Resumen
The continued computerization of health care records has enabled easier sampling and analysis of large sets of medical records, making it easier than ever for Medicare, Medicaid and other private insurers to use statistical audits to determine and demand return of alleged overpayments to health care providers. However, there are sometimes statistical difficulties with the audits, and there is frequently not sufficient transparency in the procedures or their application to reproduce the results in order to determine whether they have been carried out correctly. This paper addresses concerns in sampling and analysis of data records by looking at the case of a specific audit of a medical practice carried out by a private insurer. If done properly, statistical audits can be a very useful tool, but often the methodologies are vague and the implementation is either wrong or not explained fully enough to reproduce and analyze.