A recent study by Zack Cooper, Stuart Craig, Martin Gaynor, and John Van Reenen has documented the remarkable variations across regions and age groups, and within regions, in U.S. health care spending. Previously, the Dartmouth Atlas project studied variations using Medicare data on people over 65, and the 2013 Institute of Medicine study showed no association between spending and quality in both over-65 Medicare and under-65 private insurance markets.
The Cooper study steps up the game by using a massive data set on utilization and prices for the 14 percent of Americans covered by UnitedHealthCare, Aetna, and Humana, combined with Medicare claims data. Highly recommended is the terrific graphical interface developed by The New York Times to see what’s happening in your world.
Three Reasons Why This Study Is Important
1. Cooper and coauthors find that prices paid by insurance companies to providers vary dramatically both across regions, and within regions.
This is an eye-opening result that lays out the lack of transparency in prices in the under-65 private insurance markets. First, Cooper and associates show that markets with more hospitals exhibit lower prices — competition works.
Second, even within markets, prices are all over the map, even for common procedures such as colonoscopies. This has little to do with increasing the number of hospitals, but everything to do with the lack of pricing transparency, like ordering from a menu without any prices.
2. The researchers find a remarkable degree of variation in health care utilization across regions, both in the over-65 and under-65 population.
When Dr. John E. Wennberg started collecting data on hospital and physician utilization patterns in rural Vermont during the 1960s, he found large variations in tonsillectomy rates (for children) and hospital bed use (for older people). The Cooper study finds as much variation in health care utilization — that is, how doctors treat patients — in the under-65 private insurance markets as there is in the elderly Medicare population. (Technically, they find the coefficient of variation in expenditures, holding prices constant, for the under-65 is 0.32; among the over-65 it is 0.30.) Note it is important to recognize that when you hold prices constant, “expenditures” are a measure of health care utilization.
Furthermore, the correlation between what doctors do in the under-65 and over-65 cohorts is high; the Cooper group finds a correlation of 0.6 (where 1.0 is a perfect correlation). Other studies, such as this report comparing under-65 Blue-Cross Blue-Shield medical hospitalization rates with corresponding over-65 Medicare rates in Michigan regions (Figure 1), have found even stronger correlations. That is, regional utilization patterns for elderly Medicare patients provides a remarkably strong predictor of utilization rates in the under-65 private insurance population. Not perfect, perhaps—tonsillectomy rates in children could differ from hospice use among the elderly—but pretty close.
Figure 1: The Association Between Medicare Medical Discharges (1996) and Medical Discharges of Adult Blue Cross Blue Shield Members (1997) among Michigan Hospital Service Areas. Source: Dartmouth Atlas of Health Care in Michigan.
What The Cooper Study Does Not Say
A casual reader of The New York Times article by Kevin Quealy and Margot Sanger-Katz, “The Experts Were Wrong about the Best Places for Better and Cheaper Health Care,” would be forgiven if they missed this critical point. By focusing only on total spending—price times quantity—they give the impression that Medicare data tells us nothing about private insurance markets. Not so.
We know that when Medicare spending is high, privately insured quantities tend to be high, as shown in Figure 1. Correspondingly, when prices in private insurance markets are high, Medicare expenditures are low. It’s not entirely clear why this latter effect holds, although one study finds that incentives to treat privately insured patients (as there would be if provider prices are higher), lead to less care for Medicare patients, with no discernable impact on patient health.
The article further suggests that when President Obama visited Grand Junction, he was visiting the wrong town, because prices paid by local employers for their employee health care were so high. However, President Obama is chief executive of the largest medical insurance plan on the planet—Medicare—and he should be visiting Grand Junction to discover how to prevent Medicare from bankrupting the federal government. Of course, this doesn’t preclude adding Rochester, NY to the President’s itinerary — a city identified by the Cooper study as being low for both private and Medicare spending.
3. The study highlights that policy solutions to address variations in health care quantities are very different from policy solutions to fix variations in health care prices.
Every year, patients are harmed because of too many procedures: they are exposed to radiation from excessive CT scans, or scarred for life from unneeded surgical procedures. Patients are also harmed by receiving too few effective treatments. The health impact of high prices is less direct; employers drop insurance coverage, or patients can’t afford their drug regimens. Thus, policy solutions for excess prices are different from policy solutions for quantities that are either too high or low.
Fixing price variation is already on the radar screen of many, including firms like Castlight that encourage patients to shop around for the cheapest prices. We cannot resist suggesting a more draconian solution, which is to limit privately insured prices to whatever Medicare charges in the region, plus 25 percent. (See this Health Affairs Blog post for details.) Under a price cap, hospitals and physicians could consolidate to reduce fragmented care; they just couldn’t jack up their prices.
Fixing variations in quantities is trickier. There are by now 777 Accountable Care Organizations (ACOs) in the U.S. whose goal is to both improve quality of care and reduce excess utilization. Some (but not all) are showing great progress towards that goal. A greater reliance on shared decision making to ensure that patients are getting treatments that they want, coupled with a focus on reducing potentially harmful treatments, are two additional approaches to scaling back excess care. Indeed, the “experts” were right that we can learn quite a lot from places like Grand Junction and La Crosse, WI about restructuring primary care and improving the quality of end-of-life care.
The Cooper study takes dramatic strides forward by making private sector utilization and pricing data much more transparent than they have ever been. But more needs to be done.
The plans that shared their data should be commended. Those that have yet to do so should be encouraged (or forced) to do so, given the importance of the policy issues at stake and the need for price transparency so that patients can make better informed choices. Transparency is also important to ensure that newly formed ACOs don’t take advantage of their market consolidation by raising prices. It is worth noting that a pending Supreme Court decision could make matters worse with regard to transparency, not better.
We need to make progress not only in reducing unwarranted variations in how much care people receive, but also in reducing unwarranted variations in prices. We need better care and lower costs.