Low-Value Care Costs $98–131B. What Actually Reduces It?
by Evan Brociner, Data Product Lead
TL;DR: A landmark review estimates that low-value care costs the U.S. $98–131 billion a year in 2026 dollars. “Low value” does not mean fraudulent or necessarily poorly delivered: it means care that is unlikely to help a patient in a particular situation. Because the right choice depends on clinical context, enforcement alone cannot solve it. The best evidence supports finding specific patterns, giving clinicians useful feedback, and involving patients when appropriate. For organizations financially accountable for the total cost of care, the practical starting point is reliable claims data paired with a structured, human-led review.
Low-value care is a test, treatment, or procedure whose likely benefit is small—or whose potential harms, cost, and inconvenience outweigh that benefit—for a particular patient. It is not a label for a bad clinician or an automatic verdict on a single claim. The same service can be appropriate for one patient and low value for another.
This is the idea behind Choosing Wisely, an ABIM Foundation initiative in which medical specialty societies identified commonly used tests and treatments whose necessity should be discussed with patients. Choosing Wisely is a starting point for asking whether a service is evidence-based, necessary, nonduplicative, and right for the individual patient. A 2021 systematic review of 131 articles found that interventions built on Choosing Wisely recommendations can change practice patterns and reduce low-value care. The interventions were more effective when they targeted clinicians and used more than one component. Publishing recommendations alone was less likely to change care, which is why the implementation matters.
The six domains of waste
The most widely cited framework for healthcare waste comes from Shrank, Rogstad, and Parekh in JAMA (2019), synthesizing decades of estimates. Their total was $760–935 billion a year, roughly a quarter of all U.S. healthcare spending. Expressed in 2026 dollars using general inflation since 2019 (about 30%), that is $980 billion to $1.2 trillion, split across six domains:
| Domain | 2019 estimate, expressed in 2026 $ |
|---|---|
| Administrative complexity | $343B |
| Pricing failure | $298–311B |
| Failure of care delivery | $132–214B |
| Overtreatment / low-value care | $98–131B |
| Fraud and abuse | $75–108B |
| Failure of care coordination | $35–101B |
These are 2019 estimates adjusted using cumulative CPI growth (~30%), not a new 2026 measurement of healthcare waste. The ranges exist because the underlying studies used different methods and data. Utilization, prices, and the underlying evidence may have changed.
Fraud is real and worth pursuing. But it is a different problem from low-value care. The estimates overlap, so they are not a precise ranking, but they put both problems in the same general range.
Low-value care can include imaging for uncomplicated low-back pain, antibiotics for viral infections, or repeat testing unlikely to change management. In these cases, the claim may be valid: there may be no billing or coverage violation to deny or recover. The question is whether the care was likely to help the patient.
Why low-value care happens
Low-value care rarely comes from one bad decision or one bad actor. It grows out of uncertainty, habit, patient expectations, fragmented information, time pressure, and incentives that can make doing more feel safer than doing less.
A clinician may order a test to rule out a serious condition, to meet a patient's expectation, or because a prior result is hard to find. A patient may reasonably want reassurance. And in a fee-for-service system, doing more is often easier to reward than taking the time to explain why less is appropriate. That is why the solution cannot be a blunt denial rule: it has to give clinicians and patients enough context to make a better decision in the moment.
Why low-value care is harder to move
The two levers people reach for first—prior authorization and cost-sharing—have important limits:
- Broad prior authorization requires a clinician to obtain approval before a plan pays for a service. It can be appropriate for clearly defined coverage rules, but it can also add delays and administrative work. A systematic review published online in 2025 in The American Journal of Medicine found that authorization requirements were associated with worse clinical outcomes in the included studies.
- Blunt cost-sharing means asking patients to pay more out of pocket, such as through a high-deductible plan. It can reduce use, but not necessarily the right use. In Brot-Goldberg and colleagues' study of one firm switching to a high-deductible plan (Quarterly Journal of Economics, 2017), spending fell 11.8–13.8% because people used fewer services across the board, with no evidence they shopped on price.
What the research says does work
Targeted feedback can change routine decisions
In a 2016 JAMA cluster-randomized trial, 248 clinicians across 47 primary-care practices received behavioral interventions to reduce unnecessary antibiotics for selected respiratory infections. Compared with control practices, peer-comparison emails reduced inappropriate prescribing by 5.2 percentage points, while requiring a written justification in the electronic record reduced it by 7.0 points. Put another way, for every 100 comparable visits, these interventions were associated with about five to seven fewer unnecessary antibiotic prescriptions than would otherwise have occurred. This was not a general reminder to prescribe less. It focused on a narrow behavior, showed clinicians how they compared with peers, and asked for a reason at the moment an antibiotic was ordered.
The pattern appears beyond antibiotics. A 2022 JAMA cluster-randomized trial sent individualized audit-and-feedback reports to Australian primary-care doctors who ordered musculoskeletal imaging frequently. Across 3,660 clinicians, doctors who received feedback ordered 27.7 targeted scans per 1,000 visits, compared with 30.4 per 1,000 in the control group over 12 months—about three fewer scans for every 1,000 visits. The setting was Australian primary care, so this is not a prediction of a U.S. payer program’s savings. It is evidence that specific feedback to high-ordering clinicians can modestly change ordering behavior at scale.
More recent U.S. trials point in the same direction. In a 2024 study of eight primary-care clinics, clinicians committed to Choosing Wisely recommendations for older adults, supported by patient materials and reminders. The intervention reduced low-value care across three common clinical situations and increased the deintensification of diabetes medications. In a separate 2024 trial across 19 Northern California hospitals, peer-comparison emails reduced above-guideline postoperative opioid prescribing by 5.8 percentage points; guideline-based feedback reduced it by 4.7 points.
Patient involvement can strengthen the intervention
The evidence for patient-level education is less U.S.-specific, but it is promising. A 2025 systematic review and meta-analysis of 140 primary-care randomized trials found that patient education alone was associated with a 16% relative reduction in the targeted low-value service. For example, a 16% relative reduction would move a rate of 10 services per 100 patients to about 8.4—not to zero. Combining patient and provider education was associated with a 30% reduction, and adding audit and feedback was associated with a 34% reduction. The authors rated these patient-inclusive estimates as low certainty, meaning future, better studies could find a meaningfully different effect. Still, the results suggest that patient-level education is worth including alongside clinician-focused work. The practical implication is straightforward: give clinicians a credible signal, then give care teams a way to have the right conversation with the patient.
Why some programs do not work
The counterexamples matter too. In one randomized study that introduced the program to different clinics at different times, combining clinician precommitment, reminders, patient handouts, and email resources did not reduce low-value orders overall, and produced only a small, temporary reduction for low-back pain.
Implementation research helps explain why. Clinicians need to see a comparison as fair, understand why it applies to their patients, and retain room to identify legitimate exceptions. Programs can stall when data feel invalid, feedback feels punitive or intrusive, teams lack time to act, or a plan is made but never followed through. Patient expectations, diagnostic uncertainty, fear of litigation, and fee-for-service incentives can reinforce the existing default. Research on barriers to stopping low-value care and how to put audit-and-feedback programs into practice point to the same practical lesson: a measurement is useful only when a team can understand it, trust it, and act on it.
What a workable program looks like
A workable program follows a deliberate loop:
- Choose a clinically defensible measure with a clear alternative, rather than a broad category of utilization.
- Identify cases carefully and review exceptions so the signal holds up to clinical scrutiny.
- Find meaningful variation—for example, clinicians whose rate differs from comparable peers or from their own baseline.
- Deliver nonpunitive, actionable feedback with a clear next step for the clinician, care team, and, when appropriate, patient.
- Track utilization and balancing measures: whether the targeted service falls, whether appropriate care or outcomes worsen, and whether the change lasts.
That is closer to a professional conversation and a well-run care-management process than to enforcement.
How Falcon thinks about it
Reducing low-value care means finding where potential overuse concentrates and then putting that comparison in front of the two people who can change it: the ordering clinician and the patient.
From there the work splits across two products.
Sentinel handles the provider side. It automatically finds providers with a meaningful pattern of low-value-care signals, then gives the team one clean place to review the claims, the relevant measure, and the clinical context. When outreach makes sense, it helps turn that review into a clear documentation request or compliance letter that the client can approve and send.
Helm is the view providers and care teams use to manage their population. It shows the attributed members with low-value-care signals alongside the rest of the quality, utilization, and care-gap work—so the team can see who may need a conversation, follow-up, or a care plan. That puts a practical intervention next to an ACO’s financial accountability.
If your organization is financially accountable for the total cost of care and has not looked at its low-value-care footprint, get in touch. We can show you the opportunity set in your population—from provider review in Sentinel to member follow-up in Helm.