A new study of random-sample auditing of health plans says the method doesn't detect hundreds of millions of wasted dollars.
Researchers for Healthcare Data Management, Inc., a provider of health plan audits and analytics, say they can conclusively prove that the random-sample methodology is missing most of the money that it's supposed to find.
The study, conducted on behalf of HDM by Ronald Klimberg and George Sillup from the Haub School of Business at St. Joseph's University in Philadelphia, compared the relative effectiveness of random-sample auditing vs. a method that relies on a 100-percent-of-claims analysis as a core element.
Using actual paid medical claim data sets from two unidentified HDM Fortune 100 client companies, researchers ran 100 simulations each of 300- and 400-claim sample sizes among only the errors that HDM found with its proprietary five-step protocol for analyzing 100 percent of paid claims.
According to Klimberg, "regardless of the sample size, the best the random sample simulations could do was catch less than 10 percent of the errors that the 100-percent of claims methodology found."
"Translated into dollars, from $200,000 to over three quarters of a million dollars in over- and underpayment errors were simply missed by the random-sample simulations," Klimberg said.
"With the common use of retrospective random-sampling as a health plan auditing tool, it is not far fetched to say that, in aggregate, hundreds of millions of dollars in wasted claims expense is being missed annually by organizations relying on the random-sample methodology for all-important health plan auditing," Sillup said.