As we continue our review of the National Bureau of Economic Research paper on geographic health spending variation (link in yesterday’s post), we now focus on hospitals part in contributing to that variation. First, the paper notes that in general there is a different role in price and utilization in Medicare compared to private spending variation. Focusing on hospital inpatient spending, dispersion in such spending is higher for private plan members than for Medicare beneficiaries, but both have a fair amount of variation. For private plans, price plays a greater role in spending variation than does quantity of care. For Medicare the opposite is true. The quantity of care, or utilization, is a much greater contributor to spending variation than is price, which should be the case since Medicare basically fixes what it pays hospitals with a national formula that recognizes some regional input cost differences.
Looking next at private payer hospital price differences, even on a risk-adjusted basis there is great dispersion, with the highest HRR having prices four times greater than the lowest for inpatient care. For the seven individual procedures studied, the variation is even greater, from around 6 times to almost 12 times. That is an astounding amount of price difference. And even within HRRs there is great variation in general inpatient and procedure specific payments to hospitals, with price ranges of 3 times or more being common. In Denver, for example, the ratio of maximum to minimum payments for knee replacement is 3.09, for PTCA 2.83 and for MRI, 2.87. So even within an HRR you can see the enormous value of helping patients go to hospitals with lower reimbursements. The savings from getting patients to use even middle-of-the-range hospitals instead of expensive ones could be a 20% or more reduction in inpatient spending.
What accounts for this difference in hospital prices? Labor costs are a significant input cost to hospitals, but adjusting the analysis for this results in only slightly lower price variation across HRRs, and obviously labor costs are similar for hospitals within an HRR. Quality might also have an impact on price, with higher quality potentially costing more to deliver and with patients and payers possibly being willing to pay more for quality. There is significant variation on outcomes believed to be related to quality, but there is not that much correlation between quality as measured by these outcomes and price. Other factors, such as hospital size, teaching status, number of poor patients served, number of high-tech services and ownership status, might affect price. Standard economic theory, however, suggests that a dominant factor is likely to be market structure–how many competing hospitals are there in a particular geography, and how many competing private health plans.
The researchers did an analysis of all these possible price influencers and gee, shock of shocks, guess which one seems to be most responsible for price variation for general inpatient services? You got it, market structure. In bivariate analyses, higher Medicare base payment rates, the number of technologies offered, the number of beds, US News and World Reports ranking were associated with higher prices, while a higher Medicare share of patients and higher concentration of health plan lives at the county level were associated with lower prices. Hospitals in monopoly markets and duopoly markets also had higher prices. But in a multivariate regression, hospitals that have a monopoly have 26% higher prices than do hospitals in markets with four or more competitors. Having a concentrated health plan market ameliorates these effects somewhat. Brand reputation, reflected in the US News rankings, is associated with a 13% increase. (note that this has little to do with quality in most cases, but is purely a result of brand marketing.) Good performance on actual quality measures leads to only slightly higher prices. Being bigger leads to higher prices, so so much for any notion that scale efficiencies might lead to lower prices. The results are similar for the individual procedures.
What policy conclusions can we draw from this research? Our lax antitrust enforcement toward hospital mergers over several decades clearly has resulted in higher health spending. No hospital mergers should be permitted at this point and large systems should be actively broken up. Where there are three or less hospital competitors in a geographic area, charges to private payers should be limited to a set percent above the hospital’s Medicare payment rates, no more than 20% above and really only 10%. And payers should encourage patients to use lower-priced hospitals by reducing cost-sharing amounts at those facilities.