How much geographic variation is there in US health spending? Is that variation the same for various payers? For different diseases? What is the cause of the variation–is it primarily unit price or utilization driven? Is the variation justified by different patient needs? Does it result in different health outcomes? Is geographic variation in spending even worth studying at all? All big questions which still don’t have clear answers. A new National Bureau of Economic Research paper adds to our knowledge on these issues. (NBER Paper) The researchers used the Medicare databases and the large commercial payer database maintained by the Health Care Cost Institute. Private insurance covers 60% of Americans and accounts for a third of all health spending. Medicare covers 16% of our population and 20% of all health spending. The HCCI database includes information on 28% of the privately insured and claims from 2007 to 2011 were analyzed for this study. Hospital spending was a particular focus of the analysis because hospital care is 31% of all health spending, with an average inpatient cost of $12,976. Hospital prices for Medicare and for private payers were analyzed in general and for seven specific services: hip replacement, knee replacement, cesarean delivery, normal delivery, PTCA, colonoscopy and MRI. The researchers also looked at “list” prices for these services. It is fairly well-understood now that there is not a strong relationship between a hospital’s list price and what a private payer actually reimburses and this study reinforces that, finding a correlation of .25 to .48 across the procedures.
In general, the difference between Medicare payment rates for hospital inpatient care are 53% of what private health plans pay, they are 55% for hip replacement, 56% for knee replacement, 67% for cesarean delivery, 65% for vaginal delivery, 52% for PTCA and 27% for MRI. Since most hospitals have been found to be making money, albeit not a lot, on Medicare reimbursement, you can see that they are making very large margins on private care, particularly since there is no reason to think that it costs hospitals more to service a private health plan patient than a Medicare one. As the authors suggest, just reducing this reimbursement gap, much less eliminating it, would greatly reduce spending by commercial health plans and in the nation as a whole–a savings which would be passed on to employers, employees and individuals in the form of lower premiums. A reasonable reduction of the gap could reduce private health plan inpatient spending by 20%, or 7% of total spending. Bringing Medicare and private prices even closer together could double those savings.
The next finding was that there is substantial variation across the country in private spending, as there is with Medicare spending. Using the hospital referral region as the unit of geography, across the 306 regions private spending varies by over three times. The highest spending area in Napa, California, has per private covered person annual spending of $5516, while the lowest spending area of Honolulu is at $1707. The 90th percentile to 10th percentile ratio is 1.53. For Medicare the ratio is 1.45. So both payer types experience large geographic variation, but correlation between private and Medicare spending is low across the geographies, .14 overall and .27 for inpatient spending. Generally, an HRR with high Medicare spending does not have high private spending and vice versa. Some poster cities for low Medicare spending are high for private spending. Grand Junction, Colorado is the third-lowest spending Medicare HRR in 2011 but the 43rd highest private plan one. Rochester, Minnesota has the 14th lowest Medicare spending but the 11th highest private plan per capita spend. And LaCrosse, Wisconsin has the lowest Medicare per beneficiary spending, but the 22nd highest commercial plan spend. So obviously whatever is causing this variation appears to be different for Medicare and private health plans. More on that tomorrow.