Do You Collect the Right Data For Defending Your Paid Claims?

The 72-hour rule regarding surgical cases and “exact” ICD-9 match for separate billing-Recently, RACs have been analyzing data related to the 72-hour rule. The rule states that “all diagnostic services provided three calendar days prior to the calendar day of the admission are bundled and paid as part of the admission.” The non-diagnostic pre-admission services are only bundled if they are related to the admission and “related” means an exact match of ICD-9 digits of the principal diagnosis for both the admission and the pre-admit outpatient (OP) services. The problem is that hospitals typically code an admission as one encounter and do not code the OP services separately. RACs are looking for admissions with an OP procedure and an unrelated diagnosis code. The potentially higher paying surgical diagnosis-related group (DRG) is changed to a lower paying medical DRG. As the auditors become more sophisticated, providers must become better prepared. Capturing multiple diagnoses and being able to analyze the status of the patient is becoming more and more important in data management.

Continuity of data integrity when warehousing-In the constant attempt to save space and minimize file size, data elements are often eliminated or abbreviated. Modifiers are often not saved, line item payment amounts are bundled into a total claim payment amount for OP services, denial reason codes are ignored, and other relevant data are not captured and stored. Then, when analyzing past data, the most pertinent information is sometimes missing, hindering an efficient and accurate analysis. While capturing correct data is critical, storing the complete data is now even more important.

STAYING AHEAD WITH DATA MINING

Another area where health informatics have evolved is in the identification of risk areas for audits-either by employing methodologies that mimic those publicly announced by the auditors, or by drilling down into levels of detail that support line-items on claims known to have been targeted.

The Program for Evaluating Payment Patterns Electronic Report (PEPPER), a Microsoft Excel file containing hospital-specific data statistics for CMS target areas that are often associated with Medicare payment errors, compares a hospital's performance across a dozen or so defined target areas, relative to other hospitals in the state, Medicare Administrative Contractor (MAC) region, and the nation. The statistics from each defined target area are calculated from a specific set of claims. A hospital should be able to identify these claims in their data warehouse and conduct their own analyses on the drivers for any occurrences appearing to be an “outlier” statistic.

When the RACs and other auditors submit their demands for supporting documentation as part of a “complex” review, and ultimately demanding repayments, it's a good idea to begin and continue conducting reviews on the completeness of responses for those claims and also for claims that have similar characteristics.

Data mining also can be used for anticipating and ultimately defending the “automated” reviews of RACs and other auditors. For example, claims with an inordinate number of time-based procedure codes in a 24-hour period can be flagged for internal review. And, the implementation of claim “scrubbers” can prevent the submission of National Correct Coding Initiative coding pairs or the duplicate billing of codes for the same patient on the same date-of-service.

In conclusion, a different adage about collecting data-“What can be measured, will be audited”-for healthcare insurance claims applies better to the new era. The question is whether a hospital is saving the right data to successfully defend paid claims and avoid potential repayments. Other benefits of doing so include greater efficiency at less cost when appealing later demands for re-payment and enhancing the reporting of quality and outcome initiatives.

Phil Hurd is a director at Navigant Consulting, based in Baltimore, Md
Bo Martin, Ph.D, is an associate director and statistician at Navigant Consulting, Chicago Healthcare Informatics 2010 October;27(10):48-50