Medicare Advantage (MA) risk adjustment payments have come under increasing scrutiny in recent years. In the wake of rising MA enrollments, increasing expenditures, and a number of recent lawsuits around HCC coding practices, payer and provider organizations are feeling additional pressure to ensure they’re performing due diligence on submitted codes.

Today, MA risk adjustment teams recognize the need for robust compliance programs. However, implementing internal audits has become an operational challenge as the volume of patients, claims, and charts increases year after year. Many teams are under-resourced and struggling to keep pace.

Compliance demands won’t likely slow down any time soon, but there’s hope for risk adjustment teams. Artificial intelligence (AI) can help internal reviewers focus their attention on areas with the greatest audit risk for their organization and ensure payment accuracy.

MA Risk Adjustment Requirements

Every MA plan must comply with program guidelines when they submit HCC codes to CMS. As part of these guidelines, plans are required to provide appropriate documentation for each HCC code, ensure every code follows MEAT (Managed, Evaluated, Assessed, Treated) guidelines, and confirm that supporting clinical documents are signed by a qualified physician type as part of a face-to-face encounter.

Additionally, each MA organization is responsible for implementing a compliance program that monitors compliance risks, puts systems in place for responding quickly to compliance issues, takes appropriate corrective actions in response to compliance violations, and reports potential fraud or misconduct to CMS.

Growing Regulatory Focus on RA Compliance

Regulators are paying close attention to MA program compliance, particularly risk adjustment coding practices. In 2017 alone, CMS estimated it made $14.4 billion in improper payments to MA plans. As more government dollars are paid out to private insurers every year, CMS wants to make sure those dollars are being spent managing real conditions for real patients. Regulators have put pressure on MA compliance in two ways: False Claims Act (FCA) lawsuits and annual RADV audits.

FCA Lawsuits

Over the past decade, CMS has engaged with several large health plans and provider systems over suspected fraudulent risk adjustment submissions. A few of the largest recent cases include:

SCAN Health Plan paid out $3.8 million in 2012 to settle a federal FCA whistleblower lawsuit related to inflated risk adjustment scores.
• UnitedHealth fielded FCA whistleblower lawsuits in 2011 and 2017 for upcoding patient condition codes.
• Sutter Health was involved in a 2018 FCA lawsuit involving the submission of inflated or false patient risk scores.
• DaVita paid $270 million back to Medicare in 2018 for upcoded services.
• Anthem fought with the DOJ in 2018, refusing to turn over risk adjustment coding data related to an in-progress FCA lawsuit.

RADV Audits

With Medicare spending clocking in at $705.9 billion in 2017 and projected to increase to $1.4 trillion by 2027, CMS is exploring every avenue to contain costs. One of these cost containment avenues is RADV audits, which require a subset of MA program participants to provide supporting documentation for codes submitted through claims. Additionally, CMS conducts targeted RADV audits with MA plans who’ve raised red flags due to large increases or decreases in risk scores from one year to the next. Plans must pay back revenue associated with unsupported codes found during the audit process.

Currently, CMS claws back payments only for the subset of codes reviewed as part of the RADV audit process. It also applies FFS adjusters that calculate a permissible level of payment error and limit RADV audit recovery to payments made above the defined error rate. To recover more improper risk adjustment payments, CMS has proposed new RADV audit requirements that involve extrapolating audit findings to all risk adjustment payments submitted by a plan, as well as removing FFS adjusters. If these changes are enacted, RADV audits will put even more pressure on plans to prove the accuracy of their code submissions.

Impact on Health Plans

What does this heightened regulatory focus on MA risk adjustment mean for participating health plans? In a nutshell: It’s more important than ever for HCC codes to be correct out of the gate. CMS is keeping a close eye on notable changes in patient risk, so appropriate documentation is necessary to support audits and related inquiries. Additionally, plans need to pay close attention to the accuracy of clinical documentation and claims submissions, particularly for composite or high-cost HCCs where clawbacks will have a large revenue impact. Lastly, plans need to brace themselves for more frequent audits and broader financial consequences for incorrect submissions in the event that new RADV requirements are implemented for 2020.

How AI Can Help Plans Manage RA Audit Risk

Most MA plans use a manual risk adjustment audit process. Reviewers look at the codes submitted on claims, pull patient charts related to each HCC, and then conduct a review to see if there’s sufficient evidence to support each submission. This is an extremely costly and time intensive process that requires a team of people to execute well. This process also doesn’t scale well with increasing membership and care utilization.

While technology will never be able to supplant human judgment, AI can help plans streamline their risk adjustment auditing process by identifying codes without proper supporting evidence to meet CMS guidelines. Well-trained AI models can:

• Confirm supported HCCs. AI can do the heavy lifting to confirm HCC codes that have appropriate documentation  and pull them out of the review queue so audit reviewers can focus on the codes that need attention.
• Flag codes with insufficient documentation. If documentation related to a HCC doesn’t support the submitted code, AI can flag it for investigation by a reviewer. If the reviewer confirms there’s insufficient documentation, they can use that data to close the documentation gap with the provider or delete the code from the RAPS file.
• Identify high-risk coding issues. AI models can locate specific conditions or providers with persistent coding errors for further training and support to improve physician claims coding efforts.

By augmenting the auditing process with AI, plans can perform more thorough, more frequent reviews of submitted codes to maintain a high level of accuracy throughout the year, and prepare their organizations for potential RADV audits. Ultimately, more efficient reviews will reduce audit risk for plans and financial waste for CMS. Additionally, AI-powered audit processes can scale with increasing enrollments and changing member health needs without requiring a huge team of internal staff to keep pace with the demands of compliance reviews, saving plans time and money in the long run.

Want to learn more about the benefits of AI for risk adjustment compliance programs? Request a demo of HCC Auditor, Apixio’s auditing solution for MA and Commercial plans.