How to close 2026 strong while building a more accurate, audit-ready program for 2027.
For Medicare Advantage (MA) risk adjustment leaders, the second half of 2026 is a balancing act. Teams are working against submission deadlines, monitoring performance, and closing documentation gaps while preparing for policy changes that will affect CY 2026 service-year data, CY 2027 risk-score calculations, coding operations, and audit readiness.
In this article, we speak with Amanda Liu, senior director at Cotiviti, who outlines how MA organizations can strengthen year-end performance while preparing for the policy, audit, coding, and technology changes ahead.
Priorities for the final stretch of 2026
At this point in the year, risk adjustment teams face no shortage of competing demands, Liu explains. But the most important work can be organized around five actions drawn from the policy, audit, and technology developments already shaping 2026 decisions—and the way MA coding and submission programs will need to operate in 2027:
- Quantify exposure. Measure reliance on unlinked chart review records (CRRs) for 2026 dates of service and identify where diagnoses lack an appropriate encounter link or may not meet other risk-adjustment eligibility requirements.
- Protect the last mile. Strengthen post-visit, pre-claim controls that can identify supported additions, potential documentation gaps that warrant an appropriate provider query, potentially unsupported diagnoses, and document issues before claims and encounter data move downstream.
- Strengthen two-way validation. Pair programs that find supported diagnoses that were missed with processes that validate reported diagnoses with consistent rigor and initiate timely corrections or deletions when support is insufficient.
- Target independent review. Apply independent second-level review based on diagnosis error risk, complexity, anomalies, prior findings, and coder or vendor performance, with payment impact as context rather than the driver, to complement rather than replace broader sampling.
- Connect intelligence to action. Coordinate risk and quality opportunities across provider and member workflows, then measure the effect on documentation, encounter acceptance, risk-score accuracy, and downstream outcomes.
None of these actions are about choosing between financial performance and compliance, Liu explains. They're about improving risk-score accuracy, revenue integrity, and audit defensibility together. Some can be sharpened before year-end with the capabilities you already have; others are about building the program for 2027. Organizations should begin with the priority that has the shortest runway—and for many, that may mean quantifying their exposure to unlinked chart review records.
Quantify exposure to unlinked chart review records
The urgency comes down to timing, according to Liu. In the CY 2027 Rate Announcement, the Centers for Medicare & Medicaid Services (CMS) finalized a payment policy that excludes diagnoses from unlinked CRRs from CY 2027 risk-score calculations for MA organizations other than PACE, with a narrow exception for beneficiaries who move from an MA contract in one parent organization to an MA contract in a different parent organization. CMS will begin applying the exclusion with the CY 2027 mid-year run, and CY 2026 service-year data will feed the CY 2027 mid-year and final risk-score runs. This is therefore a current operational issue, not a future-year planning exercise, Liu says.
Organizations can still submit unlinked CRRs, but those diagnoses will not be used in CY 2027 risk-score calculation unless the exception applies. The impact depends on how much an organization relies on unlinked records today and how well it can appropriately link diagnoses to the encounters where they were documented. Quantifying exposure now gives teams time to address linkage and submission issues for CY 2026 dates of service. Any link should reflect the encounter where the diagnosis was actually documented; linking alone does not make a diagnosis eligible.
Protect the last mile before data moves downstream
Protecting the last mile means identifying problems before a claim or encounter has already gone out, Liu explains. Shifting some of that work earlier—into prospective, concurrent, and especially post-visit, pre-claim workflows—adds a control point while the encounter evidence is still current. Post-visit, pre-claim review can distinguish among three situations before data leaves the organization: a supported addition, where a code was omitted even though the documentation and coding rules appear to support it; an ambiguous or incomplete record that warrants an appropriate provider query; and a diagnosis that does not appear supported and needs review and correction through established processes.
Providers remain responsible for the clinical diagnosis and its documentation, and qualified coders assign codes under official guidelines. Any documentation amendment must follow proper documentation practices; it cannot manufacture support after the fact. Paired with more connected data and automation, these controls can reduce administrative burden, reflect member acuity accurately and sooner, and give the program better intelligence to act on early.
Strengthen two-way validation and year-round audit readiness
Audit-ready validation works best as a year-round discipline rather than a periodic exercise, Liu says. In May 2025, CMS announced that all eligible MA contracts would be included in newly initiated Risk Adjustment Data Validation (RADV) audits for each payment year. Separately, recent Department of Justice False Claims Act settlements have involved allegations not only that organizations submitted unsupported diagnoses, but that they failed to delete or withdraw diagnoses their own reviews showed were unsupported. That second responsibility is often the harder one to operationalize, Liu notes.
Programs are better positioned if they support both sides of coding accuracy at once: identifying and validating clinically supported diagnoses and correcting or withdrawing diagnoses that are not supported. Every diagnosis used for payment should be traceable to an acceptable medical record, a qualifying face-to-face encounter with an acceptable provider type, the date of service, and the coding guidance in effect for that date. This two-way validation pairs programs that find missing diagnoses with processes that validate what has already been submitted and initiate timely corrections or withdrawals when support is insufficient, according to Liu.
Target independent review and govern automation
Continuous audit readiness works best with dedicated workflows for validation, correction, and deletion, backed by independent second-level review of selected high-risk diagnoses and first-pass coding results. A broad, statistically valid sample shows how the program performs overall, but it says less about the diagnoses, coders, or vendors that carry the most risk; the two approaches complement each other. Organizations can target second-level review using diagnosis error risk, complexity, anomalies, prior findings, and coder or vendor performance, with payment impact as context rather than the driver. Potential deletions should receive the same rigor as potential additions.
The same discipline needs to extend to automation, Liu explains. Automation can help prioritize records, spot anomalies, and direct limited coding resources, but its governance is becoming as important as its efficiency. OIG’s February 2026 MA compliance guidance, which is voluntary and nonbinding, flags the risk of prompts—including AI-generated prompts—that encourage unsupported diagnoses or diagnoses that did not affect a member’s care, treatment, or management. Its recommendations include educating staff and vendors on appropriate use of queries and prompts; reviewing the software and systems used by plans, vendors, and providers; monitoring vendor activities; using data-filtering logic and analytics to identify anomalies and outliers; and auditing diagnosis data before and after submission. Automation can accelerate the work; disciplined governance makes the output explainable, reproducible, and defensible.
Connect intelligence to action across provider and member workflows
Coordinated provider and member engagement complements the other four priorities and helps connect insights to clinically appropriate care, Liu says. Provider engagement remains where documentation accuracy starts: targeted education, outreach, and documentation review help ensure that conditions evaluated and addressed during a qualifying encounter are represented appropriately, while the clinical diagnosis and its documentation remain with the provider.
Member engagement is becoming harder at the same time it is becoming more important. Schedules are busier, outreach fatigue is real, and appointments become harder to secure as the year winds down. In an encounter-linked environment, coordinated engagement creates clinically appropriate opportunities for providers to evaluate and document conditions during qualifying encounters. The most effective outreach begins well before deadlines and is backed by regularly refreshed visibility into performance, emerging gaps, and provider and member needs.
When outreach comes from separate programs, providers and members can receive multiple, sometimes conflicting requests, creating confusion and fatigue. Connecting intelligence to action means coordinating risk and quality opportunities across provider and member workflows, then measuring the effects on documentation, encounter acceptance, risk-score accuracy, and downstream outcomes. Measurement turns coordination from a sound idea into a program that can keep improving.
Tips to prepare coding operations for 2027
CMS is continuing the fully implemented 2024 CMS-HCC risk adjustment model (V28) for non-PACE organizations in CY 2027. Beginning with the CY 2027 midyear risk-score run, CMS will exclude diagnoses from audio-only services for both PACE and non-PACE organizations when every risk-adjustment-eligible service line on the record carries modifier 93 or FQ. Organizations should therefore confirm that their coding, encounter-data, and validation workflows can identify affected records before submission.
The FY 2027 ICD-10-CM code set also took effect October 1, 2026, so CY 2026 dates of service span more than one release, including the April 1, 2026, update and the new FY 2027 edition, and some conditions may be reported differently before and after that date. Coding, edits, and HCC mappings need to follow the release that applies to each date of service. Not every code change affects HCC assignment or payment, but misalignment can create validation errors, rejections, mapping discrepancies, or reconciliation work. This effort is particularly significant in conjunction with the planned expansion of CMS RADV audits to all eligible MA contracts.
Build year-round habits instead of relying on a year-end push: The central principle is simple: do not wait, Liu says. Pre-claim controls deliver the most value when they run throughout the year rather than being saved for the final quarter, and outreach works best when it begins before schedules tighten. Programs that make both part of their regular cadence tend to carry less risk into the final stretch. For most teams, the practical step this quarter is to tune the controls they already have; building new integrated workflows is work for 2027.
Closing 2026 strong is part of preparing for 2027: Closing the year strong and preparing for 2027 are the same job, not competing ones, Liu says. The most common missed opportunity is treating audit readiness as a once-a-year cleanup instead of a year-round discipline of validation, correction, and deletion. As coding requirements evolve and audit scrutiny increases, organizations that sharpen existing controls now and continue investing through 2027 in encounter-linked documentation, integrated workflows, governed automation, two-way coding controls, and continuous audit readiness will be better positioned to adapt.
Those capabilities support clinically appropriate care and accurate, supportable diagnosis data while improving payment accuracy, revenue integrity, and audit defensibility. Acting on these five priorities now can help risk adjustment leaders move beyond reacting to deadlines and build programs that deliver accurate, supportable results under both today’s requirements and tomorrow’s expectations.
About Amanda Liu
As senior director of product commercialization and market interface for Cotiviti, Amanda guides new risk adjustment product offerings and ensures the delivery of value to existing customers. With a strong background in data-driven strategy, product management, and advanced analytics, Amanda leverages data insights to shape solution strategies, and enhance outcomes for health plan clients.
About Cotiviti
Cotiviti’s infrastructure platform enables healthcare organizations to deliver better care at lower cost through advanced technology and data analytics that improve the quality and sustainability of healthcare in the United States. Cotiviti’s solutions increase transparency and collaboration between payers and providers while empowering them to reduce medical and administrative costs, enable better health, improve claims payment efficiency, streamline operations, drive interoperability, and advance value-based care. Its customers serve the majority of U.S. healthcare consumers, providing coverage and care for over 300 million members and patients. For more information, visit www.cotiviti.com.