The federal government could improve the risk-adjustment model for commercial insurers participating in the marketplace exchanges by making efforts to understand the demographics of enrollees better and considering prescription drug costs in estimations, according to a report (PDF) Thursday from Avalere Health.
It came the same day as a CMS white paper (PDF) examining similar issues in how the Affordable Care Act program could be changed to work better for insurance companies and the overall market. HHS will meet next week to discuss potential changes.
The risk-adjustment program for the individual and small group commercial market offsets the incentive for plans to enroll healthier and less costly people by paying the companies more for beneficiaries who need more care. This is intended to create a level playing field where companies compete based on product quality and patient satisfaction.
The efforts have struggled. One component of the model is risk corridor payments to insurance companies that lose money on plans in the first few years of exchange operation. In November, the CMS said it will only be able to pay out about 13% of those claims, shorting already struggling companies by $2.5 billion.
The risk-adjustment program, however, uses incomplete data and has methodology flaws that threaten to create marketplace volatility, said Caroline Pearson, senior vice president of Avalere.
“This is a new market that is not as stable as we'd like right now,” she said.
The cost estimates are off because the people enrolling in these markets have different demographics than those used to create the estimates. They tend to be older and have a lower income and therefore have more complex illnesses, she said.
This isn't particularly surprising, but more accurate data was not available before, Pearson said.
“One of the reasons they didn't use real demographics is they didn't have these people insured and didn't know what they looked like,” she said.
The estimates also predict the number of people who are enrolled for only part of a year. The system assumes a person will receive their care throughout the year and pay a full set of premiums. In practice, however, some people enroll for part of year and receive the care all at once.
Another key missing piece has been prescription drugs. While the model is used to predict total healthcare costs, including medication, it does not use prescription data.
Pearson said this can skew the numbers because some people with the same condition have vastly different drug costs. A person with early stage or rheumatoid arthritis, for example, typically takes a relatively inexpensive drug. As the disease progresses, however, they eventually take far more expensive biologics. Estimating payments based just on diagnostic codes does not take this into account.
Also, using prescription data could create more complete records. A diagnosis code may be missed, but if a person is receiving medication treatment, it could be included, she said.
The CMS said it has heard concern that companies would shift people to more expensive drugs to try to increase estimates and risk corridor payments. It is considering several model changes that would include varying amounts of prescription data, according to its report.
