Post Payment Reviews by Healthcare Payors: Lies, Damn Lies, and Statistics*

by Gregory J. Pepe

December 21, 2022

Many physicians face retrospective audits of reimbursements they were paid by private and public payors. The audits are performed after the physicians receive the money, often years after. The typical audit involves an examination of a limited number of patient records during a prescribed period. Typically, the auditor either arrives at the physician’s office for the review or asks for copies to be sent to the auditor’s business. The number of patient records sought can vary from audit to audit but is supposed to reflect the type of issue involved in the audit. Most often, following the auditor’s examination of the patient files, the auditor will seek an “Exit Conference”, and during that time present an audit report showing the reimbursements that were “disallowed” based on the records reviewed. The next step is when the “magic” of mathematics happens.

Based on the percentage of the patient records reviewed vs. the numbers of those records disallowed, the auditor uses that percentage across the entire universe of claims and reimbursements for the same period. Thus, based on an audit of 100 billing records and supporting patient charts, spanning a three-year period, in which 40 are disallowed, an auditor would disallow 40% of all claims and payments over the same three year period for the same issues alleged in the disallowances. In a recent case handled in our office, there were 150 records reviewed involving 3 CPT codes, and a finding of overpayment of $2,138.09 for those 150 records. Extrapolated over the universe of claims for the same codes, which totaled 42,315 payments, the amount claimed for repayment was $3,128,495.54. Here is the math….

Universe size, NI= 42,315   (Strata sizes: Ni = 1,594, N2 = 27,261, N3 = 13,460)

Sample size, n=150  (Sample strata sizes: ni = 30, ny = 90, n3 = 30)

Total Sample Error Dollars = $2,138.09

Total Universe Amount Paid = $3,128,495.54

Weights: NVN = 1,594/42,315, N2/N = 27,261/42,315, N3/N = 13,460/42,315

Strata means: % = 37.2800,512= 10.5936, y3 = 2.2090

Stratified estimate of the total dollars in error: P(st) = N y(st)

The 95% confidence interval for the total dollars in error is: $230,963.15 to $524,933.60

A successful challenge of the statistical extrapolation involves the retention of an experienced expert, and unfortunately, most physicians only seek out such expertise after the fact. Virtually all statistics experts seek a calculation where the “confidence level” for the calculation is 95% or greater. Retaining an expert earlier in the process, before turning over any files to a payor, could help narrow the scope of the sampling, and thus reduce the huge numbers that result from a more inclusive sampling. Although many believe that the use of extrapolation is fundamentally unfair in American jurisprudence, courts in most jurisdictions permit the use where large numbers of claims are involved. Physicians are advised to consult with knowledgeable professionals at the time a payor makes a request for a review of billing and patient records. As Mark Twain famously said*: “There are three kinds of lies; lies, damned lies and statistics.” Act with caution.

Disclaimer: This article is for educational purposes only and to give you a general understanding of the law, not to provide specific legal advice. No attorney-client relationship exists by reading this article. This article should not be used as a substitute for legal advice from a licensed professional attorney in your state.

---