Q&A: Choosing sample cases for a coding audit

Q: I'm in charge of a coding audit for our department for the first time and trying to determine how many cases to use as a sample. What factors should I consider?

A: Sample sizes may vary depending on the purpose of the audit. A commonly recommended audit sample size is 30 cases for routine quality auditing with no known serious compliance issues. Conducting an audit of this magnitude for every coder every month can be a daunting challenge. Assess the purpose of a particular audit and consider modifying the sample size and/or allowing more time to conduct a review.

For example, when auditing for discharge disposition accuracy, review several discharges per coder every week, and then evaluate the monthly or quarterly results. Auditing inpatient coding may take 20–40 minutes depending on the length of stay, copy-forward use, and number of conditions. It may take less time if the audit is assessing only whether the principal diagnosis was correctly assigned and that at least one comorbidity/complication (CC) or major CC (MCC) was identified. In this type of audit, the auditor is not assessing all the potential conditions that could be coded and validated.

Spreading the review over a quarter of the year facilitates reviewing 30 cases per coder. Additionally, management should ensure that the coding performed by contracted coders is evaluated as well.

Auditors reviewing a new coder’s work may find that a sample of 30 cases is not adequate and may need to conduct a review of all cases coded until such time that the new coder demonstrates a grasp of the coding guidelines and can accurately assign codes. Or, in the case of an experienced coder new to an organization, an audit of a substantial sample may be necessary until the auditor believes the new person understands the idiosyncrasies of the organization, its systems, and unique coding processes.

When auditing all coders for an Office of Inspector General (OIG) Work Plan item and nothing suggests that one or more coders have errors in the topic selected, consider reviewing a smaller sample, such as five encounters per coder. Increase the sample size to 10 for any coders with errors in the Work Plan issue. Continue to expand the sample if more errors are found. This auditing by exception approach limits reviews to only those coders with errors.

Editor’s Note: This article originally appeared in Revenue Cycle Advisor. This answer is excerpted from JustCoding's Essential Guide to Coding Audits by Rose T. Dunn, MBA, RHIA, CPA/CGMA, FACHE, FHFMA, CHPS.