Two CY 2027 proposed rules are open at once, and for the first time one of them tries to answer a question practices have been asking since ambient documentation and autonomous coding arrived: how does Medicare pay for software that does clinical work?
In the CY 2027 Hospital OPPS/ASC proposed rule (CMS-1850-P), released July 2, 2026, CMS proposes creating a payment pathway for what it calls Software as a Medical Service (SaMS) — clinical software that applies an algorithm to patient data and produces a diagnosis, a risk score, or a treatment recommendation. CMS proposes a new status indicator “O1” and would designate 36 HCPCS codes as SaMS, moving 21 of them out of standard clinical APC groups and into New Technology APCs. In the companion CY 2027 Physician Fee Schedule proposed rule, CMS proposes renaming the existing “Software-as-a-Service (SaaS)” code family to SaMS for alignment, and paying for SaMS analyses performed on laboratory tests under the PFS rather than the Clinical Laboratory Fee Schedule.
The CY 2027 PFS proposed rule, issued July 14 and published in the Federal Register on July 16, sets two conversion factors:
| Track | CY 2026 | Proposed CY 2027 | Change |
|---|---|---|---|
| Qualifying APM participant | $33.5675 | $33.1693 | −1.19% |
| Non-qualifying APM participant | $33.4009 | $32.8409 | −1.68% |
| Anesthesia (qualifying APM) | — | $20.4165 | — |
The mechanics are statutory. Congress granted a one-year 2.5% conversion factor increase for CY 2026 only; it expires December 31, 2026, and the unwind is automatic. Everything discretionary in the rule pushes the number partway back up — and it still lands negative.
CMS proposes a practice expense stabilization adjustment capping annual PE RVU movement at ±5% for most existing services, while phasing out outdated specialty survey data over a two-year transition. Read the cap for what it is: a shock absorber, not a reprieve. Specialties whose published 2027 impact looks survivable may be seeing only the first five points of a larger multi-year repricing.
The OPPS comment window closes August 31 and the PFS window closes September 14. Most practices comment on neither. If your practice bills any of the 36 proposed SaMS codes, uses AI-assisted diagnostics, or performs office-based procedures affected by the practice expense re-basing, the comment record is the only lever left before the November final rules.
CMS proposes to sunset traditional MIPS beginning with the 2029 performance year, making MIPS Value Pathways the only reporting option for MIPS-eligible clinicians from CY 2029 forward. For 2027, CMS proposes three new MVPs — Diabetic Disease, Hypertension, and Hospitalist — and would modify every MVP to include MIPS core measures, requiring each clinician to report at least one measure considered fundamental to their specialty and patient population.
CMS also proposes a new incentive rewarding responsible clinician use of AI. To qualify, a clinician must maintain organizational policies governing how AI-enabled tools are evaluated, deployed, and monitored — or help create or pilot such tools. Risk-stratification models, clinical decision support, and predictive analytics for chronic disease management all count.
2027 is the year Medicare stops treating clinical software as an unpriced input and starts treating it as a billable service. The SaMS framework is narrow today — 36 codes, mostly hospital outpatient — but the direction is unambiguous, and the CY 2027 QPP AI incentive puts a governance requirement on the physician side of the same trend.
Prior authorization reform is now running on two separate tracks, and practices need a different playbook for each.
CMS-0057-F is live. As of January 1, 2026, impacted payers must decide standard prior authorization requests within 7 calendar days and expedited requests within 72 hours, and denial notices must state a patient-specific clinical reason rather than pointing at a policy number. CMS has now proposed extending that architecture to pharmacy with the 2026 CMS Interoperability Standards and Prior Authorization for Drugs proposed rule (CMS-0062-P), which would require impacted payers to support electronic prior authorization for drugs, decide within shortened timeframes aligned to CMS programs, and increase PA transparency. HHS is separately proposing to adopt HL7 FHIR standards for prior authorization transactions under HIPAA Administrative Simplification — a scope that would reach all HIPAA covered entities exchanging PA electronically, not just the payers named in CMS-0057-F.
Health plans covering roughly 270 million Americans signed AHIP’s multi-year commitments, and the first public progress reporting landed in 2026.
| Payer Action | Detail |
|---|---|
| UnitedHealthcare Gold Card | Reduced PA volume an average of 30% for eligible provider groups in 2025; national Gold Card code list updated effective April 1, 2026 |
| UnitedHealthcare standardization | More than 70% of UHC prior authorizations to move to a standardized submission process by the end of 2026 |
| Cigna | 96 codes removed May 31, 2025; 1,100+ total PA code removals since 2020, with roughly 500 more targeted in Medicare Advantage |
| Aetna | PA rolled back on cataract surgery, video EEG, and some home infusion drugs |
| Industry 2027 target | 80% of electronic PA approvals with complete clinical documentation answered in real time, with FHIR APIs across all markets |
Progress is real and uneven. Medicare Advantage prior authorization denial rates still diverge sharply by carrier, and appeal windows are not converging at all.
A single missed calendar difference converts a winnable appeal into a write-off. The 65-day UHC window is roughly one third of the commercial standard, and it is the deadline most often missed by practices running a single generic appeal queue.
Build a payer-specific appeal deadline matrix into your denial worklist this week, keyed to the 65 / 180 / 60-day split, and sort the queue so the shortest clock surfaces first. Then pull the current UHC Gold Card code list and the Cigna removal list and reconcile them against the PA-triggering rules in your practice management system. Practices are still requesting authorizations for services the payer stopped requiring — staff hours that produce nothing at all.
The 2026 adoption data tells a consistent story across every survey: buying AI is nearly universal, operationalizing it is not.
Read the last two figures together. The single highest-yield AI application in the revenue cycle is also the least deployed. Meanwhile 80% of health systems are exploring, piloting, or implementing generative AI for revenue cycle work — up from 58% two years earlier — and the industry pays $57.23 in administrative cost per denied claim, up from $43.84 the prior year, to rework claims that pre-submission scoring could have flagged.
The Wasteful and Inappropriate Service Reduction (WISeR) Model launched January 1, 2026 across six states — New Jersey, Ohio, Oklahoma, Texas, Arizona, and Washington — using AI and machine learning to screen prior authorization requests for services at higher risk of fraud, waste, and abuse. All denials require licensed clinician review, and decisions are due within 72 hours (48 for expedited requests).
Six months in, CMS ordered one of its own WISeR AI contractors to take corrective action. Providers in Arizona reported weeks-long delays overturning denials; in Ohio, the request submission portal was not operational for months after launch.
The WISeR corrective action is the most useful AI governance artifact of the year, and it came from the regulator. Whatever AI you deploy against claims, denials, or coding, instrument it the way CMS is now being forced to: measured turnaround time, measured overturn rate, a named human reviewer on every adverse action, and a documented escalation path for when the model is wrong. The CY 2027 QPP AI incentive will ask whether those policies exist.
CMS has posted the FY 2027 ICD-10-CM update: 238 new codes, 4 revised codes, and 21 deletions, effective for encounters on or after October 1, 2026 — 55 days from today. There is no grace period.
Changes to load into the encounter form and EHR problem list now:
Effective March 1, 2026, UnitedHealthcare expanded its Excludes1 editing policy to cover both outpatient and professional claim types. Combine that with a heavy FY 2027 instructional-note revision cycle and October 1 becomes a denial-spike risk, not just a code-load task. Run your top 100 diagnosis pairs against the FY 2027 Excludes1 changes before the effective date.
| Change | Detail | Effective |
|---|---|---|
| Maternity care restructuring | Moves away from the traditional global obstetric billing model toward a granular structure separately reporting antepartum care, labor management, delivery, and postpartum services | Jan 1, 2027 |
| 58 new Category III codes | Approved at the February 2026 CPT Editorial Panel meeting for the 2027 production cycle | Jan 1, 2027 |
| 28 Category III codes | Accepted at the September 2025 meeting — already billable | July 1, 2026 |
| Adaptive behavior analysis | Most significant ABA revision since 2019: six new codes, revisions to existing codes, and deletion of two temporary Category III codes | Jan 1, 2027 |
| 2027 CPT Professional edition | Publication expected in fall 2026 | Fall 2026 |
For OB practices, the maternity restructuring is a contract renegotiation event, not a coding update. Global-package fee schedules negotiated as a single obstetric bundle will need to be rebuilt component by component, and any payer contract that references the global codes by number needs to be reopened before January 1.
| Metric | 2026 Figure |
|---|---|
| Industry-wide initial denial rate | 11.8% (up from 10.2%) |
| Commercial payer first-pass denial rate | 13.9% |
| Medicare Advantage first-pass denial rate | 15.7% |
| Providers running denial rates at or above 10% | 41% (up from 30% in 2022) |
| Administrative cost per denied claim | $57.23 |
| Net patient revenue lost to denials, underpayments, and missed filing windows | 3–4% for roughly half of practices |
Kodiak Solutions’ benchmarking across more than 2,300 hospitals and 375,000 physicians found the median final denial rate rose from 2.5% to 2.7%, and that those organizations lost more than $48 billion in net revenue in 2025 to final denials and uncollected patient balances combined.
Practices are getting measurably better at collecting at the desk and losing ground on everything that leaves the building. Front-end financial engagement is working; back-end patient A/R is not.
The CAQH Index economics on eligibility verification remain the most lopsided numbers in the revenue cycle:
Spending on eligibility and benefits transactions has risen 60%, to $43 billion annually. And even though 96% of benefit verifications are now transacted electronically in aggregate, 23% of practices still name eligibility and benefits verification as a top front-end problem — because the electronic transaction returns a coverage answer, not a benefits answer. Staff fall back to the phone for the deductible, the copay tier, and the PA requirement.
| KPI | Industry Median | Target |
|---|---|---|
| Clean claim rate (first submission) | 90–94% | 95–98% |
| First-pass denial rate | 11.8% | Under 5% |
| Days in A/R | 40–45 | 30–35 |
| A/R over 90 days | 15–20% | Under 10% |
| Cost to collect | ~5% of collections | Under 4% |
| Point-of-service collection rate | 24.8% | Over 35% |
A practice billing $1.5 million a year at the 11.8% industry denial rate is carrying $50,000 to $120,000 in recoverable revenue it never collects. Moving first-pass yield out of the 78–82% range and into the low 90s does not require an AI platform. It requires an automated eligibility check with benefits-level detail at scheduling, a rules-based scrubber before submission, and a denial worklist sorted by dollar value rather than by date.
Now in its 28th year and drawing on tens of thousands of provider evaluations, the 2026 Best in KLAS report named winners across nine revenue cycle technology categories and fourteen RCM services categories.
| Category | Winner | Detail |
|---|---|---|
| RCM Suites | Waystar | Overall “A” — the highest grade in the report; cited for cohesive UI across acquired modules, bundled pricing that reduces nickel-and-diming, and high adoption of AI-generated appeals |
| End-to-End RCM Outsourcing | Ensemble | 96.9 points, outpacing R1 RCM by 42.9 points |
| Extended Business Office (<200 beds), Government Reimbursement, Underpayment Recovery | R1 RCM | Three Best in KLAS category wins |
| Revenue-cycle-adjacent technology | Abridge, Microsoft | Top spots for ambient documentation and CDI |
Black Book Research separately published its first industry-wide evaluation of AI-driven RCM solutions, and Becker’s listed 385+ RCM companies to know in 2026 — a vendor field that has grown faster than any practice’s ability to evaluate it.
PitchBook’s Q2 2026 analysis carries a blunt title: AI Kills the RCM Star. The thesis is that agentic AI compresses the labor arbitrage underpinning percentage-of-collections outsourcing contracts, and that private equity buyers of RCM service platforms are underwriting take rates that automation is actively eroding.
This is a contracting insight before it is a market insight. If your outsourced billing vendor’s underlying cost structure is falling because it automated eligibility, claim status, and payment posting, your percentage-of-collections rate is renegotiable at renewal. Ask what share of your account’s workflow is now automated, and price accordingly.
RPA is the layer most independent practices skip on the way to buying generative AI. It is also the layer with the shortest payback and the least implementation risk.
Every vendor will claim agentic AI by Q3 2026. The only differentiator worth diligence is task completion rate without human touch — ask for it by workflow, measured over 90 days, on accounts that resemble yours. A vendor that cannot produce that number for eligibility, claim status, or payment posting is selling RPA with a new label.
HHS has moved the proposed HIPAA Security Rule overhaul to its Long-Term Actions agenda, with July 2027 as the anticipated timeframe for final action. A final rule had been expected as early as May 2026.
Do not read the delay as a reprieve. The proposal eliminates the distinction between “addressable” and “required” implementation specifications — every control becomes mandatory — and would make encryption, multi-factor authentication, and vulnerability scanning explicit requirements. Once finalized, the expected sequence is effective 60 days after publication, compliance 180 days after that (roughly 240 days total), with business associate agreements updated within one year of the effective date.
For a practice, the compliance-critical detail is the BAA clock. Every clearinghouse, billing company, RCM vendor, and AI coding platform you use is a business associate. Refreshing that entire contract portfolio inside twelve months is a project, not a task.
The CY 2027 OPPS proposed rule includes a Request for Information on strengthening machine-readable file requirements — standardizing hospital pricing data, improving the accuracy and completeness of free-text field reporting, and enhancing the comparability and usefulness of the consumer-friendly display. Comments close August 31, 2026. Practices with hospital affiliations, joint ventures, or provider-based departments should read the MRF direction as a preview of where physician-side transparency expectations go next.
The updated 42 CFR Part 2 rule took effect February 16, 2026, aligning substance use disorder record consent with HIPAA and permitting a single consent covering treatment, payment, and health care operations. Behavioral health practices that built segregated release workflows can simplify — but only after updating the Notice of Privacy Practices and the consent form itself.
Inventory every business associate touching PHI in your revenue cycle — clearinghouse, billing vendor, AI coding tool, patient payment processor, transcription service, collection agency. Date-stamp each BAA. When the Security Rule finalizes, you will have twelve months to re-paper all of them, and the list is always longer than practices expect.
A March 2026 Veradigm survey of 360 independent practice leaders produced the most useful single statistic in independent practice finance this year: only 24% report high or complete visibility into where they are losing revenue.
That is the whole problem in one number. Three-quarters of independent practices are managing margin without instrumentation — which means the response to a 1.68% conversion factor cut, a practice expense re-basing, or a 7% specialty RVU decline is guesswork rather than targeted action.
The share of physician-owned practices fell 12.8% in the last two years, and nearly all independent practices report being approached about acquisition or consolidation in the past two years.
But the exit intent is not there. Among primary care clinicians surveyed January 31 – February 23, 2026, more than 80% are concerned about their practice’s financial stability over the next one to three years — and only 2% are considering leaving independent practice. A separate March 2026 survey found 54% reporting that financial pressure increased over the prior twelve months.
Independent physicians are not choosing employment. They are being outlasted by administrative cost. MGMA data puts the median investment required to support a single physician FTE at $343,128 as of Q4 2025, with support staff salaries and benefits consuming roughly 25% of total practice revenue — about half of all overhead.
| Metric | Typical Independent Practice | Target | Priority |
|---|---|---|---|
| Visibility into revenue leakage | 24% report high visibility | Monthly leakage report by payer, CPT, and denial reason | Highest |
| Eligibility verification | Manual or partial | 100% automated at scheduling, benefits-level detail | Highest ROI |
| First-pass denial rate | 11–14% | Under 5% | High |
| Days in A/R | 42–48 | Under 35 | High |
| Clean claim rate | 88–92% | 95% or better | High |
| Cost to collect | 5–7% of collections | Under 4% | Medium |
| Point-of-service collection rate | ~25% | Over 35% | Medium |
| Support staff cost | ~25% of revenue | 22% or less with automation offset | Medium |
| BAA inventory current and date-stamped | Rare | 100% | Rising |
The practices that stay independent through 2027 will not be the ones that found more revenue. They will be the ones that could see their revenue. Automating eligibility verification and building a denial report segmented by payer and reason code costs a fraction of what an EHR module costs — and it converts the 24% visibility statistic from a description of you into a description of your competitors.
Every specialty average below conceals a far wider code-level distribution than the headline suggests. Budget from your own top-25 code mix repriced at the proposed 2027 rates.
| Specialty | What Changed | The Number | What To Do |
|---|---|---|---|
| Primary Care | G2211 converts from a flat add-on code to a modifier on the base office/outpatient E/M; APCM matures with new behavioral health add-ons | +16% on the associated E/M (+32% for MSSP and LEAD ACO clinicians). APCM codes G0556 / G0557 / G0558 pay roughly $16–$117 PMPM; 2026 BH add-ons G0568 / G0569 / G0570 stack to roughly $263 PMPM | Model the modifier conversion against current G2211 volume — a percentage scales with visit level where a flat rate does not. Confirm APCM is not billed concurrently with CCM, PCM, or TCM; it can stack with RPM |
| Cardiology | CY 2027 PFS lands modestly positive on the specialty overall; the Ambulatory Specialty Model adds heart failure accountability in 2027 | ~+1% overall for cardiovascular services vs. 2026. ASM puts 9% of Part B payment at risk starting in the 2027 performance year, rising to 12% by 2031 | Run the ACC’s 2027 PFS calculator against your own code mix — the specialty average hides wide service-level variation. Begin ASM readiness if you treat heart failure |
| Orthopedics | The hardest-hit specialty in the CY 2027 PFS, before the same-day E/M policy is layered in | −7% overall; −5% non-facility, −8% facility | Quantify same-day E/M-plus-procedure volume. Orthopedic surgery also appears in the ASM low back pain cohort alongside anesthesiology, pain management, neurosurgery, and PM&R |
| Oncology — Radiation | Treatment delivery codes gain sharply while planning and management are cut; 340B redistribution favors non-drug services | CMS estimates +3% overall (+5% freestanding); ASCO models +1.5%. 77402 +23.11%, 77412 +21.81%, 77407 +7.73%; planning, management, and SBRT −3% to −6%; 340B budget-neutral redistribution estimated at +8.44% | Re-forecast by code, not by specialty average. A center weighted toward planning and SBRT can land negative inside a “positive” specialty |
| Oncology — Medical | 340B drug payment proposed at ASP−33.4% in the OPPS rule, with dollars redistributed to non-drug services under budget neutrality | Margin compression on infused and injected drugs — the dominant outpatient oncology revenue stream | Model the site-of-care shift. Integrate PA tracking for immunotherapies and targeted therapies directly into the billing workflow rather than a separate clinical queue |
| Radiology | Positive at the specialty level with violent code-level swings | +2% radiology, +2% nuclear medicine, +3% interventional radiology. CT angiography of the chest (71275) +149% to $83.09; CT upper extremity without contrast (73200) −15% | Pull the ACR impact tables and reprice your top 25 codes individually. Radiology also leads autonomous-coding adoption — the volume and template profile fit |
| Neurology | Payers are actively auditing bundling and AI-assisted interpretation | Scrutiny on EEG/EMG bundling, modifier 25 on same-day imaging plus E/M, and chemodenervation frequency documentation per treatment cycle. AI-assisted EEG interpretation and automated neuroimaging reports are explicit audit targets | Document medical necessity for each chemodenervation cycle. If you bill AI-assisted interpretation, keep the human interpretive attestation in the record |
| Mental / Behavioral Health | Consent simplification plus new stackable revenue | 42 CFR Part 2 aligned to HIPAA effective Feb 16, 2026 — single consent for treatment, payment, and operations. All major commercial payers now cover BH telehealth, though codes, modifiers, and rates vary by plan. New APCM BH add-ons G0568 / G0569 / G0570 | Rebuild the consent form and Notice of Privacy Practices. For CoCM and BHI programs, layer the APCM add-ons rather than billing legacy codes in parallel |
| Gastroenterology | Anorectal testing modernized; bariatric endoscopy gets a permanent code | 91120 and 91122 deleted, replaced by 91124 and 91125. 43889 added for endoscopic sleeve gastroplasty. GI carries among the highest outpatient denial rates in 2026, driven by aggressive NCCI bundling edits | Load the anorectal replacements before the next cycle. Enforce the screening-to-diagnostic conversion rule: modifier PT for Medicare, modifier 33 for commercial preventive designation |
Radiology is “+2%” and contains a +149% code and a −15% code. Radiation oncology is “+3%” and contains +23% delivery codes and −6% planning codes. The specialty impact table tells you how your society will lobby; it does not tell you what your practice will collect. Reprice your own top 25.
Eleven items, each tied to something in this issue. The first three are date-bound.