Operations Playbook — Table of Contents

• The Documentation Gap Competitors Miss: Why TheraNest Progress Notes Fail Golden Thread Audits Without Intervention-to-Goal Binding

• Scribing.io Clinical Logic: How a 10-Therapist Medicaid Clinic Reversed a $19,800 Recoupment

• Workflow Comparison: TheraNest Native vs. Sunoh.ai vs. Scribing.io Golden Thread Mapper

• Technical Reference: ICD-10 Documentation Standards for F33.1 and F41.1

• 90785 Interactive Complexity: The Revenue Line Your Clinic Is Leaving Uncoded

• 90-Day Plan Review Attestation Engine: Eliminating the Second-Most-Common Audit Finding

• EHR Integration Architecture: TheraNest, athenahealth, and Epic Deployment Models

• 10-Minute Medicaid ROI Model: Run It With Your Own Denial Data

• Implementation Protocol: Week-by-Week Deployment for Clinical Directors

Best AI Scribe for TheraNest: Behavioral Health ROI — The Clinical Director's Operations Playbook

Twelve percent. That is the flag rate a Clinical Director should expect when an MCO auditor runs Golden Thread extraction logic against TheraNest progress notes that were written without explicit intervention-to-objective binding. Not because the clinicians are incompetent. Because TheraNest's progress note module does not auto-insert Treatment Plan Goal/Objective IDs into session documentation, and no amount of narrative clinical skill compensates for a structural data field that the auditor's software cannot find. Scribing.io exists to close that structural gap—parsing active TheraNest treatment plans, auto-tagging every documented intervention with its GoalID/ObjectiveID, capturing measurable client response data, triggering 90-day plan review alerts, and prompting 90785 interactive complexity justification when session dynamics warrant it.

This playbook is not a product brochure. It is the operational reference a Clinical Director (LCSW) at a Medicaid-heavy outpatient clinic needs to (a) understand exactly why Golden Thread audit failures happen at the data-structure level, (b) see the step-by-step clinical logic Scribing.io uses to prevent them, and (c) model the financial impact against their own denial data. Every claim in this document maps to a verifiable audit standard, a published CMS billing rule, or a reproducible clinical workflow.

The Documentation Gap Competitors Miss: Why TheraNest Progress Notes Fail Golden Thread Audits Without Intervention-to-Goal Binding

What Sunoh.ai and Generic AI Scribes Get Wrong

The competitor landscape for behavioral health AI scribes—exemplified by platforms like Sunoh.ai—markets itself on "real-time transcription," "customizable templates," and "EHR-agnostic integration." These are table-stakes features. They solve the speed problem. They do not solve the compliance problem.

Here is what every Clinical Director running a Medicaid-heavy outpatient practice needs to internalize:

TheraNest progress notes do not automatically bind each session's Therapeutic Intervention to the exact Treatment Plan Goal/Objective ID that state DHHS Medicaid audit tools explicitly look for during Golden Thread verification.

This is not a formatting preference. It is the structural fault line that causes recoupments. And it persists whether you are integrating with TheraNest directly, routing through an athenahealth API bridge, or managing multi-EHR environments that include Epic Integration alongside your behavioral health stack.

The Anatomy of the Golden Thread

The "Golden Thread" is the auditable chain connecting five discrete documentation elements. State Medicaid agencies and their delegated MCOs verify this chain using automated extraction logic and manual sampling per CMS Program Integrity guidelines:

1. Diagnosis — ICD-10 code at maximum specificity (e.g., F33.1 - Major depressive disorder, recurrent, moderate; F41.1 - Generalized anxiety disorder)

2. Treatment Plan Goals and Objectives — with unique identifiers (Goal 1.2, Objective 1.2.b) and measurable criteria

3. Session Interventions — the specific evidence-based technique applied during each encounter

4. Client Response and Measurable Progress — behavioral observations tied to the objective's defined metric

5. Plan Review Attestations — typically required every 90 days per state Medicaid program rules, documented within the chart

State DHHS audit platforms use automated extraction logic that searches for explicit Goal/Objective ID references within progress notes. When these identifiers are absent, the note is flagged regardless of how clinically rich the narrative may be. The HHS Office of Inspector General has repeatedly identified behavioral health documentation as a high-risk area for improper payments precisely because of this structural gap.

What TheraNest Does (and Does Not Do)

TheraNest provides robust practice management: scheduling, billing, intake forms, and a progress note editor with DAP/SOAP templates. Its treatment plan module allows clinicians to create goals and objectives with identifiers. However, the critical handoff—automatically inserting those Goal/Objective IDs into each session's progress note and mapping the specific intervention used to the specific objective addressed—does not occur natively.

The clinician is left to manually type or copy-paste references like "Objective 1.2.b" into every progress note. Across a 10-therapist clinic averaging 40+ clients per therapist per week, this manual process degrades within days. Notes default to generic language: "Continued to work on treatment plan goals." That language fails audit.

What Generic AI Scribes Miss

An AI scribe that transcribes session audio and generates a DAP note—even a well-structured one—still produces documentation that is Golden Thread orphaned unless it has parsed the active treatment plan and mapped the detected intervention to a specific GoalID/ObjectiveID. Platforms like Sunoh.ai describe their behavioral health capabilities in terms of transcription accuracy and template customization. Neither capability addresses the structural compliance requirement. Research published through the National Library of Medicine on AI-assisted clinical documentation consistently identifies structured data linkage—not narrative generation—as the critical factor in audit readiness.

Current clinical benchmarks indicate that behavioral health Medicaid audits flag between 8–15% of progress notes for Golden Thread deficiencies. The most common finding is the absence of explicit intervention-to-objective linkage—not missing signatures, not late entries, but the structural binding itself.

Scribing.io Clinical Logic: How a 10-Therapist Medicaid Clinic Reversed a $19,800 Recoupment and Achieved Zero Follow-Up Audit Findings

This section models the exact scenario a Clinical Director (LCSW) faces and demonstrates Scribing.io's clinical decision logic at each processing step.

The Scenario

A Clinical Director (LCSW) operates a 10-therapist Medicaid outpatient clinic using TheraNest. The clinic undergoes an MCO audit covering 24 months of claims. Auditors apply standard Golden Thread verification protocols consistent with CMS Medicaid Integrity standards.

Audit Findings (Pre-Scribing.io):

• 12% of 90837 sessions (53-minute individual psychotherapy) flagged

• Primary deficiency: Lacking explicit linkage of interventions to Treatment Plan Objective IDs

• Secondary deficiency: Missing 90-day plan review attestations

• Recoupment issued: $19,800 (approximately 12 sessions × $1,650 average Medicaid reimbursement per flagged unit)

• Lookback exposure: State permits 5-year retrospective audits per federal Medicaid rules

Why the Notes Failed

The therapists were clinically competent. Their notes described interventions accurately—CBT cognitive restructuring, motivational interviewing, exposure-based techniques. A representative note read:

"Therapist utilized cognitive restructuring to address client's negative thought patterns related to depressive symptoms. Client demonstrated moderate engagement."

The auditor's extraction tool searched for:

• A specific Treatment Plan Goal ID (e.g., "Goal 1.2")

• A specific Objective ID (e.g., "Objective 1.2.b: Client will identify and challenge ≥3 cognitive distortions per week as measured by thought records")

• A measurable progress metric (e.g., "Client completed 2 of 3 targeted thought records this week")

• 90-day plan review documentation within the required window

None of these structured data points were present. The notes were clinically adequate but audit-structurally deficient.

Scribing.io's Golden Thread Mapper: Step-by-Step Clinical Logic

The Outcome (Post-Scribing.io Implementation)

• Follow-up audit findings: Zero

• Previously denied units: Recovered through appeal using Scribing.io-formatted documentation containing explicit GoalID/ObjectiveID linkage, measurable progress data, and 90-day review attestations

• 90785 revenue capture: The clinic identified an estimated 18% of sessions qualifying for interactive complexity add-on billing that had previously gone uncoded

• Clinician documentation time: Reduced from 12–15 minutes per note to 3–5 minutes of review and attestation

• 5-year lookback exposure: Eliminated prospectively for all new documentation

Workflow Comparison: TheraNest Native vs. Sunoh.ai vs. Scribing.io Golden Thread Mapper

Technical Reference: ICD-10 Documentation Standards for F33.1 and F41.1 in Medicaid Behavioral Health

Precise ICD-10 coding is the first link in the Golden Thread. Documentation that does not support the selected code at the specificity level required by the payor triggers denials before the auditor ever reads the progress note. The AMA's ICD-10 coding guidelines and CMS's ICD-10-CM Official Guidelines for Coding and Reporting both require that the medical record substantiate every axis of the selected code.

F33.1 — Major Depressive Disorder, Recurrent, Moderate

F41.1 — Generalized Anxiety Disorder

Scribing.io ensures that F33.1 - Major depressive disorder and F41.1 - Generalized anxiety disorder reach maximum coding specificity by embedding the documentation elements that substantiate each axis of the code directly into the progress note—episode characterization, severity qualifier, validated measure scores, functional impairment language, and comorbidity differentiation. This is not post-hoc coding support. It is concurrent documentation that makes the code defensible at the moment of signing.

90785 Interactive Complexity: The Revenue Line Your Clinic Is Leaving Uncoded

CPT 90785 is an add-on code reported with primary psychotherapy codes (90834, 90837) when specific complicating factors are present during the session. Per the AMA CPT codebook, qualifying factors include:

• The need to manage maladaptive communication among participants (e.g., caregiver-client conflict)

• Involvement of third parties (school personnel, CPS, legal guardians) requiring the therapist to adjust the treatment interaction

• Evidence of a communication barrier (use of interpreter, cognitive impairment, sensory deficit)

• Emotional or behavioral dysregulation that requires the therapist to manage the therapy interaction itself before addressing the treatment plan content

In Medicaid behavioral health—particularly child/adolescent and complex adult populations—these factors are present in an estimated 15–25% of sessions. Most go uncoded because:

1. The clinician does not recognize the session qualifies during note-writing

2. The documentation does not contain the specific justification language payors require

3. The practice's billing workflow does not have a systematic trigger for 90785 review

Scribing.io's 90785 Detection Logic

During session transcription, Scribing.io's NLP engine monitors for linguistic markers of interactive complexity: caregiver interjections, interpreter pauses, therapist redirect language addressing dysregulation, references to third-party communications. When detected, the system does not auto-bill. It prompts the clinician:

"Interactive complexity factors detected in this session: [specific factor identified]. Do you wish to document 90785 justification? [Yes — Generate justification language] [No — Not applicable]"

If affirmed, Scribing.io generates payor-compliant justification language specific to the detected factor, embedded within the progress note before the clinician's final attestation. The clinician retains full authority to accept, modify, or reject.

For a 10-therapist clinic: if 18% of sessions qualify and the average 90785 reimbursement is $15–$25 per unit, the annualized revenue recovery ranges from $28,000–$47,000—revenue that was clinically earned but never billed.

90-Day Plan Review Attestation Engine: Eliminating the Second-Most-Common Audit Finding

After intervention-to-objective binding, the most frequent Golden Thread audit deficiency is the missing or late 90-day treatment plan review. Most state Medicaid programs require that treatment plans be reviewed and updated at minimum every 90 calendar days. Some states (e.g., North Carolina, Ohio, New York) impose stricter timelines for specific populations.

TheraNest does not include automated 90-day plan review tracking. The responsibility falls on the clinician or administrative staff to manually track review dates—a system that fails reliably across any practice with more than a handful of active clients.

Scribing.io's Review Tracking Architecture

• Day 0: Treatment plan creation/last review date is indexed from TheraNest.

• Day 75: First alert appears in the clinician's Scribing.io dashboard and within the next session note draft for that client: "Treatment plan review due in 15 days."

• Day 85: Escalated alert: "Treatment plan review due in 5 days. Schedule review session or document concurrent review."

• Day 91+ (overdue): Hard flag on every subsequent note: "⚠ Treatment plan review overdue. Documentation for this session may be subject to audit finding. Complete plan review and attestation."

When the clinician completes the review—whether during a dedicated session or as a concurrent administrative action—Scribing.io generates the attestation block:

"Treatment Plan Reviewed: [DATE]. Goals and objectives reviewed with client. Goal 1.2 maintained — client continues to demonstrate need. Objective 1.2.b modified: target increased from 3 to 4 cognitive distortions identified per week based on progress. Next review due: [DATE + 90 days]. Clinician signature: [NAME, CREDENTIALS]."

This attestation is written into TheraNest and indexed for audit retrieval.

EHR Integration Architecture: TheraNest, athenahealth, and Epic Deployment Models

Scribing.io's Golden Thread Mapper is not a standalone documentation tool. It operates as an integration layer that reads from and writes to your EHR. The architecture varies by platform:

For multi-EHR environments—common in organizations where behavioral health operates on TheraNest while primary care runs on athenahealth or Epic—Scribing.io maintains Golden Thread integrity across platforms by normalizing treatment plan data into a unified GoalID/ObjectiveID schema regardless of the source EHR's native data structure.

10-Minute Medicaid ROI Model: Run It With Your Own Denial Data

This model requires three data points from your practice. Pull them from your TheraNest billing reports and your most recent MCO remittance advice.

Calculation

• Monthly lost revenue: A × B × C = 400 × 0.08 × $110 = $3,520/month

• Annual lost revenue: $3,520 × 12 = $42,240/year

• 5-year lookback exposure: $42,240 × 5 = $211,200

• 90785 uncaptured revenue (estimated): A × 18% × $20 avg = 400 × 0.18 × $20 = $1,440/month = $17,280/year

• Total annual recoverable revenue: $42,240 + $17,280 = $59,520

Compare that figure against Scribing.io's licensing cost for your practice. In every Medicaid-heavy behavioral health clinic we have modeled, the ROI exceeds 400% in year one before accounting for the clinician time savings (7–10 minutes per note × therapist hourly rate × sessions per week).

Implementation Protocol: Week-by-Week Deployment for Clinical Directors

See TheraNest Golden Thread auto-mapping (Goal/Objectives IDs), real-time 90785 prompts, and a 10-minute Medicaid ROI model using your denial data. Request a clinical demonstration from Scribing.io →

This Operations Playbook is maintained by the Clinical Documentation Integrity team at Scribing.io. Clinical standards referenced herein reflect CMS Medicaid guidelines, AMA CPT conventions, DSM-5-TR diagnostic criteria, and state DHHS audit protocols as of 2026. Clinicians should verify state-specific Medicaid requirements, as plan review intervals and documentation standards vary by jurisdiction. Nothing in this playbook constitutes legal advice or guarantees specific audit outcomes.