Original story posted on: January 17, 2012

Predictive Modeling: A Silver Lining in the Perfect Storm

By
Updated on: June 22, 2012

As the provisions of the Patient Protection and Affordable Care Act begin to go into effect, the healthcare industry is headed toward what many consider to be the perfect healthcare storm. Providers, and physicians in particular, are facing uncertainty with respect to what they can expect in terms of reimbursement for the care they provide and even the way they practice medicine.

 

These tumultuous times come on the heels of a decade of what was supposed to be an era of transformation in healthcare, characterized by compliance and regulatory regimes meant to curb escalating costs.

 

Working in what is considered one of the nation's most highly fragmented and heavily regulated environments, physicians have scrambled for advice on achieving compliance with complicated rules and regulations while trying to avoid costly and time-consuming scrutiny and audits. Many providers have experienced these audits, or at least know stories about their colleagues being subject to them for nothing more than inadvertent billing errors. While many new physicians are wondering what they have gotten themselves into, however, their more seasoned counterparts worry about the future. Adding to their fears is the mixed message sent by government agencies responsible for enforcing rules against fraud and abuse.

 

On the one hand, the U.S. Department of Health and Human Services (HHS) Office of Inspector General (OIG) praises physicians, finding the majority to be honest and that very few are actually investigated for fraud - while on the other hand, the agency reports that hundreds of millions of dollars are lost to healthcare fraud, waste and abuse annually. Providers and their patients would benefit if investigators could more effectively and efficiently identify providers worthy of more detailed scrutiny.

 

Rules-based Methods

 

As long as current rules-based methods are used by payers to identify fraud, all providers will be subject to the same treatment - and constant worry that one day they too will face a fraud investigation. Payer inability to prioritize investigations means that all providers are treated with the same degree of scrutiny and all claims are reviewed with the same intensity. Provider stress is intensified because they find themselves dealing with administrative details rather than focused on patient care. The 2009 federal legislation providing for the use of predictive modeling to identify suspect claims before they are paid, and the suspension of payments pending investigation, may represent a silver lining to providers, however.

 

Predictive Model

 

A predictive model is designed to identify provider behavior characteristics likely to influence or suggest future behavior. In predictive modeling, data is collected for relevant characteristics, a statistical model is formulated, predictions are made and the model is validated or revised as additional information becomes available. The model may employ a scoring formula that evaluates claims at the time of processing and before payment. If a claim appears to be high-risk or out of the ordinary, the claim will be flagged for further scrutiny. The scoring model provides multiple reasons why claims become flagged for review. This flagging and application of codes is done in an automated manner.

 

Know the Score

 

Another innovation in predictive modeling is the ability to perform real-time assessments of every single transaction. The scoring model is capable of processing the heavy volume of healthcare transactions that currently exists in the medical provider and insurance communities.

 

A workflow management and workstation methodology efficiently presents only high-risk transactions to the system for automatic rejection or review by a fraud analyst. Real-time scoring has two important benefits.

 

One, it has the potential to reduce the incidence of fraud and abuse that must be investigated because it detects unusual behavior patterns prior to payment; therefore, it prevents the completion of fraudulent or abusive transactions. Second, it helps providers who simply make mistakes correct them while the claims are still available and familiar.

 

A key capability of a predictive modeling approach is that it is designed to "learn" as it is being used to review claims. Aberrational behaviors routinely are reviewed and calibrated against verified instances of fraud and abuse by scientists trained in analytics. Detection system models then can be refined as part of an ongoing "learning" process. This process periodically is incorporated into new versions of scores that show improved performance and adjustments to changing conditions.

 


Know When to Hold

 

Predictive modeling also can provide automated decisions to hold, deny or allow payments depending on uniqueness of the transaction in question. Proponents of using predictive modeling to detect healthcare fraud and abuse believe it is an effective means to reduce fraud, an efficient method of implementing targeted educational tools, and a benefit to the vast majority of "good" providers who currently are being treated nearly the same as "bad" providers.

 

Automated fraud prevention and analytics likely would yield more programmatic savings than the current pay-and-chase methods used by Medicare.  Predictive model methodology seeks to identify the likelihood that a provider will have future unusual behavior patterns (or is "bad," per se) by evaluating current and historical behavior and assigning a score.

 

Generally, a high score indicates high risk of fraud, abuse or abnormal behavior. Any provider below a score threshold or cut-off value is considered to be generally compliant, or "not bad." It can be expected that between 90 and 95 percent of all healthcare providers are among the "good guys" and submit claims that are not fraudulent, wasteful or abusive.

 

Automated Predictive Modeling

 

In its most conservative implementation, automated predictive modeling will complement rules and edits by identifying new, emerging and unknown patterns of aberrant behavior and preventing fraud and abuse before payment. In a pilot demonstration project recently completed on Medicare providers, many of the high-scoring providers (in which a high score is "bad") scored high because of overutilization, mistakes or duplicate services.

 

Understanding patterns of billing errors and overutilization can be very useful for physicians who are going to have to reduce costs and increase the quality of outcomes in order to get paid the "incentives" being offered by new payment models.

 

Again, much of predictive modeling relies on identifying provider behavior characteristics and patterns that indicate fraudulent, abusive or incorrect practices. Physicians historically have detested traditional "profiling" because it generally is perceived to be a method that insurance companies use to compare their performance to their peers and exclude them from their networks. If physicians decide to embrace this new scoring technology, which only attempts to identify behavior that appears to be fraudulent or abusive, they end up not just saving money by avoiding wasting their time jumping through administrative and compliance hoops, but they take better care of their patients because they are able to focus on what they are trained to do: practice medicine.

 

About the Author

 

Karen Mandelbaum, J.D., MHA, is an attorney with both technical and policy experience in the healthcare industry. Karen is an associate attorney at the law firm of Tilton & Dunn PLLP. Karen has a strong background in the contracting, compliance and regulatory affairs issues that uniquely affect healthcare organizations. She has extensive experience representing clients who have been subjects of both public and private payer fraud investigations and recoupment actions. Karen earned her law degree in 2005 from William Mitchell College of Law. In addition, she holds a master's degree in healthcare administration from the University of Minnesota's Carlson School of Management.

 

Contact the Author

 

karen.mandelbaum@fortelanalytics.com

 

To comment on this article please go to editor@racmonitor.com

 

The Use of Observation in Patients Undergoing Outpatient Procedure

 

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