Differential coding practices led to a significant increase in Medicare Advantage risk scores and an estimated $33 billion in additional payments to plans in 2021, according to a March study published in the Annals of Internal Medicine.
The study analyzed 697 MA contracts across 193 insurers using data from CMS for calendar years 2015 through 2021, covering a total of more than 27 million lives. It looked at risk scores for the MA plans, factoring in various health diagnoses, then calculated the impact of differential persistence and new incidence on risk scores. The analysis also included other factors, such as the number of years a plan has been in the program and the proportion of enrollees with special needs or who receive Medicaid benefits. The method assumed that plans don’t adjust their contract bids in response to coding intensity. If plans do adjust their bids, the actual effect of differential coding on revenues could be even higher.
Though research into MA coding patterns is extensive, no research has estimated the extent to how individual insurers code or the amount of revenue they secure through differential coding.
Six notes:
1. In 2021, the average MA risk score was 1.26, 18.5% higher than 1.07 for traditional Medicare (TM). The researchers said the higher risk scores were due to higher rates of ongoing health conditions being recorded year after year within MA plans and more new health conditions being added to members’ records.
2. In the top 10 diagnostic groups, MA’s persistence was 78.1%, compared to 72% in TM. Cumulative new incidence in MA was 46%, compared to 33% in TM.
3. The effect of differential coding increased MA risk scores by 0.19 points, leading to an estimated $33 billion in additional payments to MA plans in 2021.
4. UnitedHealth accounted for $13.9 billion of that total, or 42% of the additional payments. In 2021, the company had a 27% share of total MA membership. This breaks down to $1,863 per member in revenue, compared to the industry average of $1,220 per member.
5. Chart reviews and health risk assessments were responsible for about half of the increased persistence and new incidence in MA, meaning that differential coding, not disease burden, is the likely cause. The researchers noted that differential coding doesn’t necessarily mean MA plans are overcoding, and TM is undercoding, but the practice does result in higher payments for MA plans.
6. To address differential coding, CMS began phasing in the V28 model in 2024 to reduce the risk score difference between MA and TM. This process will continue through 2026.
“CMS could almost completely neutralize the effects of differential coding in MA by excluding an additional 10 V28 HCCs in a hypothetical “V29,” the researchers wrote. “Alternatively, CMS could use other sources of data, such as survey data from the Consumer Assessment of Healthcare Providers and Systems, to measure risk. Further research is needed to assess the advantages and disadvantages of this approach.”
