At some point in your professional health career, you might be the recipient of a dreaded post-audit letter like the following from the fictional “PayLess Health Plan”:
Dear Doctor,
Thank you for your cooperation with our recent audit of our payments to you. We reviewed 20 of your patient records. We found that 4 do not have sufficient supporting documentation as required by our Provider Manual rules. We are therefore retroactively denying $1,020 of the $5,100 in payments for those 20 patient encounters.
Please be advised that, as allowed by the PayLess Health Plan Provider Manual, any disallowance of payments can be extrapolated, so that we can seek return of previous payments to you. We are exercising our right to extrapolate from the audit results. As we disallowed 20% of payments for to the 20 audited patients, we are similarly disallowing 20% of the last three years of all payments. Your payments for the last three years total $520,000; therefore kindly remit $104,000 (20%) within 10 business days. Failure to do so could result in our cancellation of your provider contract, and our taking further legal action to recoup the overpayments. We trust that you will not force us to take these steps.
Very truly yours,
Auditing Department
PayLess Health Plan
If you did not get legal advice (as you should have) when you received your first audit notice, you should be seeking legal advice now. Extrapolation is an extremely powerful tool in provider audits. But like any powerful tool in the hands of a regulator (or as in the case of a private health plan, a “pseudo-regulator”), it is subject to abuse. There are several ways to challenge the validity of an extrapolation.
The first question to ask is, “Is the sample valid?” If you open a bag of M&M’s, and remove two which happen to be brown and red, you obviously cannot conclude that the rest of the M&M’s in the bag are half brown and half red. Two M&M’s selected from the bag is not a valid sample, upon which you can base any conclusions at all.
Similarly, if someone tells you that the first 10 M&M’s they took from the bag were green, you would question whether the sample they took from the bag was “random.”
Sample size and randomness issues just scratch the surface of potential sampling problems which can form the basis of a challenge to extrapolation. It can be advisable to engage a statistician to review the sample and compare it to the extrapolation cohort. You, your lawyer or your statistician might uncover any or all of the following types of problems:
- If patients were sampled from more than one office locale, are some locations over or under-represented in the sample?
- If more than one provider is involved in the audit, are some providers over or under-represented in the sample?
- Were audit dates selected at random?Perhaps there was one location with documentation problems concentrated into one period of time, and that time period was over-sampled.
- Is there a diagnosis over- or under-represented in the audit?
- Are there patients over- or under-represented by demographic (gender, age, domicile)?
Always be vigilant in determining whether there is some confounding factor responsible for a heightened error rate in the audited sample. Did one location not get proper training? Is there one provider who is inadvertently failing to complete charts properly?
Assuming the sample is valid, there are additional issues that are raised by the practice of extrapolation. Many if not all payors allow a grace period during which a claim can be re-submitted to correct honest documentation errors. But what if the extrapolation period is extended to include some or all of the re-submission period? Clearly, it would be inequitable to lose the opportunity to re-submit, just because an audit took place. Even before the initial audit is complete, the provider should be considering a self-audit of any claims that are about pass into the staleness category – before they are too old to re-submit.
Another basis for challenging the validity of extrapolation is whether after the sample period there was an intervening “system” change (e.g., installation of EMR or change of EMR vendor) which dramatically changed the recordkeeping environment.
Auditing and extrapolation of audited results are not always performed by persons trained in the nuances of statistics and the dangers of putting a powerful magnifying glass to an invalid sample. As the provider, it is up to you to assure that payors are charged properly, but it is also up to you to assure that payors who question their payments to you have a valid basis for doing so.