Medical Homes Fail Yet Another “Natural Experiment”
Three “natural experiments,” three failures. Such is the fate of patient-centered medical homes (PCMH), a well-intentioned but unsuccessful innovation now kept afloat by the interaction of promoter study design sleight-of-hand with customer innumeracy.
By way of review, a natural experiment is an experiment in which the design is outside the control of investigators, yet mimics an experiment. The first two natural experiments below involve applying the intervention across entire states. The third involves a stimulus-response experiment in one specific community.
Statewide Natural Experiments: North Carolina and Vermont
In North Carolina, a statewide Medicaid PCMH was implemented years ago and steadily expanded until most Medicaid recipients belonged to one. There was no reduction in relevant event rates (for ambulatory care-sensitive admissions) and costs increased. While the overall Medicaid budgets were routinely exceeded and that should have caused legislators to realize that something in their PCMH was amiss, Milliman fabricated data to pretend the PCMH program was a success. Milliman got caught making up data (and ignoring other data that quite definitively invalidated its conclusion, and changed their story 180 degrees, a tacit admission that they lied. And shortly thereafter (at least “shortly” by the standards of state government), North Carolina announced that it is abandoning this failed experiment.
In Vermont, the making-up-data-to-disguise-failure story was similar, following the implementation of a statewide medical home model. Even though there was no change in relevant hospital event rates (ambulatory care-sensitive medical admissions), the PCMH model was initially alleged to save $120,000,000, or a whopping $600/eligible person. Had this figure been real instead of completely fabricated, Vermont could have financed its long-planned move to single-payer simply by projecting that savings to the entire population. Embarrassingly, the state was so taken with the faulty analysis that they didn’t back off their single-payer plan for a year after the savings were shown to be made up.
These two results are supported by the findings from Oregon Medicaid’s healthcare “lottery” program. Oregon Medicaid was not a “natural experiment” in PCMH’s, but rather in whether increased access to care made a difference in outcomes. It turned out going from no insurance to insurance offering free primary care had no effect on physical health after two years. The Law of Diminishing Returns would imply that adding even more primary care would therefore also have no effect. We’re not counting that as a natural experiment, but rather as inferential support for the two natural experiments.
The response of the PCMH lobby to these natural proofs? Forced to defend the indefensible against an onslaught of both fifth-grade arithmetic and actual events proving the fifth grade arithmetic accurate, the head of the PCMH special interest group, Marci Nielsen, simply made ad hominem attacks on me. (These attacks were not html and hence not linkable. Rather there is a healthcare policy email group to which industry thought leaders–and apparently the occasional lobbyist–are invited to participate.)
Lest I give the impression these childish displays of churlishness bother me, quite the opposite. I’m bummed the thread isn’t linkable. In the immortal words of the great philosopher George W. Bush, bring ’em on. The fact that these people can’t actually construct a coherent argument demonstrates the vacuity of their position. Literally, one of Ms. Nielsen’s arguments was: “That’s the pot calling the kettle black.”
I revel in that kind of criticism. I was hoping she would accuse me of having small hands.
Flint: The Most Recent Natural Experiment
There are no redeeming features in the Flint debacle, but creating a “natural experiment” in the PCP-patient relationship – the underlying linchpin of the PCMH model — is as close as it gets.
Roughly 50 primary care physicians practice in and around Flint. Only one figured out the obvious: their patients were all drinking poisoned water. That means 49 of them – 98% — had so little connection to their patients’ day-to-day lives that they didn’t make their patients aware of the obvious and substantial harm visible to the naked eye. All these doctors had to do was look at the water and then get it tested. But only one did.
Yet the entire PCMH model is based on the implicit assumption that primary care physicians should be the stars in the healthcare firmament. That hypothesis was given a simple natural experiment – helping patients avoid or at least identify obvious harms — and totally flunked.
Presumably some of those doctors in Flint had PCMH-levelrelationships (even if not a formal PCMH) with some of their patients. It didn’t help. What we learned in Flint is that the idea of “building a relationship” with a doctor by getting a checkup may be like building a relationship with your Amway rep by purchasing Artistry®LuXury Eye Cream.
This is not to denigrate the role of the PCP at all. That is people’s first line of care, especially for people who have enough health problems to merit regular visits, as well as for patients with acute needs. However, the two ideas that:(1) sending more people to the doctor builds relationships and relationships matter a lot; and that (2) PCPs somehow get it right/save money both fail these three natural experiments. No natural experiment has ever shown the opposite.
So — just like in wellness where it turns out the way to save money is not to spend more of it on prevention –it’s time to reconsider the entire PCMH assumption that spending more money on PCPs saves money somewhere else. In other words, it’s time to re-think the entire PCMH concept.
And, just for the record, my hands are fine.
Al Lewis is the co-founder of Quizzify and the author of Why Nobody Believes the Numbers.