Chart Recoding and Reprocessing Yields Substantial Gain
A regional health plan asked Change Healthcare to perform recoding/reprocessing on a sample selection of 147 Medicare Advantage members. Chart reviews—conducted with best-in-class, AI-driven coding capabilities—identified new or additive hierarchical condition categories (HCC) totaling $30,291, or about $357,000 on an annualized basis.
Regional health plan
Ensure Appropriate Risk Adjustment Scores by Harnessing Automation to Retrospectively Review Medicare Advantage Charts for Missing HCCs.
AI-Driven Coding Solution
New or additive HCCs totaling $30,291—a 20%-30% improvement over incumbent manual HCC review performance
When it comes to Medicare Advantage plans, the frequent failure by physicians to annually redocument a member’s chronic-disease acuity and/or new and existing comorbidities can skew risk-adjustment scores lower and cost plans significant amounts of money.
The ongoing financial impact of missed hierarchical condition categories (HCCs) has been exacerbated by the pandemic, which dramatically reduced encounter and claim volume through 2020 and into 2021, thus reducing access to medical records.
In early 2021, a regional health plan that had long engaged Change Healthcare to provide risk assessment for its Affordable Care Act (ACA) population requested a sample retrospective assessment of its Medicare Advantage (MA) HCC/ICD-10 coding.
The plan, which has about 8,500 MA members, was contracting with another vendor to provide HCC retrospective analysis. However, the organization was interested in determining whether Change Healthcare could meet or exceed the incumbent’s performance by reviewing charts the existing vendor had already processed.
Change Healthcare’s AI-driven coding solution uses artificial intelligence, natural language processing, and machine learning to scrutinize existing member data and flag cases where key information may have been omitted. AI-driven coding capabilities include scanning medical-record documentation for clinical terms that describe the assessment and treatment of chronic conditions to identify instances where appropriate care was not redocumented for risk-adjustment purposes. The coding solution is aware of what risk gaps have already been reported and what is new or would close an open gap.
In this instance, Change’s expert risk-adjustment team looked at 147 members, then deployed natural language processing to carefully review the population’s chart information to determine whether undercoding may have occurred. Natural language processing can assess all available data, including unstructured clinical information, to provide a complete snapshot of the risk-adjustment value present in medical records.
The result was a 20%-30% improvement of standard manual HCC review results.
All told, approximately 1,000 charts were reviewed. The result was the identification of an additional $30,291 in HCCs, or about $357,000 on an annualized basis. The amount represented a 20%-30% improvement of standard manual HCC review results produced by the incumbent.
With the documentation provided by Change Healthcare, the plan was able to resubmit the supplemental data on encounters to increase risk adjustment scores for the population in question. This not only boosted revenue in the near term, it created the foundation for more accurate risk scoring going forward.