What Complete HCC Code Capture Means
Complete HCC code capture means every diagnosis the medical record supports, and that the payer's risk-adjustment model recognises, is documented, coded and submitted for the period in which it was addressed. The two ways to get this wrong are symmetrical. A code the record does not support is a compliance exposure that an audit can unwind. A code the record does support, but which never reaches the claim, is revenue the agency earned and did not collect.
In home health the second failure is the more common of the two, and it is rarely a coding-knowledge problem. It is a workflow problem: the diagnosis was there in the visit notes and never travelled the rest of the way to the claim. This guide covers where codes are lost between the visit note and the submitted claim, how to read a record the way a coder must, and the pre-submission review that closes the gap.
The Four Points Where HCC Codes Are Lost
A diagnosis can disappear at four distinct points in an episode, and each one has a different fix. Knowing which point is costing you the most matters, because tightening the wrong step changes nothing.
The first is the note itself. A chronic condition the clinician actively managed during the visit, whether that meant adjusting a medication, reviewing labs or evaluating a symptom, may never be written as an assessment. Nothing downstream can recover a diagnosis that is not in the record at all, and no coder or query process can invent it.
The second is specificity. A condition can be documented and still not be codeable to the level the classification system requires. Diabetes cannot become a code until the type and any associated manifestation are stated. Heart failure needs its type and acuity. The condition is on the page, but it stops short of the detail that determines which code, and sometimes which payment group, applies.
The third is the source of the diagnosis list. When that list is seeded from the referral, the hospital discharge summary, the previous episode's claim or a document template, the coder never sees the conditions the current period's visit notes actually support. This is the least visible of the four losses, because the claim looks complete and internally consistent.
The fourth is post-coding loss. Claim scrubbing edits, payer-specific edits and well-intentioned cleanup can strip a diagnosis that was correctly coded because it looked unfamiliar or made the claim appear to be an outlier against the agency's own history.
Read the Record the Way a Coder Must
HCC capture begins with specificity. A condition is only fully captured when the record states what the coder needs in order to select the most specific code available: type or cause, site, laterality, severity or stage, and current status. A non-healing wound and a diabetic foot ulcer of the right heel with a documented stage are not two ways of writing the same thing. They are different codes, and often the difference between a condition the risk model recognises and one it does not.
Linkage carries as much weight as detail. A complication has to be documented as a complication of the condition it belongs to. When neuropathy and diabetes sit in the record as two unrelated entries, a coder cannot build the combination code without assuming a relationship the documentation does not establish.
Status language decides whether the condition belongs on the claim at all. In home health, a diagnosis documented as suspected, probable or ruled out is not coded as though it were confirmed, and the record is coded to the highest degree of certainty it actually establishes. Equally, a condition that appears only in the past medical history line, with no assessment, monitoring or management during the period, is the classic weak capture: present in the record, absent from the episode's care.
That last distinction is the heart of complete capture. Code capture follows clinical attention. What the team assessed, evaluated, monitored or treated during the 30-day period is what that period's coding should reflect, and what the documentation has to show is the attention, not the label.
The Home Health Wrinkles: OASIS, PDGM and the Claim
Home health adds two layers that most risk-adjustment workflows never have to handle: the OASIS assessment and the 30-day payment period.
The primary diagnosis drives PDGM clinical grouping and the period's case-mix weight, so a more specific primary diagnosis can move an episode into a different group entirely. Secondary diagnoses feed the PDGM comorbidity adjustment and, on Medicare Advantage and other risk-bearing contracts, they are the raw material for HCC capture. The same missing diagnosis therefore has two different consequences: under traditional Medicare it can cost a comorbidity adjustment, and under Medicare Advantage it can understate the patient's risk score.
The practical trap is that the diagnosis list on the claim is built from the OASIS data set rather than from the visit notes directly. A condition the clinician documented and managed during the period, but which never made it into the assessment's diagnosis items, tends not to reach the claim at all. Reconciliation between the visit notes and the OASIS diagnosis items is where most recoverable capture actually sits.
The second wrinkle is period discipline. Capture is per period, not per patient. Carrying a diagnosis forward because it was coded last period is not the same as capturing what this period's documentation supports, and it is exactly the pattern that makes a claim hard to defend later. Each period's coding should stand on that period's record.
A Pre-Submission Capture Review That Holds Up
A capture review that pays for itself is short, sequenced and finished before the claim is submitted. Start the diagnosis list from the visit notes for the period rather than from the previous claim, and pull every condition the clinicians documented as assessed, monitored, evaluated or treated across those 30 days. That candidate list, not last period's claim, is the starting point.
Reconcile the candidate list against the OASIS diagnosis items, the referral and any discharge summary. Anything on the candidate list that is missing from the assessment is either a documentation gap or an assessment gap, and both are worth knowing before the claim leaves the building.
Then check specificity on every candidate. Where the record establishes the condition but not the detail the code requires, the answer is a focused query back to the clinician who saw the patient, never an assumption by the coder. Let the coder sequence and validate the final list as well, because the primary diagnosis decision is a medical coding decision and should be made against the record rather than inherited from a template.
Finally, close the loop inside the same period. When a query comes back answered, the answer is coded into that period's record and captured on that period's claim. Queries that resolve after submission recover the documentation, but they rarely recover the payment it would have supported.
Queries Beat Assumptions
The queries that recover the most capture are the ones a clinician can answer from what they already know, phrased so that no is a valid answer. Is the peripheral neuropathy diabetic in origin? Is the COPD being actively managed this period or is it historical? Was the patient still on the anticoagulant during these visits? Each of those turns a vague entry into a codeable one without asking anyone to add a diagnosis the record does not support, which is the line a capture programme must never cross.
The inverse is worth saying plainly, because it is the risk that follows every conversation about capture: a diagnosis added to the claim because it was likely present, or because it appeared in a prior year, is not capture. It is a compliance finding waiting to be written.
Measuring Capture With Your Own Numbers
Most agencies cannot measure capture because they only look at the claim, and the claim shows what was submitted rather than what was available to submit. The measures that make the gap visible are all internal: average secondary diagnoses reported per 30-day period; the share of episodes where a condition documented in the visit notes never appears on the claim; query volume and, more usefully, query response time; and the movement in risk scores for a fixed cohort of patients between two periods.
Trended against your own history, those four numbers tell you whether capture is improving and whether a change you made had any effect. Compared against a published industry benchmark they tell you very little, because the agencies behind that benchmark do not share your payer mix, referral sources, geography or documentation system.
Where a Second Review Earns Its Cost
Capture reviews come in two kinds and they are not interchangeable. A pre-submission review protects the claim: it checks specificity, linkage and sequencing before submission, and it is judged by denied claims avoided. A risk-adjustment review works the same record from the other direction, asking which conditions that shape the patient's risk profile are documented and codeable but absent from the claim, and it is judged by risk score accuracy.
Agencies in Medicare Advantage or other value-based contracts generally need both, because a diagnosis that is defensible for payment and a diagnosis that is accurate for risk are the same diagnosis examined through different questions. Buying either one alone leaves half the gap open, and both sit inside the same revenue cycle management workflow as eligibility, submission and follow-up.
Start With One Closed Period
Pick a single closed 30-day period, pull the visit notes, and compare the conditions the team documented against the diagnosis list that reached the claim. Count the ones you can defend from the record as it stands. That number is your capture gap for the period, and it is the only baseline you need before deciding whether to tighten the process internally or hand it to a partner.
It also tells you which of the four loss points is costing the most, because the fix for a missing assessment line, a missing detail, a wrong source list and a scrubber edit are four different fixes. That comparison is where our risk adjustment coding and clinical documentation review work starts on every engagement, and it can be produced from records you already have.