AI and Legislative Actions Aim to Address Medicare Advantage Coding Disputes

Medicare Advantage (MA) plans have faced scrutiny from lawmakers and regulators for aggressive medical coding practices that affect risk-adjusted payments. The Centers for Medicare & Medicaid Services (CMS) launched the Value-Based Insurance Design (VBID) pilot over eight years ago to incentivize high-value care through Medicare Advantage programs. However, plans exploiting coding to maximize payments led the Biden administration to terminate the VBID program early, citing $2.3 billion in excess costs in 2021 alone.

At the core of the controversy is the system of medical billing codes that underpin payments. CMS adjusts payments to MA plans based on the coded health status of enrollees; therefore, plans have incentives to identify and code as many comorbidities as possible, a practice that can inflate the apparent illness burden of patients compared with traditional Medicare. Investigations by the HHS Office of Inspector General revealed billions paid out due to diagnoses derived solely from chart reviews, raising concerns about coding practices and potential overpayments.

Legislators such as Senator Jeff Merkley and Senator Bill Cassidy have proposed legislation—the No UPCODE Act—to curb excessive coding by adjusting CMS risk models and restricting the use of diagnoses obtained from chart reviews. This legislation aims to mitigate incentives that lead to inflated coding and overpayments, which some view as destabilizing Medicare systems.

Opposing views argue that higher risk scores in Medicare Advantage reflect a sicker enrolled population rather than manipulative coding. Reports from advocacy groups and data analyses indicate that new MA enrollees frequently have multiple chronic conditions and higher mortality risks than traditional Medicare counterparts. Furthermore, proponents highlight that diagnostic codes across Medicare and Medicare Advantage must adhere to uniform standards.

Experts acknowledge a fundamental challenge in accurately assessing patient health and assigning value to care based on diagnostic codes. The existing ambiguity contributes to coding discrepancies, prompting calls for technological innovations such as artificial intelligence (AI) to enhance accuracy and transparency in coding.

AI applications in healthcare can assist clinicians by identifying care gaps, streamlining coding processes, and providing evidence-based recommendations during patient encounters. These tools offer audit capabilities and transparency features, enabling better detection of coding anomalies and ensuring coding conforms to clinical evidence.

Leaders in the field emphasize that AI should serve as a decision support tool, keeping clinicians in control of coding and care decisions to avoid misaligned incentives. Proper integration of AI can enhance patient data accuracy and potentially reduce administrative burdens like prior authorization delays.

Addressing risks such as biased data in AI training sets, experts recommend strict auditing, clear transparency, and policies retaining clinicians as ultimate decision-makers to harness AI's benefits while minimizing unintended consequences.

In conclusion, while challenges persist in balancing Medicare Advantage coding practices with accurate risk adjustment and payment models, AI represents a promising avenue to improve compliance, coding integrity, and patient outcomes within Medicare systems. Legislative efforts and technological advances together shape the future framework of value-based care reimbursement.