100% error rate audits challenge credibility.
Over the past five or six years, I have worked as a statistical expert on hundreds of extrapolation audits. And at least a couple dozen of these were based on 100 percent error rates. That means that the auditor determined that, of all of the claims that were selected for review, every single one of them was coded and/or billed in error.
These hundred-percent error rate audits present some very serious and difficult challenges to the provider. From the statistician’s perspective, it makes it more difficult to challenge the statistical method used for sampling and extrapolation, because the fact that the auditor determined that 100 percent of the sampling units were coded and billed in error is the big elephant in the room. And that makes it easy to ignore all of the other issues.
I am always suspicious when I see an audit like this. It’s easier to believe when the auditor has selected a nonrandom sample, like for some specific procedure code that they already know in advance was likely billed in error. But when they claim that this occurred with a random sample, the hair on the back of my neck stands up. I mean, if you just think about this from a standpoint of common sense, who would believe that a medical provider in practice for some given period of time does everything wrong, all the time, when it comes to coding and billing? From a probabilistic standpoint, it is as improbable as it gets.
Now, I am not saying that one shouldn’t challenge the statistics involved in sample selection and extrapolation; I am saying that it is just as, if not more, important to challenge the findings themselves. One of the big flaws that exists within almost every audit I have ever seen is that the auditor determines sample size based on the paid amount. The U.S. Department of Health and Human Services (HHS) Office of Inspector General (OIG) has issued specific guidance on this. They have emphatically said that this is absolutely the wrong variable of interest. In fact, on their frequently asked questions web page for corporate integrity agreements, they say that we already know what the paid amount is, what we are interested in is the overpaid amount. But in the program integrity manual, it says that we can use the paid amount (or any variable of interest, for that matter) if we can show that it’s highly correlated to the overpaid amount. And in an audit where there is a 100 percent error rate, at least within the sample, there is 100 percent correlation between the paid and the overpaid amount.
The auditor also can get jammed up with hundred-percent error rate audits. For example, if the paid amount average for the sample is even a dollar more than the paid amount average for the sample frame, it results in an overpaid point estimate that is greater than that paid-to-point estimate. And that never looks good, when presented to a judge.
Let’s say, for example, that you have a sample frame of 10,000 claims and that the average paid amount for those 10,000 claims is $100 each. That means that the total paid amount for the sample frame is $1 million. Let’s say that the average paid amount for the sample is $105 per unit – and this happens quite often, because it is extremely rare that the average paid amount for the sample will be exactly the same as the average paid amount for the sampling frame. Under most circumstances, this is absorbed into the error rates and the confidence intervals. But that is not the case with the hundred-percent error rate. In our example, where the overpaid amount is equal to the paid amount because every single claim is determined to have been billed in error, you multiply the average overpaid amount of $105 by the sampling frame of 10,000 claims, and you get a point estimate of $1,050,000. In this case, it means that the overpaid point estimate for the sampling frame is actually more than what the practice was paid. Even though this is defensible, it looks bad for the auditor.
So now I get to put on my old-man cynical hat. But when you’ve been through as many audits and appeals as I have, I believe you’ve earned the right to wear that hat. The truth is, I have never seen an audit that started out with a 100 percent error rate wind up with a 100 percent error rate. The problem is that the majority of the reversals occur at the administrative law judge (ALJ) level. I don’t think I am alone in my opinion that the first two levels of appeal are basically worthless, useless wastes of time. Redetermination is a joke. It is basically a rubber stamp of the auditor’s findings; only in a very, very low percentage of cases have I ever seen anything good for the provider come out of redetermination. But alas, it is still a hoop we have to jump through. Reconsideration is almost as bad; however, I have seen some partially favorable results come out of that. And in a partially favorable result, where some number of claims are reversed back in favor of the provider, the error rate is then reduced below 100 percent. And when that happens, it makes challenging the statistical process a whole lot easier. For one, not only do we lose the strength of the correlation between the paid and the overpaid amounts, but it also impacts the precision rate, which is very important whenever we talk about inferential statistics of any kind.
But the big impact occurs at the ALJ level. In fact, every case that I have worked on where there was a determination of a 100 percent error rate ended up with a significantly lower error rate after the ALJ hearing. It’s really quite interesting, when you think about it. How is it, that at redetermination and reconsideration, very little occurs that favors the provider, but at the ALJ level, nearly 70 percent or more of the findings of error are reversed in favor of the provider? The difference is that the ALJ is an independent arbitrator and is not under the thumb of CMS, like we find with the Medicare Administrative Contractor (MAC) and the Qualified Independent Contractor (QIC). The sad part is that I am seeing these types of audits increasing in number. In fact, just this year alone I have seen half as many as I have during the prior five years combined. Counting the ones I am working on now, I believe that I have been the statistical expert on over a dozen audits for which the auditor filed a determination of a 100 percent paid error rate. What’s really sad is that after redetermination, the provider often has to make a full payment even before the due process of the ALJ hearing occurs.
So I don’t know if this is a conspiracy or whether the auditors have tasted blood and see that this is a great way to extract as much money as possible from a practice or hospital or some other healthcare provider. But I do know that it is creating havoc among our clients. The bottom line is that you have to take this seriously. I’ve had a couple of clients that decided they just wanted to pay back the demand amount without a fight. I guess they felt that it would be more expensive to fight it than it would be just to pay back the money. The problem is that in doing this, what you have admitted is that you do everything wrong, all the time – and I think that’s just a terrible and dangerous precedent to set. Particularly when combined with extrapolation, hundred-percent error rate audits can be financially devastating to any healthcare provider.
Personally, I don’t believe that any provider bills everything wrong all the time, and apparently, neither do independent judges. I accept the fact that we may not be right all the time, but I refuse to accept the fact that we do things wrong all the time.
And that’s the world according to Frank.