Cost Nudges Had Minimal Effect on Clinic Selection as Tiered Benefit Design May Already Drive Smarter Choices: Tim McDonald, PhD, MPP, and Bryan E. Dowd, PhD

The American Journal of Managed Care^® (AJMC^®) recently spoke with Tim McDonald, PhD, MPP, and Bryan E. Dowd, PhD, 2 authors of "Benefit Design and Consumer Information: Results From a Randomized Trial," a study published in the April issue.

This randomized controlled trial (RCT) evaluated whether tailored cost information could reduce consumer inertia in selecting primary care clinics (PCCs) within the Minnesota State Employee Group Insurance Program’s (SEGIP) tiered cost-sharing system.

During 2020 open enrollment, members in the treatment group received emails highlighting local PCC tier information, but the intervention had only a small effect on clinic selection and no effect on health plan changes. Nearly 85% of members were already choosing PCCs in the 2 lowest cost-sharing tiers before the intervention, suggesting the existing tiered design may have already encouraged lower-cost choices.

In this excerpt from the interview, McDonald and Dowd discuss the study’s main findings, the limits of information-based interventions, and future directions for improving consumer choice.

Listen to the full conversation here.

This transcript has been lightly edited for clarity.

AJMC: Can you summarize the study's main findings? Were there any results that particularly stood out to either of you?

McDonald: The headline finding is that there was only a marginal effect on where people chose to get their primary care. We found a couple statistically significant effects. Members with family coverage were more likely to move to a lower tier than those with single coverage, and the intuition could be that there could be higher switching costs if you have a family, but they also could save significantly more. Also, members in tiers 3 and 4, which are the higher-cost tiers, if they changed their clinic, were more likely to move to a lower tier vs a clinic in their current tier. But otherwise, very small, often not statistically significant effects.

On one hand, you could see this, and it could kind of be a bummer. You'd say, “Well, we did this big RCT that was statewide, and we didn't see any effects.” But we find it kind of interesting, because as part of this effort, we first established a baseline to understand where the members were currently choosing clinics, and 85% of members in this state were choosing clinics in tiers 1 and 2, which is kind of a strong indicator that the benefit design might already be doing a lot of the work.

At baseline, if 85% are choosing clinics in tiers 1 and 2, the tiering system and the way that they need to make an affirmative decision every year, and then they see when they make the decision of their clinic the different cost-sharing, that could already be leading to fairly efficient decision-making. In other studies, we've interviewed and surveyed the members and asked why they choose clinics when they're in more expensive tiers, and they have reasons that they do that: convenience, specific physicians, or longevity of relationships. There are nonfinancial reasons why people choose clinics in tiers 3 and 4.

That's one big insight: the tiering design could already be efficient. Another insight is that members with family coverage are more likely to choose a lower tier, suggesting that the potential economic savings are significant. Together, these suggest that in the tiering design, even though we saw a small effect from providing the nudge to the members, in many ways, there wasn't much room for people to move further.

AJMC: Based on your findings, what lessons emerge about the limits of information-based interventions, and what do they suggest about the broader role of information in health care decision-making?

Dowd: Let me be clear that our study concerned only the choice of medical clinics. It didn't concern the choice among different types of medical care. We found that employees really liked getting information, even though there were relatively few changes. SEGIP took that finding to heart, and they've now started giving employees that information and even more about quality during the open enrollment period each year.

Now, when consumer information doesn't lead to massive changes in choices, that doesn't mean that the information is worthless. For example, the information might help a consumer confirm that they actually did make the best choice, and that would have value to the consumer.

Finally, I've learned that we really have to be realistic about the amount of change that we should expect from better information. If anyone has ever looked at a consumer report on automobiles, what they'll see, generally speaking, is that there is one brand of automobile that, if you take any message away from Consumer Reports, it would be, “Don’t buy that brand of automobile.” They emphasize it over and over and over again with different models of cars. It's not unique to a single model, and yet, you go out on the highway and the highway is jammed with people driving that particular brand of automobile. They love that brand of automobile, and that's perfectly okay.

As Tim said, there's a lot of stuff we don't get to see about why people pick particular brands. But again, that doesn't mean that Consumer Reports is worthless. It's actually worth a great deal to consumers who are in the market for automobiles.

AJMC: What directions do you see as most promising for improving consumer choice in the future?

McDonald: The really big opportunity is in a comprehensive approach to addressing the market failures. This is not about choice or no choice, consumers or no consumers, fee-for-service or not fee-for-service. The debate often gets stuck on a level that's 1 or 2 steps above the foundational level. The foundational level is getting really powerful incentives to reward measurably better care at less cost. You need really powerful upside incentives. Walt McClure has a line: if we're going to ask clinics and providers to make less per patient, they need to have more patients as a result. If we become more efficient in the health care system, we will be making less revenue per patient, so you need to get more patients as a result.

If you can get more patient volume, that can be a viable upside incentive. A possible really powerful downside incentive would be if, on an apples-to-apples comparison, risk-adjusted, you're unable to be as efficient as more efficient providers, then you lose patient volume to those providers. This is not a marginal change per patient. You're losing a patient and the entirety of the revenue. We need really powerful upside and downside incentives to drive the kind of transformation that's needed. That's just kind of a basic macro-market design reality. Almost all the discussion around reforms doesn't get to that fundamental diagnosis of how we get really powerful incentives.

There is just a huge opportunity in doing that, because we hypothesize, we have this theory of the case that a comprehensive approach to addressing the information incentive problems via something like the tiered networks could move us in that direction. We're not positing this as the only approach, but the reason we find the SEGIP model interesting is that it takes a comprehensive approach to addressing all the different market failures, and a lot of the evidence we see merging, in scientific terms, is supporting evidence of the theory of the case. It doesn't prove it, but it's all supporting evidence.

SEGIP is the largest employer purchaser of health care in Minnesota, but it's one program. If Medicare were to adopt the design principles that the SEGIP system is based on, if other large employers were to adopt the design principles, we might be able to then see more marketwide effects, and that would be really interesting.

Dowd: I'm beginning to think about health care reform in terms of sort of axiomatic assumptions; let me suggest them and see if you would agree.

One is that physicians remain the most trusted source of information, not health plans, not the government, not medical societies, not artificial intelligence, at least for the time being; it's your physician whom you trust for medical information. Second, there's wide existing variation in prices and practice styles among medical practices. Third, a concept of what consumers are actually looking for when they go into the market for health care services. I would assert that they're looking for effective preventive care, an accurate diagnosis when they have a medical problem, and a prescription of effective treatment, and they would like all of that at competitive prices.

Now, if you agree with those axioms, they point you in a particular direction. That is, consumers simply need to be told how well medical practices perform on each of those measures, and when consumers choose lower-cost medical practices, they need to share in the same things. It's difficult to see how it could be simpler, and if you believe the axioms, they point you toward a particular set of reforms.