Insufficient Documentation Is the Single Largest Home Health Payment Error
Insufficient documentation is 49.4% of home health improper payments. The facts that support those claims are usually established on a phone call.
Insufficient Documentation Is the Single Largest Home Health Payment Error
When a home health claim is found improper, the most likely reason is not that the care was unnecessary, not that the coding was wrong, and not that documentation was missing entirely. It is that documentation existed and was not sufficient to support the claim.
That is not an impression. It is the largest single line in CMS's own error breakdown, and it points at a specific operational problem: the facts that support a home health claim are frequently established in a conversation, and conversations are the least reliably documented thing an agency does.
The numbers
CMS measures Medicare fee-for-service payment accuracy through the Comprehensive Error Rate Testing program and publishes a detailed annual breakdown. The current edition is the 2025 Medicare Fee-for-Service Supplemental Improper Payment Data, released January 2026, covering claims submitted between July 1, 2023 and June 30, 2024.
For home health, Table D3 reports $1,070,619,633 in projected improper payments at a 6.9 percent improper payment rate, with a 95 percent confidence interval of 5.3 to 8.5 percent. The error mix for that billion dollars:
| Error type | Share of home health improper payments | Approximate dollars |
|---|---|---|
| Insufficient documentation | 49.4% | $0.5B |
| Medical necessity | 30.4% | $0.3B |
| No documentation | 11.8% | $0.1B |
| Other | 4.8% | $0.1B |
| Incorrect coding | 3.6% | — |
Nearly half. And note that "no documentation" — the failure everyone pictures when they hear the phrase documentation error — is the smaller problem by a factor of four.
The pattern is not unique to home health. Across all of Medicare FFS, the FY2025 improper payment rate was 6.55 percent, or $28.83 billion, down from 7.66 percent and $31.70 billion the prior year. Insufficient documentation accounts for 53.0 percent of that total, roughly $15.3 billion.
One precision note before you quote any of this
The same report contains a 6.3 percent figure that is frequently misattributed to home health. It belongs to the "Home Health & Hospice" claim type in Table A7 — an aggregate covering $44.5 billion in payments and $2.8 billion in improper payments across both service lines. Home health alone is 6.9 percent on $1.07 billion. Pairing "6.3 percent" with "$1.1 billion" mixes two denominators. If you are putting a number in a board deck, use $1.07 billion at 6.9 percent for home health, or $2.8 billion at 6.3 percent for home health and hospice together.
What "insufficient" actually means
The distinction that matters most is the one between the top two rows of that table. Medical necessity is its own error category, at 30.4 percent. So an insufficient documentation error is not a determination that the care was unnecessary. It is a determination that the record, as submitted, did not let the reviewer confirm that the requirements were met.
The care may have been perfectly appropriate. The patient may have been unambiguously homebound. The physician may have had a genuine face-to-face encounter. If the record does not demonstrate it, the payment is improper anyway.
HHS OIG's ongoing work on home health compliance with Medicare requirements names the recurring deficiency types directly: face-to-face encounter requirements, plan of care requirements, unsupported codes, and beneficiaries who were not homebound or did not require skilled services.
Look at that list as an operator rather than a coder. Each item is a fact about the patient's circumstances — not a clinical measurement taken in the home, but a status that gets established through conversation. Whether the patient can leave the home without considerable and taxing effort. Whether the physician encounter happened and what it addressed. What the household situation actually is. These are things someone finds out by talking to a referral source, a physician's office, a patient, or a family member.
The gap between where the fact is learned and where it is recorded
Trace a typical intake. A discharge planner calls at 4:40 on a Friday. The intake coordinator takes the referral, asks about mobility, hears "she can't really get out on her own, her daughter has to drive her and it wears her out," confirms the physician, notes the diagnosis, and enters what fits into the referral form.
That sentence about mobility is the homebound determination in raw form. It came in over the phone, from someone who will not remember saying it, and if it does not make it into the chart in usable language, the only durable record of it is a recording nobody will ever find.
Multiply that across every referral. The claim is built weeks later from a chart, and the chart contains what the coordinator had time to type — not what the coordinator was told. The delta between those two things is where insufficient documentation errors live.
The same dynamic runs through the whole episode. Changes in condition, refused visits, family concerns, and physician instructions all arrive by phone. Some are documented well. Some are documented in a free-text field. Some are documented nowhere, because the person who took the call resolved the issue and moved on.
What to do this week
You cannot fix an agency-wide documentation gap in a week. You can find out how big yours is.
- Pull ten recent referrals and trace one fact each. Pick homebound status. For each, find where in the chart the supporting detail lives and how specific it is. "Homebound: yes" is a checkbox, not evidence. If eight of ten are checkboxes, you have your answer.
- Read the free-text fields. Every intake system has one, and it is where the good information goes to die. If your coordinators are recording meaningful clinical and eligibility detail in a notes box that never reaches the claim record, that is a routing problem you can fix with a form change.
- Time-box the face-to-face check. Not whether you have the document, but whether the documentation supports the clinical condition for which home health was ordered. Agencies usually discover the form is present and thin.
- Ask your intake team what they get told that they have nowhere to put. They will have a list. It is usually short, specific, and immediately actionable.
A hypothetical, for the shape of it
An agency with clean survey history and no billing complaints reviews twenty of its own charts against the OIG deficiency list. Nineteen have a face-to-face document on file. Eleven of those documents reference the encounter and the diagnosis but do not connect the clinical findings to the need for skilled care in the home. No care was inappropriate. No one did anything wrong. On an external review, a meaningful share of those eleven would be scored insufficient.
Nothing about that agency looks like a compliance problem from the inside. That is precisely why the category is the largest one in the table.
Making the conversation part of the record
The structural fix is to stop treating phone conversations as ephemeral. If the facts that support your claims are established on calls, then calls need to be retrievable, searchable, and reviewable — the same way a visit note is. This is the documentation half of an audit-ready contact center, and it is the half that gets skipped because the compliance conversation usually starts and ends with recording and retention.
It also connects to a broader point about what agencies already possess and do not use: the information needed to support most of these claims was already spoken aloud, on a line you may already be recording. The problem is retrieval, not capture. That is the same reason most call archives are storage rather than evidence.
Where tooling fits
Turning conversations into usable documentation is a search and structure problem: transcription, categorization, and the ability to locate the call where a specific fact was established without listening to a hundred others. SurfacerIQ works in that category of call intelligence for healthcare contact centers. Whatever you evaluate, the capability that matters for this particular problem is retrieval — whether you can find the conversation that supports a claim, months later, on demand.
This post is general information for operators, not legal advice; documentation sufficiency for any specific claim is a determination your compliance counsel and clinical leadership should make.
If you want to work through where your intake conversations are and are not making it into the record, talk to us.
Keep exploring
Related resources
Compliance monitoring →
Flag policy and regulatory issues on every call, not a sample.
Home care & healthcare →
Purpose-built for regulated call handling in home care and health.
HIPAA at SurfacerIQ →
How SurfacerIQ handles PHI, BAAs, and audit trails.
More from Compliance & HIPAA
- The 2026 Fraud Takedown Included a Hospice Kickback Scheme. It Ran on Conversations.7 min
- Your Compliance Program Needs an Audit Trail Built From Conversations7 min
- A Business Associate Was Penalized $10,000 for a 15-Million-Record Breach. Read the Field List.7 min
See SurfacerIQ in action
Calls in. Tickets out. Automatically. See how it works on a real call.
Keep reading
The 2026 Fraud Takedown Included a Hospice Kickback Scheme. It Ran on Conversations.
The 2026 National Health Care Fraud Takedown charged 455 defendants in $6.5B of alleged fraud, including a $27M hospice case. Hospice fraud is structurally a conversation problem.
Your Compliance Program Needs an Audit Trail Built From Conversations
42 CFR 484.50(e) requires you to document that a complaint existed and how it ended. Neither it nor 484.110 requires call recording. Here is what a real audit trail contains.
A Business Associate Was Penalized $10,000 for a 15-Million-Record Breach. Read the Field List.
OCR's MMG Fusion settlement exposed 15 million records and cost the vendor $10,000. For agencies, the field list and the notification clause matter far more than the fine.