IAMI CPMA common mistakes
IAMI CPMA · Understand the traps and how to avoid them.
Coding the encounter from the physician's narrative summary alone and never confirming that each documented diagnosis or procedure is supported by the body of the record (progress notes, operative report, pathology, orders).
Candidates treat the assessment/plan line as authoritative because it is concise and reads like a summary, and they are under time pressure; they skip the cross-check against the operative note or results that actually establish the code.
What to do instead
Adopt a fixed review order for every record: identify the service, then locate the primary evidence (operative report, pathology, imaging, medication administration) before assigning any code. Write a one-line evidence citation next to each code you assign so unsupported codes surface immediately. Practice on de-identified records where the summary and the body disagree.
Applying a code from memory or from a prior year's code set without verifying the current tabular entry, includes/excludes notes, and specificity requirements.
Experienced candidates rely on habit and pattern recognition; code sets change annually, and excludes notes are easy to overlook when you already 'know' the code.
What to do instead
Build a habit of opening the current tabular list and index for every code you assign, and read the includes/excludes and 'code also' notes aloud. Track a personal list of codes you use most and re-verify them at the start of each code year. In practice questions, treat any code assigned without a cited guideline as provisional.
Designing an audit sample by grabbing the first 30 charts in the work queue or whichever records are easiest to retrieve, then reporting results as if they represent the population.
Convenience sampling feels efficient and the sample size looks defensible; candidates underestimate how selection bias inflates or deflates error rates and makes extrapolation invalid.
What to do instead
State the population definition and sampling frame explicitly before selecting anything, then use a documented method (simple random via random number generator, systematic with a random start, or stratified by provider/payer). Record the seed or selection log so the sample is reproducible. When asked to extrapolate, confirm the sample was probability-based; if not, report descriptive findings only.
Choosing sample size by rule of thumb ('always pull 30') without reference to the expected error rate, desired confidence level, precision, or the audit's objective.
The number 30 is memorable and appears in many informal guides; candidates confuse a statistical convention with a calculated sample size and never connect it to the precision the stakeholder needs.
What to do instead
For each audit, write the objective, confidence level, tolerable error rate, and expected error rate first, then compute or look up the sample size from those inputs. Show the arithmetic in your work. If the objective is education rather than measurement, say so and label the sample as non-statistical.
Treating a documentation discrepancy as confirmed fraud and writing the finding up with fraud language before performing any intent analysis or provider query.
Candidates conflate 'error' with 'intent' because the coding difference is objective and dramatic (for example, a high-level E/M billed for a low-complexity visit), and they want the report to sound decisive.
What to do instead
Classify every finding on an error-versus-abuse-versus-fraud spectrum using defined criteria, and document the evidence for intent separately from the evidence for the coding difference. Include a provider query or education step in the corrective action unless facts clearly indicate deliberate conduct. Route suspected fraud to the appropriate referral process rather than into a routine audit report.
Ignoring the seven elements of an effective compliance program when evaluating an organization, and reporting only on coding accuracy rates.
Coding accuracy is measurable and familiar, while program infrastructure (written policies, compliance officer and committee, training and education, lines of communication, internal monitoring, enforcement and discipline, response and prevention) is diffuse and harder to score.
What to do instead
Use the seven elements as a fixed checklist for every program review and record a rating plus evidence for each. When an accuracy finding appears, trace it back to the element that failed (for example, no training on a new guideline, or no monitoring of a high-risk code). Present both the accuracy result and the program-element assessment in the report.
Performing a medical record review without confirming the record is complete and authenticated, then drawing conclusions from missing pages, unsigned addenda, or late entries.
Reviewers focus on the clinical content and assume the chart they received is the whole chart; incomplete releases and unsigned documentation are common and quietly distort findings.
What to do instead
Before scoring, verify page count continuity, authentication (signature and credentials), and the presence of required components such as history, exam, and results. Log any missing or late documentation as a distinct data-quality finding rather than scoring it as a coding error, and request the complete record before finalizing conclusions.
Writing audit findings that describe what was found but never state the criteria, the condition, the cause, the effect, or a specific recommendation with an owner and date.
Candidates narrate the chart review chronologically and stop when the discrepancy is described; they have not been taught to convert an observation into an actionable finding, and recommendations feel like someone else's job.
What to do instead
Use a fixed finding template: criteria (the rule or guideline), condition (what the record shows), cause, effect (financial or compliance exposure), and recommendation with responsible party and target date. Draft each finding in that structure and have a peer check that all five parts are present before the report is issued.
Accepting a provider's explanation or a stakeholder's preferred conclusion without independent verification, or auditing a process the auditor personally designed or operates.
Access and rapport make it easy to take an explanation at face value, and auditors are often embedded in the same team, so independence threats go unrecognized.
What to do instead
Independently corroborate every explanation against source documentation and record the corroboration in the workpapers. Before each engagement, complete an independence and conflict-of-interest check; if you designed or operate the process, disclose it and have the engagement reassigned or independently reviewed. Document the threat and the safeguard applied.
Extrapolating a sample error rate to the full population and presenting a projected overpayment without documenting the projection method, confidence interval, or the assumption that the sample was representative.
Stakeholders want a dollar figure, and a point estimate is easy to produce; candidates skip the statistical disclosure because they are unsure how to express uncertainty.
What to do instead
When projecting, state the population size and value, sample size and method, point estimate, and confidence interval, and note any exclusions. If the sample was not probability-based or the population was not homogeneous, decline to project and report the sample results only. Keep the projection worksheet in the workpapers so the number can be reproduced.