Assessing the 2015 MA Stars ratings
This Intelligence Brief was updated on November 21, 2014 to correct an error in Aetna’s Star rating. Aetna has an enrollment-weighted average rating of 4 Stars, not 3.
We analyzed CMS’s data covering 691 MA plan contracts across the 50 states to determine which types of products had achieved the highest average Star ratings. Three key observations emerged from our analysis:
- Average Star ratings differ by product type. Health maintenance organization (HMO) products performed the best, with an enrollment-weighted average Star rating of 3.96
- Plans built around integrated delivery networks (IDNs)1 continue to receive a higher weighted average rating (4.43) than commercial plans (3.81) and Blues plans (3.76)
- The weighted average Star rating for plans offered by commercial and Blues carriers is rising, but commercial carriers are improving at nearly twice the rate of the Blues carriers
Average Star ratings vary by product type
Overall, about 40 percent of the plans achieved a Star rating of 4 or higher. HMO products achieved the highest enrollment-weighted average Star rating (3.96).2 However, if Kaiser Permanente, which received a 5-Star rating, is removed from the analysis, the HMO average drops by 0.15, and local PPO plans have a higher enrollment-weighted average Star rating (3.87). Regional PPO plans achieved an enrollment-weighted average rating of 3.53, a significant improvement from their 2012 rating of 2.99 (Exhibit). A county-contract level regression analysis3 showed that HMO and local PPO plans are associated with a 0.2 to 0.3 improvement in the enrollment-weighted average Star rating compared with regional PPOs.Exhibit 1
- Includes both provider-led integrated delivery networks (IDNs) as well as payor-led IDNs. The analysis includes Kaiser Permanente as an integrated carrier.
- Methodology used to calculate enrollment-weighted average is described in the appendix.
- A multiple linear regression model was developed to explain 2015 Star rating for a contract at a given county, using CMS historical MA plan data. Key independent variables considered in the model are plan type, carrier type, contract age, urbanity, low-income subsidy, plan size and growth rate, county-level Medicare size, and MA penetration. Weighted least square was used for variable coefficient estimate. The weight variable was 2014 MA enrollment size for a contract at a given county. Only variables with p-value <0.01 are retained in the final model.