Geometric Shape

Blog Post

Navigating Medicare Disruption

by Steve Runfeldt and Michael Blix

In August, the U.S. Congress passed the Inflation Reduction Act, which promises to lower costs and improve outcomes for American seniors. Beginning in 2026, the Secretary of the Department of Health and Human Services (HHS) will choose a selection from the most expensive Part D drugs each year to come under negotiated price restrictions. Other changes are pending.

These changes will have a disruptive impact on the consumers’ process of shopping and selecting a Medicare plan. Already about 10% of Medigap or Medicare Supplement switchers each year move to Medicare Advantage (about 1% of MedSupp members overall). Those people who base their selection on plans with the lowest cost for their most expensive drugs will likely begin to base their selection on other criteria. But how can an insurer predict which benefits will become most important?

Conjoint Analysis

In recent years, one of the most effective predictive tools at our disposal has been consumer surveys that employ a combination of Conjoint and Max-Diff Analysis.

Conjoint Analysis presents consumers with a series of choice tasks. Respondents are asked to select between sets of test plans composed of randomly selected levels of standard benefits. Benefits tested by Conjoint Analysis typically include brand, premium, drug deductibles, maximum out of pocket, doctor copays, drug copays, formulary, hospitalization benefits and more recently dental, vision and hearing benefits, among others.

Conjoint Analysis looks at the contribution of each benefit to the consumer’s final plan choice. Conjoint results are most effectively viewed in an online simulator that allows the user to test different benefit levels in order to produce health plans that are optimally attractive to most Medicare consumers. Further, the survey respondents can be grouped by demographic, health or attitudinal factors to show the most attractive plans for specific market segments.

Conjoint Limitations and Max-Diff

Conjoint Analysis works most effectively when the number of attributes/benefits tested is no more than 10 to 12. If more benefits are tested, the reliability of the statistical models starts to become unstable. But health plans have far more than 12 options to consider.

One solution to this limitation is the use of the Max-Diff methodology. In Max-Diff, consumers are presented with groups of five or six items at a time and are asked to select the benefit they most prefer and the one they least prefer. An anchoring question is often included as an extra measurement of preference—do they actually like the item they most prefer in that task? Do they actually dislike the one they least prefer? On the basis of their selections, the Max-Diff model can provide optimal combinations of benefits that are likely to appeal to the largest group of people.

Compared to Conjoint Analysis, the limitation of Max-Diff is that it cannot easily discern between different levels of a benefit. It can tell whether a dental benefit combined with a vision benefit is better or worse than a transportation benefit combined with an OTC allowance, but it cannot tell the user whether a $200 dental benefit combined with a $500 vision benefit is better or worse than a $500 dental benefit combined with a $200 vision benefit.

Adaptive Conjoint

The most cutting-edge methodology for dealing with a changing Medicare marketplace is Adaptive Choice-Based Conjoint (CBC) analysis. Adaptive CBC is capable of handling a much larger number of plan attributes and levels than is traditional Conjoint. As the name suggests, this method adapts to the stated needs of the consumer. In addition to handling a larger number of attributes, the adaptive nature of the exercise also leads to better quality data due to respondents only seeing options that are relevant to their tastes.

Adaptive Conjoint often starts with a larger number of plan options and asks the survey respondent to indicate which benefits are most important to them in their shopping for a new Medicare plan. It then includes only those benefits that are of the greatest interest to that survey respondent. Survey respondents can also be asked about their interest in different levels of a plan attribute with the tasks modified accordingly. For example, someone who is only interested in zero-dollar premium plans may only see those plans in their tasks. This allows for inclusion of more attributes than traditional Conjoint.

In addition to narrowing the tasks based on prior questions, Adaptive Conjoint dynamically focuses the test on the respondents’ favored benefit levels. As the conjoint tasks proceed, the program narrows the selections to those which previous responses indicate are likely to be more attractive. In the end, the definition of what respondents are most likely to buy becomes increasingly refined.

As with traditional Conjoint, Adaptive CBC is also most useful to the end user in an online simulator, where competitive “what-if” scenarios that quantify the impact of subtle benefit changes on consumer preference can be modeled.

Predicting the Future in the Midst of Disruption

As the Medicare market continues to change due to new regulations, Deft Research can design surveys using these tools (Conjoint Analysis, Max-Diff and Adaptive CBC) that take into account likely future disruption.

For example, with an Adaptive CBC design we can ask people about their current plan and their current prescriptions. Based on data from CMS, we can group people according to those who take only Tier 1 generic drugs, brand drugs and those who take more expensive Tier 3 and specialty drugs. Then we can see how their plan choices change when the cost of the most expensive drugs comes down. Will people no longer be willing to pay a higher premium if their drug costs are the same on another plan?

To learn more about our predictive tools or to commission custom research. please submit your contact information below: