In May 2016, a landmark healthcare fraud case resulted in a federal jury convicting Florida-based Dr. Ona Colasante of 162 counts of healthcare fraud.
The government’s claim was that she billed for services that were either unnecessary or not provided at all, and purchased drugs that, while identical to those labeled for sale in the U.S. were, in fact, purchased outside the country, excluding their packaging for approval from the Food and Drug Administration (FDA) for sale in the United States.
In a trial such as this, the government has the responsibility to propose to the court some amount of money that should be imposed on the defendant as both loss and restitution, which account for funds improperly paid to the physician as well as any other costs the government deems recoverable.
In its motion, the government reported that the face value of the payments for the procedures included in the guilty verdicts amounted to $23,575.13. But the government was proposing a total loss estimate of $6.9 million, or nearly 300 times as much.
Where did this figure come from? Well, I can tell you, because I was the statistical expert hired to come up with it. In 2011, SafeGuard, LLC, the Zone 7 Zone Program Integrity Contractor (ZPIC), initiated a coding and billing compliance audit on one of the clinics owned by the defendant. A sample of 350 claims was drawn from a universe of some 65,000 claims, and the government audited those 350 to determine whether the associated line items met the standards for documentation and medical necessity. The outcome was a proposed paid error rate of around 75 percent, or $5.6 million. This was extrapolated from an actual overpayment amount of $34,524.22. It’s a long way from $35,000 to nearly $6 million, but the government apparently felt comfortable making that stretch.
Next, as a result of the conviction, the government conducted a second audit. This audit included many of the same patients as the prior audit and covered a completely different time frame. In this case, instead of using claims, the government statisticians chose the beneficiary as the audit unit and drew a stratified sample of 40 beneficiaries from a universe of 508. The statistician chose to break the sample into five strata and audited eight beneficiaries from each. The result: a demand for repayment of $1.2 million based on an actual overpayment amount of $246,000.
In total, the government was demanding that Dr. Colasante repay the government $6.867 million, or the sum of the extrapolated overpayment projections for the two audits. Now, I know that I have talked about extrapolation a lot in the past, but maybe I haven’t talked about it enough – at least not from a critical position.
My job, as the defense statistical expert, was to review the work that the government statisticians had done to determine whether, in my opinion, the audits met the standards outlined in the Centers for Medicare & Medicaid Services (CMS) Program Integrity Manual (PIM). In general, Section 8.4.2 of the PIM (and I am paraphrasing now) permits the use of stratification as long as the sample is a statistically valid random sample – and this is defined by six criteria.
In this Section, the government says that extrapolation can occur as long as:
- An appropriate universe is utilized
- The sample frame is constructed properly
- The sample units are appropriate
- The randomization method is correct
- The calculations for point estimate and sample error are done properly
- The extrapolation formula is correct for the type of audit
As long as these six criteria are met, at least from the government’s position, the extrapolation can proceed and should be accepted.
Well, not so fast there, Bucko!
Even though, in my humble opinion, the PIM does not always represent the best in standards of statistical practice, it is what it is – and that is the standard by which the government stands. If that is the case, then one should be able to at least hold the government to those standards. And if one can prove that the government did not in fact abide by those standards, then the extrapolation should not be permitted to stand. And the judge in this case apparently agreed with that position.
While it would not be appropriate to discuss the details here, I can tell you that in my expert opinion, the government did not use the proper universe, did not create an appropriate sample frame, did not select the appropriate units for audit, did not use the correct calculations for sample size, sample error, or precision, and did not use the correct formulae to create the extrapolation. At least that is what I expressed to the judge. In Herrington v. Richter (562 U.S. 86, 111 (2011)), Strickland did not enact Newton’s Third Law for the Presentation of Evidence, requiring that for every prosecution expert, there is an equal and opposite expert from the defense – and in this case, the government had two to the defendant’s one.
We both presented our case to the judge and ended with our summary opinions, and I am happy to say that in this case, the defendant won out and the judge determined that the statistical portions of the audits were fatally flawed and excluded them in their entirety from consideration during the sentencing phase of the trial.
Just last week, on Aug. 18, after five days of testimony, the judge rendered his final determination in the trial; Dr. Colasante was ordered to spend one year and one day in prison, and the government now is entitled to collect back $1.02 million, or over $5 million less than they had proposed.
Now, I think I know what some may be thinking. Why would you defend someone that was already found guilty of fraud? Well, I wasn’t defending the physician on the fraud counts, but rather on whether the government should be allowed, under any circumstances, to violate its own standards in order to inflate damages on anyone, guilty or not. It was and still is my opinion that it should not.
Every year, billions of dollars are demanded in repayment from healthcare providers, with the government using extrapolation as a powerful multiplier of actual overpayment findings. And if we don’t hold the government to its own standards, then everyone suffers.
Whether or not you agree with the punishment for physicians who commit fraud, it doesn’t justify the government also committing wrongdoing. Because remember, two wrongs don’t make a right.
And that’s the world according to Frank.
For more on this breaking news story listen to Frank Cohen during today’s edition of Monitor Mondays.