Written by xScion
on December 12, 2017


With healthcare spending at $3 trillion a year, rising medical costs and increased competition, it is more important than ever for healthcare payers to validate that claims payments are accurate. Payment Integrity refers to the prevention of inaccurate or “improper” claims payments due to a variety of reasons, e.g. fraud, waste and abuse, duplicate payments, and a lack of coordination of benefits. Consider that 3%, or $60 billion annually, of all healthcare spending in the U.S. is lost to fraudulent claims spending alone. Additionally, 3-5% of all medical claims payments in the U.S. are paid by insurers that are the responsibility of other payers.


Traditional Claims Initiatives

Traditional efforts to reduce improperly paid claims and associated costs focus on the “pay-and-chase” post-payment recovery model (think phone calls, letters and even legal action). In the post-payment recovery model, payers leverage rules-based technology, applying “if-then” scenarios to analyze paid medical claims in order to identify improper payments and data patterns. This strategy has had some success, but it loses its value since the analysis is based on paid claims data and requires recovery efforts for reimbursement. Recovery attempts after the fact are ineffective and costly due to the many, often manual, steps that are required in order to evaluate, request and seek reimbursement - ultimately eroding the value of the original claim. 

In addition to the post-payment recovery model, many payers have implemented costly clinical code edit technologies that support nationally recognized coding guidelines in order to identify and prevent improper payments before a claim is paid (e.g. claims with procedure codes that are not appropriate based on a member’s diagnosis). While these rules-based technologies can represent a significant technology investment, companies that offer clinical code editing technologies boast a significant ROI that pays for itself within a few years of use by reducing the total number of improperly paid claims.  These technologies, however, come with expensive upfront costs.


The Good News: Data-driven Instead of Rules-based

As technology continues to advance, there are exciting new capabilities in data-driven predictive analytics and machine learning that can be applied within healthcare claims processing to prevent improper payments in a way that the traditional rules-based approaches cannot. By analyzing historical claims data and applying predictive analytics/modeling and machine learning, potential improper claims are identified and flagged during pre-adjudication. This ensures that additional checks are performed before a claim can continue through the process and be paid (e.g. to ensure a claim for the same service does not already exist as a duplicate claim or to ensure that coordination of benefits has been performed for a member with secondary insurance).  And, because these new capabilities provide insights into the root causes of improper payments, the reoccurrence of the same improper claims scenarios can be eliminated entirely by addressing the underlying problem. Payers can update, for example, existing rules to prevent the same types of errors in future claims.


The Benefits of Predictive Modeling & Machine Learning

Payers that implement these new capabilities, in addition to the traditional approaches described above, will be uniquely poised to achieve additional, immediate benefits to their bottom line through significant reductions in improper claims payments, the elimination of repeat scenario improper claims payments, and reductions in the need for recovery and reimbursement efforts.

If you are a payer and are not yet convinced that data-driven capabilities to improve payment integrity is a worthy investment, consider that projections indicate that they will be more impactful and less costly than traditional pre-payment clinical code editing solutions.

Contact us to learn how your organization can benefit by applying these exciting new capabilities in predictive analytics and machine learning.