The Best Outcomes In Addiction Medicine Belong To Doctors. Here’s Why.

There is a group of people in the United States with unusually good long-term outcomes after addiction treatment. They are not in luxury facilities and they are not unusually motivated.

They are doctors — and specifically, doctors treated under a state Physician Health Program.

The reason this matters to a family choosing care has nothing to do with medicine as a profession. It is that the PHP model is the closest thing the field has to a controlled demonstration of what actually makes the difference, and almost none of it is what treatment marketing sells.

What the research found

McLellan and colleagues (2008) followed a large cohort of physicians treated through PHPs across many US states and reported outcomes at five years. DuPont and colleagues (2009) surveyed those programmes directly to establish what they actually consist of, and set out the model in a companion paper the same year.

The headline figures, from 802 physicians followed for five years:

It is worth being precise about what that first number is, because it is routinely misquoted — including in the first draft of this article. 78.7% is licensed and working. It is not an abstinence rate.

Among the 647 who completed treatment and returned to practice under monitoring, 126 — 19% — had alcohol or drug use detected by urine testing at some point across the five years. A third of those tested positive again.

So the honest summary is not “four in five stayed sober.” It is: four in five were still practising medicine five years later, and one in five was caught using at some point along the way and was still practising medicine five years later.

That second reading is arguably the more useful one, and we will come back to it.

The obvious explanation is that physicians are different: more educated, more health-literate, more to lose. That explanation is worth taking seriously — and it is also insufficient, because the programmes they enter are structurally unlike ordinary treatment in ways that map onto other evidence about what works.

The five things a PHP actually does

It runs for five years, not thirty days. Monitoring and support continue for about five years after the formal treatment episode ends. The residential part is a small early component of a much longer arrangement.

Testing is frequent, random and witnessed. Physicians were typically tested around four times a month in the first year — roughly 48 tests — easing to about 20 a year by the fifth. They call a number every working day and are told whether to report that day. Being tested yesterday does not exempt you today.

It is written down. A signed contract sets out the care, the monitoring, the reporting, and what happens if the terms are not met.

A positive test triggers more care, not expulsion. This is the part most people get wrong about the model. When testing detected use, the usual response was more intensive treatment and closer monitoring — which is exactly why one in five being caught is compatible with four in five still practising.

Entering carried its own leverage. Signing the contract generally caused pending legal, employment or family sanctions to be deferred, for as long as the terms were met.

Now compare that to the ordinary experience: thirty days somewhere, a discharge plan, a weekly outpatient appointment for a while, and a phone number. Same person, same illness, radically different structure.

The contingency piece is not a moral point

It would be easy to read the licence condition as “doctors behave because they are threatened,” which sounds unattractive and misses what is happening.

There is a substantial evidence base on contingency management — attaching consistent, timely consequences and rewards to verified behaviour — as a treatment approach in its own right. Prendergast and colleagues (2006) pooled that literature in a meta-analysis and found it to be one of the better-supported behavioural approaches in the field.

And note what the physician data actually shows about consequences. Detected use rarely happened in the context of patient care: of 159 documented incidents across the cohort, 10 — 6% — occurred on duty or on call, with a single recorded episode of patient harm. The monitoring was catching people in their private lives, early, and responding before it reached their work. That is what a system designed to notice looks like, as opposed to one designed to punish.

The PHP arrangement is, in effect, contingency management with unusually high stakes and unusually reliable verification. What it demonstrates is not that fear works. It is that structure, verification and rapid response work — which is exactly what the contingency literature predicts.

Why almost nobody else gets this

If this model produces the best outcomes in the field, the obvious question is why it is not simply how addiction is treated.

The answer is mundane. It is expensive, it lasts for years, and nothing pays for it. Insurance reimburses acute episodes, not five years of case management and random testing. And most people have no equivalent of a medical licence — no external body with both the standing and the motivation to require it and check.

Physicians get this care because a regulator insists on it and a profession funds it. Everyone else gets thirty days.

What this means if you are paying privately

Here is the part that is genuinely useful, and it is also the argument for what we do, so weigh it accordingly.

If you are paying for care yourself, you are not constrained by what insurance reimburses. That means you can buy the parts of the PHP model that matter — and they are not the parts that show up in a brochure.

Duration over luxury. Two years of light-touch case management is probably a better purchase than an additional two weeks somewhere beautiful.

Active management over availability. Somebody whose job is to notice, proactively, on a schedule.

Verification, if the person will agree to it. Testing is not punitive when it is agreed in advance as part of a structure, and when a positive result brings more support rather than a penalty. Many people in recovery describe it as a relief — it externalises the question and removes the daily negotiation.

Speed of response. Whatever arrangement you make, ask how fast someone reacts to a warning sign. Days, not months.

That is essentially what case management is: a privately-funded version of the structure that produces the best documented outcomes in the field, for someone who does not have a licensing board to provide it.

The honest caveats

Physicians are not a random sample, and no one has run the trial that would settle how much of the effect is the population and how much is the programme. The comparison to the general treatment population is not a controlled one.

The contingency in a PHP is unusually powerful and cannot be fully replicated privately — a family’s agreement is not a licensing board.

These are observational findings, so they establish that the model is associated with strong outcomes, not that any one element causes them.

And the outcome measured is licensure and occupational status, which is not the same thing as recovery. A physician can be licensed and working and still be struggling.

None of that undermines the practical conclusion, which is modest: the structure most strongly associated with long-term recovery is long, monitored and actively managed, and almost nothing about it is the part of treatment that gets photographed.

References

• McLellan AT, Skipper GS, Campbell M, DuPont RL (2008). Five year outcomes in a cohort study of physicians treated for substance use disorders in the United States. BMJ, 337, a2038. 10.1136/bmj.a2038
• DuPont RL, McLellan AT, White WL, Merlo LJ, Gold MS (2009). How are addicted physicians treated? A national survey of Physician Health Programs. Journal of Substance Abuse Treatment, 37, 1-7. 10.1016/j.jsat.2009.03.010
• DuPont RL, McLellan AT, Carr G, Gendel M, Skipper GE (2009). Setting the Standard for Recovery: Physicians’ Health Programs. Journal of Medical Regulation, 95. 10.30770/2572-1852-95.4.10
• Prendergast M, Podus D, Finney J, Greenwell L, Roll J (2006). Contingency management for treatment of substance use disorders: a meta-analysis. Addiction, 101, 1546-1560. 10.1111/j.1360-0443.2006.01581.x