27 Dec 2014

Riddle Me This: How Can Distance to Old Providers Go Down By Switching Insurance Plans?

Health Legislation 11 Comments

I’m reading this Working Paper by Jon Gruber and Robin McKnight, and am baffled by one of their findings. First, the context:

There is one important outcome that is the current focus of much debate over limited network plans, however: patient access to providers as proxied by distance traveled.

A major concern raised about limited network plans is that it will lead patients to have to travel much further to see their providers. We can address this concern with our data by examining the distance between patients and the providers they do see when they join limited network plans. To do so, we use the distance between the centroid of patient and provider zip codes in our data, for every provider‐patient pair that we observe.

Then they go on to write: “We find that those patients who continue to see their old providers are traveling shorter distances to do so, but that those patients who see new providers are traveling farther” (p. 24).

I believe if Bill O’Reilly were here, he’d say, “You can’t explain that.” Anyone?

(In case it’s not clear: How can it be possible that people who continued to see the same provider as they did before the policy change went into effect, now drive a shorter distance to their provider?)

11 Responses to “Riddle Me This: How Can Distance to Old Providers Go Down By Switching Insurance Plans?”

  1. Major.Freedom says:

    Might it have to do with the composition of the new plans being different in terms of primary and tertiary care, whereby the distances traveled is different?

  2. Mogden says:

    Perhaps it means that they are traveling shorter distances relative to those who are seeing new providers.

  3. Andrew_M_Garland says:

    A possibility. Joe has chosen a primary care doctor Dr. Family near his home. Later, he is referred to specialists that are located more randomly. The centroid for this group is farther from his home than Dr. Family.

    ObamaCare cuts his connection to some or all of his specialists. He doesn’t replace them immediately, so his provider network is pruned, but he keeps Dr. Family. His average distance to care goes down.

    Mike was unlucky. He lost his connection to his primary doctor and his specialists. His options are limited, so his new primary doctor is farther from his home, and his specialists may be also. A limited group of specialists is going to be farther away from patients on average than a larger group.

    Statistics can be interesting, but they tell us nothing without careful matching to a mechanism. Not a proposed mechanism, but an investigated and verified one.

    The problem with government programs and ObamaCare is that they are mostly statistics filled out with speculation. The mechanisms are largely unknown. The government doesn’t know why it has those statistics. It often doesn’t even verify the statistics it uses.

  4. Darien says:

    My read on it is that it’s oddly-worded, but mainly it’s saying that, if you compare the set of people keeping their old providers post-Obamacare with the set of all insured patients pre-Obamacare, the mean travel time to the provider will be lower in the first group than in the second.

    • Daniel Kuehn says:

      No it’s DID. You’re differencing two pre-post changes across state employee groups.

  5. Andrew Keen says:

    I think it’s saying (in a poorly worded way) that people who got new healthcare providers as a result of obamacare are traveling farther to see their new healthcare providers than those who kept the same healthcare providers are traveling to see theirs.

    It’s not a before/after comparison (as the wording would lead you to believe). It’s comparing group A (same provider) vs group B (new provider) after obamacare went into affect.

    • Daniel Kuehn says:

      I don’t think it’s a comparison of old provider to new provider users – the type of provider distance is the outcome and the comparison is a DID design over time and across state/non-state employees. I’m not sure what’s going on but it’s presumably got something to do with an interaction between those variables and the other demographic characteristics and the outcome. Not exactly clear.

  6. Dan says:

    Easy. Obamacare made them so rich that they moved to a new house closer to their doctor.

  7. AcePL says:

    I believe this is a case when Gruber and his team provided “complex analysis and/or modelling” overnight, when [some governmental unit dealing with statistics or some other data crunching] required weeks for the same work”.

    In other words: bullshit peddled by “policy writers” to the Congress by providing doctored data.

    C’mon, guys… how hard can it be?

  8. Tel says:

    I believe if Bill O’Reilly were here, he’d say, “You can’t explain that.”

    I believe if Obama were he he would say, “Yes we can.”

    We find that overall there is no significant impact on distance traveled for an office visit. But we find that this masks important heterogeneity by type of office visit: distance traveled for primary care visits falls by 0.65 miles in our reduced form estimates, or by about 5.6 miles as our implied IV coefficient; this is more than half of the baseline distance traveled for primary care. [footnote 12].

    On the other hand, distance traveled rises for specialists.

    Then in footnote 12, bottom of page 23:

    It is possible to observe a negative impact on primary care distance as a result of compositional change in primary care visits. We observe that limited network enrollment is correlated with an increase in primary care visits and, to the extent that those additional visits are to providers who are relatively short distances away from the patients, it is possible to find that limited network enrollment is associated with a decline in average primary care distance.

    So Gruber does explain how it works, although in this case the “shorter distance” kind of sounds like it’s a good thing for the patient, but then when you look closely at how they get this result you might question whether it’s really that good.

  9. Yancey Ward says:

    If I had to guess, the wording is hiding something bad. This has Grubering all over it.

Leave a Reply to Daniel Kuehn

Cancel Reply