New State AI-in-UM Law Reshapes Payer Obligations in Your Market

(Inferred) A California Knox-Keene-licensed MA or commercial payer issues a PA denial for an inpatient admission or elective procedure citing InterQual or MCG criteria outputs but lacks any documented licensed physician attestation of final review. Payer may claim internal workflow satisfies SB 1120 but cannot produce evidence of human reviewer identity, credentials, or individualized clinical rationale.

Upon receipt of any PA denial from a California-regulated plan effective Jan 1, 2025, UR team must request in writing the name, NPI, and licensure of the reviewing clinician. If the denial cites only algorithmic or criteria-tool output without named physician attestation, flag as SB 1120 non-compliant and escalate to physician advisor. File a concurrent DMHC complaint (Knox-Keene) or CDI complaint (disability insurers) if human review cannot be documented. Preserve all denial correspondence for audit.

'Per California Health & Safety Code §1367.01 as amended by SB 1120, effective Jan 1, 2025, a final determination on medical necessity may not be made solely by an AI or algorithmic tool — a licensed physician or qualified health care professional must make the final call. I am requesting confirmation of the reviewing clinician's name, licensure, and the individualized clinical basis for this denial specific to this enrollee's clinical history.'

(Inferred) A California commercial payer or Medi-Cal managed care plan issues a concurrent or retrospective downgrade from inpatient to observation, citing AI-assisted review of MCG thresholds. The payer asserts attending documentation did not meet population-level inpatient benchmarks embedded in the tool, without demonstrating that the enrollee's individual clinical circumstances were evaluated by a licensed physician.

Ensure every California-enrolled inpatient admission includes an explicit attending attestation: 'Inpatient admission medically necessary based on [specific individual clinical findings] which exceed observation-level management given [specific risk factors] — not based solely on population criteria.' Concurrent-review nurse must document payer reviewer name and credentials on each contact. Any downgrade must trigger physician advisor review of SB 1120 compliance before acceptance.

'Under H&S §1367.01 as amended by SB 1120, any modification of services based on medical necessity must reflect the enrollee's specific medical and clinical history and individual clinical circumstances — not solely group datasets. I am requesting confirmation that a licensed physician made this downgrade based on a review of this patient's individual record, and I would like to discuss the specific clinical basis for observation status being sufficient.'

(Inferred) A California Knox-Keene behavioral health managed care plan denies continued inpatient psychiatric or residential treatment, citing standardized algorithmic assessment (e.g., LOCUS/CALOCUS scores processed through AI-assisted UR) and aggregate population discharge benchmarks rather than documented individualized review of this enrollee's psychiatric history, current acuity, and treatment response.

Implement a BH-specific UR template requiring the attending psychiatrist to document individual risk factors, specific barriers to step-down, treatment response to date, and explicit statement that level-of-care change would pose individualized clinical risk. On any BH denial from a CA plan, request reviewer name, NPI, and licensure — if not a licensed psychiatrist or doctoral-level psychologist, flag for SB 1120 'qualified health care professional' challenge. Submit DMHC complaint if individualized human review cannot be documented.

'This continued-stay determination involves a psychiatric presentation with individualized clinical factors including [specific symptoms, risk factors, treatment barriers] in the submitted record. SB 1120 requires AI tool determinations to be based on this enrollee's specific clinical history and individual circumstances — not solely on group datasets. I am requesting a P2P with the licensed clinician who made this determination and documentation that this was not an AI-only determination.'