Does regional variation in health care spending reflect patients’ preferences, legitimate differences over what is appropriate care or doctors’ responding to their own financial incentives? It appears to be the case that even after adjustment for health status, demographic and other factors there is significant geographic variation in health spending and if those regions spending more don’t have better outcomes for patients, then there is little reason to allow it to continue, although it is difficult to imagine the mechanism by which care for hundreds of millions of patients would be that closely monitored, particularly in real-time. A new study published in the New England Journal of Medicine looks a geographical differences in Medicare drug spending.
The Medicare drug benefit was added a few years ago and most beneficiaries have taken it up. Medicare Advantage enrollees typically already had a drug benefit and continue to do so. The authors used data from 2008 to examine overall drug use and spending, as well as that in three specific categories. The primary end-points were per capita drug volume and cost per prescription. The geographic unit was the typical hospital referral region. Overall per capita spending ranged from $2413 in the lowest quintile of regions to $3008 in the highest. Most of the difference, about 76%, was due to differences in cost per prescription. This means that drug volume was responsible for less than a fourth of the spending difference.
The ratio of branded drug use to generic drug use had a range across regions of .24 to .45, with a range of .24 to .55 for common blood pressure drugs; .29 to .60 for statins; and .15 to .51 for anti-depressant and other mental-health drugs. It is apparent that some areas are resistant to the use of generics, which is pretty much completely unjustified. It would seem to be relatively easily fixable by CMS as well, with the potential to save several billion dollars annually. This study also reflects the more general truth that our spending problem is one of unit cost, not utilization, and that is reflected in many variation studies.