Did direct primary care save Union County $1,280,000? Don’t bet on it. [Updated 1/6/18. Re-updated 1/13/20]

See 2020 update below.


A 4+ – year age gap between two health care enrollee pools explains a lot.

DPC advocates find their poster child.

A group of self-declared “fighters for health care freedom”, such as North Carolina’s John Locke Foundation and its Director of Health Policy, Katherine Restrepo, have been heavily promoting a health care delivery vehicle known as direct primary care, or DPC. They claim that direct primary care reduces health care costs and improves health care quality, primarily by having patients receive primary care through an insurance-free clinic. The savings in insurance overhead will presumably allow direct providers more time for patient primary care and that, in turn, curbs clinic members’ need for additional services outside the clinic like expensive specialists, emergency rooms, hospitals, and costly medications.

Ms. Restrepo found her poster child for direct primary care in tiny Union County, North Carolina. She tell us that, when compared to County’s average costs for the 1120 employee and dependent lives in the County’s more traditional insurance plan without DPC, direct primary care yields the County an impressive 23% cost savings amounting to a $1.28 million dollars a year, or $260 per employee per month. By another measure, the county is said to have 28% cost savings.

But there is good reason to believe that the reported savings are simply the result of selection bias. A an age gap of over four years between the two pools seems to explain a lot.

Here are just five of many possible avenues for selection bias that seem likely to affect Union County enrollment choices.

Beginning in the county’s 2015-16 fiscal year, Union County employees were given, for the first time, a DPC alternative to the county’s more traditional insurance plan.  If they elected the DPC, they received unlimited primary care, with no deductibles or copayments, but only through one of two doctors at the DPC clinic. If they declined DPC membership, they were allowed to see any primary care physician they wished, but they did so in traditional pay as you go practices subject to deductibles and coinsurance.

Despite DPC’s considerable financial inducements, 59% of Union County employees elected a traditional pay-as-you-go-with-doctor-choice plan over DPC’s all-you-can-eat approach. A highly probable explanation for this tilt to the traditional plan is that the narrow selection of physicians in the clinic is a barrier to enrollees unwilling to sever established relationships with their previous primary care physicians.

The strength of bonds to prior PCPs are likely to increase for patients who have had close relationship with their PCP, one fostered either by a history of serious health problems or simply by having had the same doctor for a long time. The narrow physician panel of the direct primary care clinic, therefore, could be a potent generator of age and health status selection bias adverse to the traditional plan.

Consider also that every new employee entering the county’s health plan has, since DPC became available, been defaulted into direct primary care. New employees are apt to be younger than the existing employee complement, so decisional inertia may be a potent source of age selection bias.

Furthermore, as part of her case for direct primary care, Ms. Restrepo points out that the DPC clinic is located near county offices, making appointments convenient for the hundreds of county employees working there. While this is likely to attract current employees into DPC, Union County’s health enrollment includes about 150 retired former employees, not yet eligible for Medicare, for whom the clinic’s location is no particular advantage. Since current employees are usually younger on average than retired employees, the clinic’s location may be a yet another significant source of age selection bias.

Next, consider the selection bias resulting from the special attractiveness to families with multiple children of DPC’s absence of copayments. Premiums (in either the DPC plan or the traditional plan) paid by employees with dependent children are the same for any number of children. But the larger an employee’s family is, the more significant the absence of copayments under DPC becomes. The DPC pool skews young by pulling in families with more children per employee.

Finally, at every signup or renewal point, there will be a certain number of potential members who know they are all but certain to exhaust their out-of-pocket maxima during the upcoming coverage period. Some may be in a hospital at the very moment of renewal. Some may be awaiting an already scheduled  surgery, a planned course of chemotherapy, or the delivery of a child. Some will just have chronic illness that exhaust their OOP year in year out.  Such members will anticipate having as many primary care visits as they wish without paying a financial penalty, even under traditional fee for service. For them, higher health care costs are themselves the source of a reduced incentive to choose direct primary care. 

And some of those who will expect to exhaust OOPs are likely to have particularly high claim costs. About 1.2 % of claimants in group insurance pools are so-called “high cost claimants”, with annual claims in excess of $50,000; Their average annual claims exceed $120,000 per person and combine to account for over 30% of all claims. Each high cost claimant that has leaned away from direct primary care will account for 10% of the claimed $1.28 million savings.

Are DPC advocates oblivious to age-cost curves or merely pretending to be so?

Unless asked a direct question, DPC advocates withhold that there is an age gap between the younger direct primary care pool and the older traditional insurance pool that exceeds four years.

Age matters though, and it matters a lot.

Age-cost curves for health care are steep. Many conservatives argue that the costs for 64 year olds are five fold higher than costs for 21 year olds; they argue that premiums should reflect this 5:1 ratio and nothing less. Of the 3:1 ratio mandated by the Affordable Care Act, JLF’s Restrepo personally complained that “Obamacare causes chest pain” because “[t]he old and sick [ ] benefit at the expense of the young and healthy”.

As an interim step pending legislation to establish a 5:1 ratio, the Trump administration once floated the idea of moving by administrative fiat to an age-premium curve of 3.49:1. On that curve, the age gap between the two Union City groups in age would explain every penny of a 23% difference between the health claims experience of the two populations.

A 5:1 curve would imply that offering the DPC plan cost the county nearly $600,000. Now Ms. Restrepo, the biggest booster of the Union County DPC, has gone silent on age-cost curves. Really.

[If direct primary care is indeed no better than traditional care at reducing overall costs, this is a very plausible number. The county paid the direct primary care provider an annual membership fee of $1500 per adult. Enrollees in the traditional plan received instead a health reimbursement account of not more than $750. Put otherwise, the county’s upfront pre-claims commitment on the direct primary care side is twice the size of its pre-claims commitment on the other side. If direct care had no economic advantage, half of the roughly $1,000,000 in clinics fees was simply wasted.]

Simply oblivious?

Neither high cost claims data that might cast light on risk distribution nor the age differential are addressed in current Union County DPC advocacy. Not even in an article, written for the Georgia Public Policy Foundation by Ms. Restrepo, purporting to “tackle the skepticism” over the “myth” of DPC cherry-picking – an article expressly citing the Union County experience for its “compelling numbers”.

Instead of addressing age differences or high cost claimants, Ms. Restrepo painstakingly breaks down Union County’s DPC enrollees by the number and frequency of their chronic conditions. But what does her article say about how traditional plan enrollees’ health status compares on these same measures? Not a word.

Even if DPC enrollees once picked turned out to somewhat sour cherries, they could easily be more palatable than the bitter lemons and high cost claimants that might have been found in the traditional insurance pool had anyone bothered to look.

Stop the rush to direct primary care.

Let’s not bet the health care of county enrollees, or anyone else, on the idea that little Union County won big savings by offering direct primary care. A far safer bet is that all Union County’s decisionmakers actually managed to do was segment their enrollee population based on health status, then proclaim an unjustifiable win for an unproven health care concept.

Note: for supporting calculations and other information please see certain Google “Sheets” here and here or use the “Contact” menu item above.

Update: See this post for some cost-adjusted data that confirms signficant selection bias, while still suggesting that direct primary care has net positive effects.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: