Paying a Doctor the Same Amount Whether You Get Sick or Not
Under capitation a provider receives a fixed monthly payment per enrolled patient and delivers whatever care that patient needs. It reverses the incentive of fee for service and creates a different one.
Two Ways to Pay for Care
Under fee for service, a provider is paid for each visit, test, and procedure. More activity produces more revenue, and the provider bears no financial risk for how much care a patient needs.
Under capitation, the provider receives a fixed amount per enrolled patient per month, frequently abbreviated as a rate per member per month, and is responsible for delivering the agreed scope of care. The payment arrives whether the patient is seen ten times or never.
The reversal is complete. Fee for service makes a patient who needs nothing a lost opportunity. Capitation makes that patient the most profitable one on the panel.
The Incentives Each Creates
| Fee for Service | Capitation | |
|---|---|---|
| Revenue rises with | Volume of services | Number of patients enrolled |
| Incentive on unnecessary care | Encourages it | Discourages it |
| Incentive on necessary care | Encourages it | Discourages it |
| Incentive on prevention | Weak, prevention reduces revenue | Strong, prevention reduces cost |
| Who bears cost risk | The payer | The provider |
Neither payment model is aligned with good care on its own. One pays for doing things and the other pays for not doing them. Everything depends on the quality measurement bolted onto whichever is chosen.
The Prevention Argument
The strongest case for capitation is that it makes prevention profitable, which fee for service structurally cannot.
Under fee for service, a practice that successfully manages a patient diabetes so they never require hospitalisation has reduced its own future revenue. The savings accrue to the insurer.
Under capitation, that same practice keeps the difference. Investing in care coordination, medication adherence programmes, nutrition support, and home visits becomes a return generating activity rather than an uncompensated one.
This is why capitated organisations look operationally different: they employ care managers, run outreach programmes, and invest in identifying high risk patients before they deteriorate. Those activities have no billing code and would not exist under fee for service.
The Risk Selection Problem
The obvious hazard is that a provider paid a flat amount per patient prefers healthy patients, and the response is the same mechanism used in insurance generally.
Risk adjustment varies the payment according to the expected cost of each enrolled patient, based on demographics and documented conditions. A practice enrolling sicker patients receives more per member.
That solves the selection problem and imports a different one, since payment now depends on documented diagnoses, which the provider influences. The incentive shifts toward thorough documentation of conditions, which is legitimate when the conditions are real and treated, and is a revenue exercise when they are neither.
This is the same tension that arises in health plan risk adjustment generally, relocated one level down to the provider.
Degrees of Risk
Capitation is not binary, and the gradations matter enormously to whether a provider can survive it.
Primary care capitation covers only services the practice delivers itself, leaving specialist care, hospitalisation, and drugs with the payer. Risk is limited and manageable.
Global capitation covers the total cost of care for the patient, including hospitalisation. A single catastrophic case can exceed a small practice entire capitated revenue, which is why global arrangements require either substantial scale or reinsurance, typically through stop loss coverage above a threshold per patient.
Many arrangements sit between, using shared savings or shared risk structures where the provider participates in a portion of the difference between actual and expected cost rather than bearing all of it.
The Historical Failure
Capitation was widely adopted in the 1990s through managed care organisations and produced a serious backlash. Patients experienced denied referrals, restricted access to specialists, and short appointments, and attributed it correctly to a payment model rewarding less care.
The retreat that followed was substantial, and the lesson drawn was not that capitation is unworkable but that capitation without robust quality measurement and patient protection produces exactly what it pays for.
Current arrangements therefore attach quality measures, patient experience surveys, and minimum access standards, with a portion of payment contingent on them. Whether those measures are adequate to counterbalance the underlying incentive is the central open question, and the honest answer is that measurement of clinical quality remains imperfect enough that the question is not settled.
What It Requires to Work
Several conditions recur in arrangements that function. Sufficient panel size so that random variation in patient need averages out, since a small panel makes the payment a gamble. Accurate risk adjustment, without which the provider is penalised for treating sicker patients. Stop loss protection against catastrophic individual cases. Meaningful quality measurement tied to payment. And data infrastructure allowing the provider to see the total cost of care for its patients, including services delivered elsewhere, which many practices simply do not have.
The last item is why capitation has been adopted most successfully by large integrated organisations and least successfully by independent small practices.
The Bottom Line
Capitation reverses the incentive of fee for service, replacing a reward for volume with a reward for keeping patients healthy and, unavoidably, a reward for doing less. It makes prevention economically rational for the first time, which is a genuine achievement, and it requires risk adjustment, stop loss, scale, and quality measurement to avoid producing the access problems that discredited it once already. The choice between the two models is not a choice between a good incentive and a bad one. It is a choice about which distortion you would rather measure and manage.