Skip to content

CMS & Regulatory Compliance

California SB 1120

California SB 1120: Cardiology AI Review Compliance

Document discrete LVEF, NYHA class, and GDMT titration so SB 1120 physician review holds. Book your audit at https://cal.com/merryai/demo.

Key Takeaways
  • California SB 1120 amends Health and Safety Code §1367.01 and Insurance Code §10123.135, requiring payer AI tools to base utilization review on documented clinical history, with medical necessity decisions reserved for licensed physicians
  • Discrete LVEF percentage, NYHA functional class, and pillar-level GDMT titration are the statutory inputs payer algorithms read for ICD, CRT, LVAD, and imaging requests
  • Clinician attestation safeguards keep every AI-suggested value flagged as a proposal until the cardiologist confirms, edits, and signs
  • Field-level provenance logs support HIPAA audit controls under 45 CFR §164.312(b) and give reviewers traceable documentation
  • Validation via [Access Clinical Prompts at templates.scribing.io](https://templates.scribing.io)

Executive Key Takeaways

  • California SB 1120 amends Health and Safety Code §1367.01 and Insurance Code §10123.135, requiring payer AI tools to base utilization review on documented clinical history, with medical necessity decisions reserved for licensed physicians
  • Discrete LVEF percentage, NYHA functional class, and pillar-level GDMT titration are the statutory inputs payer algorithms read for ICD, CRT, LVAD, and imaging requests
  • Clinician attestation safeguards keep every AI-suggested value flagged as a proposal until the cardiologist confirms, edits, and signs
  • Field-level provenance logs support HIPAA audit controls under 45 CFR §164.312(b) and give reviewers traceable documentation
  • Validation via [Access Clinical Prompts at templates.scribing.io](https://templates.scribing.io)
Regulatory Verification Framework
2026 Audit Ready
Cal. Health & Safety Code §1367.01(k)Verified Compliant

AI, algorithm, or software tools used in utilization review must base determinations on the enrollee's medical or clinical history and individual clinical circumstances, and may not supplant provider decision-making

AI, algorithm, or software tools used in utilization review must base determinations on the enrollee's medical or clinical history and individual clinical circumstances, and may not supplant provider decision-making

Cal. Health & Safety Code §1367.01(e)Verified Compliant

only a licensed physician or licensed health care professional competent in the specific clinical issue may deny, delay, or modify requests on medical necessity grounds

only a licensed physician or licensed health care professional competent in the specific clinical issue may deny, delay, or modify requests on medical necessity grounds

Cal. Insurance Code §10123.135(j)Verified Compliant

parallel AI utilization review obligations applied to disability insurers, including written policies, disclosure, and periodic performance review

parallel AI utilization review obligations applied to disability insurers, including written policies, disclosure, and periodic performance review

Cal. Health & Safety Code §1367.01(h)Verified Compliant

utilization review decision timeframes that AI-assisted determinations must satisfy, including 72-hour review for urgent requests

utilization review decision timeframes that AI-assisted determinations must satisfy, including 72-hour review for urgent requests

Statutory Compliance Architecture for California SB 1120

Documenting cardiology encounters under California SB 1120 begins with the statutory text. The Physicians Make Decisions Act amended Health and Safety Code §1367.01 and Insurance Code §10123.135, effective January 1, 2025. Any artificial intelligence, algorithm, or other software tool that a health plan or disability insurer uses for utilization review must base its determination on the enrollee's medical or clinical history, the individual clinical circumstances presented by the requesting provider, and other relevant clinical information in the record. The tool may not supplant provider decision-making, and it may not deny, delay, or modify care on medical necessity grounds.

Only a licensed physician or a licensed health care professional competent to evaluate the specific clinical issue may make that determination under §1367.01(e). Plans must also maintain written policies, disclose tool use, and periodically review tool performance for accuracy and reliability. For cardiology, the implication is direct: the chart is the dataset the payer algorithm reads, and the physician reviewer must be able to verify every value the algorithm relied upon.

Discrete Cardiology Parameters as Statutory Inputs

Left ventricular ejection fraction must be recorded as a numeric percentage with modality and study date, for example LVEF 30% by transthoracic echocardiogram on a specified date. Narrative phrasing such as reduced function cannot be mapped against payer thresholds for ICD or CRT implantation, LVAD evaluation, or advanced imaging. It invites an algorithmic mismatch that a reviewer must then reconstruct by hand.

NYHA functional class and GDMT titration history complete the clinical picture the statute requires plans to consider. Document class I through IV with the symptom anchor, then list each pillar with current dose and trajectory: ARNI, ACE inhibitor, or ARB; evidence-based beta-blocker; mineralocorticoid receptor antagonist; and SGLT2 inhibitor. Record intolerance, contraindications, and duration at maximally tolerated dose, because most device criteria assume at least three months of optimized therapy.

EHR Field Architecture and Clinician Attestation

Merry writes dictated cardiology values into the discrete fields of the active EHR browser window through Chrome extension DOM placement, including web-rendered Epic surfaces, without API configuration. When a cardiologist dictates NYHA class III with an ejection fraction of 30% on maximally tolerated GDMT, the engine maps each parameter to its version-controlled field selector for LVEF, NYHA class, and the heart failure therapy section. It restricts write-back to the active assessment context.

Human-in-the-Loop Attestation Controls

Every injected value remains a proposal until the cardiologist reviews, edits, and signs. Suggested fields carry a visual flag that clears only on explicit confirmation, so documentation never presents as an automated necessity determination, which §1367.01(e) reserves for licensed clinicians. Injection occurs only after a deliberate user action, such as applying the note, rather than on passive page load.

Field-level provenance logs record the timestamp, source dictation segment, target field, and confirmation status for each write. These records support HIPAA audit controls under 45 CFR §164.312(b). They also give plans the traceable reviewer documentation that emerging SB 1120 implementation guidance expects to be retained for at least three years.

Comparative Liability Across Documentation Models

The table below contrasts three documentation approaches against the statutory dimensions a California plan reviewer will examine.

Compliance DimensionManual ChartingStandard Generic AI ScribesMerry AI Compliance Architecture
LVEF capture (§1367.01(k))Often narrative, variable completenessEmbedded in note textDiscrete numeric field with modality and date
NYHA class and GDMT historyScattered across encountersSummarized, rarely dose-levelStructured class plus pillar-by-pillar titration
Physician determination (§1367.01(e))Clinician-authoredUnflagged AI text risks ambiguityFlagged proposals requiring attestation
Provenance and audit trailSignature onlyNote-level or noneField-level log per injected value
EHR write-backManual entryCopy-paste or narrative pushDOM placement into discrete fields
Denial rebuttal readinessChart reconstruction requiredReviewer must parse narrativeVerifiable values mapped to payer criteria

Generic scribes produce readable prose but leave discrete cardiology parameters buried in narrative text, where payer tools misread them and reviewers must search. Manual charting preserves authorship but rarely achieves consistent structure across a panel of heart failure patients.

Overturning Algorithmic Denials With Verifiable Evidence

When a payer algorithm flags an ICD, CRT, or advanced imaging request, SB 1120 entitles the requesting cardiologist to a determination grounded in the documented record and made by a competent licensed reviewer. Discrete, attested values in the assessment fields let that reviewer confirm LVEF, NYHA class, and GDMT duration within seconds. This converts a peer-to-peer discussion into a verification step rather than a records search.

Pair each necessity statement with its objective anchor: NT-proBNP value, echocardiographic parameters, cardiopulmonary exercise data, or hospitalization history. Link the statement to the requested service and the relevant guideline threshold, so any modification or denial must articulate why documented criteria were not met.

CMS Coding Alignment: G2211, Modifiers, and ICD-10

Longitudinal heart failure management supports HCPCS G2211 when the cardiologist furnishes ongoing care for a single, serious, or complex condition, and GDMT titration documentation substantiates that relationship. CMS limits G2211 when the office E/M carries modifier 25, with narrow preventive-service exceptions, so separate same-day procedural work must be documented distinctly. Code heart failure to full specificity, such as I50.22 for chronic systolic heart failure, using the current code set at https://www.cms.gov/medicare/coding-billing/icd-10-codes.

Practices preparing for California plan audits can review cardiology field maps, attestation flows, and provenance exports with our informatics team. Schedule a 15-Minute Workflow Audit to evaluate your documentation against the statute.

Regulatory & Compliance FAQ

Does California SB 1120 prohibit payers from using AI to deny cardiology prior authorization requests?

Not entirely. Health and Safety Code §1367.01(k) and Insurance Code §10123.135(j) permit AI tools in utilization review but prohibit them from denying, delaying, or modifying care based on medical necessity. Under §1367.01(e), a licensed physician competent in the clinical issue, such as a cardiologist evaluating CRT eligibility, must make that determination. The tool must also rely on the enrollee's documented clinical history, including discrete LVEF, NYHA class, and GDMT records.

Which cardiology data elements should be documented discretely to satisfy SB 1120 utilization review?

Record LVEF as a numeric percentage with modality and study date, NYHA class I through IV with the symptom anchor, and GDMT titration by pillar: ARNI, ACE inhibitor, or ARB; beta-blocker; MRA; and SGLT2 inhibitor, including doses, intolerances, and duration. Because §1367.01(k) requires determinations based on individual clinical circumstances, discrete values give the physician reviewer verifiable inputs and reduce algorithmic misreads of narrative text.

How does Merry keep AI-suggested cardiology values from appearing as medical necessity determinations?

Merry injects dictated values into discrete EHR fields only after a deliberate user action, then flags each value as a proposal until the cardiologist confirms, edits, or rejects it before signing. Field-level logs capture timestamp, source dictation segment, target field, and confirmation status. This preserves clinician authorship consistent with §1367.01(e), supports HIPAA audit controls under 45 CFR §164.312(b), and keeps Merry positioned as documentation support, not a decision engine.

How should cardiology documentation respond when a payer algorithm modifies or denies a device request?

Ensure the assessment fields contain attested LVEF, NYHA class, and at least three months of documented maximally tolerated GDMT, paired with objective anchors such as NT-proBNP, echocardiographic data, or hospitalization history. Link each necessity statement to the requested ICD, CRT, or imaging service and its guideline threshold. Because SB 1120 requires physician review grounded in the record, verifiable discrete evidence obligates the reviewer to articulate which documented criteria were unmet.

Where can clinics verify compliant templates?

Clinics can review SB 1120-aligned cardiology prompt structures at [https://templates.scribing.io](https://templates.scribing.io), including heart failure assessment templates with discrete LVEF, NYHA class, and GDMT titration fields. Compare each template against your EHR field map and payer criteria before deployment, confirm that attestation steps are enforced, and validate ICD-10 specificity against the CMS code set. For a structured review of your current workflow, schedule an audit at https://cal.com/merryai/demo.

Explore More Knowledge

Related Clinical Rules & Compliance Guides

View all clinical rules & compliance
California AB 3030 Compliance for Primary Care
CMS & Regulatory Compliance

California AB 3030 Compliance for Primary Care

Append AB 3030 AI notices and clinician contact steps, then timestamp review. Book your audit at https://cal.com/merryai/demo.

Dr. Sarah Chen, MD•7 min read
Read guide
UDS / HRSA Compliance for FQHC Documentation
CMS & Regulatory Compliance

UDS / HRSA Compliance for FQHC Documentation

How FQHCs map tobacco, blood-pressure, and screening evidence to UDS measures with source traceability and human validation. Book your workflow audit at https://cal.com/merryai/demo.

Dr. Sarah Chen, MD•7 min read
Read guide
Florida 934.03 Compliance for Multi-Specialty Groups
CMS & Regulatory Compliance

Florida 934.03 Compliance for Multi-Specialty Groups

How outpatient groups satisfy Florida's all-party consent felony statute for ambient AI audio capture. Book your workflow audit at https://cal.com/merryai/demo.

Dr. Sarah Chen, MD•7 min read
Read guide