How Much Does Low-Value Healthcare Spending Cost, and Can Clinician-Level Measurement Reduce It?

Direct Answer: The Cost Is Large, but No Single Number Is Definitive

U.S. healthcare spending has reached a scale at which even small inefficiencies matter. The Centers for Medicare & Medicaid Services reported national health expenditures of about $4.9 trillion in 2023, equal to approximately 17.6% of gross domestic product and roughly $14,570 per person. Estimates of spending on low-value care commonly fall between roughly $400 billion and more than $1 trillion per year, but that range is not a literal estimate of money that could be removed next year. The range combines studies using different definitions, time periods, populations, and counterfactual assumptions about what care would have happened if clinicians had selected differently.

Also worth reading: How do organizations accurately calculate AI benefits consulting ROI measurement in healthcare settings? · How can organizations effectively approach optimizing healthcare spending 2026 to combat rising costs? · What Are the Biggest Healthcare AI Privacy Risks and How Can Health Organizations Reduce Them?

Low-value care includes services with little or no net health benefit for a particular patient, as well as potentially beneficial services delivered at excessive cost, in the wrong setting, or at the wrong time. Examples can include unnecessary imaging for uncomplicated low-back pain, early stress testing in selected asymptomatic patients, avoidable emergency-department visits, high-cost imaging that does not change management, and treatments whose harms outweigh benefits. Savings estimates also depend on whether the analysis compares spending with a clinical alternative or removes spending altogether.

Clinician-level measurement can identify patterns and improve targeting, but it cannot solve every source of waste. It works best when paired with clinical judgment, shared decision-making, workflow redesign, payment reform, and operational improvements. A dashboard alone may produce “cost awareness” without changing behavior, while poorly designed incentives can encourage undertreatment. The practical objective is therefore not simply to minimize spending; it is to reduce spending on care that does not improve outcomes or patient experience.

How Healthcare Spending Is Measured

Healthcare spend measurement begins with gross national health expenditures, which divide spending into categories such as hospital care, physician and clinical services, prescription drugs, nursing care, dental services, home health, and administration. National totals provide a consistent macroeconomic view but are too aggregated to tell a hospital or physician why spending increased. Analysts therefore use claims, encounter records, cost reports, insurance payments, and patient-level clinical data to attribute costs to diagnoses, procedures, clinicians, facilities, and episodes of care.

Claims data remain a common starting point because they are widely available and include diagnosis and procedure codes. They are not a complete record of the patient’s condition: coding practices vary, out-of-network costs may be missing, and a clinician cannot always be identified precisely because services are attributed to a group or facility. Claims-based measures may also fail to capture whether a service was appropriate in context. An emergency visit may appear unnecessary until the patient’s symptoms and access alternatives are considered.

Clinical, operational, and patient-reported data add necessary context. A rigorous system might combine allowed amounts and paid claims with medication orders, lab results, referrals, prior authorization, appointment access, discharge information, and standardized patient-reported outcomes. In 2023, approximately 92% of Americans had health coverage for some or all of the year, so multiple payers contribute to the spending picture. A measure that works in a commercial plan may not behave identically in Medicare or Medicaid, making payer mix and benefit design important variables.

An economic principle worth separating is potential savings from realizable savings. If an unnecessary $1 billion service category is identified, the organization should not assume that it can recover the full $1 billion. A portion may represent fixed costs, some patients may receive alternative services, and clinicians may not have the time or authority to act. Valid business cases use historical spending, response rates, implementation expense, time to benefit, and confidence in causal effect.

Why Low-Value Care Persists Despite Known Problems

Low-value care persists because the healthcare system rewards volume in ways that are imperfect but measurable. Fee-for-service payments often pay more when more procedures are performed, while many organizations still organize around departments, locations, and individual transactions rather than patient outcomes. Hospitals incur costs before a clinician decides whether a service is necessary, creating a structural conflict between stewardship and financial performance.

Patients also face legitimate reasons for seeking or receiving additional care. Fear of missing a serious diagnosis, limited health literacy, transportation problems, fragmented records, and the expectation that more testing equals better care can all increase utilization. Clinicians may lack evidence at the point of decision, experience alert fatigue, or treat under uncertainty. In that context, ordering an unnecessary test is sometimes a low-cost way to reduce perceived diagnostic risk, even though the cumulative system effect is expensive.

The cost of low-value care is not limited to the procedure’s price. Unnecessary imaging can lead to incidental findings, repeat testing, anxiety, and cascades of intervention. A visit may be inexpensive but still expose a patient to harm, opportunity cost, or later treatment. Conversely, an expensive service can be high-value when it prevents hospitalization, disability, or a costly complication. Measurement must therefore consider outcomes and harm, not just line-item cost.

What Clinician-Level Measurement Can and Cannot Do

Clinician-level measurement can bring spending patterns closer to the decisions that can change them. A dashboard can show a clinician’s imaging rate for uncomplicated low-back pain, use of broad-spectrum antibiotics without a qualifying diagnosis, or spending on a selected episode of care. Comparisons should be risk-adjusted and use a sufficiently large patient panel; without those safeguards, legitimate differences in case mix can look like inefficiency. For example, a clinician treating older, medically complex patients will appropriately use more services than one treating a healthier population.

The strongest dashboards display actionable measures rather than every available statistic. Measures should be evidence-linked, measurable, understandable, modifiable, and connected to a decision. A primary care panel might review avoidable imaging before ordering, while a cardiology group might examine the appropriateness and timing of diagnostic tests. Giving clinicians a peer benchmark and asking them to review outliers can promote reflection, but ranking individuals without reviewing clinical context is likely to damage trust.

Measurement can also identify organizational bottlenecks. A low adherence rate may reflect not only clinician preference but also insufficient appointment supply, an unclear referral pathway, or a hospital outside the clinician’s control. Conversely, a clinician with a modest dashboard score may have limited influence over facility fees or inpatient utilization. Responsible measurement separates controllable decisions from system-level factors and avoids presenting all spending variation as individual waste.

The central risk is that cost reduction becomes a substitute for quality. A medical group that postpones diagnostic work, reduces referrals, or avoids high-risk patients can appear efficient while producing worse outcomes. Measurement should therefore pair cost with clinical outcomes, access, patient experience, equity, and adverse events. Savings claimed without these balancing measures should be treated cautiously.

Practical Steps for Building a Useful Measurement Program

A first step is to define the decision and patient population precisely. “Reduce healthcare spending” is too broad; a workable initial target might be selected MRI episodes for patients with uncomplicated acute low-back pain, excluding patients with trauma, neurological deficits, cancer, immunosuppression, or other defined exceptions. The baseline should state the number of eligible patients, the total allowed spending, the outcome measures, and the period from which data were drawn. A one-year pre-intervention year and at least one post-intervention quarter are useful starting points, although longer measurement is preferable where clinical effects are delayed.

The second step is to select claims or encounters for baseline analysis, then validate them against chart review. An apparent outlier may reflect a coding error, a different site of service, or a clinically necessary exception. The third step is to co-design the intervention with frontline clinicians, pharmacists, nurses, utilization-management staff, and patients. An evidence reminder alone may help with clear-cut decisions, but ordering pathways, decision support, referral standards, and follow-up capacity may be necessary for complex decisions.

The fourth step is to compare results with a credible baseline or control group and report confidence intervals where possible. A 12% reduction in a $10 million category saves $1.2 million in gross terms, but the net result must subtract program expense and account for displaced services. A pragmatic pilot may be appropriate when evidence is strong and risk is low; a wider deployment should wait when the relationship between spending and outcomes is uncertain. Independent evaluation is valuable when the sponsoring organization controls both the intervention and the reporting.

Finally, decisions should be revisited quarterly during a 12-month pilot and annually afterward. Rapid improvement is not automatically durable because staffing, payer contracts, and clinical guidance may change. Organizations should publish the numerator, denominator, inclusion rules, missing-data rate, and balance measures. Transparency makes it easier to distinguish a real clinical improvement from a temporary coding change or patient-selection effect.

Comparison: Measurement, Payment Reform, and Operational Improvement

These approaches are alternatives only in a limited sense. A primary care organization might use a dashboard to identify unnecessary imaging, redesign the ordering pathway, and then negotiate a quality-linked payment arrangement that supports the change. Conversely, a capitated model without operational capacity can leave a clinician responsible for costs they cannot control. The intervention should match the mechanism producing the spending, rather than selecting a dashboard simply because it is easy to launch.

Pricing, Benefits, and Return on Investment

Pricing varies by the scope of the work. A limited educational analysis using de-identified claims may be inexpensive, while a full economic evaluation of a hospital system, multi-payer physician group, or clinical intervention can require licensed data, software, statisticians, clinical reviewers, and implementation funding. A consultant’s fee should not be confused with the healthcare savings created by the program. A vendor may price a diagnostic assessment, dashboard subscription, implementation package, and ongoing monitoring separately, with additional costs for data integration, security review, and outcome validation.

The return on investment should be calculated conservatively. A simple formula is: net benefit equals attributable savings plus avoided future costs, minus software, labor, training, integration, and management costs. Attributable savings should exclude fixed expenses that would remain even if volume declined. In many organizations, the first year focuses on data acquisition, workflow changes, and clinician education; financial returns may appear later, and some interventions may be justified primarily by patient safety or experience rather than immediate savings.

The scale makes even a small reduction meaningful, but percentages can be deceptive. A 20% reduction in a $1 million category produces $200,000, whereas a 2% reduction in a $2 billion category produces $40 million. Organizations should compare absolute dollars, implementation cost, and health outcomes rather than selecting the largest percentage. They should also test whether savings recur after incentives end.

Common Mistakes and the Conditions for Acting

A common mistake is treating all cost as waste. A higher-cost site may provide better access, shorter waits, or fewer complications. Another is comparing clinicians without adjusting for patient risk, social conditions, referral patterns, or facility fees. Several organizations also begin with dozens of measures instead of one bounded clinical decision, making the program too complex to sustain.

Avoid assuming that an AI-generated recommendation is correct. The tool can be useful for summarizing records, identifying duplicate tests, or suggesting evidence-linked questions, but it may rely on inaccurate coding, incomplete data, or models evaluated on a different population. A human review pathway is still needed for high-impact decisions, and performance should be tested across age, race, language, disability, and clinically relevant risk groups. Privacy, security, procurement, and change-management requirements apply as well.

Action is most appropriate when a pattern is frequent, the clinical evidence is reasonably clear, the target population is defined, and the organization can monitor outcomes. Organizations should not wait for perfect certainty when a low-risk change can remove clear waste, but they should pause broad rollout when savings are promised without evidence of clinical effect. The strongest program treats measurement as a feedback system, not a disciplinary system: diagnose, intervene, measure, review exceptions, and improve again.

What Success Looks Like by 2026 and Beyond

By September 2026, a successful healthcare spend measurement program is likely to be less about a single “cost per patient” score and more about a connected set of decisions, outcomes, and safeguards. Health systems are increasingly adopting agentic AI and advanced analytics, yet automation does not eliminate the need for accountable governance. The practical test is whether an organization can move from a population-level signal to a clinically appropriate action and then determine what happened.

A credible first year might establish a baseline, validate ten to twenty exception rules, pilot one or two high-volume pathways, and review results quarterly. The organization should report gross and net savings, eligible patient counts, utilization, quality outcomes, adverse events, patient experience, and subgroup effects. It should also identify cases where clinicians appropriately departed from the pathway, because those exceptions can reveal evidence gaps or operational problems.

The answer is therefore affirmative but conditional. Low-value care likely accounts for hundreds of billions of dollars annually within a U.S. health system spending roughly $4.9 trillion, yet available estimates should be presented as a range rather than a cashable savings target. Clinician-level measurement can reduce some waste when it is specific, risk-adjusted, paired with workflow changes, and balanced by quality controls. Used alone, it can merely explain spending; used responsibly, it can connect resource decisions to better outcomes and a more sustainable healthcare system.