February 18, 2015

Retrospective Baseline Probe Audits: Problems and Solutions to an Outdated Methodology


Based on surveys conducted by this author and others, it has been determined that the average large practice (greater than 150 physicians) relies upon retroactive baseline probe audits (RBPA) in order to identify patterns of compliance risk among their provider populations.

In general, the methodology for a RBPA involves the services of an experienced procedural coder who is tasked with conducting a complex review of a static number of individual charts for some number of providers within a physician organization.

The number of providers that are included in a RBPA is normally established based upon budgetary and personnel limitations. For example, based on the complexity involved in reviewing a chart for errors, an experienced coder will average around four charts per hour, or 8,000 charts per year (assuming that they commit 100 percent of their time to the baseline audit, which is quite unrealistic).

In fact, if you consider the other responsibilities of a coder within a typical physician organization, you will find that they spend nearly half of their time on education, remediation, and initial coding for procedure-based services. According to the AAPC, the average annual salary for an experienced CPC is $53,489. Therefore, the average cost per chart for one FTE CPC for a baseline audit is around $6.70. 

Looking at the surveys referenced above, the same average large practice reports auditing approximately 60 charts per physician per year, so at 4,000 charts per year allocated for conducting the baseline audit, one FTE CPC can handle the audits for approximately 67 providers per year. At a conservative level of only 25 charts per provider per year, one FTE CPC can support a RBPA for approximately 160 physicians. 

Good News, Bad News

This is the good news. The bad news is that the methodology for conducting the baseline probe audit is fatally flawed when compared against the goals and objectives of that audit, which is to get some representative idea of patterns of potential billing and coding risk within the organization. There are some 15,000 individual procedure codes listed in the physician fee schedule database, and additionally, between HPCPS level 1 (CPT) and HCPCS level 2 codes, there are some 300 modifiers. In theory, there are more possible combinations of procedure codes and modifiers than there are protons in the known universe.

For a typical internal medicine physician, however, 19 individual procedure codes account for the top 50 percent of what he or she reports, and it takes 77 procedure codes to reach the top 80-percent mark. If you were to randomly pull 20 individual charts for a given internal medicine physician, you have a near zero chance of reviewing at least half of the codes they report – and even if this were to occur, reviewing one chart per procedure code is wholly inadequate for the purpose of establishing a pattern of risk. This is why it is my opinion that the retroactive baseline probe audit is both fatally flawed and nearly useless.

Do the Math

Consider this: the typical physician generates approximately 200 claims per month, or 2,400 claims per year. At 8,000 claims per year, then, the average CPC dedicating 100 percent of his or her time to the RBPA could retrospectively review all of the claims for just over three physicians. Therefore, at even 100 physicians, the healthcare organization would require 30 full-time CPC coders for a total of somewhere in the neighborhood of $1.6 million in salary alone. This is, of course, a ridiculous scenario, and in fact, many organizations not only limit the number of claims per physician, but also the number of physicians in total. For example, if an organization has three CPC coders than can commit 50 percent of their time to retrospective baseline probe audits, they have a total capacity of auditing 12,000 charts per year. For a 400-provider organization, at an average of 60 charts per physician per year, they would be able to include only 100 (25 percent) of their physicians in the baseline audit. In essence, they would have the capacity to review 60 charts for every physician once every four years, which most compliance experts would tell you is simply too insignificant of a sample to come to any accurate conclusion regarding billing risk. 

Another downside of the RBPA is the additional financial liability it may impose on an organization. Let’s say that you were to audit a statistically valid random sample of charts (say, 30) for a given provider, and let’s say that you were to find some rate of billing error. Not only would you have an obligation to repay the carrier for the amount overpaid, you could face the obligation to self-disclose, which carries with it the need to extrapolate the error rate to the universe from which the sample of charts was drawn, increasing the potential overpayment damage to the organization. 

Predictive Analytics

The solution to this dilemma is to employ statistical and predictive analytical models. In July 2011, the Centers for Medicare & Medicaid Services (CMS) started using predictive analytical algorithms created in conjunction with Northrup Grumman and Verizon to assess the potential risk of any given claim being subject to fraud, waste, or abuse.

In fact, based on their reporting, 100 percent of all claims submitted to Medicare are now evaluated by this system. Statistical and analytical models can be used to replace the baseline probe audit, and rather than reviewing a critically small number of procedure codes, statistical analyses allow for an automated review of 100 percent of all billing activity. The ROI is actually quite clear: at only 25 charts per provider per year, an analytical engine can replace one FTE coder for every 320 physicians. At the average of 60 charts per physician per year, it can replace the work required by one FTE coder for every 133 physicians. At the average salary of $53,489 for one FTE CPC coder, statistical modeling can save the organization $402 per physician/chart per year while providing approximately 40 times the amount of claims coverage. 

In essence, statistical systems offer a variety of cost savings and increased efficiency for any organization. It is unrealistic to believe that this type of a statistical analysis would completely supplant the need to conduct retroactive chart reviews, and clearly it does not do that. In fact, it encourages chart reviews, but rather than relying on a strategy of shooting in the dark, it providers clear guidance to the coding team regarding which procedure codes and/or modifiers present the most risk for external audit. And once the identification occurs at this granular level, based on accepted statistical practices, the number of chart audits required to identify intrinsic risk is significantly less than that required for the RBPA. 

In fact, I recommend reviewing between three and five charts per code/modifier risk event, which translates to between five and 25 charts per physician-at-risk per year, which is approximately 25 percent of the providers within the average larger organization.

Under this common scenario, most organizations will end up conducting a complex chart review on some six charts per physician per year, which allows for a single dedicated coder to manage the retrospective audit for 1,300 physicians. 

About the Author

Frank Cohen is the director of analytics and business intelligence for DoctorsManagement. He is a healthcare consultant who specializes in data mining, applied statistics, practice analytics, decision support, and process improvement.  Mr. Cohen is also a member of the National Society of Certified Healthcare Business Consultants (NSCHBC.ORG). 

Contact the Author


Comment on this article


This email address is being protected from spambots. You need JavaScript enabled to view it.