The Coding Audit That Found Too Much
A 12-physician orthopedic practice conducts an internal coding audit of 60 randomly selected evaluation-and-management (E/M) encounters. The auditor, a certified coder, finds that 9 of 60 (15%) were upcoded - mostly level-4 visits documented at a level-3 support level. The practice's compliance plan sets a 5% error-rate threshold for 'focused review' and a 10% threshold for 'mandatory repayment analysis.' The managing partner argues the sample is too small to mean anything, notes the auditor is 'just one person's opinion,' and points out that the practice has never had a payer audit. The compliance officer wants to expand the review and consider a self-disclosure.
1. Evaluate the managing partner's three objections and determine whether the 15% finding crosses the plan's thresholds.
Read the worked answer
Take each objection in turn:
- 'Sample too small.' A 60-chart random sample yielding 9 errors is not nothing. The point estimate is 15%; even with wide confidence bounds, the lower bound is plausibly above the 5% focused-review threshold and near or above the 10% repayment threshold. The correct response to sampling uncertainty is to expand the sample, not to dismiss the finding. Small samples justify more data, not inaction.
- 'One person's opinion.' Coding determinations are judgment-based, but that is why the fix is a second-level review (inter-rater check) and use of the same published guidelines the payer uses - not discounting the result. If two independent coders disagree, that disagreement is itself a finding about documentation clarity.
- 'Never had a payer audit.' Absence of enforcement is not evidence of compliance; it is luck plus lag. Extrapolated over the practice's full E/M volume, a 15% upcoding rate is a material overpayment exposure regardless of whether anyone has looked yet.
Threshold determination: 15% exceeds both the 5% focused-review and 10% repayment-analysis thresholds. The plan's own policy is triggered; the partner's objections do not override the written threshold. The CO should document that the thresholds were met and proceed.
2. Design the expanded review: what population, sample size approach, and review method would give a defensible estimate?
Read the worked answer
A defensible expansion:
- Population: all E/M claims for the look-back period (e.g., 12–24 months, consistent with payer timely-filing and the plan's policy), stratified by physician and by code level so high-level codes are not diluted.
- Sampling: use a statistically valid random sample sized to estimate the error rate with a reasonable confidence interval (e.g., 95% confidence, ±5% precision) rather than an arbitrary count; stratify so each physician has enough charts to detect outliers. If the first sample already shows a high rate, a larger sample tightens the estimate and supports extrapolation for repayment.
- Review method: two independent certified coders on a subset to measure inter-rater reliability; adjudicate disagreements with a defined rule (e.g., documentation supports the lower level). Use the same coding guidelines and code sets the payers apply.
- Documentation: record the sampling method, the random-selection seed/process, reviewer credentials, and adjudication rules so the estimate is reproducible and defensible to a payer or the government.
- Output: error rate with confidence bounds, dollar overpayment estimate, and physician-level breakdown to target education.
3. Under what conditions should the practice pursue self-disclosure versus internal correction and repayment?
Read the worked answer
Self-disclosure is not automatic; it is a judgment tied to the facts. Consider:
Factors favoring formal self-disclosure (e.g., payer disclosure or a government self-disclosure protocol):
- The error rate is high and extrapolates to a material overpayment across many claims.
- The conduct may be more than isolated error - e.g., a pattern suggesting intentional upcoding, or a physician who ignored prior education.
- There is evidence of a potential violation of law (false claims) rather than a pure documentation/coding disagreement.
- The practice wants the credit for cooperating and limiting exposure.
Factors favoring internal correction + repayment without formal disclosure:
- The errors are isolated, clearly unintentional documentation issues.
- The exposure is small and can be repaid through normal claims-adjustment channels.
- The payer has a routine refund process that fits.
Regardless of path: the practice must (1) stop the conduct (education, documentation templates, pre-bill review for high-level codes), (2) quantify and repay identified overpayments within required timeframes, and (3) document the analysis. The CO should present both options with the facts to the managing partner and, if the pattern suggests intent, escalate to counsel. The decision and rationale must be documented - the failure mode is not choosing 'wrong,' it is choosing nothing and having no record.