In the recent ICD10monitor article ICD-10 is Risky Business, Holly Louie states that “the best way to fix claim errors is to not make them in the first place.” Since the Oct. 1 implementation of ICD-10 has passed, how will your organization know when and where errors are likely to occur so you can take action beforehand?
Predictive analytics can provide the insights necessary to more effectively manage claim denials. Think of it as an early warning system for new patterns of denials and appeals.
Early Warning System
Many of the claims risk management algorithms currently in use are based on rules learned over decades of experience processing claims coded with ICD-9. With the introduction of ICD-10, these algorithms may not perform as well in detecting problems. ICD-9 code-specific quality assurance rules must either be bypassed or modified for ICD-10.
Bypassed rules result in less information for automated systems to identify errors and denial risks. Modified rules may be flawed and result in reduced problem detection effectiveness and higher rates of false alerts. Add in the unpredictable changes in behavior exhibited by both providers and payers as the ICD-10 conversion gets underway, and you have a good recipe for a substantial decline in performance of these rule-based systems.
How can providers make sense of the new rules of the game during this period of rapid change and intransparency?
Predictive analytics can reliably identify claims that are most likely to be denied and/or delayed due to a variety of errors and payer disputes. When predictive analytics is implemented using machine learning techniques, accelerated fact-finding can be achieved as well. That’s because machine learning excels in recognizing patterns.
An automated claims data analysis process supported with machine learning will rapidly detect new or emerging patterns in denied or contested claims. With the transition to ICD-10, the ability to immediately discover these patterns enables providers to respond by educating staff, developing corrective procedures, and targeting training efforts.
The ability to rapidly detect such patterns also enables providers to have statistically supported discussions with payers about observable trends in their denial and dispute behaviors. Payer processing rules and procedures are continuously evolving, but they are expected to be even more volatile during the ICD-10 conversion period.
In short, targeted and effective machine learning implementation can provide a solid early warning system to detect and respond to emerging changes. They also continuously adapt claims risk identification without relying exclusively on rules developed through trial and error associated with historical processes, with ICD-9 as the prevailing coding standard. With this adaptive approach to prioritizing claims reviews and targeting education and training efforts, providers can mitigate financial risks associated with ICD-10 conversion.
Operational Opportunities with Predictive Analytics
As it pertains to claim denial management, the use of predictive analytics can be applied to numerous functional areas. The following are a few examples:
- Authorization: Evaluate referrals and authorizations for corresponding future denial risk earlier in the process.
- Eligibility:Highlight patient-centric denial risk factors during data entry and analyze patient demographics (with automated demographic enrichment). By enriching existing data sets with additional external patient information that is typically not utilized by healthcare providers, new insights and causative factors will emerge in efforts to identify denial risk prior to filing a claim.
- Coding/Medical Necessity: Identify patient, clinical, insurance, and provider inter-relationships that influence denial risk for targeted intervention.
- Revenue Cycle Management
- Claim Processing: Raise alert at claim submission, before claims scrubber/clearinghouse processing.
- Claim Management:Prioritize submitted but unresolved claims for intervention based on increasing risk of denial over time and funds at risk.
The Benefits of Operationalized Predictive Analytics
Hospitals have been struggling with claim denials for decades, and ICD-10 likely will exacerbate such issues. Current claim denial management processes, systems, tools, and reporting in many cases are not adequate to eliminate preventable denials. A new approach is required – predictive analytics. Predictive analytics can benefit the hospital by:
- Predicting where denials are likely to occur next and driving the operational change needed to reduce denial rates;
- Providing understanding of what non-obvious factors are correlated and contributing to your initial denial rate;
- Improving denial overturn ratesand increasing associated earned revenue, accelerating cash flow, and reducing days in AR and costs to manage denials;
- Improving patient, employee, and physician satisfaction;
- Offering predictive insights (e.g., automatically routing denials and underpayments to the right individual/team for faster resolution); and
- Pinpointing denial and underpayment root causes for continuous AR process improvement
Numerous industry associations and publications, along with the Centers for Medicare & Medicaid Services (CMS) have detailed the potential adverse consequences that ICD-10 could have on revenue cycle and claim denials, specifically. Using predictive analytics can help your organization better manage the risks associated with ICD-10.
About the Authors
Tony Broyles is the vice president of healthcare solutions for Illuminate360. He leads Illuminate360’s business development efforts where he applies his 30-plus years of experience shaping the predictive insights solutions for healthcare organizations. Prior to joining Illuminate360, Tony was the chief executive officer at Georgia Lung Associates.
Tripp Cox is the chief technology officer (CTO) and managing partner at Illuminate360. As CTO, Tripp leads Illuminate360’s product strategy, R&D, and technical operations. Prior to Illuminate360, Tripp served in CTO and R&D executive roles for Damballa, EarthLink/MindSpring and Sun Trust Bank.
Comment on this Article