AI Scribing for Tele-Consult vs. In-Person Logic: Fixing the 2025-2026 Billing Divide

TL;DR — Tele-Consult vs. In-Person Billing Logic

The 2025–2026 telehealth coding landscape is fractured: the AMA adopted new telemedicine CPT codes (98000–98016), but CMS declined to recognize them, instructing providers to keep using in-person E/M codes with modifiers and Place of Service (POS) codes. This dual-track reality is exactly where geographic billing fraud, denials, and recoupments originate.

Scribing.io closes the gap deterministically. We bind patient location and modality to the FHIR R4 Encounter resource in real time, using Scribing.io Ambient Clinical Intelligence:

• POS 10 versus POS 02 — resolved via US Core Location type=PTRES for patient residence.

• Modifier 93 versus modifier 95 — resolved via the HL7 Virtual Service backport extension for audio modality.

• Attestation of Location captured — stored as a FHIR QuestionnaireResponse, linked via Provenance, auto-inserted into the note.

For Clinical Operations Directors this eliminates the single most common telehealth audit failure — POS/location mismatch — before the claim ever leaves the encounter.

• Jump to sections below:

• The 2026 Telehealth Coding Fracture

• Deterministic FHIR Location Logic

• Cardiology Multi-State Audit Resolution

• ICD-10 Documentation Standards

• Integration and Deployment Workflow

• Operations Director Adoption Checklist

The 2026 Telehealth Coding Fracture: Why AMA CPT and CMS POS Rules Disagree

CLINICAL UPDATE 2026: Revised for new CMS CPT G2211 standards, SB 1120 compliance, and FHIR interoperability.

Any Clinical Operations Director architecting a compliant telehealth billing workflow in 2026 is managing a structural contradiction. As of the AMA's 2025 CPT revisions, the CPT Editorial Panel introduced telemedicine-specific service codes — 98000–98007 for synchronous audio-video E/M, 98008–98015 for synchronous audio-only E/M, and 98016 for brief synchronous communication — while deleting the legacy audio-only codes 99441–99443.

Here is the operational problem: CMS did not adopt CPT 98000–98015. Medicare instructs providers to continue billing in-person E/M codes (99202–99215) with the appropriate POS code and modifier to indicate the visit was conducted via video or audio-only. This means the same clinical encounter can require two different code stacks depending on the payer.

The AMA handbook is explicit that "there may not be specific guidance on how to appropriately report POS" for the new codes, and it advises confirming reporting standards with each payer. What that reference cannot solve — because it is a coding table, not a documentation engine — is deterministic capture of where the patient actually was at the moment of service.

That geographic gap is where audit exposure originates. A coding reference tells you which code exists; it does not bind the physical location to an immutable, timestamped source. This playbook addresses that binding directly.

Scribing.io Deterministic Logic: Binding Location and Modality to FHIR R4

This is the Information Gain pillar of the playbook — the mechanism reference handbooks do not implement. Medical AI Scribing under Scribing.io does not suggest a POS code to a biller after the fact. We resolve it deterministically at the point of encounter by reading structured FHIR data.

The Resolution Chain Explained

1. Location binding at check-in: the patient's self-attested address is written to Encounter.location.location. If the resolved type includes PTRES from the US Core Location profile, the engine sets POS 10; otherwise it sets POS 02.

2. Modality binding on the resource: the engine reads the HL7 Virtual Service backport extension on the Encounter or Appointment to distinguish audio-only from audio-video. Audio-only maps to modifier 93; audio-video maps to modifier 95.

3. Cryptographic attestation linkage: the self-attested location is captured as a FHIR QuestionnaireResponse and linked via a Provenance resource to the Encounter, producing the Attestation of Location auto-inserted into the note.

What the reference handbook misses: the AMA guidance describes which code to use but relies on the human coder to independently establish patient location. There is no linkage between the documented claim and an immutable source of truth.

The Provenance linkage closes that gap and aligns with CMS telehealth documentation requirements extended through 2026. This is the practical difference between a defensible claim and a recoupment target under post-payment review.

Clinical Logic: Resolving a Multi-State Cardiology Group's $84K Recoupment

This is the centerpiece scenario Clinical Operations Directors evaluate during deployment discovery. It illustrates the exact audit mechanics of geographic billing fraud.

The Audit Situation Summarized

A multi-state cardiology group — 12 providers, 38% tele-consult volume — was audited after payers flagged a spike in 99213–99215 telehealth claims. The pattern triggered a targeted post-payment review across all telehealth-participating payers.

The Documented Root Cause

Patients located at home were billed with POS 02 plus modifier 95 instead of the correct POS 10, and no explicit location attestation existed in the note. The absence of the Attestation of Location made every claim unsubstantiated on geographic grounds.

The Scribing.io Resolution Steps

• Check-in prompts wrote the patient's self-attested address to Encounter.location with type=PTRES.

• An attestation sentence was inserted directly into the encounter note at signature time.

• The engine auto-selected POS 10 versus 02 based on the resolved location type.

• Modifier 93 was applied automatically when sessions were audio-only.

Denials fell below one percent, and — critically for a longitudinal specialty like cardiology — the group safely appended G2211 on eligible longitudinal visits without triggering payer edits. The underlying POS and modality data were now internally consistent.

Operations leaders modeling the financial impact of eliminating a recurring five-figure quarterly recoupment can run the numbers with our AI Medical Scribe ROI Calculator.

Technical Reference: ICD-10 Documentation Standards

Correct telehealth billing logic is only durable when the underlying diagnosis coding is specific and audit-defensible. Cardiology and chronic-disease tele-consults most frequently carry two high-volume diagnoses.

Ambient Clinical Intelligence structures documentation so each diagnosis carries the specificity CMS reviewers demand alongside the POS and modality evidence.

Diagnosis specificity matters for G2211. The complexity add-on is defensible only when longitudinal management of a chronic condition is documented across encounters. Clinical-Grade Scribing carries diagnosis continuity into every tele-consult note automatically.

Integration and Deployment Workflow Across EHR Platforms

The deterministic logic engine deploys through SMART on FHIR, so it operates inside existing charting rather than requiring a parallel system. Deployment reads and writes the same Encounter resource your billing team already reconciles.

Location and modality macros vary by platform, and the automation logic adapts to each host. Reference implementations are documented per system for operations teams planning rollout.

• Chiropractic and outpatient macro logic is detailed in the Scribing.io Chirotouch Automating Macro Logic Reference.

• Long-term and post-acute care logic is documented in the Scribing.io Pointclickcare Long Term Care Logic Reference.

For specialty-specific configuration details, review the relevant workflow under Scribing.io specialty configurations. Each specialty inherits the same location-binding chain with tailored documentation prompts.

State-level telehealth rules also apply, including SB 1120 governance of automated clinical decision logic. Compliance mapping is maintained under the Scribing.io AI scribe law reference and Scribing.io integration documentation.

Operations Director Adoption Checklist and Next Steps

Before you approve any telehealth billing workflow for 2026, confirm that each of the following controls is enforced at the encounter level rather than at the biller's desk.

1. Verify location binding is deterministic — POS 10 versus 02 must derive from Encounter.location.type, not manual default.

2. Confirm modality resolution is automated — modifier 93 versus 95 must read from the Virtual Service extension.

3. Require Provenance-linked attestation — every tele-consult note must carry the auto-inserted Attestation of Location.

4. Validate G2211 dependency logic — the add-on should append only when POS and modality data are internally consistent.

Pricing and deployment tiers are published transparently for operations budgeting. Review the Scribing.io Pricing & Plans to align seat counts with your tele-consult volume.

The structural fracture between AMA CPT and CMS POS rules will persist through 2026. Binding location and modality to FHIR at the point of encounter is the defensible path — and the one that keeps your telehealth claims out of the recoupment queue.