Off-Label Audits

Although lacking in ICD-9 diagnostic information, pharmacy claims are commonly used in various sectors of the health care industry to study the prevalence of diseases and specific diagnoses.

There are 2 principal reasons for this practice, the first being that pharmacy claims are available to payers in virtually real-time, as PBM adjudication systems capture prescriptions at the point-of-sale and immediately archive these based on payer, benefit design, eligibility, cost-share mandates in the beneficiary contract, prescriber, member demographics, and an additional 5-8 fields capturing drug and prescription detail. PBM pharmacy claims are uploaded into payer data warehouses within 2 weeks of the adjudication event. In contrast, medical claims typically have a lag time of weeks to months before capture in the payer data warehouse.

Secondly, although the claim, a prescription, does not bear a diagnosis (except in the uncommon domain of Specialty pharmacy), drug indications are typically narrow, in most cases enabling an “inferred diagnosis” associated with the claim. Thus, insulin and other anti-diabetic drugs may be assume to be prescribed for a diabetic with a high degree of confidence. (In some cases, the approved range of indications may be very broad, making the assignment of an inferred diagnosis more of a risk. The best example would be the ACE inhibitors, which may have up to 5 different indications that are mutually exclusive disease processes.) The payer practice of “Disease Management” typically relies heavily on pharmacy claims to provide sentinel information on patients of interest with specific conditions.

When investigating unlawful off-label promotion in violation of the False Claims Act, investigations rely heavily upon the same inferred diagnosis logic. The general approach is the following:

1) a query must be created against the CMS Data Warehouse. Typically, pharmacy and all non-pharmacy claims (referred to as “Medical Claims”) are stored together in a master data warehouse or in data marts which may be cross-queried.

2) In order to have a high degree of confidence in “claim completion”, medical claims should be at least 45-60 days old. That is, hospital and professional provider claims typically are not submitted in real time, but depend on claim cycles which may introduce delays. In the current era of electronic claims transmissions, this completion time may be less, but usually is not less than 30 days from the service date.

3) It is estimated that as many as 50% of pharmacy claims in the US are for off-label indications. Prior to the last 20 years and enhanced payer (Federal and Commercial) sensitivity to health care costs, audits of prescriber records were not routinely conducted. Because of sensitivity around improved chart documentation stemming from litigation, providers generally have adopted more fastidious office record-keeping to justify written diagnoses. As long as the physician is licensed and bears Federal and State controlled substances permits, he/she may prescribe off-label, but this does not give pharmaceutical companies the right to promote a drug off-label for uses for which there is inadequate clinical trial evidence to support. Payers may build edits into adjudication systems limiting payment for certain classes of highly specialized drugs to certain specialist prescribers (i.e., cancer drugs, transplant drugs, etc), but this does not counter unlawful off-label promotional efforts.

4) To capture possible off-label prescription claims, the drug of interest would be queried against a defined member/beneficiary population. Once the beneficiary and drug have been linked by an adjudicated prescription, then medical claims are queried for the same beneficiary (by SS #) for a period of 45 days prior to and sometimes, following the pharmacy claim. Diagnoses and procedural codes are collected from the medical claims and matched to the suspect prescriptions.

5) If the diagnosis and procedural code array does not include any of the FDA label approved indications for the drug, then an off-label event is assume to have occurred.

6) Final validation of this, to the extent necessary, is then carried out by soliciting the medical record from the prescriber for the dates of service of the query. If the medical record does not provide an appropriate on-label diagnosis, and the medical claim lacks this, then a valid off-label event can be concluded with a high degree of certainty.