Tag Archives: morbidity

Addressing health plan selection bias through risk adjustment

When employers facilitate health plan choices, the method for setting employee premium contributions can create selection bias toward certain options. Selection bias happens when a sicker and more costly population tends to choose one option over another. A common example is when a more open network preferred provider organization (PPO) attracts those employees who want a broader range of providers and use their benefits more than those choosing a limited network health maintenance organization (HMO).

To reduce the selection bias, employers should adjust each option for morbidity. Because selection bias does not change the overall morbidity of the group, it is important to set the premium contributions without consideration for how healthy the subpopulation is within any one option. Otherwise, the option with the sicker subgroup will become more and more costly.

Risk adjustment is used to adjust the costs of two or more cohorts of people so all cohorts can be compared as if each had the same morbidity. A risk score is calculated for each person using age, gender, and medical conditions. Risk adjustment is a way to level the playing field so that the cost differences among options reflect benefit differences as well as network and operational performance differences, but not morbidity differences.

In this paper, Milliman’s Will Fox, Brent Jensen, and Ben Diederich explain in more detail how employers can address health plan selection bias using risk adjustment.

2015 individual market pricing: Morbidity and other considerations

There are several known regulatory and market changes that will affect the pricing of plans sold on the state exchanges in 2015. Items such as the reduction in the federal transitional reinsurance program and increases in the Patient Protection and Affordable Care Act (ACA) insurer fees will influence pricing. Lack of morbidity data related to new populations of insureds creates a level of uncertainty when pricing risk beyond 2014. However, there are ways to gain additional understanding on morbidity for 2015 pricing compared to the approaches used for 2014.

In this new article, Scott Katterman discusses how insurers can take advantage of data sources to gain some perspective into future pricing. Here is an excerpt from the article:

Reduction of federal transitional reinsurance
To ease the transition to a guaranteed issue environment, ACA includes a federal reinsurance program to help mitigate the risk of extremely high-cost individuals entering the individual market. While the reinsurance program will be a significant help to insurers in 2014, its impact will shrink over time. The budget allocated for reinsurance payments will drop from $10 billion in 2014 to $6 billion in 2015. All else being equal, this will reduce reinsurance payments to individual health plans by 40%. Actuaries estimate that the reinsurance program has reduced premiums by 10% to 15% in 2014, so slimming it down will have a certain and material impact on pricing in 2015.

The bottom-line impact of this shift will depend on the size of the individual market in 2015. If individual market growth is significant in 2015, the transitional reinsurance program payments into the individual market may well go down more than 40% on a per-member basis, simply because the $6 billion budget will be spread across a larger individual market in 2015 relative to 2014. As a result, actuaries need to estimate the size of the (non-grandfathered) individual health insurance market in 2015.

There are reasons to think the market may increase materially, such as increasing tax penalties for remaining uninsured in 2015, as well as organic growth of the market as knowledge of and familiarity with the exchanges increases.

Cohort studies: estimating morbidity using actual 2014 data
A technique that we believe will be useful to many carriers is examining the risk scores of cohorts of the individuals which make up their 2014 insured populations. Risk scores based on prescription drug data are likely to provide the most accurate estimates, since that data will be far more complete at that point compared to medical claims data. Prescription drug risk scores can be calculated for each enrollee based on the few months of 2014 data that will be available. These partial-year risk scores can then be adjusted based on analyses of risk score development and completion patterns from prior years, taking into consideration potential differences in completion patterns for new vs. existing enrollees. These adjustments will allow comparisons of 2014 risk scores to prior year risk scores on a more “apples to apples” basis.

Once risk scores are estimated, the membership can be split into cohorts of new vs. existing members for both 2014 and prior year data. Comparisons can then be performed to help answer the following questions:

• How has the overall risk score of our 2014 population changed compared to 2013?
• How do the risk scores of new 2014 enrollees compare to existing members?
• How do the risk scores of new 2014 enrollees compare to new enrollees in 2013 and prior years?
• Has our existing member block risk score increased significantly due to lapses of younger/healthier individuals?

The answers will be very valuable when developing 2015 premium rates.