CRC (Other exams) common mistakes
CRC · Understand the traps and how to avoid them.
Coding diagnoses from the encounter's problem list, superbill, EHR 'active diagnosis' banner, or a vendor-supplied suspect/HCC list instead of reading the body of the provider note
Problem lists and suspect lists look authoritative, are pre-populated with the conditions that map to HCCs, and save time under exam or production pressure; candidates assume anything on the list was 'documented' for the encounter
What to do instead
Adopt a strict abstraction order: read the chief complaint, history, exam, and Assessment/Plan in full, then code only conditions the provider documents in that encounter. Treat problem lists, billing summaries, and suspect lists as prompts to go looking in the note, never as the source of an HCC
Abstracting only the primary/first-listed diagnosis on a claim and ignoring secondary chronic conditions the provider also documented and managed during the same encounter
Habits carried over from outpatient claim coding, where the principal diagnosis drives the claim, lead candidates to assume only that one code 'matters'
What to do instead
Remember risk adjustment captures every documented, MEAT-supported condition on the claim, not just the first-listed one; before finalizing, re-scan the Assessment/Plan for each additional chronic condition and confirm it meets capture criteria
Assuming a chronic condition documented and coded last year automatically carries forward, and treating the current-year submission as optional once the condition has been captured once
The phrase 'chronic condition' implies permanence, and candidates confuse the clinical concept of a chronic disease with the annual payment requirement of the risk adjustment model
What to do instead
Anchor on the recapture rule: a diagnosis must be supported by a face-to-face encounter in the applicable data-collection/look-back period for the payment year being submitted. Build the mental checklist 'documented AND dated within the collection period AND MEAT-supported' for every chronic condition
Applying CMS-HCC condition-category and hierarchy rules to an HHS-HCC (commercial/marketplace) scenario, or assuming a code that maps to an HCC in one model version maps identically in another
Study material is dominated by the Medicare model, so candidates over-generalize it; model versions and populations differ in condition categories, hierarchies, and included diagnoses
What to do instead
First identify which model and version the question is testing (Medicare Advantage CMS-HCC vs. commercial/Medicaid HHS-HCC), then apply that model's categories. When a question cites a model version, use that version's mapping rather than memory of another year's model
Coding a condition the note describes as 'history of,' 'resolved,' 's/p,' or 'no longer on treatment' as an active condition because the disease name appears in the text
'History of MI' or 'history of stroke' contains the same disease words as an active diagnosis, and copy-forward problem lists blur the distinction between past and current status
What to do instead
Look for language of current activity (treatment, monitoring, worsening, ongoing management) tied to the condition. If only historical language is present, do not assign an active diagnosis code; note that history Z codes generally do not risk-adjust and document the finding as not captured
Reporting multiple codes within the same disease family as if each adds separately, e.g., coding both 'diabetes with complication' and 'diabetes without complications,' or both a stroke/hemiplegia pairing and its less severe counterpart, when the hierarchy truncates the family
Candidates equate 'more codes documented' with 'more HCCs captured' and don't check whether the model assigns one condition category per family with a severity hierarchy
What to do instead
Before adding a code, ask whether the family hierarchy is already satisfied by a more severe category. Code the most specific, highest-severity presentation the documentation supports, and drop the redundant lower-severity code for scoring purposes while still coding what is clinically documented per ICD-10-CM rules
Coding an etiology and its manifestation as unrelated standalone codes, or omitting the manifestation entirely, instead of using the combination code or required sequencing (e.g., diabetes with CKD, hypertensive heart disease with heart failure)
Candidates code each condition independently as it appears in the note and stop at the first tabular option without checking the ICD-10-CM Index and the 'code also'/'use additional code' notes
What to do instead
Look up the systemic condition in the Index and follow the manifestation pointer; check tabular notes for combination codes. Verify that the provider linked the two conditions in the documentation before combining them - do not assume linkage that the provider did not state
Assigning a confirmed diagnosis code for a condition documented as 'probable,' 'suspected,' 'possible,' 'rule out,' or 'working diagnosis'
The condition often does map to an HCC, so it feels like a capture opportunity, and inpatient habits (where uncertain diagnoses may be coded) bleed into the outpatient risk adjustment setting
What to do instead
Apply the outpatient guideline: do not code uncertain diagnoses as confirmed. Record the condition on the suspect/follow-up list only, and do not report it on the claim until the provider documents it as confirmed
Sourcing a diagnosis from a provider attestation, a query response alone, or a leading query that suggests the desired diagnosis or HCC, and accepting late addenda written to create documentation after the encounter
Capture pressure and productivity targets make the query feel like a legitimate shortcut, and candidates underestimate how narrowly program integrity rules define acceptable documentation
What to do instead
Query only to clarify or resolve inconsistencies in existing documentation, keep queries non-leading (open-ended, with clinical indicators cited), and remember that under RADV the medical record itself must support each submitted diagnosis; a query response or attestation that is not supported by the record is not acceptable validation
Inferring a diagnosis from a symptom, lab value, or medication list - coding diabetes from an elevated glucose, CKD from a single 'renal insufficiency' note or one creatinine, COPD from 'dyspnea,' or CHF from a diuretic on the med list
Candidates recognize the clinical pattern but not the diagnostic criteria or documentation threshold, and lab trends feel like proof of disease
What to do instead
Require the provider to state the diagnosis; use symptoms, labs, and medications only as indicators that prompt a record review. Know key thresholds (e.g., CKD requires documented stage based on GFR/albuminuria criteria, not just the phrase 'renal insufficiency'), and treat unstated diagnoses as suspect findings, not captures