Auditing and Monitoring: Building a Defensible Sample and Following Through
Auditing is not monitoring
Monitoring is ongoing, built into the workflow, and preventive - pre-bill edits, monthly supervisor chart checks, dashboards. Auditing is retrospective, independent, and systematic: a defined population, a defined period, a documented methodology. Questions often describe an ongoing operational process and label it an audit.
Plan from risk
A defensible audit plan begins with a risk assessment: high-dollar services, new providers, elevated denial rates, prior findings, payer mix, and regulatory focus areas. Then set scope (codes, providers, dates), sampling method, sample size, and data source - and write it down before pulling records.
Sampling essentials
- Define the population precisely; one claim may contain several line items.
- Random sampling supports projection; targeted sampling does not.
- A probe sample of roughly 20–30 records can indicate whether to expand, but it cannot be extrapolated.
- Know the difference between a discovery sample and a statistical sample with stated confidence level and precision.
Execute and follow through
Use standardized audit tools, retain work papers, and have a second reviewer check a subset of the sample. Findings without a corrective action plan, an owner, and a due date are just complaints - and you must re-audit to confirm the fix held.
Worked example
A 30-record probe finds 6 errors (20%). Because the sample was targeted at high-risk codes, the 20% figure describes those codes only. Expand to a random sample across the full population before quoting any organization-wide error rate.
Rule of thumb
You may only project to the population you randomly sampled from - no exceptions and no rounding up.