Miskeying, Miscoding and Mismanagement: The Costly Mis-takes of Risk Adjustment

It has been a busy six months in the Medicare Advantage (MA) risk adjustment space, with plenty to keep the industry’s best and brightest thinking about what changes are in store for PY 2024 and beyond related to the risk model transition from v24 to v28, and the RADV Final Rule. Much of this prodigious brainpower is focused on regulatory compliance, as evidenced by the vast amount of media attention being paid to the question at its foundation – is there an overcoding problem in MA, or an undercoding problem in traditional Medicare (TM)? The answer depends upon whom you ask, and in the end, the Office of the Inspector General (OIG) will have the last word as it makes it recommendations for compliance program improvements that are focused on diagnoses that are at high risk for miscoding or miskeying. In this article/blog, we’ll talk about the answers to this question according to experts, as well as CMS/OIG, and then discuss what health plans and other risk-bearing entities need to consider for enriching their compliance programs in the manner OIG expects.

As we consider the answer to the question about over or undercoding, let’s first see what some of the industry’s leading experts have to say. In a recent debate related to MA reform, two of the industry’s top experts, Rick Kronick of UC San Diego and Tom Kornfield of Avalere had some very different opinions about the upcoming changes¹. Kronick asserts that MA coding artificially inflates patient acuity, and points to his own studies that show the average risk score for a MA member in 2020 was 20% higher than an average risk score for a traditional Medicare fee-for-service beneficiary. His conclusion is that this is attributable to aggressive MA coding practices. He also says that MA plans have a more favorable selection of patients and are probably somewhat healthier. His overall impression is that the changes CMS announced for PY 2024 will not go far enough to address the alleged aggressive coding practices. However, in another study to which Kronick contributed, approximately 40% of members coded with quadriplegia in a 12-month period do not have this diagnosis on a claim in the subsequent 12 months. Therefore, it is hard to argue overcoding in MA. Providing the opposing viewpoint to Kronick is Kornfield’s assertion that CMS should be basing their rate changes on MA data and not fee-for-service (FFS) data. CMS has been collecting encounter data in MA for over ten years now, and the GAO has periodically audited CMS’ encounter data collection practices to ensure that there is adequate oversight and accountability for the completeness and accuracy of this data.  There is also the fact that FFS providers are incentivized to report procedures and not diagnoses, since MA providers are paid on a capitated basis and in FFS providers are paid per procedure. Thus, there are stronger incentives in MA to code accurately.

The OIG certainly has a lot to contribute to the answer to the question about over/undercoding as the industry’s “watchdog.” Anyone who has ever undergone AAPC's training courses for their coding, auditing, documentation and compliance credentials knows that the AAPC includes content related to the OIG Work Plan and why any health plan’s highest priority is to monitor the OIG Work Plan and consider its content in the development and maintenance of an overall corporate compliance plan, inclusive of medical coding. Starting in 2018, OIG began targeted, RADV-like compliance audits² of six to ten high-risk diagnosis code groups for some of the nation's largest health plans for PY 2014, 2015 and/or 2016. They were specifically looking at the medical record documentation to determine if the diagnosis was properly reported for risk adjustment, and they also looked at diagnoses that are commonly miskeyed due to transposition errors in billing data entry on claims:

• Major depressive disorder: typically treated with medicine but no corresponding prescription
• Acute stroke: diagnosis on a physician claim but no corresponding inpatient claim
• Acute MI (heart attack): diagnosis on a physician claim but no corresponding inpatient claim
• Vascular claudication (clots): typically treated with medicine but no corresponding prescription
• Embolism (also clots): typically treated with medicine but no corresponding prescription
• Breast/colon/lung/prostate cancer: a diagnosis that did not have surgical, radiation therapy or chemotherapy within 6 months preceding or following the diagnosis
• Acute stroke and acute MI combination

In these audits, OIG found for six of the larger plans audited, 66% of the HCCs were not validated, which correlated to $30,340,000 in overpayments to these health plans. OIG stated that most of these health plans' compliance programs did not have any practices that focus on these diagnosis groups that are at high risk for being miscoded, and that was their main recommendation for compliance program improvement. Plans acknowledged that it is not clear how they could identify problematic diagnosis codes without selecting and reviewing the claim information. OIG said that is exactly why they made the recommendation and that this type of targeted auditing on high-risk and commonly miskeyed diagnoses should be a part of their compliance program. OIG further stated that such auditing falls within the scope of Section 42 CFR § 422.503(b) Federal Regulations Regarding Compliance Programs That Medicare Advantage Organizations Must Follow. Furthermore, there is increased OIG scrutiny of diagnosis data collected via member health risk assessments and submitted as an unlinked chart review but those diagnoses not being reported on any other provider encounter record for the members. During these 17 OIG audits that took place starting in 2019, they found no support for nearly 69% of diagnoses used for risk adjustment from these assessments, resulting in $113 million in overpayments.