Read on Twitter
New Paper Alert (actually, old paper that just got published, but hey)@Michael_Chernew @ProfFionasm Columbia PhD student Eugene Larsen-Hallock and I examine whether health care is 'shoppable' by looking at how patients consume lower-limb MRI scans. https://www.sciencedirect.com/science/article/pii/S0167629621000126?dgcid=author
This dovetails nicely with paper out by @amitabhchandra2 and @oziadias earlier this week. Predictions from price theory are that if you ratchet up cost-sharing and give patients transparency tools, they'll consume health care services they way they consume peppers.
A bunch of work shows that this just isn't the case. We were interested in why patients tend not to price shop.
Focusing on how patients consume planned lower-limb MRIs is a useful setting because this is really a pure commodity with (as we show), no quality variation.
Focusing on how patients consume planned lower-limb MRIs is a useful setting because this is really a pure commodity with (as we show), no quality variation.
This is really a 'fact' paper. Fact 1, patients bypass lots of low-priced options en route to receive care. This the price dispersion in a major city. You see that the highest priced provider is also the highest volume provider.
My fav fact from the paper is that patients bypass six lower-priced providers on average between their home and where they receive care.
Patient choices were such that spending on lower-limb MRI scans would 
by 35.83 percent if patients accessed the lowest priced provider available within the same drive time as the facility where they received care. That is, we can save money without more travel.
by 35.83 percent if patients accessed the lowest priced provider available within the same drive time as the facility where they received care. That is, we can save money without more travel.
We then decompose the factors that influence where patients receive care. Fact 3: Referring physicians explain the largest variation in where patients receive a scan (by a wide margin). It explains more than zip code fixed effects, cost sharing, or patient characteristics.
Fact 4: physicians refer narrowly. For example, while the median patient had 16 MRI providers within a 30-minute drive from her home, the median referring orthopedic surgeon in our sample had 80 percent of her patients receive care from a single imaging provider.
As a result, to save money, patients have to diverge from physicians' usual referral patterns. We find that few patients use price transparency tools and qualitative evidence suggests patients just listen to their MD on where to go.
Collectively, this reveals a crucial shortcoming of cost sharing: patients basically just listen to their doctor, so cost sharing hits them on the head with economic pain (which @amitabhchandra2 and co find harms health outcomes) and it does little to influence choice.
We also find that vertically integrated physicians are more likely to send their patients to a hospital for care. This increases MRI spending. As a result, we should be quite concerned with hospital acquisition of physician practices.
So what to do? We argue that rather than targeting patients, we should target referring physicians with information and incentives to steer their patients towards high value care. Great work by Kate Ho and Ariel Pakes highlight this can be quite effective. https://www.aeaweb.org/articles?id=10.1257/aer.104.12.3841
. @afrakt did a great job telling the story of this paper in @UpshotNYT https://www.nytimes.com/2018/07/30/upshot/shopping-for-health-care-simply-doesnt-work-so-what-might.html
. @porszag had a great take in @bopinion on what these results mean for policy. https://www.bloomberg.com/opinion/articles/2018-08-08/consumer-driven-health-care-is-unlikely-to-save-costs
My favorite anecdote about the paper was how it threw a @McKinsey consultant into a blind rage, such that she walked out of a conference where I was presenting it. There were a bunch of exec there to whom she was advising to raise cost sharing and focus on 'consumerism'.
Collectively, I think we know pretty clearly now that the gains from 'consumer-driven' health care are tiny. High cost sharing doesn't do much to shape patient consumption patterns and leads to worse outcomes. Instead, let's incentivize providers, use narrow networks, etc.
My own suspicion is that we can have big gains by shaping physician referral patterns. We need to figure out how to get internists etc to steer me to the highest value care and get patients to physicians who refer well.
