A major them at the AMA is the severity of the gap between inflation and physician payments based in RVUs. This affects a lot of things that are paid outside of hospital budgets - for example, community imaging services and community pathology services.
I would argue this also impacts how AMA views innovation, such as artificial intelligence-based diagnostics in radiology and cardiology. One issue is whether automation will speed up physician work (and therefore shrink time-based reimbursement). Another issue is whether AI-based services will get expanded payments in a way that literally "shrinks" the available RVUs.
Here are some 2025 entry points.
- Medicare experts back tying physician payments to inflation. Here.
- AMA web resources on shrinking RVUs relative to inflation. Here.
- Medicare volume groowth; payments shift to non-physician practitioners. Christensen et al. here.
- Doctors slam 2026 pay cuts. Here.
- AMA slams two CMS proposals; August 2026, here.
- Kaiser Briefing: How CMS pays physicians. Here.
- Medpac 2025 re: Physician payments. Here.
- Why Medicare reimbursement keeps declining. Here.
- How Inflation impacts the Medicare physician fee schedule. (Amer Coll Radiol), here.
- Cigna downcoding here.
- Reimburesment in the age of AI radiology, Dogra et al. Here.
Based on the sources you provided regarding Medicare physician pay, there are significant implications for specialties like radiology and pathology, especially in the context of artificial intelligence (AI) development and integration. The current Medicare Physician Fee Schedule (MPFS) operates under a budget neutrality requirement, meaning any increase in payments for one area must be offset by decreases elsewhere, often through a reduction in the monetary conversion factor that translates Relative Value Units (RVUs) into payments.
This system has led to a substantial decline in real, inflation-adjusted physician payments over the past two decades. From 2001 to 2025, Medicare physician pay has effectively dropped by 33% when adjusted for practice cost inflation. The core of your concern lies in how AI could exacerbate this deflationary pressure on specialties like radiology and pathology through two primary mechanisms.
Concern 1: AI-Driven Efficiency Leading to RVU Reductions
Your first concern is that AI could make physician work more efficient, and this increased efficiency could lead to future reductions in the work RVUs assigned to various procedures. The sources strongly support the basis for this concern.
- Foundation of RVUs: The MPFS and its underlying Resource-Based Relative Value Scale (RBRVS) are designed to measure the resources necessary for a service, including physician time, technical skill, mental effort, and stress. The "work" component of RVUs is meant to reflect the relative levels of these inputs.
- CMS's Efficiency Adjustment: The Centers for Medicare & Medicaid Services (CMS) already operates on the assumption that physicians become more efficient over time with experience and improved technology. The proposed 2026 MPFS includes an "efficiency adjustment" that would cut work RVUs by 2.5% for over 7,000 services, based on the unproven assumption of greater efficiency and less time involved in each service. This proposal highlights CMS's willingness to reduce RVUs based on perceived efficiency gains, even without new data to support it.
- AI's Impact on Work: AI tools, particularly assistive or augmentative AI, could streamline workflows in radiology and pathology. If these tools reduce the time, mental effort, or technical skill required from a physician to interpret an image or a digital pathology slide, future surveys conducted by the American Medical Association/Specialty Society RVS Update Committee (RUC) could reflect this reduced physician "work". As a result, the work RVUs for these services would likely be revised downwards in subsequent fee schedules. This is a significant threat, as even a successful AI tool that improves care could inadvertently lead to lower reimbursement for the physicians using it.
Concern 2: Successful AI Reimbursement Squeezing Existing Payments
Your second concern is that if AI algorithms successfully obtain their own CPT codes and reimbursement, the new spending on these AI tools could trigger budget neutrality adjustments, leading to cuts in existing physician payments within those specialties.
- The Zero-Sum Game of Budget Neutrality: The MPFS is a zero-sum system. A provision in the Omnibus Budget Reconciliation Act of 1989 requires that any changes to the fee schedule that increase spending by more than $20 million must be offset by cuts elsewhere. These cuts are typically achieved by reducing the conversion factor, which lowers payments for all services on the fee schedule.
- AI Coding and Reimbursement Pathways: The sources detail the pathways for AI reimbursement, including new Category I and Category III CPT codes for "Software as a Medical Device" (SaMD). As AI tools get their own codes and valuations, the payments for these new services add to the total Medicare spending under the MPFS.
- Redistribution of Payments: If new AI services in radiology and pathology become widely adopted and billed, the resulting increase in overall Medicare spending will trigger these budget neutrality adjustments. This would lead to a reduction in the conversion factor, effectively taking money from existing physician services to pay for the new AI services.
- Historical Precedent: This is not a theoretical concern. In 2021, when CMS increased the RVUs for evaluation and management (E&M) services to better value primary care, it caused an offsetting budget neutrality adjustment to the conversion factor of -6.8% to -10.20%, which cut payments for all other services. Similarly, successful AI reimbursement could lead to a redistribution of payments away from physicians and toward the AI tools themselves, especially since software costs are difficult to characterize within the current practice expense methodology.
Implications for Future Pathology Coding Systems
These two concerns are highly relevant to discussions about new coding systems for pathology, particularly in the era of digital pathology and AI.
- Valuing Physician Work vs. Machine Work: As pathology digitizes, new coding systems will need to grapple with how to value the distinct contributions of the pathologist versus the AI algorithm. If a new coding system bundles the AI's contribution with the pathologist's work, there's a risk that the pathologist's work component will be devalued over time due to efficiency gains, as described in the first concern.
- Avoiding Cannibalization: If the system creates separate codes for AI analysis, it raises the second concern: these new codes could siphon funds from traditional pathology services due to budget neutrality. Stakeholders in pathology will need to carefully consider coding frameworks that recognize the value of AI without inadvertently triggering across-the-board cuts for pathologists. This may involve exploring alternative payment models that are not strictly tied to the fee-for-service, budget-neutral MPFS, such as value-based models that are better suited for situations where AI functions cannot be divided into discrete services.
In summary, the structure of Medicare reimbursement, particularly its budget neutrality requirement and its focus on physician work as a component of value, creates a precarious situation for specialties like radiology and pathology that are on the cusp of significant AI-driven transformation. The very success of AI, whether through improving physician efficiency or securing its own reimbursement, paradoxically threatens to accelerate the decline in real physician payment within the current system. These dynamics must be a central consideration in the development of future coding and reimbursement policies for these fields.
Here’s a detailed, integrated analysis of physician pay cuts and their intersection with AI, particularly for radiology and pathology, based on the uploaded articles and current policy dynamics:
1. Context: Structural Decline in Physician Payment
Medicare physician reimbursement has lost over 20% of its value (inflation-adjusted) since 2001, with another 4% drop between 2021 and 2023. This is largely due to budget neutrality rules in the Medicare Physician Fee Schedule (MPFS), which require offsetting any $20M+ increase in spending (e.g., new codes, E/M increases) with across-the-board cuts through the conversion factor. Unlike hospitals, skilled nursing facilities, and other Part A providers (which get inflationary updates), physician payment updates are frozen or capped at 0.25–0.75% starting in 2026 — well below medical inflation.
This is why radiologists and pathologists have seen a 24.9% decline in per-beneficiary reimbursement since 2005 (inflation-adjusted) even as productivity rose — a silent “deflationary” force that pressures practice sustainability.
2. The Two AI-Driven Risks for Radiology and Pathology
(A) RVU Compression Risk
-
CMS’s 2026 proposal includes a 2.5% “efficiency adjustment” cut to work RVUs for >7,000 services, justified by an assumption that clinicians are faster and more efficient — even though there is no new time-study data and some procedures are becoming more complex.
-
If AI tools demonstrably speed up image reads or slide reviews, future RUC time surveys could ratchet RVUs downward, embedding efficiency gains as permanent payment reductions — the same way E/M updates permanently lowered the conversion factor for all physicians in 2021–2022.
(B) Code Volume & Budget Neutrality Risk
-
Paradoxically, if AI generates new reimbursable codes (e.g., CPT codes for digital pathology AI triage, AI-detected incidental findings), this adds “new dollars” to the MPFS — which triggers budget neutrality and lowers the conversion factor for all physicians.
-
This is exactly what happened with the 2021 E/M code revaluation: primary care gained, but specialties like radiology and pathology saw net cuts to maintain budget neutrality.
In other words, radiology and pathology face a “heads I lose, tails I lose” scenario:
-
Efficiency from AI → fewer RVUs per case.
-
Growth of AI-billed services → lower conversion factor for every other service.
3. Strategic Implications for New Coding Systems in Pathology
Because of these dynamics, the profession has an incentive to be cautious about how it codes AI:
-
Bundling vs. Unbundling: New AI codes could be bundled into existing pathology services (no new dollars, thus no conversion-factor pressure) rather than stand-alone payable codes that expand spending.
-
Appropriate Valuation: If pathology societies seek new RVUs for AI-assisted services, they may need to simultaneously argue that these should not trigger downward adjustment to existing slide review RVUs (e.g., via an add-on code rather than a revaluation).
-
Policy Advocacy: Specialty societies (CAP, ACR) may push for:
-
MEI-linked annual updates (MedPAC proposal: MEI – 1%) to break the structural cycle of deflation.
-
Exemptions from budget neutrality for digital/AI transformation codes, analogous to how telehealth waivers were handled during COVID.
-
Guardrails against automatic “efficiency adjustments” unless backed by robust time data — to avoid preemptive payment erosion based on theoretical AI gains.
-
4. Broader Policy Considerations
This debate is already visible in AMA and MedPAC commentary:
-
The AMA warns that automatic efficiency cuts “are not grounded in new data” and could “encourage consolidation and reduce access”.
-
MedPAC supports re-basing RVUs with timely cost data, arguing current allocations use 2006 practice-expense data that may misvalue facility-based specialties.
-
Stakeholders fear that chronic underpayment plus AI-driven deflation could accelerate radiology/pathology mergers into large health systems — reducing independent practice and possibly raising system-wide costs.
5. Key Takeaways
-
AI magnifies the deflationary dynamics already baked into the MPFS. Without policy reform, any efficiency gains or new AI services risk translating into net pay cuts for radiologists and pathologists.
-
Strategic coding and advocacy are critical. Specialty societies should shape coding pathways that recognize AI’s value without automatically triggering conversion factor cuts.
-
Long-term fix required. Linking updates to MEI (even at MEI–1%) could stabilize physician payment and prevent AI from becoming a perverse driver of reimbursement decline.
