When Every Scan Is Free: The Tragedy of the Diagnostic Commons

Overuse in medicine is obvious to anyone who is paying attention. The overuse that irks me the most is the use of diagnostic testing in situations where it is extremely unlikely to change care. These tests are costly, expose patients to unnecessary harm, and frequently present clinicians with Decontextualized Risk Information. Today, Drs. Mitchell and Daniel Belkin approach the problem by changing incentives. I don’t agree with everything here, but I certainly appreciate the effort, and the piece did get me thinking.

Adam Cifu

Clinicians make diagnostic decisions in an environment where additional testing appears free at the moment it is ordered. In reality, radiology capacity, including scanner time, radiologist attention, and downstream follow-up, is finite. However, the ordering clinician experiences no marginal cost for another study. Here, we propose an internal market mechanism, Rad Dollars, to introduce a real-time signal of scarcity without restricting access to medically necessary care.

The Problem of Zero-Marginal-Cost Decisions

In an emergency department or inpatient service, ordering occurs under three compounding pressures: diagnostic uncertainty, throughput expectations, and litigation risk. Because ordering another study appears costless to the clinician, the rational strategy is to add information even if the expected value is low.

Existing tools, such as electronic medical record decision-support prompts and utilization review, interrupt ordering but do not change the underlying incentives.

The Hidden Cost: Attention Dilution and Incidentalomas

The financial costs of low-value imaging are well-documented. Less appreciated is the capacity cost. Radiology departments operate under hard constraints: a finite number of scanners, a finite number of radiologist hours, and a finite amount of cognitive attention per shift. Every low-yield study that enters the queue consumes a share of that capacity.

Whole-body trauma CT provides a poignant example of the downstream burden of broad imaging. In hemodynamically stable patients with low-risk mechanisms, whole-body trauma CT increasingly substitutes for the physical exam, despite limited evidence that this strategy improves outcomes as compared to targeted imaging. The result is the identification of substantially more incidental findings, necessitating additional work-up.

It is possible to decrease the use of radiological services without harming patients. In the ICU, switching from routine daily chest X-rays to an “on-demand” strategy reduced imaging by roughly one-third without affecting mortality, length of stay, or duration of mechanical ventilation. Adoption of clinical decision-making for imaging after ankle, neck, and head injury has also reduced the need for imaging.

Not Unique to Medicine

This dynamic is not unique to healthcare. When a resource appears free at the point of use, it becomes predictably overconsumed. Other industries have addressed similar problems by making opportunity costs visible.

Carbon Credits and Cap-and-Trade:
For decades, the atmosphere served as a zero-marginal-cost sink for emissions, resulting in large negative externalities. Cap-and-trade systems addressed this by setting an overall emissions cap and allocating or auctioning a limited number of permits. Firms may emit only to the extent they hold allowances, and because permits are tradable, emissions acquire a market price.

Feeding America and the Choice System:

A similar dynamic emerged in the Feeding America food bank network. Before 2005, donated food was allocated through a centralized queuing system in which all shipments were treated as roughly equivalent, and food banks were penalized for declining loads. The result was predictable misallocation: some regions accumulated surplus while others lacked basic items. Feeding America replaced this system with an internal currency, “shares,” allocated based on local need, and allowed food banks to bid for food in regular auctions. Prices revealed which items were most valued in real time, improving matching and reducing waste despite initial ideological resistance.

The Rad Dollar Mechanism and Expected Effects

We propose an individual-level resource budget of Rad Dollars to manage access to finite diagnostic capacity. Each ordering physician will receive a monthly allocation of Rad Dollars adjusted for their expected utilization based on their personal historical case mix and patient acuity. When ordering a specific exam through an electronic health record, the clinician would be spending their allotted Rad Dollars to purchase each individual study. If a clinician exhausts their allotment of Rad Dollars, they can purchase additional Rad Dollars with personal funds at a transparent USD-to-Rad Dollar exchange rate.

Rad Dollar prices should reflect real-time demand on capacity. A study ordered when scanners are idle, and many radiologists are at work, imposes far less opportunity cost than the same study ordered overnight when a single resident is on call. Rad Dollar prices would therefore fluctuate with a radiology department’s burden.

This dynamic pricing shifts discretionary, lower-acuity imaging toward times when it does not compete with emergent studies for the same finite attention, just as congestion pricing shifts low-value car trips away from rush hour without preventing anyone from driving.

The expected effect of Rad Dollar implementation would be to encourage clinicians to reserve imaging for situations where results are likely to change management, maximizing diagnostic yield per study. A physician who manages their allocation efficiently and ends the month with a surplus could potentially convert unused Rad Dollars back into compensation, creating a reward for disciplined ordering rather than only a penalty for excess.

Potential Objections

What about stroke and trauma patients who present to a physician who has run out of Rad Dollars? There are no exemptions to Rad Dollars for high acuity patients. A physician who needs to image a stroke patient would spend their Rad Dollars without hesitation because the clinical value is self-evident. If they have run out of Rad Dollars, they would, without hesitation, pay a nominal fee for patient care. This is not a limitation of the system. It is evidence that the system is functioning as designed.

Making physicians pay their own personal money for patient care is punitive. The conversion rate from USD to Rad Dollars need not be punitive. It just needs to be non-zero. The health economics literature on copays is clear: even trivially small out-of-pocket costs produce meaningful reductions in utilization. A physician paying five dollars for an additional chest radiograph or fifty dollars for an additional CT is not bearing a significant financial burden. They are losing the ability to order without feeling any marginal cost.

Radiologist misaligned incentives. Radiologists are often incentivized to maximize RVUs, which are generated per study read. A system that reduces scan volumes would decrease radiology productivity metrics. Any real-world implementation would likely face resistance from radiology departments unless radiologists are brought in as explicit partners in the allocation-setting process. This could require restructured RVUs to account for other patient care commitments such as tumor board participation, clinician consultations, and protocolling imaging studies.

Conclusion

Tradeoffs always exist. In medicine, the only question is whether we see them or ignore them. Every imaging order has an opportunity cost: radiology capacity that could have been used for another potentially more urgent patient. The current system does not eliminate that tradeoff; it only hides who bears it.

Rad Dollars do not introduce scarcity into radiology. They introduce visibility. Every ordering physician would, for the first time, encounter the question that every other allocator of finite resources must answer: Is this use worth what it costs? The fact that this question has never been asked at the point of ordering is not evidence that it need not be. It is evidence of how long the problem has been ignored.

The authors thank Alex Tabarrok and Arjun Byju for their helpful comments on an earlier draft.

Mitchell Belkin and Daniel Belkin are diagnostic radiology residents in Baltimore, Maryland. They co-host the External Medicine Podcast.