Overzealous and hyper-aggressive audit tactics are scaring providers away from self-governing.
The U.S. Department of Health and Human Services (HHS) strikes again, and this time, based on an Office of Inspector General (OIG) audit, Saint Francis Health Center has been asked to refund $5.5 million back to the Medicare contractor.
St. Francis is a 253-bed hospital located in Topeka, Kansas. According to the OIG report (citing Centers for Medicare & Medicaid Services, or CMS, national claims history data), Medicare paid the hospital some $74.5 million for 7,807 inpatient claims between Jan. 1, 2015 and Dec. 31, 2016. OIG has stated that they selected a stratified random sample of 100 inpatient claims with payments totaling $1.4 million for a two-year audit period. So, the value of the sample was around 25 percent of what the OIG wants them to refund. As the audit covered $16 million in Medicare payments for the six case-mix groups (CMGs) at risk, the refund accounts for over a third of those payments. That’s a pretty hefty extrapolation, to be sure.
Not being an auditor, I am not going to get into the reasons that the OIG found St. Francis wanting, in regard to their audit results. From what I read, there were a number of different issues, and as one would expect, the hospital vigorously defended itself when it came to the appropriateness of the claims. In general, the OIG stated that the hospital “did not fully comply with Medicare billing requirements for . . . 51 claims, resulting in overpayment of $707,118.”
Let’s do the math; the face value of the overpayments (according to the OIG-contracted auditor) was $707,000, and the extrapolated refund amount was calculated to be $5.5 million, which is almost eight times the actual overpayment finding. No wonder healthcare providers are petrified of extrapolation. And because (at least in my experience) the government rarely follows standards of statistical practice when creating the sample and the extrapolation, those fears are absolutely justified.
In fact, one of the defenses put forth by the hospital challenges the Medicare Program Integrity Manual (PIM) statement that, in order to perform an extrapolation, the contractor must determine that “there is a sustained (finding) for high level of payment error” or “a documented educational intervention has failed to correct the payment error.” St. Francis argued that, since, in their opinion, none of the 70 inpatient rehabilitation facility (IRF) claims were miscoded, there could not have been a “high level of payment error,” since the error rate would have been zero.
One problem with this is that St. Francis is arguing that 100 percent of the findings by the government contractor were wrong, and it is not often that this level of an argument ensues. In addition, for the other stratum, they did agree that, contrary to the OIG finding, only four of the 30 high-severity claims were incorrectly coded, which resulted in a very low face value ($5,566) and subsequently, a low payment error rate of only 2.32 percent.
There are a couple of problems with this argument. First, and this has been a very longstanding problem, there is no legal definition for either “high” or “sustained,” as it pertains to this statute. In fact, Section 184.108.40.206 of the PIM states the following:
“For purposes of extrapolation, a sustained or high level of payment error shall be determined to exist through a variety of means, including but not limited to:
- High error rate determinations by the contractor or by other medical reviews (i.e. greater than or equal to 50 percent from a previous pre- or post-payment review)
- Provider/supplier history (i.e., prior history of non-compliance for the same or similar billing issues, or historical pattern of non-compliant billing practices);
- CMS approval provided in connection to a payment suspension;
- Information from law enforcement investigations;
- Allegations of wrongdoing by current or former employees of a provider/supplier; and/or
- Audits or evaluations conducted by the OIG.
From what appeared in the OIG report, none of these requirements were satisfied, but most egregious is the first bullet point. CMS actually uses an example of 50 percent as a “high” error rate to justify extrapolation, yet in this case, it was barely 2 percent (at least, according to the hospital). So, once again, there is no actual definition provided, so we are all left guessing. The second problem, and perhaps the most difficult to overcome (and I know that I have written about this before), is found in section 220.127.116.11 of the PIM. Here, it clearly states that “by law, the determination that a sustained or high level of payment error exists is not subject to administrative or judicial review.”
In essence, this means that, even if the error rate was ridiculously low, as it might have been here, the hospital is not in a position to challenge whether it is defined as “high” or not. Rather, and I do this day in and day out, one needs to challenge the statistical sampling and overpayment methods that were used. In this case, there is a bit of a conundrum, because at least for one of the strata, the hospital is claiming that the error rate is zero – and, just based on common sense, one would thus assume that the use of extrapolation would be excluded. But this is the government, and I don’t think I have to tell anyone that it is not a hotbed of common sense. I have seen this argument unfold dozens of times, and have yet to see a provider prevail, no matter how small the payment error rate. So, again, I focus on finding errors within the sampling and extrapolation methods, if they exist. And they almost always do.
One of the statistical challenges by the hospital did make perfect sense. The OIG included in the sample frame (and subsequently the sample) two claims that should have been excluded because they had been reviewed previously. By failing to remove these two claims, each of the units within the sample frame (or at least those that should have been in the sample frame) no longer had an equal and non-zero probability of being selected. I know that this sounds a bit like a technicality, but the fact is, the rules are the rules. CMS has been quite aggressive in its arguments that as long as the sample is defined as random (each unit has an equal and non-zero chance of selection), then you can’t deny that it is a probability sample – and it becomes, according the PIM, appropriate for extrapolation. But in this case, because two claims were included that should not have been, the sample was no longer a probability sample, and as such, technicality or not, the extrapolation should not have been tossed.
In their defense, the OIG said that they fixed this problem by excluding those two claims from an overpayment determination, but still kept them in the sample. In essence, they included two claims that should not have been included, and rather than removing them (which would have rendered the sample useless for extrapolation), they just manipulated the overpayment amount, which (you guessed it) also rendered the sample useless for extrapolation. The truth is, by pre-determining that they would not assess an overpayment for those two claims, they did, in effect, remove them from the sample. Therefore, they only reviewed 68 of the 70 claims in the sample, which violates the standards of statistical practice, as well as the PIM. It’s pretty simple from there; no probability sample, no extrapolation. Or, to put it more colloquially, what’s good for the goose is good for the gander. In fact, it is quite astonishing to me that OIG somehow feels that it is exempt from these basic statistical standards, and that they can bend the rules as they see fit. Ask 100 statisticians if it’s OK to remove a couple of units from a sample and still consider it a statistically valid random sample, and I would be willing to bet that you would get 100 statisticians that said no.
While I was not privy to what may have been a more detailed statistical analysis and report, the OIG report, in my opinion, makes clear that, once again, HHS is acting like a bully trying to get lunch money from some weaker kid, rather than working hard to earn his own. It is my opinion, based on hundreds of conversations with healthcare providers, that the government’s overzealous and hyper-aggressive audit tactics are scaring providers away from self-governing because of the fear that, no matter what they find, HHS will come in and try to extract another pound of flesh.
The joke goes like this:
A farmer who has been involved in a terrible road accident with a large truck ends up in court fighting for a big compensation claim. “I understand you’re claiming damages for the injuries you’re supposed to have suffered?” stated the counsel for the insurance company. “Yes, that’s right,” replied the farmer, nodding his head. “You claim you were injured in the accident, yet I have a signed police statement that says that when the attending police officer asked you how you were feeling, you replied, ‘I’ve never felt better in my life.’ Is that the case?” “Yeah, but,” stammered the farmer. “A simple ‘yes’ or ‘no’ will suffice,” counsel interrupted quickly. “Yes,” replied the farmer. Then it was the turn of the farmer’s counsel to ask him questions. “Please tell the court the exact circumstance of events following the accident when you made your statement of health,” his lawyer said. “Certainly,” replied the farmer. “After the accident, my horse was thrashing around with a broken leg and my poor old dog was howling in pain. This cop comes along, takes one look at my horse, and shoots him dead. “Then he goes over to my dog, looks at him, and shoots him dead too. Then he comes straight over to me and asked me how I was feeling. Now, mate, what the heck would you have said to him?”
If the government really wants cooperation from providers regarding coding and billing compliance, they need to stop shooting the injured ones in the head.
And that’s the world according to Frank.