Intrigued by many things in my first few days in the U.S., what perplexed me the most was that there seemed to be a DaVita Dialysis wherever I went; in malls, in the mainstreet of West Philadelphia, near high rises and near lower rises. I felt that I was being ominously followed by nephrologists. How on earth could providers of renal replacement therapy have a similar spatial distribution as McDonalds?
After reading Friedrich Hayek’s essay, Use of Knowledge in Society, I realized why. In stead of building a multiplex for dialysis, which has shops selling pulmonary edema-inducing fried chicken, DaVita set shop where people lived or hung out. It wasn’t a terribly clever business plan but its genius was its simplicity, its humility. If the mountain will not come to Muhammed, Muhammed must go to the mountain. DaVita went to the masses.
The link between Hayek’s wisdom and DaVita’s business plan may seem tenuous. But Hayek has been misunderstood, particularly in healthcare. Many a times and oft in the policy world Hayek has been rated about money and usances. This is because of a misperception that Hayek was all about profit and loss, which are anathema to healthcare. Hayek’s message was simple: local knowledge can’t be aggregated. From this premise sprouts others – dispersed agents in certain times and places possess fragments of knowledge which don’t come easily to central planners.
For Hayek, socialism and capitalism weren’t moral but epistemic issues. Socialism would fail because of a coordination problem – markets would succeed because they could use price signals to coordinate. Healthcare doesn’t use price signals to coordinate, not explicitly, at least. Nor does it capitalize on dispersed agents – on local knowledge. Hayek, a supporter of universal healthcare, didn’t specifically discuss healthcare in his essays. Nonetheless, it would be a useful intellectual exercise to speculate how Hayek might have applied his wisdom to modern healthcare.
What does local knowledge in healthcare even mean? Stated in a rather unlettered way, it is the provision of healthcare locally. AEDs are no good if they aren’t located where people congregate. The value of local presence of medical facilities, particularly in poor neighborhoods, is hardly rocket science. Just as great cities grew near rivers, great hospitals germinated in poor neighborhoods. But, with growing centralization of healthcare, with hospitals becoming multiplexes, futuristic cities with a distinct architectural phenotype, different from the neighborhoods they serve, the value of decentralization can be missed.
A recent artistic RCT underscored the value of dispersed agents in healthcare. The researchers asked if a combination of a barber and a pharmacist in-shop can manage hypertension in black males better than a combination of a barber and a doctor in their office. After screening black patrons for hypertension, barber shops were randomized to intervention and control groups. In the intervention group, trained pharmacists, with partial physician supervision, managed the hypertension in the shop – i.e. started the patron on anti-hypertensives, including long-acting diuretics, checked their metabolic profile, and titrated the drugs according to blood pressure and side effects. In the control group, barbers checked blood pressure, gave structured advice about lifestyle, but the hypertension was medically managed by physicians in their office.
Within six months, the systolic BP in the intervention group fell by 27 mm Hg, on average, and to less than 130 mm Hg, the safest space of BP, in two-thirds of patrons. This whopping treatment effect is similar to the VA hypertension study from 1970. The VA study, to recap, was a ballsie study in which veterans with hypertension were randomized to anti-hypertensives and placebo – yes you heard that right, placebo! The VA study was stopped when they discovered that anti-hypertensives halved all-cause mortality.
What about the control group in the barber study? They were no slouches, either. The systolic BP fell by 9 mm Hg, which is nothing to scoff at – renal denervation would have envied a drop of 9 mm Hg – particularly as this was achieved by barbers just talking to patrons about lifestyle and coaxing them to see their physicians.
The clever nitpicker might forage the supplementary appendix for faults in the trial, and there surely are many. Only 319 of the 4567 patrons originally screened made the cut. Barber shops had to be combined, statistically, to make a cluster. But this was also one of the rare occasions where the researchers underestimated the effect size – the actual effect size, a difference of 21 mm Hg between intervention and control groups, was three times what the researchers had estimated in their stringent power calculations.
Before getting too deep into the trial protocol, it is important stepping back and asking what is being compared. This is not a pharmacist versus physician study. This is not a study showing that non-physicians are as good as, or better than, physicians. This is a study showing that for black patrons, pharmacists IN the barber shop outperform doctors IN their offices in managing hypertension. This isn’t even about pharmacists. This is about being there, about showing up, about location, location, location.
A bird in hand is worth two in the bush. Pharmacists in the barber’s shop are worth 21 mm Hg more than physicians in their office.
The barber study quantifies the value of decentralization in healthcare in barometric units: 27 mm Hg. The blood pressure is not just a measure of decentralization but a surrogate for segregation – the tighter the racial homogeneity in poor neighborhoods, the higher the average blood pressure. BP is the hemodynamic equivalent of the Gini coefficient.
The RCT was cluster randomized – the barbershop was the unit of randomization, the smallest indivisible unit, a Hayekian nucleus. What is it about the barber which gives the barber local knowledge about the patron? Surely it is not mastery of evidence-based medicine, an understanding of bioplausibility, or an awareness of risk factors for cerebrovascular disease. Academics might use fancy terms such as “trusted networks.” But it is scarcely believable that black men trust their barbers more than their physicians about their medical problems.
What gives the barber leverage is that their patrons see them NOT to discuss their medical condition, but for a haircut and a banter. I realize this truism is so obvious that to state it is mildly insulting to one’s intelligence. But it is easy missing the paradox of decentralization – blood pressure was managed so well by barbers precisely because their primary job wasn’t managing blood pressure.
My barber, a rugged individual from South Philadelphia and an unabashed Trump supporter, has given me more insight about politics than mainstream media. I enjoy speaking to him. The barber shop is where many still enjoy life’s trivial pleasures. It’s a social hangout, like a pub, or a coffee shop, or a local diner. Its revolving chairs beat Ikea sofas in boutique medical practices not for their aesthetics or comfort, but what they are used for and, notwithstanding Sweeney Todd’s unconventional practice, the barber’s chair isn’t used for lowering blood pressure.
(There is a cute historical irony. Surgeons used to be barbers, which is why the Royal College of Physicians threw surgeons out when they applied for membership. Still smarting from the insult, surgeons in Britain drop the “Dr.” title and call themselves “Mr.” or “Ms.” once they’re anointed members of the Royal College of Surgeons. That barbers are now “internists” completes the karmic cycle.)
The fastidious might generalize the study protocol and, discovering that it is not generalizable beyond the strict trial stipulations, conclude, after generous self-congratulation, that the trial lacks external validity. To be fair, it is easy missing the essence of this study. RCTs, not known for inspiring artistic wonder, incite a scavenger hunt of the exclusion criteria for confirmation of bias. But the black barbershop study isn’t just about black barbershops.
I hope that policy wonks don’t propose barber shops chains, Haircut Hypertension, in poor neighborhoods, or a billing code for haircut-lifestyle counselling, or require that barbers have minimum CME credits to continue cutting hair. One can so easily imagine future archeologists finding beneath the rubble of barber shops a clunky electronic health record subsidized by Uncle Sam. That would be an epic disaster.
If possession is nine-tenths of the law, compliance is ten-tenths of pharmacokinetics. For many asymptomatic people, particularly from fatalistic cultures, taking pills to prevent bad things, such as stroke and aortic dissection, from happening ten years from now may not be their top priority. There’s nothing irrational about this state of affairs – i.e. there’s nothing irrational in not taking pills because one is not symptomatic.
Hypertension falls in the dominion of anticipatory medicine and, as indubitably effective as anti-hypertensives are in making people live longer, compliance is exquisitely sensitive to many factors including the mere act of turning up to the doctor’s office even, or particularly, if located in Philadelphia’s sassy Rittenhouse Square. Decentralization is particularly helpful for an asymptomatic condition such as hypertension – whether it is value for money is a legitimate, but different question.
It is hard not quibbling with the implications of the study which are both obvious and ground breaking. The black barbershop study not only showed the value of getting healthcare to the main streets of poor neighborhoods, but that trained non-physicians can with some, though sparse, supervision manage chronic conditions. The physician isn’t out of the loop but elevated to delegating and supervising a decentralized local network. In healthcare, centralization and decentralization must co-exist. You can’t have unfettered decentralization.
The study reflexively elicits a peculiar objection, particularly in the socially conscious, which is that by using non-physicians we’re short changing poor people – that is managing them on the cheap. The armchair egalitarian’s ire is roused by the inequities this study inspires – if you live in the gated, privileged community of Naval Square, Philadelphia, you have peripatetic cardiologists at your beck and call, but if you live in North Philadelphia you must do with the barber and his apprentice. The morally sophisticated might liken the dispensation of non-physicians to poor neighborhoods to the presence of fast food in these areas. But with a treatment effect of 27 mm Hg, pharmacists in barber shops aren’t akin to a cheap cheeseburger from McDonalds, but caviar in Ritz Carlton.
It takes an odd moral compass to muster greater disdain for the presence, in poor areas, of non-physicians – i.e. someone – than no healthcare provider at all. The streets of West and North Philadelphia aren’t flooded with independent doctors, for understandable economic reasons. Yes, it’s cheaper getting pharmacists to poor neighborhoods than debt-laden, MCAT-excelling physicians. This is common sense, which is disingenuous to dispute. The objectors, the “MD or no one” crowd, sound like a particularly insightopenic Marie Antoinette; “if they can’t have cake, they shouldn’t have bread either.” If physicians can’t literally get to the mainstreets of poor neighborhoods, they shouldn’t deride those who do.
There’s a lot of talk of physicians empathizing with their patients, and of medical students feeling for poor communities. Communication is an essential skill, and appearing authentic, genuine, is necessary for empathy. Barbers have much to teach doctors on how to make small talk with people.
Not all of medical care is equally susceptible to decentralization. Clearly, there are diminishing returns with decentralization – you can’t manage acute pulmonary edema in a barber’s shop. Nor would it do any good having CT scans next to barber shops. Nor is decentralization a panacea for healthcare spending though, as the direct primary care movement has shown, it can reduce the costs that the segregated information domains of excessive centralization induce.
The most unscalable criticism of the barber study, which would make Hayek squirm in his grave, is that the study is not scalable. Hayek might ask despairingly – did you not understand a single thing I said? By its very definition local knowledge can’t be aggregated, can’t be scaled – it is local information. To scale barber-pharmacists is to destroy the pristine Hayekian wilderness, it is to perforce coral atolls upon arctic tundra, it is to gallantly miss Hayek’s point.
The correct interpretation of the black barbershop study isn’t that barbers can replace doctors, or that cardiologists should offer haircuts to their patients or their services, but that decentralization, which is unique in time and place, is powerful in healthcare. If I were a proponent of markets in healthcare I’d give the researchers a standing ovation.
Saurabh Jha is a contributing editor at THCB. He can be reached @RogueRad