price and utilization each contributed to that variation. A commercially available episode grouper was used to construct episodes, which also included a four-level severity ranking. The data is a little dated, since it was from 2006 and 2007, but it seems doubtful the relative contribution of price and utilization has changed a lot for most diseases.
The top 5 disease episodes account for 21% of all spending and the top 15 for 42%. Episodes have wide variability in cost, for example hypertension is pretty low per episode but heart attacks are high. Across all episodes, inpatient services account for 18.5% of spending, outpatient hospital is 23.1%, primary care physician is 8.8%, specialists are 15.8%, pharmacy is 21.85 and other (lab, ambulance, ER, etc.) is 12%. (You can see how this type of analysis can guide care management efforts–what episodes are your costs concentrated in and what spending categories.) Most episode disease categories show a fair amount of geographic variation in spending per episode, but for some price is a more important contributor and for others, utilization is. One result of this research is that using the disease method reports reduced variation compared to what is seen, for example, in the Dartmouth Atlas reports. While there is variation across MSAs, the study also reveals significant variation within an MSA in how a disease is treated. The MSA results are very interesting. For example, Milwaukee has a very high per episode spending across all episodes, which appears to be mostly due to price, while Gary, Indiana is also high, but because it seems to have more utilization per episode. But there is also very significant variation within an MSA in regard to different disease episodes. Much of the price differences across MSAs, for all episodes, appears linked to general price differences. And very importantly, price and utilization appear negatively correlated, in other words, high prices generally go with lower utilization. It is not clear whether this is driven by physicians or patients; i.e., do doctors order and deliver more services when prices are low, so that they can meet a target income, or do patients seek fewer services when prices are high. It is likely a combination of the two. Finally, there is also significant variation and source of variation within service categories–pharmacy has the lowest variation in overall spending across MSAs and outpatient hospital and primary care visits have the highest variation. Utilization appears to have higher variation within a service category across MSAs than does price. This is an extremely thought-provoking paper.