ICD-10 K76.0: Fatty Change of Liver (NAFLD) — Clinical Documentation & AI-Assisted Coding Operations Playbook for Primary Care

TL;DR

K76.0 (Fatty [change of] liver, not elsewhere classified) is the default ICD-10-CM code for nonalcoholic fatty liver disease (NAFLD) when steatohepatitis has not been clinically confirmed. Upgrading to K75.81 (NASH) requires explicit clinician attestation of inflammatory criteria. This playbook details how Scribing.io's ambient AI scribe auto-computes the FIB-4 fibrosis index directly within the visit note, documents NASH indicator presence or absence, and generates payer-ready medical-necessity language for transient elastography (CPT 91200) — eliminating the documentation gaps that cause prior-authorization denials and delayed risk stratification for your patients with metabolic liver disease.

• Why K76.0 Documentation Demands More Than a Code Lookup

• Technical Reference: ICD-10 Documentation Standards for K76.0 and K75.81

• The FIB-4 Score: The Missing Link Between Primary Care and Hepatology Referral

• Scribing.io Clinical Logic: From Denied FibroScan to Approved Elastography in One Visit

• How Scribing.io Auto-Computes FIB-4 and Documents NASH Indicators via FHIR R4

• NAFLD-to-NASH Documentation Workflow: A Step-by-Step Decision Table

• Payer Perspectives: Medical Necessity, DRG Assignment, and Denial Prevention

• Frequently Asked Questions: K76.0, FIB-4, and AI-Assisted Documentation

Why K76.0 Documentation Demands More Than a Code Lookup

The CMS MS-DRG v44.0 Definitions Manual lists K76.0 (Fatty [change of] liver, not elsewhere classified) and K75.81 (Nonalcoholic steatohepatitis [NASH]) as principal diagnoses that group to DRG 441–443 (Disorders of Liver Except Malignancy, Cirrhosis or Alcoholic Hepatitis). That taxonomy is accurate but entirely structural. It tells a coder where the code lives in the DRG hierarchy. It tells a primary care physician almost nothing about:

• When to assign K76.0 versus K75.81 in an ambulatory encounter

• What documentation elements payers require to authorize downstream workup (elastography, hepatology referral)

• How to compute and record FIB-4, the non-invasive fibrosis index that current AASLD guidance positions as the first-line risk stratification tool in primary care

• Which metabolic co-morbidities must be explicitly documented to establish medical necessity

This gap — between code taxonomy and clinical documentation sufficiency — is precisely where denials originate and where patient care timelines fracture. Published estimates suggest that 70–75% of NAFLD in the U.S. is managed in primary care, yet documentation audits consistently reveal that fibrosis risk scores and NASH indicator attestation are absent from the majority of visit notes carrying a K76.0 diagnosis. Scribing.io exists to close this gap at the point of documentation, not after the denial letter arrives.

The CMS reference provides the DRG map. This playbook provides the clinical documentation operations layer that sits underneath it — purpose-built for internists managing metabolic liver disease in 2026. For clinics evaluating ambient AI scribes, book a 15-minute demo to see our FHIR/LOINC-driven FIB-4 autopopulation and one-click 91200 elastography prior-auth packet with MAC policy checks and audit-ready rationale embedded in your note.

Technical Reference: ICD-10 Documentation Standards for K76.0 and K75.81

Understanding the precise semantic boundary between these two codes is foundational. Mis-assignment in either direction creates downstream problems: K75.81 without supporting documentation triggers audit risk; K76.0 when NASH criteria are met understates severity and may result in denial of advanced workup. The Scribing.io ICD-10 Documentation Library maintains a continuously updated reference for liver-related codes; the critical comparison follows.

The Excludes1 Rule and Why It Matters Operationally

ICD-10-CM's Excludes1 notation between K76.0 and K75.81 means these codes are mutually exclusive — they cannot be reported on the same encounter for the same clinical finding. A patient either has simple steatosis (K76.0) or steatohepatitis (K75.81), never both simultaneously. This is not merely a coding convention; it reflects a genuine clinical distinction codified in the ICD-10-CM Official Guidelines for Coding and Reporting.

When documentation is ambiguous — when the note neither confirms nor denies NASH — coders default to K76.0, which may understate the patient's risk profile and weaken the medical-necessity argument for elastography. This is why Scribing.io's workflow requires explicit clinician confirmation or negation of NASH before allowing code assignment. The system will not auto-upgrade to K75.81 based on lab values alone. That guardrail is non-negotiable.

Reaching Maximum Code Specificity

Scribing.io ensures both K76.0 — Fatty (change of) liver and not elsewhere classified; K75.81 — Nonalcoholic steatohepatitis (NASH) reach maximum specificity through three mechanisms: (1) structured NASH indicator attestation that forces a binary clinician decision, preventing the ambiguity that collapses to unspecified coding; (2) automatic pairing of the primary liver code with all relevant metabolic comorbidity codes (E11.x for T2DM, E66.x for obesity, E78.x for dyslipidemia) to paint the full clinical picture; and (3) pre-submission audit logic that flags any K76.0 note lacking FIB-4 documentation or any K75.81 note lacking the clinician's NASH attestation timestamp. These three layers convert a code selection into a defensible clinical argument.

The FIB-4 Score: The Missing Link Between Primary Care and Hepatology Referral

The Fibrosis-4 (FIB-4) index is a validated, non-invasive serum biomarker score that estimates hepatic fibrosis using four readily available data points:

FIB-4 = [Age (years) × AST (U/L)] / [Platelet count (10⁹/L) × √ALT (U/L)]

Published validation studies — including the foundational Sterling et al. (2006) Hepatology paper and subsequent meta-analyses in JAMA — demonstrate that FIB-4 has an area under the receiver operating characteristic (AUROC) of 0.80–0.85 for excluding advanced fibrosis (≥F3) when the score falls below the low-risk cutoff. It is not a perfect test. It is the best first-line, zero-cost triage tool available in primary care, and the AASLD Practice Guidance on NAFLD (2023, reaffirmed 2025) explicitly recommends it as the initial stratification step.

Why FIB-4 Belongs in Every NAFLD Visit Note

Current AASLD and AGA clinical guidance recommends FIB-4 as the initial risk stratification tool for all patients with suspected or confirmed NAFLD in primary care. It is not a hepatology-only metric. It is the gateway score that determines whether a patient needs:

• Reassurance and lifestyle modification (low risk, FIB-4 <1.3)

• Transient elastography or specialist referral (intermediate risk, FIB-4 1.3–2.67)

• Urgent hepatology consultation (high risk, FIB-4 >2.67 or decompensation signs)

Age-Adjusted Decision Thresholds

The Documentation Gap That Causes Denials

Despite its central role in guidelines, FIB-4 is rarely computed and documented in real-time during primary care encounters. The labs required (AST, ALT, platelets) are almost always available in the EHR, but the calculation is manual, the age-adjusted thresholds are easy to misremember, and the result is almost never recorded in a structured format that payers can audit. This is the single most common reason that elastography orders and hepatology referrals are denied on first submission: the note does not contain a computed fibrosis risk score linked to the patient's own time-stamped labs.

Scribing.io Clinical Logic: From Denied FibroScan to Approved Elastography in One Visit

Clinical Scenario: A 52-year-old man (BMI 33, T2DM) has ALT 78 U/L, AST 62 U/L, platelets 180 × 10⁹/L. His PCP orders a FibroScan, but the initial request is denied because the note lacks FIB-4 and any statement on NASH. With Scribing.io, the visit note auto-pulls time-stamped labs, computes FIB-4 = 2.1 (intermediate risk), documents metabolic risk factors and imaging-confirmed steatosis, prompts the PCP to confirm NASH criteria (kept as NAFLD only), and generates a 91200 elastography order with medical-necessity text. The resubmission is approved, avoiding a 6-week delay and ensuring timely risk stratification.

Dissecting the Denial: What the Original Note Was Missing

Step-by-Step Logic Breakdown: How Scribing.io Solves This Problem

The following sequence executes automatically during the encounter. The clinician speaks naturally; Scribing.io's ambient engine handles the documentation logic in parallel.

1. FHIR R4 Lab Retrieval: When the encounter opens and NAFLD/liver is identified as a visit reason, Scribing.io queries the EHR's FHIR R4 Observation endpoint for AST (LOINC 1920-8), ALT (LOINC 1742-6), and Platelets (LOINC 777-3). A date filter restricts results to the most recent 6-month window. The system handles an Epic/Cerner API quirk: by default, both platforms page Observation resources at 50 per bundle. If the patient has extensive lab history, Scribing.io follows pagination links to ensure it retrieves all Observations from the same draw date, preventing split-date computation errors.

2. Unit Normalization: Platelet counts arrive in inconsistent units across EHR systems — sometimes as 10⁹/L (180), sometimes as 10³/µL (180), and occasionally as raw counts (180,000). Scribing.io normalizes all incoming platelet values to 10⁹/L before computation. AST and ALT are verified as U/L. If unit metadata is missing from the FHIR resource (common with scanned outside labs ingested via OCR), the system applies a heuristic range check: a platelet value of 180,000 is normalized to 180; a value of 180 is accepted as-is.

3. FIB-4 Computation: With normalized inputs — Age 52, AST 62, ALT 78, Platelets 180 — the engine computes: FIB-4 = (52 × 62) / (180 × √78) = 3224 / (180 × 8.832) = 3224 / 1589.76 = 2.03. (The scenario rounds to 2.1; the precise value may vary slightly depending on decimal handling. Scribing.io records the value to two decimal places.) The system then applies the age-adjusted threshold: patient is under 65, so the intermediate-risk band is 1.3–2.67. FIB-4 of 2.03 falls squarely in the intermediate-risk zone, triggering the elastography recommendation pathway.

4. NASH Indicator Inventory: The system scans the active problem list, medication list, and vitals for NASH-associated indicators. For this patient, it identifies: T2DM (confirmed, E11.65 on problem list), BMI 33 (obesity, Z68.33 from current vitals), persistent ALT elevation (ALT >40 on two consecutive draws ≥3 months apart — verified via historical FHIR query), imaging-confirmed steatosis (ultrasound report from 2025-09-20 containing "hepatic steatosis" or "fatty liver"), AST:ALT ratio 0.79 (below 1, which is typical of NAFLD rather than alcoholic liver disease), and GGT if available. Each indicator is documented in a structured NASH Risk Factor Checklist within the Assessment section of the note.

5. Clinician NASH Attestation Prompt: This is the critical guardrail. Scribing.io presents the PCP with a binary decision point: "Based on available data, does this patient meet criteria for nonalcoholic steatohepatitis (NASH, K75.81), or does the diagnosis remain nonalcoholic fatty liver disease (NAFLD, K76.0)?" The clinician selects NAFLD. The system timestamps this attestation and embeds it in the note: "Clinician attests: NASH criteria not met at this time. Diagnosis: NAFLD (K76.0). Rationale: No biopsy-confirmed inflammation; no clinical features consistent with steatohepatitis beyond imaging steatosis and metabolic risk factors." This explicit negation satisfies auditor requirements and prevents downstream ambiguity.

6. 91200 ServiceRequest Generation: With FIB-4 in the intermediate zone and NASH ruled out (but advanced fibrosis not yet excluded), the system generates a FHIR R4 ServiceRequest for CPT 91200 (transient elastography). The request includes: the procedure code, linked diagnoses (K76.0, E11.65, Z68.33), a medical-necessity narrative citing FIB-4 score, metabolic risk factors, imaging history, and AASLD guidance, and a reference to the relevant MAC Local Coverage Determination (LCD) if one exists for the patient's jurisdiction. The entire packet is exportable as a one-click prior-auth submission.

7. Note Finalization and Audit Trail: The completed note contains: structured lab values with LOINC codes and draw dates, computed FIB-4 with threshold interpretation, NASH indicator checklist, clinician attestation with timestamp, ICD-10 code assignment with Excludes1 compliance verification, and the 91200 order with embedded rationale. Every element is audit-ready. Every element was missing from the original denied note.

How Scribing.io Auto-Computes FIB-4 and Documents NASH Indicators via FHIR R4

The technical architecture behind the workflow described above is not a black box. Clinicians evaluating AI scribes deserve transparency about how data moves from the EHR into the note. This section provides that transparency.

FHIR R4 Observation Retrieval

Scribing.io connects to the EHR via HL7 FHIR R4 using SMART on FHIR authorization. For FIB-4 computation, the system issues three targeted Observation queries:

• AST: GET /Observation?patient=[id]&code=1920-8&date=ge2025-07-15&_sort=-date&_count=5

• ALT: GET /Observation?patient=[id]&code=1742-6&date=ge2025-07-15&_sort=-date&_count=5

• Platelets: GET /Observation?patient=[id]&code=777-3&date=ge2025-07-15&_sort=-date&_count=5

The date=ge parameter enforces the 6-month recency window. The _sort=-date ensures the most recent result is first. The _count=5 retrieves enough results to verify persistence (e.g., confirming ALT elevation on multiple draws).

The Epic/Cerner Pagination Problem

Both Epic and Cerner (now Oracle Health) default to returning 50 Observation resources per FHIR bundle page. For patients with complex lab histories — common in metabolic liver disease patients who also have T2DM and are monitored quarterly — the AST, ALT, and platelet results from a single draw date may span multiple pages if other labs from that date are also returned. Scribing.io follows Bundle.link entries with relation: "next" until all Observations from the target date are retrieved, then filters by LOINC code locally. Without this pagination handler, the system might retrieve AST and ALT from January 15 but miss the platelet count from the same draw, producing either an incomplete FIB-4 or a computation using mismatched dates.

OCR Fallback for Scanned Outside Labs

Patients frequently present with outside lab results that exist only as scanned PDFs in the EHR — not as discrete FHIR Observations. Scribing.io's OCR fallback extracts AST, ALT, and platelet values from these documents using pattern-matched text recognition, applies confidence scoring, and presents the extracted values to the clinician for confirmation before using them in computation. The note documents these as "OCR-extracted, clinician-confirmed" with the source document reference, maintaining audit integrity.

NASH Indicator Documentation Logic

Beyond FIB-4, the system inventories the following NASH-associated indicators and documents their presence or absence in a structured checklist format:

This checklist serves two purposes: it gives the clinician a complete picture to make the NASH attestation decision, and it gives the payer reviewer a structured, audit-ready inventory of the clinical rationale underlying the K76.0 (or K75.81) assignment and the elastography order.

NAFLD-to-NASH Documentation Workflow: A Step-by-Step Decision Table

The following table maps the complete clinical decision tree for metabolic liver disease documentation in primary care. Each row represents a decision point; the right column describes Scribing.io's automated action at that point.

Payer Perspectives: Medical Necessity, DRG Assignment, and Denial Prevention

Why Payers Deny Elastography Orders

Prior-auth denials for transient elastography (CPT 91200) in the NAFLD context cluster around three documentation failures, all preventable:

1. No fibrosis risk score documented. The note says "fatty liver" but provides no computed FIB-4 or equivalent. The payer cannot verify that the patient is in an intermediate or high-risk category that warrants imaging beyond standard ultrasound.

2. No NASH attestation (positive or negative). The note is silent on whether the patient has simple steatosis or steatohepatitis. Without this distinction, the payer cannot determine whether the elastography is for routine monitoring (often denied) or for actionable risk stratification in a high-risk metabolic phenotype (typically approved).

3. Diagnosis code without metabolic context. K76.0 is linked to the order, but the note does not document the co-morbidities (T2DM, obesity, dyslipidemia) that establish the metabolic syndrome context. An isolated K76.0 without metabolic co-morbidity codes looks like incidental steatosis, not a disease state requiring advanced workup.

How Scribing.io Prevents Each Denial Pattern

For failure #1, the system auto-computes FIB-4 and embeds it in the note before the order is generated — the score is inseparable from the request. For failure #2, the mandatory NASH attestation prompt ensures every note has a timestamped clinician decision. For failure #3, the 91200 ServiceRequest automatically links to all relevant metabolic co-morbidity codes identified in the NASH indicator checklist, presenting the payer with a complete diagnostic picture.

DRG Implications for Inpatient Encounters

While most NAFLD management occurs in outpatient settings, hospitalized patients with metabolic liver disease benefit from precise code assignment. Both K76.0 and K75.81 group to DRG 441–443, but K75.81's potential CC status in certain encounter contexts can affect reimbursement. Accurate documentation — driven by the same FIB-4 and NASH attestation workflow — ensures the correct code is assigned without upcoding risk. The AMA CPT guidelines and CMS MS-DRG manual remain the authoritative references for procedure and grouping logic, respectively.

Frequently Asked Questions: K76.0, FIB-4, and AI-Assisted Documentation

Can I assign K75.81 (NASH) without a liver biopsy?

Yes. ICD-10-CM does not require histologic confirmation for K75.81 assignment. However, the clinician must explicitly attest that the patient meets NASH criteria based on clinical judgment — which may incorporate persistent transaminase elevation, metabolic risk factors, imaging findings, and elastography results suggesting fibrosis. Scribing.io facilitates this attestation but never auto-assigns K75.81 without clinician confirmation. The AASLD guidance supports non-invasive diagnosis pathways in appropriate clinical contexts.

What if the patient's FIB-4 is low (<1.3) but the PCP still wants elastography?

A low FIB-4 significantly reduces the likelihood of advanced fibrosis, and most payers will deny elastography in this scenario absent additional justification. Scribing.io will still generate the order if the clinician requests it, but the medical-necessity text will flag the low-risk FIB-4 and prompt the clinician to document additional clinical rationale — such as family history of cirrhosis, prior elastography showing progression, or discordant imaging findings — that might support approval despite the low score.

Does Scribing.io replace the coder's role in ICD-10 assignment?

No. Scribing.io generates a suggested code with supporting documentation that the clinician reviews and confirms during note sign-off. In practices with dedicated coders, the coder receives a note pre-populated with structured data (FIB-4, NASH attestation, metabolic co-morbidities, Excludes1 compliance check) that dramatically accelerates and improves the accuracy of their review. The AI does the computational and data-retrieval work; the human makes the clinical and coding decisions.

How does Scribing.io handle the NAFLD-to-MASLD nomenclature transition?

The field is transitioning from NAFLD/NASH to MASLD (metabolic dysfunction-associated steatotic liver disease) and MASH (metabolic dysfunction-associated steatohepatitis) per the 2023 multi-society consensus. As of June 2026, ICD-10-CM has not yet adopted MASLD/MASH-specific codes; K76.0 and K75.81 remain the operative codes. Scribing.io's note templates use the MASLD/MASH terminology where clinically appropriate while mapping to the current K76.0/K75.81 codes, ensuring documentation reflects current clinical language while maintaining coding accuracy. When new codes are published, the system will update mappings automatically.

What LOINC codes does Scribing.io use for FIB-4 computation?

The three lab inputs are retrieved using: AST (LOINC 1920-8), ALT (LOINC 1742-6), and Platelets (LOINC 777-3). Age is pulled from the patient demographic resource. All four values are documented in the note with their LOINC identifiers to ensure traceability and payer verifiability.

Is transient elastography always CPT 91200?

CPT 91200 is the code for liver elastography, mechanically induced shear wave (i.e., FibroScan). Other elastography modalities — such as acoustic radiation force impulse (ARFI) or MR elastography — may use different CPT codes. Scribing.io maps to 91200 specifically for FibroScan-type transient elastography. If the clinician orders a different modality, the system adjusts the CPT code and corresponding medical-necessity language accordingly.

Ready to eliminate FIB-4 documentation gaps and elastography denials in your practice? Book a 15-minute demo to see our FHIR/LOINC-driven FIB-4 autopopulation and one-click 91200 elastography prior-auth packet with MAC policy checks and audit-ready rationale embedded in your note.