Try due diligence before betting the health of your employees on Lee-Gross-style Direct Primary Care

In his latest Direct Primary Care slide-show brag, attributing significant overall medical cost reduction for employees electing DPC over and FFS primary care alternative offered by the same employer, Dr Lee Gross insists that the favorable results do not reflect “cherry picking”.

And yet, Dr Gross fails to compare the health status of the DPC-covered employees and dependents and the actual FFS-covered employees and dependents of the same employer. Indeed, he does not even present the most basic demographic information about relative age of the two cohorts. Keep in mind that in the two most widely-published employer DPC studies the age differences between the cohorts (4.7 and 6.5 years) were large enough to explain a 25% or larger medical cost differential.

Gross’s purported rebuttal of “cherry picking” rests on comparing the frequency of a set of chronic conditions in the studied employer’s specific DPC population to the frequency of the same set of chronic conditions in an unidentified “national benchmark population” — eschewing, for some reason, the chronic conditions profile of the studied employer’s own non-DPC population. What can possibly explain that?

Risk adjustment adequate to the task of comparing population health care cost risks is far more complicated than simply counting the number of chronic conditions. Yet, even if risk adjustment was that simple, a meaningful analysis would require some basis for assuring that Dr Gross’s unaudited methods for identifying the presence of chronic conditions in the patients his practice serves are commensurate with the methods used by impartial analysts to establish the the levels of chronic conditions in the unidentified national benchmark population. There is, after all, a significant history of diagnosis up-coding by self-interest providers.

Whether or not Dr Gross cherry-picked the cherry-picking data (or contaminated it by upcoding), his current self-interested brag is just another in a long line that fails to meet minimal standards of analysis. Yet, again Dr Gross has produced a “case study”, adorned with an at-best naive version of risk adjustment without laying out any methodological detail in an actual masuscript.

Dr Gross’s current presentation notably differs from his 2019 presentation from the same DPC option experiment for a prior year. The 2018 data clearly showed that the employer’s DPC cohort had total medical expenses that were 15% higher than their non-DPC counterparts. Was there a sharp change in his case study methodology?

Also note, Gross’s earlier presentation counted the 2018 DPC cohort as having 35% more chronic conditions than the 21% he currently claims.

Whence the discrepancies between 2018 data and later years?

Outliers? In the current Gross slide show and video, Gross alludes to certain unidentified difficulties in 2021, while showing a slide that notes a $440 PMPM for 2021 which represents a 67% increase in DPC member costs PMPM over 2020. In an asterisked note, Gross links the cost increase to a pair of expensive outlier patients in the DPC cohort. That’s a fair point to make — but only if the analyst is equally sensitive to the possible presence of outliers on either side of the comparison being made; since Gross provides no description of his methodology, we have no idea how the broader DPC/non-DPC comparison might look with an appropriate outlier adjustment.

Outlier years? Note too that, Dr Gross’s spending charts for 2019 show almost identical total spending PMPM for DPC and non DPC patients just over $250, so the

focus of his brag is data from 2020. For that year, more than 60% of the cost difference came from hospital inpatient expenses. That same year, note, the non-DPC group’s hospital inpatient expenses soared to more than 70% above Gross’s selected national benchmark.

Given that 2020 saw a pandemic that, on a world scale, drove hospital admissions and health care needs disproportionately with increasing age, it might be helpful to know the size of the age differential, if any, between the studied DPC and non-DPC populations. More broadly, given that 2019 total costs for the DPC and non-DPC cohorts were substantially identical (and 2018 cost even lower for the non-DPC cohorts), the emergence of roughly 50% savings for the DPC cohort in pandemic times should be examined with through a pandemic-sensitive framing .

In this regard, it should be noted that the employer studied by Gross is a hospital, and all its employees are health care workers, and likely at or near the tip of the pandemic spear. Even seemingly small differences between the DPC and non-DPC cohorts might have been amplified in pandemic conditions. Perhaps hospital employees of different departments have differing inclinations toward DPC, a factor unlikely to matter unless Covid-19 risk expands specifically in those departments.

What we do know is that Dr Gross study was small, less than one-sixth the size of the Milliman DPC employer study. For 2020, based on such numbers as Gross does present, the average number of employees in each cohort was about 150. Dr Gross’s own computations relate savings that varied from 0% to 50% from one year to another; that overall patient costs in a cohort can rise by over two-thirds in a single year; and that in his computations a single pair of patients can account for 25% of an entire cohort’s costs.

As a result of all this, all of Dr Gross’s purported successes can turn on a handful of outlier enrollees and are unsteady year over year. Most particularly, the results during the brag come from two pandemic years featuring mass disruption of world health care patterns; they may mean nothing at all.

Should you work with Dr Gross, be sure to ask him how he knows your first year experience will not better reflect the studied employer’s first year of 15% losses, or the studied employer’s last pre-pandemic year of breaking even, rather than reflect the extraordinary, pandemic years for which he has apparent success. Ask him for demographic data on the studied employer’s DPC and non-DPC populations.

Don’t bet the health of your employees on Dr Gross “case study”.

Including Primary Care in Health Insurance Policy Coverage Is Reasonable

The missing part 5 of Brekke’s “Paying for Primary Care”, a comment.

Under the traditional insurance model, patients receiving covered primary will indirectly pay significant administrative costs, but they may also gain compensating financial advantages that Gayle Brekke’s multipart “Paying for Primary Care” series fails to recognize, ignores, or minimizes. At the top of the list, insurers bring the combined strength of large numbers of insureds to negotiating payment rates with all providers, including PCPs. Even though providers may have the upper hand in this battle of competing rentiers, insurers’ primary care payments rates are usually significantly cheaper than self-pay rates. That advantage alone may cover the administrative costs of insurance in their entirety. But the financial advantage of insurance does not end there.

Insurers actually pay about 85% of downstream costs. D-PCPs pay not a penny. This gives insurers a bigger financial stake than DPC clinics themselves in DPC’s loudly professed goal of controlling downstream costs. And insurers have some tools to make it happen. 

For instance, insurers may use their unique access to both downstream care and cost data and to other primary care quality metrics for each and every PCP on their roster to reward success with incumbency and, when necessary, discharge failure. Curating rosters to maintain, monitor, or upgrade the quality of their PCP team comes at a price, but the investment offers the possibility of reductions to downstream care costs through the improvement of primary care.

Direct pay has no comparable institutional mechanism. Indeed, direct primary care might even provide a haven for PCPs kicked off insurer network primary care rosters1.

Insurer’s integration of primary care and downstream care affords other opportunities for gleaning downstream cost reductions. For example, some insurers leverage integration ”upwards” by sending their PCPs timely, automated updates of individual patients’ downstream utilization. Insurers also leverage “downwards” by collecting  network wide health metrics from their PCPs  to identify actionable specific needs, then formulate responses (e.g., adjustments to drug formularies). Such gains from systematic integration will entail certain “administrative” costs, but may provide modest net benefits that are not available when primary care is delivered as a literally dis-integrated financial standalone through DPC. 

Direct primary care practitioners talk a good game about primary care reducing downstream care costs. But, since insurers write the checks for downstream care, insurers have the deepest and most direct financial incentive to see that their PCPs play a primary care game that actually leads to downstream care cost reductions.

One caveat. Any given insurer’s financial stake in taking positive action today to reduce downstream care costs later will be, to a degree, attenuated by the expectation that their insureds may have changed insurers by the time some of the fruits of “prevention” are realized. On the other hand, attenuation for that particular reason is probably considerably reduced for publicly funded health insurance and for self-insuring employers. However, even where an insurer can expect relatively high churn year-over-year, as in the individual market, the benefits of a good primary care game played today still produce good financial results next week, next month, and throughout the current policy year. On the penultimate day of a patient’s final policy year, her insurer could still have a larger direct financial stake in tomorrow’s downstream care than her D-PCP. And D-PCPs see member churn as well — one prominent DPC advocate has written a whole book about it.

All PCPs, no matter the payment model under which they practice or the compensation they accept, have the identical ethical and legal incentives to provide care that reduces the need for and cost of downstream care. To the good work of dedicated and honest PCPs, using insurance to Pay for Primary Care adds what direct pay can’t — both more of the will (though direct financial incentives) and some important ways (through institutional structure) to shape primary care for the goal of lower downstream costs. Reckoning the net financial effect of Paying for Primary Care with insurance requires attention to both administrative cost burden and to likely or potential benefit.

Inevitably, selling a direct primary care product built around unlimited numbers of nearly on demand thirty minute visits at a fixed price leads D-PCPs to claim that traditional primary care practice is a failure because it supplies a smaller quantum of primary care services than subscription DPC. 

Does the traditional model supply too little primary care? Precisely because they attach high value to primary care, the vast majority of insurers set low financial barriers to primary care access. A significant majority of insureds have primary care without application of deductibles; a majority of those both in the individual market and in employer groups see PCPs for a copay of  $40 or less. A large fraction of insurance plans provide a set number (2 – 5) of primary care visits per year with no cost-sharing of any kind. And all insured primary care comes with zero cost sharing for a battery of the most important preventative services, like immunizations (flu, covid) and screenings (e.g., PAP smears).

But because primary care is both useful and expensive, insurers seek to optimize the quantum of primary care delivered rather than simply maximizing it. So, while insurers are invested in seeing that patients get all the primary care that is medically reasonable and necessary, they also guard against services rendered in the name of primary care that is neither. For example, some D-PCPs actually commit to a minimum thirty minute visit length; at least one prominent D-PCP has set a 45 minute minimum; and a survey of DPC practices reported an average of 38 minutes. Note that by that standard, a substantial majority of all family physician visits fall well short on ever single day of work of providing what DPC docs would automatically deem appropriate care. Primary care specialists properly seek, and insurers properly allow, a “99214” or “99215” for less than forty percent of primary care visits. Just as they do for other elements of healthcare for which they are on the financial hook, insurers examine claims to squeeze out losses from abuse and overuse. It essentially impossible for self-payers to do this kind of “police work”.

As we contemplate the possibility of overuse, our attention turns to the elephant in the DPC room: unlimited visits without cost sharing. That practice entails costs inflated by moral hazard. With no one obliged to pay even a penny for an additional visit, and all paying the same flat membership, the price of membership must be high enough to pay for all needed primary care and for all the “induced utilization” that results from having unlimited visits with no cost sharing at all. 

Insurers have shown that the tension between the value of low cost barriers for primary care and the costs of moral hazard can be modulated by, inter alia, a balanced approach to cost-sharing. But this is far off-brand for subscription DPC, a model that seems almost premised on the idea that primary care can never be overused. DPC’s unique model greets moral hazard at the door, then bars the door to the insurance system’s best tools to mitigate moral hazard. 

Once again, we have an insight into why DPC is expensive. After insurers’ efforts to optimize primary care for cost-effective reduction of downstream care risk and cost, primary care total spend is about $450 per year for an adult, privately-insured patient. Grossed up at 125% to allow the maximum amount of insurer-side administrative costs and profits, it still comes to less than two-thirds of $900 spent for an annual direct primary care subscription.

As insurance companies are human institutions, not all aspects of insurance practice will have stunningly cost-effective results. The most effective cost control measure is almost certainly the ability of insurance companies to restrain, somewhat, provider reimbursement rates; these may alone have enough payoff for patents to justify all the extra costs of insurance. At the same time, because PCPs as a group are highly competent, evaluating physicians with quality metrics or downstream care cost comparisons may bring only modest net benefits (and can be dropped if not effective). Lying somewhere in the middle of the cost-effectiveness scale are insurance rules and processes that target fraud, abuse, overuse, or moral hazard; that these rules have real bite is made clear by the squeals of D-PCPs over denied claims.

In any event, Brekke’s method of accounting only for the costs of performing administrative tasks avoided with direct pay without accounting the financial benefits added when insurers perform those administrative tasks helps explain why Brekke’s conclusion is at war with observed reality. Traditional primary care at $450 per annum is half as expensive than $900 per annum subscription-based direct primary care.

The overwhelming majority of DPC advocates confess what Brekke will not. That subscription direct primary care is more expensive than insured-based primary care. Then, they argue that DPC is worth the added cost because (a) vast increases in the quantum of primary care are really, truly, cross-my-heart-and-hope-to-die cost-effective in reducing downstream care costs, so that DPC results in net savings despite high DPC fees and (b) that DPC gives a superior patient experience. 

There is essentially no evidence that the first proposition is true. In fact, there is thus far only one actuarially sound, independent investigation of DPC, and the investigators concluded that $900 a year DPC was a break even proposition. Even that study appears to have substantially overstated the case for DPC. Perhaps most tellingly, organized DPC has stood adamantly against D-PCPs actually sharing the risk of downstream care costs. 

Does DPC even deliver on the promise of expanded primary care? Only a few DPC clinics have actually reported their primary care utilization figures. One such braggart’s report actually revealed an average of well under two annual office visits per patient, with the only other live contacts being telephone calls at the rate of one every two years. The good news is that the sheer inconvenience of clinic visits may keep moral hazard in check.

The bad news is that this clinic’s annual adult fees exceed $1000 for that pretty ordinary care regime. If DPC patients get neither lower downstream costs nor significantly more primary care than they would get through a $500 annual spend under the traditional model, what do patients actually get when they pay about twice that amount directly to a D-PCP? 

The aforementioned improved patient experience, right? 

DPC patients are not just Paying for Primary Care. They are buying a bundle of medically necessary primary care paired with medically unnecessary “concierge” services. The concierge add-ons are made possible by DPCs having small, but very well-paying, patient panels. Precisely because concierge services are not medically necessary, and because insurers have not found such services helpful in reducing net costs, insurers have neither a duty nor a reason to pay for them, let alone to police their price. Concierge care is an optional convenience or luxury product, not unlike amusement park queue jumping. A PCP will try to sell it at the best price he can get directly from a patient. 

Added concierge services and/or inflated PCP compensation accounts for the vast bulk of the annual hundreds of dollars of excess cost of direct primary care over insured, medically necessary primary care. For marketing and other financial reasons, D-PCPs masquerade the concierge component of their product as an economic and medical necessity. But concierge medicine, even in the “lite” version called direct primary care, is expensive and unnecessary. It is just something nice for both patients who can afford it2 and for the physicians who for their own reasons — laudable or otherwise3 — prefer that model. 

FOOTNOTES

1 If a PCP had been formally admonished for negligent practice by his state medical board in 2009, would you be surprised to learn that he started his state’s first ever DPC practice that very year? Don’t be.

2 In discussing multiple forms of moral hazard in Part 3 of her series, Brekke wrote, “physicians may recommend more and more expensive care when they know the patient is insured.” I suggest a more inclusive rewrite. “Physicians may recommend more and more expensive care when they know they can get paid for it, whether through insurance or from direct pay patients with ample non-insurance resources.” There is no demonstrable reason to believe that D-PCPs are necessarily less inclined to profiteering than their FFS-PCP counterparts.

3 Some D-PCPs are true believers. But consider also how attractive DPC must be for those who run afoul of insurer’s effort to police abuse or overuse, or for those who might be unwelcome in networks. Sometimes PCPs ditch insurance network contracts . Sometimes insurance networks ditch PCPs. See note 1 supra.

Brekke’s “Paying for Primary Care”, Comment on Part 4

In the first three installments of her Paying for Primary Care series, actuary Gayle Brekke’s invoked actuarial principles and behavioral economics to scold coverage of primary care on the ground that the costs of primary care are “predictable, routine, likely events over which the customer has a great deal of control”. In her fourth installment, on the other hand, Brekke tenderly embraces subscription-based direct primary care which provide coverage for these exact same “uninsurable” primary care events. 

In this response to Brekke 4, we explore some of the foundations of Brekke’s enthusiasm for $900 per year subscription DPC. Doing so actually helps clarify why the traditional insurance model might deliver primary care at a $565 average annual adult cost, a lot less than DPC.  

In Part 4, Brekke develops a theoretical perspective on physician financial incentives that declares that patients’ financial interests are certain to be more impaired under the traditional model than under direct pay. In a nutshell, she tells us that too many corrupt, venal FFS-PCPs incur unnecessary visit costs because they are immune to patient dissatisfaction, while all D-PCPs are so fully sensitive to patient satisfaction that they are immune to venality and corruption.

Brekke explicitly accounts this difference to the proposition that “in exchange” for entering agreeing to the terms of an insurers network contract a traditional FFS physician receives “a full schedule of insured patients”. Accordingly, physicians have zero financial incentive to satisfy their patients. Although this is the linchpin of her financial incentives argument, neither Brekke nor other DPC advocates who make the same claim, e.g., Dr Kenneth Qiu, adduce any evidence for it.

Actual evidence points the other way, that insurers rely on patient-satisfying PCPs to keep their member panels full and growing. Almost every insurance plan election platform includes an engine allowing prospective plan members to determine the network status of preferred PCPs.

Brekke, like other DPC advocates, also seems to feel that D-PCPs are ethically superior to FFS-PCPs; there is certainly no evidence of that. What is clear is that all PCPs, regardless of who writes the checks, owe precisely the same duty of patient care, as demanded by professional ethics and enforceable by law.

At her most vivid, Brekke proposes that too many traditional physicians will even “require another visit when the patient has a second concern so that they can charge another fee”. (How is this necessary, or even possible, with those full schedules?) I hear Brekke’s suggestion that a corrupt PCP would require an unjustified second visit just to collect an extra fee. What Brekke cannot hear is this inner voice of that same PCP:

How sad to be a petty grifter sneaking an extra primary care visit charge every now and then, and hoping the insurer doesn’t catch on.

Do’h! I actually need to show up for those visits to make that grift work. 

I need a better and bigger grift.

I should be selling — in bulk — prepaid visits, some of which will never even happen, by offering “unlimited” visit primary care subscription packages, at a $900 annual rate for adults, knowing full well that an average adult can be expected to use only about $450 worth of my services and is unlikely to attend more than two visits a year. 

Ima Goniff, MD

There is more than one way to violate the duty of putting patients first. As a new and different payment model, subscription DPC necessarily presents those who so incline with new and different avenues for gouging. But where there is a corrupt will, there will be a corrupt way. 

One DPC thought leader, at least, has recently asserted that subscription-based direct primary care is, in the hands of some, nothing more than a vehicle for venality; Douglas Farrago, author of the best-selling book on DPC, is actively rallying “push back” against what he regards as multiple, insufficiently virtuous iterations of the subscription-based direct primary care model. 

If Dr Farrago is committed to using administrative savings from direct pay to improve patient care, good on him. But merely switching from FFS to DPC is not sufficient to make that happen. Any PCP operating under any payment model who secures overhead savings or who can merely talk his way into higher rates for his services is free to elect any combination of increased personal compensation, reduced personal effort, or increased patent access.

Many will make a laudable decision. Still, whenever some D-PCPs declare it likely that significant numbers of FFS physicians bill for unneeded services just to get more money, it brings to mind the childhood taunt: “It takes one to know one.”

In Parts 1 and 2, Brekke exaggerated by three-fold both PCP-side and insurer-side administrative cost costs associated with paying for primary care under an insurance model. A realistic figure for these two elements combined is about 15% or less. In Part 3, Brekke addressed certain cost-increasing behavioral phenomena, but my analysis suggested that, in today’s actual insurance markets, those factors actually favor the insurance model over DPC. 

In Part 4, Brekke identified theoretical incentives that might lead D-PCPs to do good and FFS-PCPs to do bad, but wholly ignored similar incentives for D-PCPs to do bad and FFS-PCPs to do good. Unsurprisingly, either payment model can be abused to extract more money from patients.  

If you are wondering where abuse might lie, then, consider following the money. DPC clinics charge an adult $900 a year for primary care. Under the traditional model, primary care providers receive an annual average annual adult spend for primary care of $450. 

Most DPC advocates confess the reality that DPC is more expensive, then seek to justify the added costs in one way or another. I can not explain why Brekke insists that paying for primary care with insurance inevitably adds over fifty percent to the cost of primary care. Her four part series fails to supply evidence that this is anywhere close to true.

Moreover, Brekke’s series has an immense blind spot. She made no effort to address whether there are any ways in which “Paying for Primary Care” under the traditional insurance model might, even while increasing some cost components of primary care relative to direct pay, might reduce other components. To address whether there are positive financial “features” of the insurance model that make Paying for Primary Care easier and financial “bugs” of direct pay that make Paying for Primary Care more expensive, Brekke’s Paying for Primary Care needed a Part 5, and a open mind.

I will deliver my take on a Part 5 at this link when it is ready. That post will further explain why, in the real world, the cost of subscription DPC is 50% more than the cost for primary care under an insurance model. I invite Ms. Brekke to respond to Part 5 or, if she prefers, to anticipate it. I look forward to hearing from her.

 

Brekke’s “Paying for Primary Care”, Comment on Part 3

In Part 3 of Paying for Primary Care, Gayle Brekke discourses on the behavioral economics of shared health cost arrangements to conclude that insuring primary care adds costs not seen in direct pay. These cost, she contends, simply add on to the 50% administrative cost burden of insurance she had already she had already declared in her Parts 1 and 2. Even though these prior parts featured large overestimates of the burden addressed, at least her assessments in those two parts pointed in the correct direction. In Part 3, Brekke changes heading to go in the wrong direction.

Brekke correctly notes that paying for one’s own health care, and only for one’s own health care, will (almost entirely) avoid a cost pooling process in which payers are at risk of a net transfer of wealth to those who receive more of the covered care than themselves. Because health insurance member payments are pooled and priced in a way that does not directly reflect the varying needs and preferences for receiving health care services, health insurance is one example of a pooling process that is potentially vulnerable to both adverse selection and moral hazard.

A second such vulnerable pooling process is subscription based direct primary care in which unlimited care is promised at a single fixed price that does not directly reflect individually varying needs and preferences.

Here, I discuss why, in today’s healthcare world, subscription based direct primary care is significantly more vulnerable to the difficulties of adverse selection and moral hazard than ordinary health insurance.

Remarkably, advocate Brekke does not admit to the possibility that the problems of behavioral economics that concern her can arise under direct primary care. But arise they will — as actuary Brekke should realize. Nor does advocate Brekke’s analysis take into account the ways in which the health insurance world currently mitigates the same set of problems, e.g., by combining guaranteed-issue, community rating, and risk adjustment under the Affordable Care Act — things we would expect actuary Brekke to address.

Neither adverse selection nor moral hazard (induced utilization) have been incurably problematic in employer sponsored health insurance (ESI). As to the latter, an employer sponsor simply determines the level of employee health care utilization for which he is willing to pay; he can address moral hazard/induced utilization, in its various forms and to the extent he deems appropriate, by adjusting employee deductibles or copayments, and even more complex arrangements like HRAs and HSAs.

As for adverse selection, ESI typically does not offer much in the way of employee choice between plans. Because an employer is generally not allowed to discriminate on the basis of employee health status, when there is employee choice the same plans are available to all employees. When choice is offered, even as the richer plans attract the sicker employees and cost the employer more, the leaner plans will attract the healthier employees and cost less. The costs and savings stay in the corporate family, as it were, and the differences do not by themselves generate net cost increases.

If, as sometimes happens, cross-subsidization between employees picking different plans becomes of concern, employers can get help from qualified actuaries in rectifying the relative pricing of the plans. An elegant extension of such “risk adjustment”, which illustrates in passing how less complex risk adjustment works, can be found here.

As it happens, adverse selection once had overwhelming salience in the market for individual insurance. But things have changed.

A sicker than expected person “adversely selecting” into a benefit rich plan certainly causes that plan’s outlays to be higher than otherwise expected, and some care costs are thereby shifted to other plan members. A shift of the costs of care, however much it may be unhappy financial news for healthier plan mates, does not by itself increase the costs of care. The wasteful, destabilizing aspect of adverse selection is that of having a health care system in which profit-requiring guarantors of the financial costs of health care costs are in a price competition and are, therefore, incentivized to recruit the healthy and reject the sick.

To survive such a competition, particularly in the individual market, insurance companies have historically engaged in risk selection (cherry-picking), incurring heavy administrative costs for underwriting and the like. If companies declined to compete for the best risks, their member pools become smaller and sicker through “death spiral” cycles of healthy member withdrawal and increasing premiums. When companies collectively compete, administrative costs grew from the weight of defensive measures. The end point of all this was the near extinction of the market for individual insurance by the first decade of this century.

Even when adverse election still loomed large, it seems unlikely that the generosity or nuances of plan coverage of primary care had much influence on the decisions of putative adverse selectors. It is the needs and decisions of the sickest patients that matter most to the financial health and stability of insurance plans. Patients so sick that they are comparing plans on the basis of their mOOPs care little about the details of the road on which they will be blowing past their deductible.

Conservatives, provided us with an object lesson, not long ago. The 104th Congress was Republicans first chance in many decades to do something about their dream to privatizie Medicare,. But making that happen is more likely when individual insurance markets are reasonably stable. To stabilize the existing situation in private health insurance, Republicans to implement a major countermeasure against adverse selection in the private-insurance-linked Medicare Advantage program. The innovation allowed the adjustment of the rates paid insurers to fairly reflect the relative health risk of the persons enrolled in their particular plans. In effect, “risk adjustment” required that insurers with low risk member panels subsidize insurers with high risk member panels. This vastly reduced the incentive to cherry-pick and, thereby, carrier vulnerability to adverse selection. Because risk-adjustment mechanisms stabilize private insurance markets they have had support across the Republican spectrum from the Heritage Foundation through Mitt Romney as well as from Democrats.

As under Medicare Advantage and RomneyCare, health insurance policies under the Affordable Care Act have guaranteed issue and are community rated. Underwriting and other practices that discriminate on the basis of health status are barred. And, the ACA extended risk adjustment of insurer payments to the entirety of the individual and small group insurance markets. It is likely the least controversial aspect of the ACA. As noted in multiple actuarial publications, ACA risk adjustment has worked effectively to significantly mitigate the problem of adverse selection. And, even the Trump Adminstration recently bragged about just how stable the individual insurance market has become.

Once the destroyer of the individual insurance marketplace, adverse selection is no longer an overwhelming driver of wasteful overhead costs in the individual insurance market. While risk-adjustment is still being refined, a major part of the work is done.

On the other hand, there is no risk adjustment apparatus to relieve the problem of adverse selection between competing direct primary care subscription plans. There is a substantial incentive for a DPC firm to seek a pricing advantage in order to grow market share; and it can do this by beating its competitors at enrolling lower risk, lower utilization members. A subscription-based D-PCP unwilling to pick cherries can easily be left with a panel of lemons. We know where enrollment races can lead.

A tacit understanding among D-PCPs to refrain from price competition could mitigate the problem, if it held. But competition for direct pay subscriptions is a growing reality; at least one DPC thought leader has already complained of some direct pay price competitors as hijackers, and he promised “push back”, equating them to a venereal disease that needs to be eradicated. I have no doubt that this summer’s Direct Primary Care Summit will have at least one speaker complaining of a “cream-skimming” competitor.

Thanks to legally mandated risk adjustment, the chance of a purchaser of individual health insurance having to bear significant, unnecessary administrative costs resulting from the phenomenon of adverse selection is small, and shrinking. On the other hand, if that phenomenon has real salience for the question of how to pay for primary care, it is because the absence of risk adjustment renders the subscription model of direct primary care subscriptions wide open to adverse selection and susceptible to cherry-picking competitions.

In sum, adverse selection is more likely to drive up the costs of paying for primary care through a direct primary care subscription than for primary care through individual market insurance.

In his seminal work, The Economics of Moral Hazard: Comment, Mark Pauly had, according to this account, the “key insight … that full coverage may not be optimal under conditions of moral hazard, that is, when consumer demand for health care responds to the reduced marginal cost of care to the individual.” In Pauly’s own terms, “[When] the cost of the individual’s excess usage is spread over all other purchasers of that insurance, the individual is not prompted to restrain his usage of care…. [Some] medical care expenses will not and should not be insured in an optimal situation.”

Pauly’s attention was directed most pointedly at routine matters like physician visits. Brekke’s scolding, repeated in Parts 1, 2, and 3, that actuarial principles command that patients should simply budget for these “predictable, routine, likely” events rather than buy coverage for them closely echos the Pauly view. Pauly was a genius to figure this out, and Brekke was wise to listen. But she has not applied Pauly’s wisdom across the board.

Pauly’s observation sparked attention to the role of health insurance deductibles as mitigators of moral hazard and seeded the movement toward high deductible health plans (HDHPs). These couple coverage for the “big things” to the use of uninsured cash pay market discipline in regard to shoppable, discretionary services. A high deductible plan is a plausible solution for consumers worried about paying for the excess utilization of primary care services of fellow plan members who succumb to moral hazard.

Given that the average deductible in the individual health care insurance market is well north of $3000 for an adult, while typical annual primary care expenditures are less than $1000, moral hazard has largely been mitigated as a contributor to costs — especially costs of primary care — in the individual insurance market.

On the other hand, a consumer who fears paying for excess utilization of primary care services by her fellow plan members had best avoid subscription-based direct primary care plans, which provides unlimited primary care visits at zero marginal cost per visit. Even low deductible insurance policies exert at least some restraint on excess primary care usage, while subscription-based DPC clinics and their most loyal supporters swoon over the complete absence of any such financial restraint. For a recent and typical example, see this blog post — by Gale Brekke! And, it’s part of this very series!

In fact, there is probably no worse primary care scenario from a moral hazard perspective than buffet style, unlimited visit direct primary care. Professor Pauly can be reached at (215) 898-5411; ask him.

Advocacy: DPC gives you unlimited primary care for a regular monthly premium subscription fee. But don’t call it insurance.

Reality: Call it what you whatever you want. Moral hazard will still inflate its costs.

Remarkably, if you peek ahead to Brekke 4, you will see that in her enthusiasm to embrace subscription-based direct primary care, Brekke turned 180 degrees and praised the fact that direct primary care clinics provide subscription-paid coverage for these exact same “uninsurable” primary care events in unlimited number without any per event charge. What was a bug when Brekke 3 applied moral hazard to insurance-based primary care became a feature in Brekke 4 when applied to direct care.

I think Brekke was probably right to the extent she concluded that lower barrier primary care is worth the risk of moral hazard.  Still, the moral hazard problem is obviously greatest when the marginal costs of additional visits is zero, as it is in subscription DPC. Brekke was therefore completely wrong in counting moral hazard costs as increasing the relative price of insured primary care over subscription based direct primary care. 

To the extent that moral hazard and/or adverse selection are relevant to the choice between paying for primary care with health insurance and paying for primary care directly on a subscription basis, each phenomena would bring higher costs to the direct primary care subscription model. In my responses to Brekke 1 and to Brekke 2, I computed that admin costs of using insurance for primary care could be as high as 15%, shrinking by two-thirds a 50% gap that Brekke had projected. Our review of the issues raised in Brekke 3 makes clear that the cost increases associated with moral hazard and adverse selection can only reduce, and may even reverse, that difference.

We are well on our way to understanding why the $900 annual cost of adult primary care delivered through subscription-based no-insurance direct primary care is more than 50% greater than the $565 annual primary care cost for an adult paying under the insurance system. My response to Brekke 4 will bring us close to home; it should appear at this link when ready.

Brekke’s “Paying for Primary Care”, Comment on Part 2

In Part 1 of “Paying for Primary Care”, actuary Gayle Brekke (mis)computed the provider side administrative cost burden of paying for primary care insurance at about 28%; in a response, I showed it likely that the true number was less than 9%, indicting that Brekke had inflated by more than three fold. Now we turn to Brekke’s Part 2, in which she raised (and compounded) the stakes by factoring in an additional 17.5% of primary care costs for insurer-side profit and other administrative expenses. I argue here that Brekke’s Part 2 analysis rests on some apparently mistaken assumptions about the nature and practical effect of the Affordable Care Act’s rules on Medical Loss Ratio (MLR). For insurer-side profit and administrative costs associated with paying for primary care with private insurance, Ms Brekke has yet again inflated reality by about three fold.

The Affordable Care Act sets a maximum allowed amount that private insurers can pass on to consumers for profit and other administrative expenses. That maximum is measured in relation to the amounts the insurers spend for medical care, its “medical losses”. In the employer insurance market on which Brekke bases her computation, insurers are required to maintain an MLR of 85% or greater. This allows them profit and other administrative expenses up to to 17.5% of their medical spend. Since non-profit insurers like Medicare and employer self-insurance have low insurer-side administrative costs (about 3% for all Medicare administrative costs), it is highly likely that somewhere between 60% and 80% of the actually usable share of potentially allowable non-MLR expenses reflects profits.

But roughly half of the people paying for primary care with FFS insurance are in non-profit plans (over 90 million in self-insured employer sponsored plans, over 30 million in traditional Medicare and over 25 million in fee for service Medicaid and CHIP). With just the one misstep of ignoring the diminished role of profit for these insurers, Brekke inflates potential insurer-side savings for the non-profit half of the market by about three-fold. And that misstep is one of many Brekke mistakes.

Within the for-profit half of the market, Brekke’s computation inflates insurer-side costs of primary care by assuming that insurers can realize the full allowable non-MLR share on every dollar of medical costs. But health insurers are currently in profit-constraining competition, even in the individual market, where prices have fallen for four consecutive years. One fairly large player in the individual and small group markets, Oscar, has never turned a profit. More broadly, a 2018 report by the National Association of Insurance Commissioners indicated that average company profits percentages were in the low single digits, (3.3% for 2018, 2.4% for 2017). Even in regard to the for-profit sector, Brekke is in the ballpark of a three-fold plus level of exaggeration.

Brekke herself explains that health insurance for the “big things” is warranted, and all DPC clinics advise subscribers to keep or acquire wrap-around coverage. Many significant contributors to insurer-side non-medical administrative costs, such as advertising, licenses, and governmental relations, are incurred on an enterprise-wide basis; others, like enrollment costs, some membership services, billing for premiums, are on a a per member or per policy basis.

In fact, for the individual and small group markets, CMS makes insurer payment adjustments of about $1.5 billion annually based on its determination that 17.5 % of MLR are attributable to those administrative costs (including profits) which are independent of claims. See rule at 81 FR 94058, at 94099 et seq. In those markets, MLR rules allows administrative costs of up 25% of MLR, suggesting a maximum 6.5% for the claims dependent component of insurer side administrative costs. For these markets, then, Brekke’s factoring in 17.5% of MLR would be a 2.7-fold exaggeration.

Administrative components are baked into all health insurance policy premiums for all policy holders, will barely budge if, say, any given number of policy holders move between a plan with a low enough deductible to effectively insure primary care and a high deductible plan that effectively excludes some coverage for primary care. Whether in the profit sector or the non-profit sector, insurer-side administrative costs that are actually salvable when PCPs ditch insurers are more or less limited to claims processing costs. A conservative estimate is that Medicare accomplishes that task for less 2.0%. Medicare pays carriers less than one dollar per adjudicated claim.

For the half of the insured in the for-profit world, based on the NAIC pegging average health insurance profit at 3.3%, 4.4 % would be a reasonably conservative estimate of profit on primary care medical spend; for the half insured with non-profits, the corresponding number is zero. The average insured patient being equally likely have for-profit or non-profit insurance, the expected value of potential insurer-side profit savings is fairly estimated at less than 2.2%. Adding that to the 2.0% in salvable insurer-side administrative costs, brings average insurer-side non-MLR costs to about 4.2%. Tack on another 25% to provide an additional margin of error, yields 5.25% and is well less than a third of the 17.5% Brekke computes. Once again, Brekke’s calculation seems to exaggerates by more than three-fold.

For Part 1, I computed provider-side administrative costs of insurance at 8.8% when an actually reasonable value of provider insurance overhead was substituted for Brekke’s first 3X+ exaggeration. Performing the second adjustment of Brekke’s methodology, but again substituting a more reasonable value (5.25%) for Brekke’s second 3X+ exaggeration, yields a net result of less than 15%.

Through two parts of Brekke’s four part analysis, using insurance to pay for primary care would seem to add less than 15% to costs, far less than the 50% Brekke hypothesizes.

In none of her four parts, by the way, does Brekke contemplate that a 15% (or even 50%) “vigorish” might pay for itself through the power of insurance companies to negotiate lower primary care pricing. In her own part 4, Brekke decries the economic power that insurers can exert on providers. To my mind, that power goes most of the way toward explaining why the $900 annual cost of adult primary care delivered through subscription-based no-insurance direct primary care exceeds by more than 50% the $565 annual primary care cost average for an adult paying under the insurance system come to half . I return to these considerations in a Part 5 response to Brekke’s Paying for Primary Care. When ready, it should appear at this link.

By the end of Part 2, still oblivious to the reality that direct primary care has proven more expensive than insured primary care, Brekke was still not done theorizing the exact opposite. Looking ahead to Part 3, in which she would apply her actuarial acumen to behavioral economics in the medical context , she promised to show us even more ways that using insurance increases the cost of primary care when compared to direct pay. Here’s my response. (Preview: the very behavioral economics and actuarial principles Brekke invokes throughout Parts 1 through 3 apply with greater force to unlimited free visit direct primary care than they do to insured primary care.)

Brekke’s “Paying for Primary Care”, Comment on Part 1

In the winter of 2021, actuary Gayle Brekke penned a four–part–blogpost–series arguing that the cost of insurance primary care delivery in the insurance system is at least 50% higher than the cost of delivering primary care through subscription model DPC. Notably, Brekke’s work was theoretical rather than empirical; she attempted to compute the relative costs of primary care under the two models, rather than simply measure them. In this post series in response to Brekke, I argue that the measured version of reality came out differently than the reality she computed because of analytical errors and omissions, mistaken presumptions, and misinterpreted references on Brekke’s part.

Health Care Cost Institute data presenting the 2017 primary care costs of millions of insureds shows an average annual primary care spend for 25-50 year olds of just less than $450. The total costs of paying for primary care are larger because they include, in addition to the spend, the amounts that insurers are allowed, under medical loss ratio (MLR) rules, to recover for associated administrative costs and profits up to 25% of the insurers’ share of that spend. As direct primary care advocates point out with remarkable regularity, however, annual deductibles result in a disproportionate and large share of primary care costs being paid by insureds rather than by insurers.

Therefore, assuming insurers’ administrative and profit allocable to primary care medical loss spend are a full 25% of 100% of primary care medical loss spend, then taking 125% of the primary care medical loss spend should be a safely over-generous estimate of the total cost of paying for primary care through insurance. That would come to $565 per year.

For the same period, the cost of direct primary care for same-aged subscribers was $900. Although you might consider direct primary care for its concierge-like (or concierge-lite) features, direct primary care hardly seems a cheap way of meeting primary care needs; in fact, primary care delivered under the $900 subscription model appears to cost at least 50% more than that under the $565 insurance-based model. But I’m not an actuary like Brekke.

Provider-side billing costs savings from direct versus insured pay.

A certain Dr Bujold made an observation relating to primary care practice overhead in a JAMA letter; specifically, he noted that when he began doing fewer hospital calls, which have low overhead,his practice-wide percentage of spending on overhead went up. Duh.

Since the same kind of shift would have occurred whether he was an FFS-PCP or D-PCP, Bujold’s observation has no implications whatsoever for determining the effect of different payment model on billing and insurance costs. There’s nothing in Bujold’s letter that indicates any effort or intention to compare insurance based and direct pay overhead costs. If you want to waste your own time trying to figure out why Ms Brekke cited Dr Bujold’s article as one of two references footnoted as if to support her computation that sets FFS-billing and insurance at nearly 60% of FFS-provider overhead, be my guest; it beats me.

The “actual” numbers on which Brekke bases her computation of provider-side billing savings come from her second, and her only other, cited source. In an AAFP practice profile, a certain Dr Forrest contrasts a somewhat widely accepted 60% figure for total overhead in family practices to what he claims are 25% overhead costs in his own, solo direct pay practice. If you know enough about accounting to giggle when you read his explanation for excluding from his computation any cost for the nurse-practitioner who works with him during patient visits, your sides may split when you also note that he actually claims that his overhead for office space is negative. If recognized principles of accounting were applied, Forrest’s overhead would plainly be vastly higher than he claims. And Dr Forrest accounts himself a significantly bad-ass cost reducer across all of the overhead board, for example, buying bargain basement equipment and supplies and taking out his own trash.

Most importantly, most of the tactics Forrest identifies for reducing overhead are independent of his payment model choice. If his fantasies were actutally true, Dr Forrest could use both zero-cost nurse-practitioners and negative overhead medical space arrangements (if such things existed) while taking his own trash out of the door of an FFS practice. Forrest gives us no basis to determine how much of the purported 60% to 25% overhead cutdown is simply an artifact of his bizarre accounting methods, how much the result of his having perfected parsimony, and how much the result of his decision to ditch insurance as a payment model.

The only thing that can reasonably be inferred from the Forrest article about how much direct pay reduced Forrest’s billing and insurance costs is that the amount was significantly smaller than a 60% to 25% cutdown.

But Brekke took the full 35% cutdown as if every penny represented provider side billing and insurance costs savings. She also assumed that Forrest’s practice (despite his claim to be paying negative office rent and taking out his own trash) was typical of DPC practices. Her resulting computation came up with a net provider-side billing and insurance cost burden of nearly 30%, a step that by itself accounts for most of the money in her final computation that insuring direct primary care imposes a 50% drag on primary care.

On this calculations page, I have set forth of Brekke’s calculation and a parallel calculation in which I have substituted a value derived from an alternative to Dr Forrest’s one-off personal saga. Look at this 2014 quantitatively detailed, peer reviewed academic study by Jiwani et al of “Billing and insurance-related administrative costs in United States health care“. Its evidence-based figure for billing and insurance-related costs in physician practices puts it at thirteen percent (13%) of gross revenues. This works out to a bit under 22% of a 60% overhead.

If you know a better study, let us all know by posting it in the comments. I will give it its own column in the calculator.

Brekke cast her computation in terms of finding a payment that would leave a direct pay practitioner with $60 in compensation; according to Brekke such a visit would cost $75 with direct pay and $96 with insured pay which she computed to be $96 using the values she derived from Forrest’s bizarre accounting. Using the values from Jiwana, I compute that such a visit would require a direct pay practitioner to charge $88.20.

Accordingly, using the result based on actual published, peer-reviewed research, additional provider side administrative costs when insurance is used would appear to be no more than 9%.

Based on her own misconstruction of already dubious figures from Forrest’s practice small solo practice, Brekke built the foundation of her argument on a number that was inflated at least three-fold.

Furthermore, the 9% figure based on the Jiwana paper, represents an upper limit on possible provider-side admin costs saved by ditching insurance. For the computation shown, I made the assumption that all billing and insurance costs would be eliminated under direct pay. But eschewing insurance pay does not remove all the provider side billing costs; in fact, direct pay can add some provider-side non-medical costs all its own.

Direct pay physicians have non-zero billing cost. For subscription direct PCP’s, there are both monthly fees to be billed and collected, and fees for primary care services not covered by the monthly fees. Direct pay physicians are not able to lean on insurers for items like enrollment and roster maintenance or patient identification cards. A look at subscription DPC social media, meanwhile, should make clear that direct pay physicians have had to be more active in recruiting patents. On the other hand, direct pay FFS-PCPs still must still bill and collect for itemized services.

Furthermore, many D-PCPs seek and enter contracts with self-insuring employers, some with multiple self-insuring employers. There’s plenty of administrative burden to go around, including reporting quality metrics to a myriad of employers.

Patients who rely on insurance for downstream care will still need their direct primary care physician, whether subscription or cash FFS, to document the clinical information needed to justify high priced downstream services for which insurers ask prior authorization. In other words, direct pay does not eliminate the PCP-side administrative costs of prior authorization.

Factors like these indicate that net PCP-side additional administrative costs when insurance is used are appreciably less than the 9% from our calculations page. Brekke’s result may be off by four-or-more-fold rather than three-fold.

Before we leave this discussion of Part 1, take a second to note that a frequent claim of direct primary care advocates is that provider-side insurance related savings are by themselves sufficient to fuel a tripling the length of primary care appointments. Even the 35% figure Ms. Brekke used falls far short in that regard.

Further commentary on Brekke’s series can be found here:

Paying for Primary Care with Insurance Makes Considerable Sense

Brekke’s “Paying for Primary Care”, Comment on Part 4

Brekke’s “Paying for Primary Care”, Comment on Part 3

Brekke’s “Paying for Primary Care”, Comment on Part 2

DPC, one year after receiving a heart transplant, prepares to have old heart reimplanted.

I’m old enough to remember pro-DPC advocacy focussed on the long, in person, face-to-face primary care visits, the very “heart of direct primary care”. Surely that’s what motivated Doctor Steve Springer when he proudly opened Southwest Louisiana’s first direct primary care clinic. I say this because it certainly wasn’t telehealth Springer had in mind. After all, he barely did any telehealth prior to two days in mid-March of 2020.

On those two days, though, he seems to have increased his telehealth practice by about 2000%, apparently going from about one virtual visit a week to about 40 visits per day. Here’s his tweet from 3/18/2020

I teased Springer for the obviousness of his opportunism in a reply tweet. Over the next few days, I watched the DPC advocacy community pull together a new spin in its long-standing pitch for DPC legislation, now presented as the definitive Congressional emergency response to the pandemic, even calling it “our bill . . . expanding virtual care to 23 million more Americans “. My favorite part of this effort was learning that the direct primary care movement had, much like Dr Springer, apparently had a sudden heart transplant, now proclaiming in near unison that “telehealth is the heart of direct primary care.”

Those 23 million Americans managed to get through the next 60 days without suffering from the lack of direct primary care’s Telehealth SuperPowers. By mid-May, as pointed out by the Larry A Green Center (a strongly pro-DPC health innovation think tank) 85% of all primary care providers of any stripe were using a significant measure of telehealth capability. Dr Springer was not the only one going from zero to sixty in a pandemic moment. Then, too, by mid-March 2021, in person visit rates had nearly returned to pre-pandemic levels.

Not that telehealth was just a flash in the pan. Some game-changing lessons for primary care were apparently learned, although not necessarily by the “traditional” direct primary care movement. One possible major development is underway at Amazon.

That company had been experimenting with providing its employees with primary care from Amazon employed or contracted doctors. The company was quick to start using telehealth during the pandemic. One thing they noticed was that the children of their employees were falling behind their vaccination schedules. They started a program of sending out vaccinators to employee homes. Then, the idea grew to embrace home delivery of many other services, like blood draws. It was, of course, particularly easy for Amazon to integrate home delivered medications and pulse-ox devices.

A year to the day after noting Dr Springer’s tweet, I saw a tweet from a national health care reporter that linked an announcement that Amazon was “expanding to bring virtual-first primary and urgent care to more Amazon employees, their families, and for U.S-based employers.” Backed by Amazon’s ability to deliver goods and services into the home, Amazon could well deliver a vastly more comprehensive “virtual-first” primary care package into the home that anything ever seen from any existing direct primary clinic.

Another staple of DPC advocate pride is the facilitation of discounted rates for downstream services like advanced radiology. But Amazon is potentially far more effective at “pricelining” MRIs and such than any network of small DPC practices, just as Expedia is more effective than private travel agencies.

And, while Amazon projects its service as “virtual-first”, the company also expects to backup with in-person sites.

The likeliest challenge facing Amazon’s service would appear to be in the area of establishing ongoing relationships between patient and a single primary care physician. Accordingly, within a few days**, I expect to hear from the DPC advocacy community that Amazon telehealth is just a shiny object of minor impact, but that what matters most are those long, in-person visits. Those visits will be born again as the heart of direct primary care.

Which is probably as it should be, because that premium, small panel service is the thing that small independent practices are likely to do fairly well. It actually makes very little sense for highly trained PCPs to spend their time drawing blood or chasing lab and MRI discounts. Whatever the semantics employed, small DPC is “concierge lite”. Premium small panel service takes a lot of valuable PCP time. That, in turn, requires DPC patients to pay significantly more money to their chosen PCP, even after insurance costs are squeezed out.

DPC docs should simply welcome that position in the health care world with, “We’re worth it!”.

** That was quick. Here’s a commentary from a DPC thought leader comparing efforts like Amazon’s to a venereal disease.

Do economic forces lead to healthier patients self-selecting to member- funded DPC practice?

Yes.
And, favorable selection to member-funded DPC is likely even greater than that already actuarially documented for employer funded DPC.

[D]o economic forces lead to healthier patients self-selecting to a DPC practice? . . .

. . . The value proposition for chronically ill patients– needing frequent visits and savings on ancillary services (labs, meds, etc.)– is obviously higher. Given this, the bias for patients seeking care in the DPC model would lean towards sicker patients; not towards healthy people who don’t expect to have great medical needs.

R. Neuhofel, 2018.

In those words, a founding member and then-current-president of the Direct Primary Care Alliance asked an important question and provided his answer. That answer has been offered time and again, before and since, by DPC advocates. It’s also an answer with some significant truth; I address below why it is also a profoundly incomplete answer. But let’s start with some relevant evidence that points away from Dr Neuhofel’s conclusion.

Dr Neuhofel mentioned that studies of the demographics and health status of DPC patients were underway. And, in May 2020, we learned the results of the only independent study of DPC conducted by actual actuaries, that by a team from Milliman on behalf of the Society of Actuaries. That study found that patients electing a direct primary care option as part of their comprehensive employer health coverage had a significantly more favorable health status than those who elected a competing fee for service primary care option. The difference, in fact, contributed substantially to the reduction of what was once the DPC world’s poster child brag to post-Milliman rubble, from “Union County saved 23%” to “Union County lost 1%” .

The Milliman’s team finding is not wholly dispositive of Dr Neuhofel’s question. For one thing, it was only a single study of a single employer’s program. Other employer’s programs might be different, although no one has expressed any general reason to expect a significant departure. In fact, the only post-Milliman report of measurement of health status of the members of an employer’s DPC plan and its FFS plan indicates a selection bias differential that falls within 10% of that observed for Union County.

More importantly, direct primary care provided through an employer as part of a comprehensive health coverage package differs in potentially consequential ways from direct primary care offered directly to consumers by DPC clinics. The Milliman team itself suggested that might be the case, largely echoing Neuhofel in its rationale — and fleshing that out with an interesting example:

Members choosing to enroll in DPC and pay on their own may be less healthy simply because they are better able to justify the recurring monthly DPC membership fee (which is likely in addition to major medical insurance premiums) than members not choosing to enroll in DPC on their own. For example, assuming copays of $35 and $50 for PCP and specialist, respectively, under traditional coverage, an individual with a chronic condition would only have to see either a PCP or specialist roughly twice a month on average (total cost of $85) to justify paying the DPC membership fee, where there are typically no copays. Members without the need for this level of recurring primary care may be less likely to see the financial value in enrolling in a DPC practice.

It is all but a given that among potential members with differing health status, if their ability to manage health costs coverage remains otherwise identical, the less healthy will have a stronger incentive to enroll in direct primary care. The problem with relying on this truism is that potential members will often not be “otherwise identical” in their ability to manage health costs, particularly when they do have differing health status.

Health status and health insurance

The sick may be “better able to justify” add on DPC fees, but as their level of sickness increases they are also increasingly “better able to justify” the higher monthly premiums of benefit rich policies, with lower cost-sharing, including lower deductibles and lower mOOPs. This is plainly an adverse selection environment for low deductible plans, the sicker one is the better the “value proposition” of gold policies over, say, bronze. This is demonstrated over millions of policies each year in the individual market.

But paying DPC fees is far less valuable for gold buyers than it is for bronze buyers. DPC is worth a lot less to those who hit their deductibles than to those who do not. So, the adverse selection into low deductible plans, leaves a pool with lowered sickness levels for DPC.

Consider the exact example offered by the Milliman team as quoted above: potential plan members who anticipate two dozen office visits in a coming year, half of those to specialists. They are so obviously not healthy that the vast majority of them likely have bigger medical cost worries than their primary care copayments. The vast majority of them, whether with employer or individual coverage, are happily avoiding high deductible plans. And even those whose coverage comes from the 20% of all employers who offer only high deductible plans are incurring medical costs at a level that not only exceeds their deductible ($2300 for an average employer HDHP plan in 2020), but approaches or exceeds their mOOP.

In any case, even among the relatively small universe of potential DPC members who are strictly limited to high deductible plans, and even given that the “least sick” have less reason to choose DPC than the “somewhat sick”, the “somewhat more sick” may see smaller net gains when from choosing between DPC and making cost-sharing payments, while the “OMG even more sick” may see net losses from the process.

So yes, even within an obligate high deductible sub-universe, there is likely to be a “Goldilocks” level of sickness at which is DPC the best value proposition. But beyond that point on the sickness axis, the net economic selection force for DPC is gradually diminished, then reversed.

Age, risk, and premiums

Under the ACA’s age banding system, even in the unsubsidized individual market, premiums for young are disproportionately high, making insurance less attractive to the young than the age-based odds warrant. In the subsided market, moreover, the premium skew against the young is actually magnified by the subsidies. Yet, for those who are both relatively young and relatively sick, insurance remains attractive. The upshot, as predicted (with a mix of crocodile tears and evident glee) by some ACA opponents, has been a disproportionate share of the young and healthy opting out of insurance coverage altogether, or choosing high deductible coverage.

DPC clinics benefit from the resulting enrichment of youth and health in the resulting pool of likely enrollees..

To be sure, of those have no (or reduced) insurance coverage, the least healthy are more likely to select DPC as Neuhofel would predict. But each of those drawn in will have been drawn from a group enriched with youth and health — they will be no more than the “sickest of the least sick”, a tail likely insufficient to wag an entire dog.

In the employer-based insurance world, age-driven effects are mitigated because employee shares of premiums are essentially level over all adult ages. DPC enrollment under employer DPC options is, therefore, not subject to the age-driven influx of good risks described above. Accordingly, we would expect consumer-paid DPC to have an even larger selection bias in their favor than the substantial selection bias already documented for employer paid DPC.

Wealth and social determinants of health

Another way in which economic factors leads to a disproportion of healthier patients in DPC panels is less direct, but likely consequential. Consumer-paid DPC is more accessible to those financially better able to add monthly DPC to whatever other financial arrangements they have for health care. Having a higher income helps in buying anything. As well, consider that high deductible plans that work best with DPC are also more financially accessible to those with accumulated savings.

In whatever form it takes, the same financial advantage that inevitably facilitates payment of DPC fees and election of high deductibles correlates with a long list of social determinants of health, such as the availability of stable housing or employment. A broad wealth/health nexus is real and well documented. That nexus pushes the mean of consumer-paid DPC panels in the direction of a healthy population.

Note, too, how this issue of economic status again distinguishes employer-paid DPC from consumer-paid DPC. When an employer offers DPC to all employees, any wealth/health effect pushing employees toward DPC is attenuated. Accordingly, on this score as well, consumer-paid DPC choices should actually skew healthy relative to comparable employer-paid DPC options already demonstrated to skew healthy.

Some DPC advocates claim to serve more than a few uninsured subscribers. As I have suggested, at any level of income at which the choice to be uninsured is voluntary, the population so choosing will skew healthy. But what of those whose low incomes make the purchase of insurance impossible?

There is a large population with incomes below the poverty line that have no realistic possibility of paying for insurance coverage. Low incomes even skew high risk. Does the low income population significantly tip the balance of DPC panels? In over two-thirds of the states, they are Medicaid eligible. But in the remaining “non-expansion” states, there is a population in poverty who are eligible for neither “expanded Medicaid” nor for ACA subsidized private insurance. How many of these can muster $75 a month for a DPC membership? Arguably, some of the most sick of these might be able to justify DPC as the closest they can get to insurance. But note that many of the sickest of the poor can eligible for traditional, “unexpanded” Medicaid and/or Medicare coverage by reason of disability.

I’ve seen no evidence that DPC seeks or obtains significant numbers of sub-poverty level subscribers, let alone whether the handful they may get skews high risk.

Casual shout out. I know D-PCPs who are doing great things for some quite sick, low income, uninsureds, even doing tasks for which they would refer their insured patients to specialists. And here’s a second shout out to FFS-PCPs who do the exact same thing.

Hard actuarially sound evidence has already shown that employer-DPC skew young and healthy. We await comparable study of the demographic and health status of consumer-paid DPC.

Still, when we put the value proposition of DPC for the sick vs DPC for the well in the context of a broad, real world dynamic of available economic factors, it is far from obvious that consumer-paid DPC skews sick. Instead, the economic considerations point toward consumer-paid DPC skewing young and healthy — to an even greater extent than already demonstrated in employer-paid DPC.

After noting, as above, how employer-paid DPC seems likely to mitigate some of the selection pressures toward the sick in consumer-paid DPC, I gave some thought as to whether there were other structural differences tied to payment source that might predict other differences in skew. Paying DPC fees out-of-pocker puts more patient “skin in the game”; one predictable consequence would be to increase the tendency of the healthiest potential to eschew DPC; on the other side of the previously suggested “Goldilocks” point, however, employer payment of the DPC fee liberates the least healthy potential enrollees from the risk of unnecessarily leaving their own money on the table. Pending an injection of wisdom from somewhere else, I will reserve opinion on the net effect of transferring this bit of skin in the game until we see some actual study results.

Even so, I’ve seen enough to now attempt to monetize this blog in the following way. I will bet that the first credible, independent academic or actuarial study of the health status of broadly representative consumer-paid DPC patients shows that as a group they are healthier than the average commercial patient population. I invite those who disagree to negotiate with me an even-money wager contract in an amount sufficient to both provide significant stakes for the winner and to fund adjudication of a win through standard arbitration procedures. I am happy to afford each of us the opportunity to put our money where our mouth is, and into a escrow account.

Fauxtrage

While strongly endorsing direct primary care, the AAFP’s May 25, 2018 position statement to CMS on direct provider contractor models invoked the structural risk that the prevailing fee for service primary care practice model incentivizes the provision of unnecessary primary care services.

As yet, the fee for service primary care physicians who comprise the vast majority of AAFP members seem to have declined to throw a hissy fit about how their ethics and professionalism had been impugned.

Less than two months later, in an opinion piece in JAMA, Adashi et al., wrote :

[DPC models] are limited by a variety of structural flaws. Foremost, DPC practices lack specific mechanisms to counteract adverse selection that threatens equity in access to care. DPC presents physicians with an incentive structure built on accepting healthier patients with limited health care needs and a willingness to pay a retainer fee. Practices directly benefit when targeting healthier patients and declining coverage to the ill. (Emphasis supplied).


Direct Primary Care: One Step Forward, Two Steps Back
Adashi, et al. JAMA (2018) 320: 637-638

Within two weeks of online prepublication of the JAMA piece, the then-President of the Direct Primary Care Alliance, signing his piece with his official title, replied:

 I find such a perspective completely out of touch and offensive to the entire primary care community. . . . disparaging the ethics and professionalism of over 1000 [DPC] physicians. 

Ryan Neuhofel, A Response to a Clumsy Critique of DPC in JAMA

For good measure, Dr Neuhofel tossed in a direct accusation of academic dishonesty.

Doth protest too much, methinks.

New DPC leader is incredible – unfortunately, not in the good way.

Let’s meet Cladogh Ryan MD, one of the new board members for DPC Alliance for 2021 who picked up the torch from some of those golden oldies.

Dr Ryan cranked up a town meeting style event to recruit some of her Cook County, IL, fee-for-service patients into her new enterprise, Cara Direct Care. She layed on a familiar pitch: “Pay me more, it will cost you less.” In this case, $4,384 less per couple!

Here’s the key content from the video of her 2017 presentation, still proudly displayed on YouTube and on Cara Direct Care’s current homepage.

Really? Taking on an additional $3500 in deductible in 2017 cut insurance premiums by over 66% — by over $7,000 for a young couple in the middle of a 19-44 years old range?

No, it didn’t. It’s not even close.

In the real world outside Dr Ryan’s town hall, the premium spread between otherwise similar plans that differ only by a low versus high deductible was slightly over ten per cent for employer sponsored coverage, per Kaiser’s Employee Health Benefits Survey. In the ACA individual market for 2017, when Ryan made her presentation, the premium spread in her own Cook County, IL, between the policy with the lowest deductible policy and an otherwise similar policy having a $3500 larger deductible was less than 20%. See for yourself,**

The supposed point of Dr Ryan’s patient pitch, of course, is that the alleged premium spread of 66%/$7084 is sufficient to cover her proposed DPC fees, and other expectable costs, and still leave thousands in resulting savings. But the real premium spread was nowhere near $7K. At 2017 pricing levels in the ACA market in Cook County, that spread was under $1700 for a couple in the age bracket Ryan discussed. Her 2017 DPC fees swallowed over two-thirds of it.

Doc Ryan’s analysis began with a premium spread error of over $5,000 for the couple. Yet, even with the head start of a $5K falsehood, Dr Ryan ended up computing typical “DPC savings” of only $4384 for that couple. With real premium numbers, Dr Ryan’s own computation means a loss on DPC, not a savings. Dr Ryan’s analysis is nonsense.

By the way, the cheapest policy in the 2017 individual marketplace for a 21 year old couple near the bottom of Ryan’s range of 19-44, $5308 per year, about $2000 and 50% higher than the figure used by the good doctor for a presumably typical HDHP policy.

I decline to accuse Dr Ryan of addressing her potential patients in bad faith in 2017. But DPC thought leaders seem almost never willing to critically examine DPC-favorable information when it comes along. And, no response has been received from Dr Ryan to dpcreferee.com’s request for her comments on the material developed for this post.

** Here are tables of relevant offerings in the ACA individual market, for Cook County IL as follows: 2017, all policies; 2017 all silver policies; 2020 silver policies. All are sorted by deductible.

2017 Cook Cty All Policies
Cook Cty Silver, 2017
Silver Policies, 2021, Cook County

Note, for example, that in the current ACA individual market, the difference between a silver plan with the smallest deductible ($2000) and a silver plan with the maximum allowable deductible of $17,100 is less than $2681 — that’s a deductible spread of $15,000 (more than four times as large as Dr Ryan’s $3500), priced at a 27% premium spread (well less than half the 66% Dr Ryan’s calculation used.)