Understanding Medicare Advantage Risk Adjustment
Medicare Advantage, an alternative to traditional Medicare, currently covers over 35 million individuals and manages more than $615 billion in annual federal payments. The payment to insurers is determined by the complexity of their enrollees' health conditions through a process known as risk adjustment. This mechanism ensures that plans providing care for sicker populations receive higher payments to manage anticipated expenses.
The system, however, has introduced a significant incentive for insurers to document a greater number of diagnoses per patient. With the advancement of AI-driven coding tools, these technologies analyze extensive clinical records and suggest diagnosis codes far more swiftly than the initial frameworks of risk adjustment could have envisaged. A critical issue has been whether these tools prioritize revenue generation over accuracy and the ambiguity over who ensures their veracity.
The Medicare Payment Advisory Commission's report from March 2026 reveals an excess expenditure of $76 billion on Medicare Advantage enrollees compared to traditional fee-for-service Medicare, with $22 billion attributed to coding intensity. This difference emerges from how Medicare Advantage plans document diagnoses compared to their fee-for-service counterparts. Until recently, coding technologies operated with minimal scrutiny regarding the precision of AI-generated codes.
Historically, oversight of AI tools in risk adjustment was minimal. Insurers acquired coding technologies from vendors, with AI systems analyzing clinical notes and suggesting diagnosis codes often adopted by human coders under tight schedules. These codes were submitted to CMS without thorough checks on the validity of the AI's logic or accuracy.
Recent findings have highlighted significant compliance issues. The Department of Justice reached over $670 million in settlements with two major healthcare organizations due to allegations of inflated payments through questionable risk adjustment practices. The Office of Inspector General’s audits indicated a high percentage of high-risk diagnosis codes lacked necessary documentation, showing complete failure rates in audits of certain acute condition categories.
Regulatory developments have been swift. The Centers for Medicare & Medicaid Services expanded its audit workforce substantially and plans regular audits on all Medicare Advantage contracts. A January 2026 memo clarified that AI should support human coders without final authority. New industry-wide compliance guidelines have been released, scrutinizing practices vulnerable to misreporting, such as add-only chart reviews.
A clearer accountability framework is emerging for AI in healthcare payments, demanding evidence-based and traceable AI recommendations. Human oversight must be meaningful, allowing coders to amend AI suggestions with documented reasoning. Insurers investing in well-governed risk adjustment platforms are laying the groundwork for accountability, while those relying on unmanaged AI face consequences through settlements and audits as governance standards are enforced.