TCORS Attorneys in New London - Norwich CT Connecticut

How Did $100 Become $100,000? The Answer: "Extrapolation"


Extrapolation is a method of using a mathematical formula that takes the audit results from a small sample of Medicaid paid claims and projects those results over a much larger universe of claims producing, in many cases, large dollar audit disallowances. A word that also brings terror and disdain to the hearts of audited Medicaid providers subject to the effects of this formula. This edition of our newsletter will attempt to de-mystify why it is used, the legal basis of its use by the state, how it works, and how it can be challenged.

History of Extrapolation in Connecticut

First let’s start with a little history and background of how and why extrapolation came to be used in the audits of Connecticut Medicaid providers. Back in the 1970’s, the audits of Medicaid provider billings were performed by auditors pouring over "green bar" computer paper lists reviewing individual Medicaid paid claims, provider by provider. If something piqued the auditors’ interest, then a field audit would be performed on a select number of claims. The number of claims reviewed was mostly dictated by how many claims an auditor could examine in a week or so of field work, which usually totaled 100 to 200 claims. After performing their claim review, the auditor would total any disallowances found on the chosen claims and assess the provider for the total disallowed amount. During this period, audit disallowances only totaled in the hundreds of dollars with some going over a thousand dollars. Audits during this time mainly concentrated on finding fraud rather than generating large disallowances. This practice, however, changed drastically around 1980 when extrapolation was introduced in Connecticut.

At that time, a new Medicaid audit director was hired at the then Department of Income Maintenance. He brought with him the idea that using random sampling and extrapolation as an audit method would be more effective and would produce results that mirrored what an audit would indeed find if the State had the resources to audit all paid claims for a finite claim universe. He convinced the Department’s administration to hire a consultant to develop statistically valid sampling and extrapolation formulas. Once those formulas were in place, he rolled out the use of extrapolation, starting with the audits of hospitals and quickly spread its use to all Medicaid provider types. Although there was initial resistance in the provider community, the use of extrapolation was imbedded within the Medicaid audit process and the formulas developed back then are still in use today.

Why Extrapolate and How It Works

The practical theory behind the use of extrapolation in Medicaid audits is simply that the program is too large to audit it any other way. In Connecticut, tens of millions of individual Medicaid claims are paid annually to thousands of Medicaid providers throughout the state, consuming billions of dollars in state revenue. Providing adequate audit coverage for a program this size would require an army of auditors if the audits had to be performed on a claim by claim basis. Connecticut believes that utilizing sampling and extrapolation provides audit coverage over a much more significant portion of the Medicaid budget and also identifies and recoups overpayments in the tens of millions of dollars a year.

As stated earlier, extrapolation is really a mathematical formula. The actual formula used by the Department of Social Services (DSS) to ensure that its extrapolation is statistically valid is fairly complex. The details of that formula, along with a discussion of confidence intervals and other statistical requirements, are too technical to be covered in this article. However, the actual mathematical calculation of the audit disallowances resulting from this formula is quite simple. To understand this computation we will offer a simple example.

An audit usually consists of a sample of 100 to 200 paid claims pulled at random from a universe of claims that normally covers a two year period. For this example, we will assume a 100 paid claim sample was selected from a universe of 20,000 claims. The audit of each of the 100 claims is performed and audit disallowances on the individual claims are totaled. In this instance, it is presumed the auditors found ten cases with errors and the dollar value of those ten errors totaled $300. This total error amount ($300) is then divided by the sample size (100) to produce an average error per sample case of $3.00. This amount ($3.00) is then multiplied by the total audit universe (20,000) to produce a total extrapolated audit adjustment of $60,000. Quite simple, yet, quite expensive to providers.

To show some actual extrapolation results, we have detailed the extrapolated audit adjustments for six recently released Medicaid audits performed by DSS. As can be seen, small total actual dollar errors found in the following chart (column 3) can result in large overall dollar adjustments when extrapolated.

Audit Universe

Sample Size

Sample Error Dollars

Extrapolated Error Amount

























Is this Legal?

One of the main questions providers ask when subject to an audit using sampling and extrapolation is: Can this be legal? Well the simple answer is, yes. Again some history is needed.

Prior to 2005, the use of extrapolation was not mentioned in any Connecticut state statute, regulation or policy. That changed when the state legislature passed Public Act No. 05-195 or what was known as the "audit bill."

This bill codified a number of issues concerning how DSS audits are to be conducted and, for the first time, set into law the use of extrapolation in these audits. Specifically, the law states: 

      "A finding of overpayment or underpayment to a provider in a program operated or administered by the department… shall not be based on extrapolated projections unless (A) there is a sustained or high level of payment error involving the provider, (B) documented educational intervention has failed to correct the level of payment error, or (C) the value of the claims in aggregate exceeds one hundred fifty thousand dollars on an annual basis."

Some audited providers misinterpret the actual meaning of this passage; they believe that DSS has to use A, B and C before it can use extrapolation. Rather, DSS can use any of the three criteria to support the use of extrapolation because the bill is written with the word "or." In the vast majority of audits performed, criterion C is used to support extrapolation. As a result, any provider with paid claims totaling over $150,000 per year, if audited, will have the results of the audit computed through the use of extrapolation.

Since the passage of the audit bill in 2005, there have been a number of attempts to amend the bill to weaken DSS’ ability to use extrapolation but all attempts to date have failed.

In addition to the passage of the audit bill, there was a Connecticut State Supreme Court case opinion in 2008 that provided additional legal support for the use of extrapolation. In the case, Goldstar Medical Services, Inc., et al. v. Department of Social Services, Goldstar, an oxygen provider, questioned the legality of DSS using extrapolation to determine "the total amount of excessive reimbursement Goldstar had received."

In this case, the Supreme Court ruled in DSS’s favor. The opinion specifically states, "the department’s use of the extrapolation method was appropriate." In reaching this conclusion, the Supreme Court cited a number of state and federal court cases that supported the use of extrapolation in Medicaid and Medicare audits.

The opinion further confirms, "It is well established that proof of damages through the use of statistics and statistical sampling has been endorsed in numerous cases involving Medicare and Medicaid overpayments."

The court also cited several federal regulations that it concluded not only allowed but mandated that DSS utilize statistical methods for determining overpayments in the Medicaid program. In such opinion the court states,

      "Federal regulatory authority thus requires states to ensure that Medicaid funds are allocated appropriately and simultaneously recognizes the impracticality of discrete assessment of claims in an effort to recoup overpayments where a multitude of claims is involved. Given the nature of the Medicaid program as a state and federal cooperative regime, it would be incongruous to interpret our statutory scheme to disallow a practice that is recognized at the federal level as the only feasible method of recouping funds that improperly have been procured."

While both the "audit bill" and the Supreme Court decision appear to strongly support the legality of the use of extrapolation in Connecticut, this does not mean that extrapolation, as a concept, cannot be challenged again in court. It is possible, if not probable, that the statistical formulas and methodology used by DSS may one day be questioned by a provider in court. Whether that happens and the challenge is successful, only time and circumstances will tell.

Can I Challenge My Extrapolated Results?

In the meantime, while the legality of the use of sampling and extrapolation are supported by state law and the courts, providers may still challenge the results of extrapolation on a variety of grounds that depend on the nature of the claims and samples selected for extrapolation. Providers still need to review both the sample and the extrapolation to ensure it produces a valid result for their particular audit. Although the actual computation of the extrapolated audit disallowance is a fairly simple mathematical exercise, there can be situations in a particular audit where the extrapolation may not work. In such circumstances, a provider will have to perform a thorough analysis of the sample, the universe and other factors, such as the error type cited in the report, in order to make its case to DSS.

In Summary

Sampling and extrapolation has been part of the Connecticut Medicaid audit process for almost 30 years and will probably be around for as long as there are Medicaid audits. Providers need to understand how it works and if it works for their particular audit situation since the financial effects of extrapolation are significant.

Click here to download a copy of the June 2011 Medicaid Audit Newsletter.

Bulletin has new feature, continues local focus
« previous
TCORS, James Wietrak, John McCormick and Paul Chausse to speak at CTHFMA Event
next »