Highlights
– In a retrospective, population-based Medicare cohort (2015–2020), treatment selection across 12 cancer indications was associated with higher clinical benefit (NCCN Evidence Blocks) but not with provider billing margin.
– Provider billing margin per course varied widely (up to $12,692) but a $100 increase in margin did not raise the odds of a treatment being chosen (OR 0.97, 95% CI 0.91–1.03).
– Findings imply that policy and price changes that alter billing margins alone are unlikely to shift oncology treatment patterns among Medicare fee‑for‑service beneficiaries in the studied indications.
Background
Rising costs of cancer care and rapidly expanding therapeutic options have prompted scrutiny of how financial incentives shape clinical decision-making. In oncology, where many agents are expensive and physicians often bill for infused therapies (the buy‑and‑bill model), there is concern that higher provider margins could influence treatment selection independent of clinical benefit. Conversely, professional norms and guideline-driven care may limit the role of financial incentives. Quantifying whether real-world treatment choices align more closely with clinical benefit than with provider profit margins informs policy debates on drug pricing, reimbursement design, and value-based care.
Study design
This retrospective, population-based cohort study used fee-for-service Medicare claims from 2015–2020 to examine whether the provider billing margin influences oncology treatment selection. Investigators identified Medicare beneficiaries with incident cancers for 12 pre-specified indications in which multiple NCCN‑recommended treatment options varied in both clinical benefit and potential billing margin. For each patient, the analysis considered the set of NCCN-recommended options available at the diagnosis date.
Key variables:
- Provider billing margin: Defined at the patient-treatment level using Medicare reimbursement rates corresponding to the patient’s diagnosis date. This reflects the difference between allowed reimbursement and estimated acquisition/administration costs as constructed by the authors.
- Clinical benefit proxy: Rank ordering of treatment options using NCCN Evidence Blocks scores applicable on the diagnosis date. Evidence Blocks synthesize efficacy, safety, evidence quality, and affordability into a comparative assessment.
- Outcome: The specific NCCN-recommended cancer treatment that each patient received.
Analytic approach: A conditional logit model was used to model the probability that a patient received each treatment option within their choice set. To reduce confounding, inverse probability-of-treatment weights were applied at the patient-treatment level to control for measured patient and provider characteristics (for example, age, comorbidity, prior therapy, and provider specialty/volume). The analysis produced odds ratios describing associations between a unit change in billing margin (per $100) or clinical benefit ranking and the likelihood that a given treatment was selected.
Key findings
The analytic sample comprised 19,397 patients across 12 cancer indications. Provider billing margin for individual courses of treatment varied substantially, from $0 to $12,692.
Main results:
- Provider billing margin: There was no statistically significant association between higher provider billing margin and the probability a treatment was chosen. Specifically, a $100 increase in billing margin had an odds ratio (OR) of 0.97 for treatment selection (95% confidence interval [CI] 0.91 to 1.03), indicating no evidence that more profitable treatments were more likely to be used.
- Clinical benefit: Higher clinical benefit—as operationalised by the NCCN Evidence Blocks ranking—was strongly associated with greater treatment use (OR 1.62, 95% CI 1.15 to 2.29). In other words, treatments judged to have higher clinical value were more likely to be selected by clinicians.
The findings were consistent across the pooled set of indications examined. Secondary analyses and sensitivity checks reported by the authors (for example, alternative model specifications or stratified analyses) did not materially change the primary conclusions.
Interpretation of effect sizes
The null association between billing margin and selection suggests that within the Medicare fee-for-service context and for the included indications, variation in per-patient provider revenue from different therapy options did not drive clinician choice. The magnitude of association for clinical benefit indicates that guideline-concordant assessment of benefit remained a meaningful predictor of treatment selection.
Expert commentary and context
This study addresses a central tension in cancer care policy: whether financial incentives embedded in payment systems meaningfully distort clinical decisions. Strengths include the use of a large, population-based claims dataset, a constrained choice-set design (patients compared only among NCCN-recommended options), and a modeling strategy that accounts for patient- and provider-level confounders.
Several contextual points are important for interpretation:
- Scope and setting: The cohort was limited to Medicare fee‑for‑service beneficiaries and indications where multiple guideline-recommended options were available. Results may not generalize to commercially insured populations, younger patients, or settings with different payment arrangements (for example, capitated systems or institutions with different buy‑and‑bill economics).
- Measurement of clinical benefit: NCCN Evidence Blocks provide a pragmatic, clinical‑facing summary of benefit, but they remain an imperfect proxy. Evidence Blocks incorporate expert judgment and multifactorial inputs; they are useful for comparative ranking but do not replace randomized comparative efficacy data.
- Billing margin construct: The authors estimated provider margin using Medicare reimbursement rates and assumptions about acquisition/administration costs. While necessary for large-scale analysis, such estimates are a proxy for real economic incentives and may not capture all contractual or institutional arrangements that affect margins at the practice level.
- Residual confounding and unmeasured factors: Despite statistical adjustment, unmeasured clinical considerations (e.g., subtle differences in patient frailty, physician judgement related to toxicities, patient preferences) could influence selection. If such factors correlate with margin estimates, they could bias results toward the null.
Taken together, the study provides evidence that, at least in the examined Medicare context, oncologists’ choices correlate more strongly with clinical benefit than with per-patient billing margin. This suggests that quality-driven practice and guideline adherence exert a detectable influence over treatment selection.
Limitations
- Observational design cannot prove causality. The null association between margin and choice may reflect residual confounding, measurement error in margins, or decision-making factors not captured in claims.
- Indication selection: The study evaluated 12 indications; findings may differ for other tumor types, lines of therapy, or when treatment options include largely palliative versus curative intent strategies.
- Excluded sectors: Patterns in private insurance, integrated delivery systems, or international health systems may differ when reimbursement and ownership structures vary.
- Potential changes over time: The study spans 2015–2020; subsequent shifts in oncology practice, drug pricing, biosimilar uptake, or payment policy could alter incentive structures.
Policy and clinical implications
From a policy perspective, these findings suggest that modifying provider margins through pricing or reimbursement adjustments may not be sufficient alone to change prescribing behavior for the studied indications in Medicare. Policies aimed at improving value should therefore remain multi-faceted: aligning evidence-based guidelines with measurement and feedback, investing in clinician decision support, coupling price reforms with quality incentives, and addressing patient-level factors such as affordability and access.
For clinicians and health system leaders, the study offers reassurance that guideline-based clinical benefit assessments appear to be influential in real-world oncology practice. Nonetheless, continuous monitoring is needed when payment reforms are implemented to ensure they do not have unintended consequences, particularly in settings where financial incentives for clinicians differ.
Conclusion
In this large, population-based analysis of Medicare beneficiaries, oncologists preferentially selected treatments with higher estimated clinical benefit, and there was no association between per-patient provider billing margin and treatment choice. While limited by observational design and by proxy measures of margin and benefit, the study suggests that altering billing margins alone is unlikely to markedly shift treatment patterns in the examined Medicare settings. Policy efforts to improve value in oncology should therefore consider multifactorial strategies that preserve guideline-concordant care while addressing affordability and system-level incentives.
Funding and registration
See the original publication for detailed funding disclosures and declarations. This study used administrative Medicare claims data and was not a clinical trial; no ClinicalTrials.gov registration applies.
References
1. Mitchell AP, Dusetzina SB, Mishra Meza A, et al. Provider billing margin and cancer treatment selection: population based cohort study. BMJ. 2025 Nov 5;391:e084729. doi: 10.1136/bmj-2025-084729.
2. National Comprehensive Cancer Network. NCCN Evidence Blocks. https://www.nccn.org/evidenceblocks (accessed 2025).
3. Centers for Medicare & Medicaid Services. Physician Fee Schedule. https://www.cms.gov/medicare/physician-fee-schedule (accessed 2025).
