Home Health PDGM Coding Audits: Diagnosis Errors That MACs and UPICs Target First

Under the Patient-Driven Groupings Model (PDGM), diagnosis codes do much more than describe a patient. The principal diagnosis on the claim assigns the 30-day period to one of 12 clinical groups, and secondary diagnoses determine whether the period receives a comorbidity adjustment. Together with admission source, timing, and the functional impairment level from OASIS, these factors place each period into one of 432 case-mix groups that set payment.

Because coding drives payment, it is a natural target for Medicare reviewers. Medicare Administrative Contractors (MACs) review coding as part of medical review, and Unified Program Integrity Contractors (UPICs) use data analysis to identify agencies whose coding patterns differ from peers. CMS has also built assumptions about coding behavior into PDGM rate setting, including assumptions related to clinical group coding and comorbidity coding.

This article explains how diagnosis coding affects PDGM payment, the coding errors reviewers target first, the data patterns that attract attention, how to audit coding, and how to fix common problems. It is educational and does not replace legal or professional advice.

How Diagnosis Coding Drives PDGM Payment

The Principal Diagnosis and Clinical Groups

The principal diagnosis reported on the claim assigns the period to a clinical group. The 12 PDGM clinical groups are:

• Musculoskeletal Rehabilitation.

• Neuro Rehabilitation.

• Wounds.

• Complex Nursing Interventions.

• Behavioral Health.

• Medication Management, Teaching, and Assessment (MMTA), divided into Surgical Aftercare; Cardiac and Circulatory; Endocrine; Gastrointestinal Tract and Genitourinary System; Infectious Disease, Neoplasms, and Blood-Forming Diseases; Respiratory; and Other.

Different clinical groups carry different case-mix weights, so the choice of principal diagnosis can significantly affect payment.

Codes That Cannot Be Grouped

CMS has identified diagnosis codes that are not accepted as the principal diagnosis under PDGM, such as many symptom codes, manifestation codes that must follow an underlying condition, and codes considered too vague to support home health. Claims with these codes as the principal diagnosis are returned to the provider. CMS publishes the PDGM grouping tool and code lists on its PDGM web page, and updates them as ICD-10-CM codes change.

Secondary Diagnoses and the Comorbidity Adjustment

Secondary diagnoses on the claim can qualify a period for a comorbidity adjustment:

• Low comorbidity adjustment: at least one secondary diagnosis in a comorbidity subgroup associated with higher resource use.

• High comorbidity adjustment: at least two secondary diagnoses in subgroups that interact to increase resource use.

CMS recalibrates comorbidity subgroups annually. The CY 2026 final rule updated them using CY 2024 data.

The Coding Behavior Assumptions

When CMS implemented PDGM, it assumed agencies would change coding behavior, including coding to place periods in higher-paying clinical groups and coding more comorbidities. CMS has used these behavior assumptions in its payment adjustments over several years. This history shows that CMS is closely watching coding patterns.

The Rules That Govern Home Health Coding

• ICD-10-CM Official Guidelines for Coding and Reporting, published by CMS and the National Center for Health Statistics.

• The OASIS Guidance Manual, which addresses diagnosis reporting on OASIS.

• CMS PDGM guidance and code lists.

• Medicare Benefit Policy Manual, Chapter 7.

• Medicare Claims Processing Manual, Chapter 10, for home health billing.

A basic principle of the Official Guidelines is that diagnosis codes must be based on documentation from the patient's provider, meaning the physician or other qualified practitioner legally accountable for establishing the diagnosis. Home health clinicians document assessments and observations, but they cannot establish new diagnoses for coding purposes.

Diagnosis Errors That Reviewers Target First

1. Principal Diagnosis Not Supported by the Practitioner's Documentation

The principal diagnosis is not documented by the referring or certifying practitioner, or the practitioner's records do not support that it is the main reason for home health. Reviewers may change the principal diagnosis, which can move the period to a different clinical group.

2. Principal Diagnosis Chosen for Payment Rather Than Focus of Care

The principal diagnosis should reflect the condition most related to the current plan of care. Selecting a code that places the period in a higher-paying clinical group when the plan of care focuses on a different condition is a common target.

Example: a patient referred for skilled nursing to manage newly diagnosed diabetes is coded with a neurological condition as principal, placing the period in Neuro Rehabilitation, even though no therapy is ordered and the plan of care focuses on diabetes teaching.

3. Comorbidity Inflation

Secondary diagnoses are coded that are not documented by the practitioner, are no longer active, or do not affect the patient's care. Coding additional conditions solely to obtain a comorbidity adjustment is a significant risk.

Examples:

• Coding a history of a condition as if it were current.

• Coding conditions found only in a medication list without practitioner documentation of the diagnosis.

• Coding resolved conditions from old hospital records.

4. Coding From Nursing or Therapy Notes Alone

A nurse observes symptoms suggesting a condition, and the coder assigns a diagnosis code without practitioner confirmation. Under the Official Guidelines, coders should query the practitioner rather than code from nursing or therapy assessments alone.

5. Aftercare and Injury Coding Errors

Injury and aftercare coding is complex. Common errors include:

• Using the wrong seventh character for injury codes. For most injuries, the subsequent encounter character is used when the patient is receiving routine care during the healing or recovery phase.

• Using aftercare Z codes when an injury code with the appropriate seventh character is required.

• Using surgical aftercare codes inconsistently with the procedure documented.

These errors can change the clinical group or cause the claim to be returned.

6. Sequencing Errors

Etiology and manifestation conventions, "code first" and "use additional code" instructions, and combination codes must be followed. Incorrect sequencing can make a claim ungroupable or change the clinical group.

7. Wound Coding Errors

Wound coding affects placement in the Wounds clinical group. Common errors include:

• Coding pressure ulcer stages inconsistent with documentation.

• Coding surgical wounds as pressure ulcers or vice versa.

• Coding wounds that have healed.

• Inconsistency between OASIS wound items and diagnosis codes.

8. Inconsistency Between the Claim, OASIS, and the Record

Diagnoses on the claim, OASIS, the plan of care, and the practitioner's records should be consistent. Differences signal possible errors.

9. Unspecified or Vague Codes When Specific Documentation Exists

Using unspecified codes when the record supports more specific codes can cause claims to be returned or grouped incorrectly. Conversely, assigning more specific codes than the documentation supports is also an error.

10. Copied Diagnoses Across Periods

Diagnoses are copied from one period to the next without reassessing whether they remain active and relevant to the plan of care.

Data Patterns That Attract MAC and UPIC Attention

Reviewers use data to identify agencies for review. Patterns that may draw attention include:

• A much higher share of periods in high-paying clinical groups than peers.

• A much higher share of periods with comorbidity adjustments than peers.

• Sudden shifts in coding patterns, such as a rapid increase in comorbidity adjustments.

• The same diagnosis combinations across many patients.

• Clinical groups that do not match the disciplines provided, such as many periods in rehabilitation groups with little therapy.

• Coding patterns that differ sharply from those of the referring practitioners.

UPICs may combine these patterns with other indicators, such as referral patterns, beneficiary interviews, and enrollment information. A UPIC review may involve statistical sampling and extrapolation, and in serious cases, referral to law enforcement.

How Coding Errors Become Overpayments

When a reviewer finds that the principal diagnosis or secondary diagnoses are not supported:

• The reviewer may recode the claim, which can change the clinical group and comorbidity adjustment.

• The payment difference becomes an overpayment.

• If coding errors are widespread, contractors may use statistical sampling and extrapolation.

• Persistent or intentional coding manipulation can lead to False Claims Act liability.

Running a PDGM Coding Audit

Step 1: Select the Sample

Focus on high-risk areas, such as high-paying clinical groups, periods with high comorbidity adjustments, new coders, or outlier data. Use a probe sample for discovery or a statistically valid random sample if results will be used to estimate overpayments.

Step 2: Gather the Records

Include the referral, the practitioner's records, hospital or facility records, the plan of care, OASIS, and visit notes.

Step 3: Recode Independently

Have a qualified coder recode the period from the documentation without looking at the original codes.

Step 4: Compare and Classify Errors

Compare the independent coding with the billed codes. Classify differences by type, such as principal diagnosis, comorbidity, sequencing, aftercare, wound, or specificity.

Step 5: Determine Payment Impact

Use the PDGM grouper to determine whether the clinical group, comorbidity adjustment, and HIPPS code would change, and calculate the payment difference.

Step 6: Analyze Root Causes

Root causes may include incomplete referral documentation, lack of practitioner queries, coder training gaps, pressure to code for payment, EHR or coding tool defaults, or copying diagnoses between periods.

Step 7: Correct and Educate

Provide coder education, improve referral documentation workflows, establish practitioner query processes, and revise coding policies.

Step 8: Address Overpayments

Evaluate obligations under the 60-day overpayment rule at 42 CFR 401.305, and work with counsel on investigation and repayment when needed.

Why Coding Problems Often Start at Intake

Many coding errors begin before a coder sees the chart. Referrals often arrive with a brief order and a diagnosis list but without the practitioner's supporting notes. Coders then rely on incomplete information, medication lists, or nursing assessments. Strengthening intake, by requesting full practitioner and facility records at referral, prevents many of the errors described above.

Building a Compliant Coding Process

• Obtain complete practitioner and facility records at referral.

• Code from practitioner documentation, not nursing observations alone.

• Query practitioners when documentation is unclear, incomplete, or conflicting, using compliant, non-leading queries.

• Select the principal diagnosis based on the focus of the plan of care.

• Code only secondary diagnoses that are documented by the practitioner and affect care, treatment, or management.

• Reassess diagnoses each period.

• Keep codes consistent across the claim, OASIS, and plan of care.

• Use qualified, credentialed coders, and provide ongoing education.

• Monitor coding patterns against peers.

• Separate coding decisions from financial pressure, and make clear that coders should code accurately, not for payment.

Practitioner Query Best Practices

A compliant query process helps coders resolve unclear documentation without leading the practitioner.

• Ask open-ended or multiple-choice questions based on clinical indicators in the record.

• Include an option such as "other" or "unable to determine."

• Do not suggest a diagnosis solely because it would increase payment.

• Document the query and the practitioner's response in the record.

• Do not code the condition until the practitioner documents it.

• Track query response times and follow up.

Coding and the Plan of Care

The plan of care should reflect the diagnoses coded on the claim. Under 42 CFR 409.43, the plan of care must include required elements, including all pertinent diagnoses. When the principal diagnosis on the claim does not match the focus of the plan of care, reviewers may question the coding. Agencies should:

• Make sure the plan of care's interventions and goals relate to the principal diagnosis.

• List secondary diagnoses that affect care, and make sure interventions address them where appropriate.

• Update the plan of care and coding when the patient's condition changes.

Coding Review Checklist

• Is the principal diagnosis documented by the practitioner?

• Is it the condition most related to the current plan of care?

• Is it an acceptable principal diagnosis under PDGM?

• Are injury codes using the correct seventh character?

• Are etiology and manifestation sequencing rules followed?

• Is each secondary diagnosis documented by the practitioner, active, and relevant to care?

• Are wound codes consistent with wound documentation and OASIS?

• Are codes as specific as the documentation supports, and no more?

• Are diagnoses consistent across the claim, OASIS, and plan of care?

• Were diagnoses reassessed for this period rather than copied?

Annual Code and Grouper Updates

ICD-10-CM codes are updated each fiscal year, and CMS updates PDGM code lists, comorbidity subgroups, and case-mix weights through annual rulemaking. The CY 2026 final rule recalibrated case-mix weights and comorbidity subgroups using CY 2024 data, and CMS released the CY 2027 final rule in November 2026. Agencies should:

• Update coding tools and grouper software on time.

• Train coders on new, revised, and deleted codes.

• Review changes in clinical group assignments and comorbidity subgroups.

• Check that claims for periods spanning update dates use the correct code set.

Outsourced Coding

Many agencies use outside coding vendors. Agencies remain responsible for the claims they submit.

• Review vendors' qualifications and quality processes.

• Audit vendor coding regularly.

• Make sure contracts require compliance with the Official Guidelines and CMS rules.

• Avoid compensation arrangements that reward vendors for increasing case-mix.

• Include a business associate agreement under HIPAA.

Responding to a Coding-Focused ADR

When a MAC or UPIC requests records for a coding review:

• Submit the full record that supports the codes, including the referral, practitioner records, hospital or facility records, plan of care, OASIS, and visit notes.

• Make sure practitioner documentation of each coded diagnosis is easy to find.

• Include query responses that support coded conditions.

• Respond on time: generally 45 days for MAC reviews and 30 days for UPIC reviews.

• Keep a copy of everything submitted and proof of submission.

• If findings recode the claims, review the recoding carefully and appeal where the record supports the original codes.

Appealing Coding Denials

• Identify where in the record each disputed diagnosis is documented by the practitioner.

• Explain why the principal diagnosis is the condition most related to the plan of care.

• Address Official Guidelines conventions directly when sequencing or seventh-character issues are disputed.

• Submit all evidence by the reconsideration level.

• If extrapolation was used, consider engaging a statistician to evaluate the sampling methodology.

Metrics to Monitor

• Distribution of periods by clinical group compared with peers.

• Share of periods with low and high comorbidity adjustments.

• Rate of returned claims for unacceptable principal diagnoses.

• Coding audit accuracy rates by coder.

• Number and type of practitioner queries.

• Changes in coding patterns over time.

A Hypothetical Example

A UPIC's data analysis shows that an agency's share of periods with a high comorbidity adjustment is far above the state average. The UPIC requests records for a sample of periods. Independent recoding finds that many secondary diagnoses came from medication lists and old hospital records without current practitioner documentation, and several principal diagnoses did not match the focus of the plan of care.

The UPIC recodes the sampled claims and extrapolates an overpayment. The agency, now working with counsel, appeals some findings where documentation was available but not submitted, and engages a statistician to evaluate the sampling. It also retrains coders, implements a practitioner query process, and begins monthly coding audits. Its comorbidity rates move closer to peers, and later reviews show improved accuracy.

Key Takeaways

• The principal diagnosis drives the clinical group, and secondary diagnoses drive the comorbidity adjustment, so coding accuracy directly affects payment.

• Code from practitioner documentation and query when documentation is unclear.

• Choose the principal diagnosis based on the focus of the plan of care, not payment.

• Code only active secondary diagnoses that affect care.

• Follow Official Guidelines conventions for sequencing, aftercare, and seventh characters.

• Keep the claim, OASIS, and plan of care consistent.

• Monitor coding patterns against peers, audit regularly, and address overpayments promptly.

Frequently Asked Questions

How does the principal diagnosis affect PDGM payment?

It assigns the period to one of 12 clinical groups, which carry different case-mix weights.

What is the comorbidity adjustment?

An adjustment based on secondary diagnoses in subgroups associated with higher resource use. A low adjustment requires at least one qualifying diagnosis, and a high adjustment requires at least two interacting diagnoses.

Can home health coders code from nursing notes?

Diagnosis codes should be based on documentation from the practitioner responsible for establishing the diagnosis. Coders should query the practitioner when nursing observations suggest an undocumented condition.

Why are some principal diagnoses not accepted?

CMS has identified codes, such as many symptom and manifestation codes, that do not support assignment to a clinical group as the principal diagnosis. Claims with these codes are returned.

What patterns trigger UPIC reviews?

Patterns such as unusually high shares of high-paying clinical groups or comorbidity adjustments, sudden coding changes, and repeated diagnosis combinations can draw attention.

Conclusion

Diagnosis coding is central to PDGM payment, and coding errors are among the first issues MACs and UPICs examine. The most common problems include unsupported principal diagnoses, principal diagnoses chosen for payment rather than the focus of care, comorbidity inflation, coding from nursing notes, aftercare and wound errors, and inconsistencies across records. By coding from practitioner documentation, querying when needed, reassessing diagnoses each period, auditing regularly, monitoring patterns against peers, and keeping coding separate from financial pressure, home health agencies can reduce audit risk and make sure their payments reflect the care they provide.

How HealthBridge Can Help

HealthBridge is a healthcare management consulting firm that helps home health agencies improve PDGM coding accuracy through coding audits, independent recoding, coder education, practitioner query processes, data pattern analysis, and vendor oversight. If you want to make sure your coding will stand up to MAC and UPIC review, please get in touch with us. We will be happy to help you with this topic.

References

• Centers for Medicare & Medicaid Services. Home Health Patient-Driven Groupings Model. https://www.cms.gov/medicare/payment/prospective-payment-systems/home-health/home-health-patient-driven-groupings-model

• Centers for Medicare & Medicaid Services. Calendar Year (CY) 2026 Home Health Prospective Payment System Final Rule (CMS-1828-F) Fact Sheet. https://www.cms.gov/newsroom/fact-sheets/calendar-year-cy-2026-home-health-prospective-payment-system-final-rule-cms-1828-f

• U.S. Government Publishing Office. Federal Register, Medicare and Medicaid Programs; Calendar Year 2026 Home Health Prospective Payment System Rate Update (December 2, 2025). https://www.govinfo.gov/content/pkg/FR-2025-12-02/html/2025-21767.htm

• Centers for Medicare & Medicaid Services. CY 2027 Home Health Prospective Payment System Regulations and Notices. https://www.cms.gov/medicare/payment/prospective-payment-systems/home-health/home-health-prospective-payment-system-regulations-and-notices/cms-1844-p

• Centers for Disease Control and Prevention, National Center for Health Statistics. ICD-10-CM and Official Guidelines for Coding and Reporting. https://www.cdc.gov/nchs/icd/icd-10-cm/index.html

• Centers for Medicare & Medicaid Services. ICD-10. https://www.cms.gov/medicare/coding-billing/icd-10-codes

• Centers for Medicare & Medicaid Services. OASIS Data Sets and OASIS Guidance Manual. https://www.cms.gov/medicare/quality/home-health/oasis-data-sets

• Centers for Medicare & Medicaid Services. Medicare Benefit Policy Manual, Chapter 7, Home Health Services. https://www.cms.gov/Regulations-and-Guidance/Guidance/Manuals/downloads/bp102c07.pdf

• Centers for Medicare & Medicaid Services. Medicare Claims Processing Manual, Chapter 10, Home Health Agency Billing. https://www.cms.gov/regulations-and-guidance/guidance/manuals/downloads/clm104c10.pdf

• Centers for Medicare & Medicaid Services. Medicare Program Integrity Manual, Chapter 4, Program Integrity. https://www.cms.gov/regulations-and-guidance/guidance/manuals/downloads/pim83c04.pdf

• Electronic Code of Federal Regulations. 42 CFR 401.305, Requirements for Reporting and Returning of Overpayments. https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-A/part-401/subpart-D/section-401.305