CMRS (ACAC) common mistakes
CMRS · Understand the traps and how to avoid them.
Citing a coding guideline, coverage rule, or documentation standard without naming its exact source (for example writing 'per CMS rules' or 'per coding guidelines' instead of naming the specific regulation, manual, or guideline provision and its applicability).
Candidates memorize the gist of rules but not the citation trail, and under time pressure they paraphrase authority generically because they cannot recall the precise source.
What to do instead
For every rule you rely on, practice stating the source at three levels: the authority (statute/regulation/manual/payer policy), the specific section or provision, and the year/version in effect for the review period. Build a one-page citation map per domain and rehearse citing from it.
Treating a documentation gap as automatically equivalent to a coding error, or conversely accepting a code because the documentation 'implies' the diagnosis without an explicit provider statement.
Candidates conflate clinical plausibility with codable documentation and skip the step of testing whether each code element is explicitly supported, linked, and consistent across the record.
What to do instead
Use a two-pass method: first extract what the provider explicitly documented and linked (diagnosis, site, laterality, causality, severity), then compare each code element against that extract. Record 'not documented' and 'documented but not linked' as distinct findings with distinct remediation.
Judging medical necessity by whether the service was clinically reasonable in general rather than by whether the record supports the payer's specific coverage criteria in effect on the date of service.
Candidates apply clinical intuition and their own facility's habits instead of the governing coverage determination, and they forget that criteria change over time and differ by payer and plan.
What to do instead
Before reviewing, identify and date-stamp the coverage criteria that applied on the date of service for that payer and plan type. Evaluate each element of the criteria against the record and document pass/fail per element, not a global impression.
Selecting a sample by convenience (for example the first records in a worklist or records already flagged as problematic) and then reporting results as if they represent the population.
Convenience samples are faster and produce dramatic findings, so candidates default to them and then overstate conclusions without acknowledging that the sample was non-random and biased.
What to do instead
State the population definition, sampling frame, method (random, stratified, or targeted), and confidence level before pulling records. When a non-statistical or judgmental sample is appropriate, label findings as limited to the sample and do not extrapolate error rates to the population.
Reporting an error rate or overpayment estimate without showing the numerator, denominator, dollar basis, and whether the figure is a sample estimate or a population projection.
Candidates focus on the headline number and omit the arithmetic and assumptions, which makes the finding unauditable and easy to challenge on appeal.
What to do instead
For every reported rate or dollar figure, include the count of records reviewed, count in error, error definition, dollar amounts per record, and the projection method with its confidence interval. Recompute the figure from the underlying worksheet before submitting.
Writing a corrective action plan that lists intentions ('staff will be re-educated', 'monitoring will increase') without an owner, a due date, a measurable target, or a validation method.
Candidates treat the CAP as a narrative summary of the finding rather than a testable plan, and they run out of time at the end of the response.
What to do instead
Use a fixed CAP template for every finding: root cause, action, responsible role, due date, measurable success threshold, validation sample size and timing, and escalation trigger if the threshold is missed. Practice writing CAPs that a stranger could audit.
Failing to distinguish fraud, waste, and abuse, and labeling every coding error or documentation weakness as potential fraud.
Candidates use the terms interchangeably and skip the intent and knowledge analysis that separates an honest documentation error from a knowingly false claim.
What to do instead
Apply a decision sequence: is there a rule or standard violated (error)? Is there unnecessary cost or inefficiency without a rule violation (waste)? Is there misuse of position or program (abuse)? Is there knowing and intentional misrepresentation (fraud)? Document the facts supporting each step and route to the correct process.
Disclosing or discussing identifiable patient or provider information in examples, screenshots, workpapers, or communications beyond the minimum necessary for the review purpose.
Candidates use real details to make answers concrete and underestimate how much identifying context survives de-identification, and they conflate internal review access with permission to share.
What to do instead
Default to synthetic or fully de-identified examples. Before sharing any work product, apply the minimum-necessary test and confirm the recipient's need to know. Treat the review's own workpapers as protected records subject to the same handling rules as source documents.
Accepting the provider's or coder's explanation of a code choice without independently verifying it against the source record and the applicable guideline.
Candidates defer to authority to avoid conflict, and deference feels like professionalism, so they skip the verification step during interviews and clarifications.
What to do instead
Treat explanations as leads, not evidence. Independently confirm each assertion against the record, and where clarification is needed, obtain a compliant written clarification that adds no new clinical information. Document what was verified and what remained unresolved.
Reaching an audit conclusion without recording the review trail: which records were selected, what was requested, what was received, what was missing, and how each finding was derived.
Candidates write up only the conclusions because the trail feels like clerical overhead, which leaves findings indefensible when challenged and prevents reproducibility.
What to do instead
Maintain a review log during, not after, the audit: record inventory, request and receipt dates, missing-item follow-ups, and the specific document and page supporting each finding. Before finalizing, have someone reproduce one finding from the log alone.