Why your doctor gets paid for procedures, not outcomes

There is a question almost nobody asks at a medical appointment, and it is not a rude one: how does this visit turn into money, and for whom?

In most of American healthcare the answer is fee-for-service. The system pays by the item. A visit is an item. A test is an item. A procedure is an item. Getting better is not an item, and it does not appear on the bill.

That is not a conspiracy. It is an accounting choice, made a long time ago, for reasons that made sense at the time. But accounting choices have consequences, and there is real research on what they are.

This is not medical advice, and it is emphatically not a reason to stop or change a medication. Nothing on this page tells you anything about your specific clinician. If you are in crisis, call or text 988 (US), 24/7, free.

What "fee-for-service" actually means

Three ways to pay a doctor, roughly:

• Fee-for-service — a payment per item of care delivered.
• Capitation — a fixed payment per patient per period, whatever happens.
• Salary — a fixed payment per unit of time.

Each one rewards something different. Fee-for-service rewards doing more things. Capitation rewards doing fewer things. Salary rewards neither, and rewards nothing in particular.

None of those is the good one. They are three different sets of pressures, and the honest version of this piece is not "fee-for-service is corrupt" — it is that whichever system you are in is pushing on the room, and you are the only person in it who is not being paid.

What the research found when payment changed

The relevant synthesis is a Cochrane review of payment systems for primary care physicians. Comparing fee-for-service against capitation and salary, it found that fee-for-service produced more visits and more diagnostic and curative services than the alternatives [2].

And then it found something that complicates the story, which is why it is worth citing rather than paraphrasing: fee-for-service also produced fewer hospital referrals and fewer repeat prescriptions [2].

Read that twice. The payment model did not simply turn the dial up on everything. It moved care toward the things the paid clinician does themselves and away from the things that hand the patient to someone else. That is a more specific and more interesting finding than "doctors do more when paid more," and it is the sort of nuance that gets flattened out of this argument every time it is made on the internet.

The limits of that review, stated plainly. It rests on only four studies, and its searches ended in 1997 [2]. It is the best synthesis of its kind and it is thin and it is old. Anyone who tells you the payment-model literature is settled has not read it. We cite it because it is what exists.

The number people quote, and what it actually says

In 2012 Donald Berwick — later the administrator of the Centers for Medicare & Medicaid Services — and Andrew Hackbarth published an estimate of waste in US healthcare in JAMA. Their estimate for overtreatment alone was between $158 billion and $226 billion in 2011 [1].

The dollar figure is the part that travels. The definition is the part that matters. Berwick and Hackbarth defined overtreatment as care driven by providers' preferences rather than those of informed patients [1].

Not fraud. Not incompetence. Not greed. Preference, in the absence of an informed patient.

That definition is unusually generous to clinicians and unusually demanding of the system. It says the waste is not mostly people doing wrong things on purpose. It is mostly what happens by default when the person receiving the care does not have enough information to have a preference of their own.

Which means the fix implied by the largest waste estimate in American healthcare is not a new payment model. It is a better-informed patient.

What this looks like in mental health

Two things are documented well enough to say out loud.

The billing unit fits some care better than others. A payment system built on discrete encounters accommodates a fifteen-minute medication review more neatly than it accommodates an hour of talking. We are not going to dress that observation up as a research finding — we have not found a clean causal study of it and we are not going to invent one. But it follows from how the unit of payment is defined, and it is worth having in mind when the length of an appointment surprises you.

The workforce grew in a very particular shape. Federal wage-and-salary employment figures show psychiatrists close to flat over a quarter-century: 17,870 in May 1999 and 27,980 in May 2025, oscillating in a narrow band for most of the years in between [4]. Over the same window, counselors, nurse practitioners, marriage and family therapists and clinical social workers all grew steeply — the counselor category alone runs to 491,930 in May 2025 [4].

Two cautions on those numbers, both material. They count wage-and-salary jobs, not people — the survey structurally excludes the self-employed, which is where a large share of psychiatrists and therapists actually sit, so the levels are roughly half the practitioner headcount for psychiatry. And the Bureau of Labor Statistics does not encourage using this survey as a time series at all, because of classification changes and a three-year pooled sample design [4]. We show the shape because the shape is informative. We are not telling you it is a headcount.

And the access gap did not close. HRSA's quarterly shortage-area reports put the share of designated mental-health need being met between 26.5% and 28.1% in every year from 2019 through 2026 — essentially flat, across a period when almost every behavioral-health occupation expanded [3]. For contrast, in the most recent quarter: mental health 26.53% of need met, primary medical 47.43%, dental 33.60% [3].

More clinicians, same gap. Whatever the system is optimising for, it does not appear to be closing that.

What this does not mean

Being precise here is the whole job.

• It does not mean your doctor is motivated by money. Population-level effects of payment systems say nothing about any individual's reasoning, and the research does not claim otherwise.
• It does not mean you were overtreated. Overtreatment is an aggregate estimate with a wide band. It is not a diagnosis of your care.
• It does not mean treatment is a scam. The same system that has these incentives also produces genuinely effective care, every day.
• It is absolutely not a reason to stop a medication. Stopping psychiatric medication abruptly can be dangerous, and several classes carry documented withdrawal or relapse risks. If anything on this page makes you want to change something, that is a conversation to have with your prescriber, with a taper plan and check-in dates agreed in advance — never a decision to make alone after reading a web page. We covered the stopping evidence in full, including withdrawal as distinct from relapse, at stopping antidepressants: withdrawal and relapse.

What to ask your prescriber

Each of these comes from a specific finding above, and none of them is confrontational:

• What are the alternatives to what you are recommending, including doing nothing for now?
• How will we know whether this is working, and by when?
• If it does not work, what is the next step — and what is the step after that?
• Is there a reason to prefer this option over one that would involve fewer appointments?
• What would you want to be true before we consider stopping?

The last one is the most useful question in this entire piece. It converts an open-ended prescription into a plan with an exit, and it is the single clearest way to become the "informed patient" whose absence Berwick and Hackbarth were measuring.

Sources

1. Berwick DM, Hackbarth AD. Eliminating waste in US health care. JAMA, 2012. doi:10.1001/jama.2012.362 — overtreatment estimated at $158–226 billion for 2011; overtreatment defined as care driven by provider preference rather than informed patient preference.
2. Gosden T, Forland F, Kristiansen IS, et al. Capitation, salary, fee-for-service and mixed systems of payment: effects on the behaviour of primary care physicians. Cochrane Database of Systematic Reviews, 2000. doi:10.1002/14651858.CD002215 — four included studies; searches ended 1997. Fee-for-service associated with more visits and more diagnostic and curative services, and with fewer hospital referrals and fewer repeat prescriptions, than capitation or salary.
3. Health Resources and Services Administration, Bureau of Health Workforce. Designated Health Professional Shortage Areas Quarterly Summary. Mental-health "percent of need met" for the quarters ending 2019-06-30 through 2026-06-30: 27.27, 26.86, 28.09, 27.66, 27.19, 26.55, 26.94, 26.53. Comparison figures for the quarter ending 2026-06-30: primary medical 47.43%, dental 33.60%. HRSA publishes only the current quarter; the multi-year run was reconstructed from Internet Archive snapshots of the same report URL. data.hrsa.gov — accessed 2026-08-25.
4. US Bureau of Labor Statistics. Occupational Employment and Wage Statistics, national cross-industry estimates, May 1999 – May 2025. Psychiatrists 17,870 (May 1999, SOC 29-1066) and 27,980 (May 2025, SOC 29-1223); substance abuse, behavioral disorder and mental health counselors 491,930 (May 2025, SOC 21-1018). BLS states it "does not encourage the use of OEWS data for time-series analysis," and the survey excludes the self-employed. bls.gov/oes — accessed 2026-08-25.

Related reading

• What FDA approval actually means — and what it doesn't
• Where diagnoses come from
• How long were these drugs actually tested?
• Who funds psychiatric drug research?