Wellness Incentives, Equity, and the 5 Groups Problem

Abstract

Wellness incentives are an increasingly popular means of encouraging participation in prevention programs, but they may not benefit all groups equally.

To assist those planning, conducting, and evaluating incentive programs, I describe the impact of incentives on 5 groups: the “lucky ones,” the “yes-I-can” group, the “I'll-do-it-tomorrow” group, the “unlucky ones,” and the “leave-me-alone” group. The 5 groups problem concerns the question of when disparities in the capacity to use incentive programs constitute unfairness and how policymakers ought to respond.

I outline 4 policy options: to continue to offer incentives universally, to offer them universally but with modifications, to offer targeted rather than universal programs, and to abandon incentive programs altogether.

Incentives aimed at individuals increasingly play a role in the organization of health care systems.^1,2 Wellness incentives are intended to encourage uptake of prevention and health promotion programs. A recent survey also found that 56% of large US employers see wellness programs as 1 of the top 3 strategies for curbing cost.³ Savings may result, for example, from reduced health care expenditure owing to a healthier workforce or from incentives structured in a way that shifts health care cost from employers to employees. The goals of health promotion and cost containment may come into conflict, and the fairness of wellness programs depends significantly on their implementation. Various ethical issues may arise, but a central concern is equity, because ideally, all who are offered incentive programs should enjoy equal opportunity to access them, especially when associated benefits are substantial.

Regulations issued by the US Departments of Labor, Treasury, and Health and Human Services in 2006 distinguish between 2 principal forms of incentives.⁴ Process incentives may offer a premium discount or rebate for participating in, for example, an exercise, weight-loss, or smoking cessation program. Outcome incentives link monetary benefits to meeting certain risk factor targets, such as body mass index (BMI) or blood pressure thresholds. The regulations impose no cap on process incentive levels, but for outcome incentives they initially specified that reimbursements must not exceed 20% of the total cost of an employee's coverage (or $965, according to the 2009 average cost of coverage of a single individual). The Patient Protection and Affordable Care Act, passed in March 2010, increased this to 30% (or $1447), with the option of 50% ($2412) in exceptional cases.^5,6

Incentives may be funded through gain sharing, but the regulations also explicitly permit cost shifting

from plan sponsors to participants who do not satisfy the standards, from participants who satisfy the standards to those who do not, or some combination of these.⁵

Depending on the exact implementation, wellness incentives may therefore lead to an increase in cost of coverage for some enrollees. Other countries, such as Germany, have had similar systems in place for some time, although incentive levels are usually much lower (generally < $100), no difference in amounts is specified between outcome and process incentives, and financing may come from gain sharing only and not from cost shifting.⁷

Individual-level incentives have been used widely outside of health policy. For example, many airlines and supermarkets provide discounts for loyal customers, and auto insurers offer lower premiums for customers with few or no claims. Program enrollment is usually straightforward, and failure to participate or to qualify for rewards typically means losing out on some benefit. Applying this model to the health care context, however, raises some concerns.

THE 5 GROUPS PROBLEM

To understand these issues, it is useful to consider the responses of 5 types of people to incentive programs that are offered universally to all enrollees of a health plan:

1. the “lucky ones,”

2. the “yes-I-can” group,

3. the “I'll-do-it-tomorrow” group,

4. the “unlucky ones,” and

5. the “leave-me-alone” group.

Depending on the exact characteristics of particular programs, the impact on these groups varies, of course. Nonetheless, this somewhat abstract model may bring clarity to the ongoing debate about the acceptability of different incentive programs, whether they focus on process or outcomes. The framework illuminates significant differences across groups of enrollees in the extent that programs succeed in promoting behavior change. It also shows that people differ in their ability to make use of incentive programs. The 5 groups problem therefore concerns this question: At what point do disparities in the capacity to use incentive programs constitute unfairness, and how should policymakers respond?

My analysis is based on a review of the public health literature on incentive use and behavior change, personal insights resulting from involvement in the evaluation of incentive programs,⁷ and conceptual analysis of the characteristics of incentive users. I began with established concepts and developed a more nuanced framework that can be applied directly to wellness programs in planning, practice, or evaluation. Differentiation between the 5 groups is not intended to provide an exhaustive and exclusive model, to suggest that beliefs are never shared across groups, or to imply individuals may not belong to different groups in different stages of their lives. My purpose is to illuminate plausible distinguishing features between different groups of users in the context of incentive programs.

Incentives give rise to a range of issues,⁸ and in addition to an equity assessment, a fuller review of their appropriateness should also take into account several other factors. These issues include evidence and rationale (What are the policy's principal goals, and are they acceptable?); intrusiveness and coerciveness (Can the objectives be achieved through less obtrusive means?); and affected third parties (Does the implementation interfere with relationships, e.g., between physicians and patients or employees and employers?).^9,10 These are important considerations but are beyond the scope of this article. In policy formation, equity issues will clearly always need to be considered alongside other important ethical concepts, such as autonomy (or self-governance), efficiency, and cost-effectiveness. In my discussion of equity, I draw primarily on the concept of horizontal equity, which can be understood as demanding that people who are the same in relevant ways, such as having the same clinical need, are entitled to equal (or equally affordable) health care.^11,12

Characteristics of the 5 Groups

The lucky ones.

Almost any incentive program will cover people who qualify for associated reimbursements without any form of behavior change. By habit, some people simply enjoy eating healthily and exercising regularly and do so quite effortlessly. Their behavior is hence compatible with the wellness program spirit, even if the incentive benefit—for example, a process-based reimbursement for going to the gym regularly or an outcome-based incentive for meeting certain BMI thresholds or for not smoking—does not lead to behavior change and, strictly speaking, does not function as an incentive. Others whose actions may remain unaffected are people whose dispositions are not as well aligned. For example, some people may eat in the most unhealthy ways and never exercise and still have favorable BMI values. Despite the dissonance between their motivations and a wellness program's spirit, they may reap the same benefits as their health-conscious counterparts, without any change in behavior or motivation (to some extent, such behavior is related to the concept of free riding in the economic literature).¹³

The yes-I-can group.

Another group of people would not normally have performed the benefit-qualifying behavior, but the incentive may be a welcome occasion—though perhaps not the sole reason—for trying to overcome inertia or lack of determination. The incentive's nudge, coupled with their underlying motivation, provides an effective basis for action. For most in this group, incentives are likely to feel like a deserved reward. The benefit may help initiate behavior change in the first place, or sustain it, where intrinsic motivation is not yet sufficiently developed.

Alignment of motivation and action cannot be taken for granted: behavior change may also occur more grudgingly, for example, where people care less about the supposed health benefit but participate mainly because they feel bribed by the level of the incentive. The yes-I-can group may therefore have subgroups: happy and grumpy. Conceptually and in practical terms, the yes-I-can group is also known as the group of responders in the literature, yet it cannot be assumed that all—or even the majority—of those offered incentive programs are, in fact, responders. Furthermore, it is plausible to assume that responders’ attitudes differ by the mode of incentive: some may find a soft process incentive most attractive and feel overly pressured by a hard outcome incentive that requires, for example, meeting BMI targets. Others may respond better to a more robust challenge that requires meeting hard thresholds. Such variation needs to be considered in designing an effective and acceptable intervention.

The I'll-do-it-tomorrow group.

Some people share the desire for behavior change with the yes-I-can group but, for a range of reasons, may not act on it. They may feel unable to try, or when they try they often fail. The reasons may stem from their everyday circumstances, such as poor availability of affordable and healthy food or insufficient time to prepare it. They may lack access and time for physical exercise in a safe environment. They may face above-average levels of professional or personal stress and resort to coping mechanisms such as smoking. Such factors can render outcome incentive programs, such as quitting smoking or achieving specific BMI values, significantly more challenging. Upbringing may also play a role: some participants likely received more encouragement than others to be self-motivated and self-efficacious. Therefore, even process incentives such as lower health care costs in return for gym attendance may be taken up more readily by some than by others. For many in this group, incentives may be extremely tempting, yet the amounts at stake can be as far out of reach as the branches of the fruit-laden trees were for the mythical Tantalus.

The unlucky ones.

For biological, medical, or other reasons that are completely external to their volition, some people face such strong constraints that, whatever they might do, they are simply unable to meet the criteria associated with specific outcome or process incentives such as BMI targets or gym participation. For example, some people with genetic mutations will always be obese, regardless of how much they exercise or control their energy intake. As with the I'll-do-it-tomorrow group, incentives that are simply out of reach will make little sense for the unlucky ones.

The leave-me-alone group.

Some people might qualify for wellness incentives but voluntarily decide not to use them. They may already meet targets or could do so easily or could effortlessly participate in incentivized activities, but still resist. They may feel patronized or “nannied” by wellness programs; they may also believe that incentives introduce an inappropriate element of competition in health plans that they think ought to be based on a principle of mutuality and fair risk sharing. Or, on quite practical grounds, they might judge the effort required to register for programs to be too burdensome.

Fairness Issues

It is clear, then, that universally offered wellness incentive programs can give rise to several equity problems:

• Some people may receive benefits, even if their motivation and behavior run counter to the spirit of wellness programs.

• Behavior change is not always required, and some people may receive benefits for default behavior—whether this is the result of deliberate previous choice or unreflective habit.

• Some people face constraints attributable to weakness of the will, poorly developed self-efficacy, or strong medical or societal constraints. Meeting targets or participating in health promotion activities requires a much greater effort of them than of others. Still, when they fail to begin or complete an incentive program, they must forgo the benefit in the same way as those who had sufficient opportunity of choice, but who voluntarily chose not to take part.

Clearly, the extent to which inequalities in incentive use occur in practice and the extent to which we might find them unfair depend critically on the way incentive programs are implemented. Many more people will typically be able to use process than outcome incentives. Therefore, an employer who uses process incentives only is more likely to enable all to secure associated monetary benefits.

One of the most important equity questions is how easy it is for employees with different backgrounds and abilities to avail themselves of the opportunities created by incentive programs. In addition, consideration should be given to the level of benefits. Disparities can become more inequitable if benefits (which, often, may only be open to some) are substantial. The recent health reforms significantly increased the reward levels for outcome incentives. This change, and the interest of employers in using incentives for cost shifting and cost reduction, may result in a scenario where only a relatively small number of people among the lucky ones and the yes-I-can group benefit from large incentives, with others, particularly the I'll-do-it-tomorrow group, at a disadvantage.

POLICY OPTIONS

The relevance of this analysis clearly needs to be ascertained in empirical studies of specific programs, and in light of the scarcity of such work to date, my framework is intended to help guide such research. Several policy options (which may change in the wake of empirical analysis) may be considered to respond to the 5 groups problem: (1) continue to offer incentives universally, (2) offer them universally but with modifications, (3) offer targeted rather than universal programs, or (4) abandon incentive programs altogether (Table 1).

TABLE 1—

Implications of Policy Responses to the 5 Groups Problem in Wellness Incentive Programs

	Groupsa
Policy Options	Lucky Ones	Yes-I-Can	I'll-Do-It-Tomorrow	Unlucky Ones	Leave-Me-Alone	Analysis
Offer universally	Benefit	Benefit	Don't benefit	Don't benefit	Don't benefit	Unlucky ones lose out. I'll-do-it-tomorrow group, treated identical to leave-me-alone group.
						Some lucky ones reap benefits even if they do not change behavior or comply with spirit of policy.
Offer universally, modified	Benefit	Benefit	May benefit	May benefit	Don't benefit	Create alternative standards for unlucky ones and I'll-do-it-tomorrow group; this can improve fairness but faces practical and arbitrariness challenges.
						Shift from offering alternative standards in response mode to proactive mode to reduce negative aspects of petitioning.
						Shift focus from outcome to process incentives.
Targeted, not universal	Don't benefit	Benefit	Benefit	May benefit	Don't benefit	No incentives for lucky ones because they do not require further encouragement.
						Potential for curbing cost, if focus is on improving health status of worst off.
						Minimizes potential for exacerbating existing disparities in health and wealth.
						Reverses financing of benefits where benefits result from cost shifting; instead of the poorer and unhealthy financing the benefits of the better off and more healthy, controversially, the opposite happens.
Abandon	Don't benefit	Don't benefit	Don't benefit	Don't benefit	Don't benefit	No unfairness from different use of incentives, but also no potential to use incentives as complement to action at the level of social determinants of health for health promotion.
						Strongest case if it can be shown that other measures to improve population health are equally or more effective.

^aLucky ones qualify for incentives without behavior change; yes-I-can only qualify if they succeed in changing behavior, which may be more likely because of the incentive; I'll-do-it-tomorrow only qualify if they change behavior but perceive obstacles to change; unlucky ones are practically impossible to qualify; leave-me-alone could qualify in principle, but refuse to do so.

Universal Incentives

As in other areas of social policy, people simply differ with regard to their uptake of opportunities, and as long as participation is voluntary, such variation may not warrant a specific policy response regarding access to wellness incentives, whether they involve process or outcome. This assumption appears to underlie the German incentive framework.⁷ However, despite the much lower levels of incentives used in Germany, proponents of this view ought to explain why it should be acceptable for the unlucky ones to lose out. It would also be desirable to justify identical treatment for the I'll-do-it-tomorrow group, who have the same aspirations as the yes-I-can group, and the leave-me-alone group.

In the United States, wellness incentives and the increased option of cost shifting have the potential of reintroducing medical underwriting by the back door, something that would be hard to reconcile with the overall spirit of the 2010 health reforms, which sought to improve access to affordable care for all Americans.

Universal Incentives With Modifications

Some will argue that differences between groups matter and that a policy response is therefore required. US policy could be altered by offering incentives in ways that are more responsive to people's circumstances and agency. The 2006 regulations specify that where achieving outcome incentive standards is “unreasonably difficult due to a medical condition … [or] medically inadvisable,”^4(p75037) a reasonable alternative standard must be provided, so that individuals can qualify for reimbursements. These provisions at least partially address the situation of the unlucky ones and make some headway toward preventing unfairness. However, providers are not required to offer alternatives proactively; rather, people who feel challenged must request them. People with self-efficacy problems may be disadvantaged by this requirement. Petitioning may also be experienced as embarrassing or humiliating. Adjustments would hence be necessary. Similar, optimized provisions should also be made for some or all in the I'll-do-it-tomorrow group.

Making adjustments for individual circumstances and capabilities may be time-consuming and costly and should avoid arbitrariness. One solution may be to change the overall focus from outcome to process incentives or to reward progress in achieving goals, rather than meeting one-size-fits-all thresholds (e.g., by abandoning the criterion of normal BMI and incentivizing weight reduction instead or by rewarding smoking cessation program participation rather than quitting itself). Careful monitoring and evaluation of programs can help significantly in designing sustainable and effective incentive programs that maximize equality of opportunity for uptake for all, while minimizing inequities in securing benefits (or avoiding penalties, where incentives are implemented by increasing insurance contributions).

Targeted Incentives

Another response would be to abandon universal in favor of targeted incentives, be they process or outcome based. This strategy addresses the argument that incentives for the lucky ones at the expense of the I'll-do-it-tomorrow group and the unlucky ones are inequitable and merely exacerbate existing disparities in health and economic status. Although I know of no data for the United States, in Germany 19% of the initial cohort of incentive program participants belonged to the most privileged quintile, and only 11% belonged to the poorest.¹⁴

Resources could be directed toward incentives that are more sensitive to the specific circumstances of the I'll-do-it-tomorrow group, for example, by combining incentive programs with improved access to healthy food or exercise opportunities. Alternative standards for the unlucky ones could be provided proactively. In addition to improving equity, this might help curb overall health care expenditure, if savings can be achieved from better health among those with proportionately higher levels of morbidity. However, this approach could reduce cost shifting to unhealthy employees; in fact, the opposite might happen, because the better off could lose out on incentives and end up subsidizing the worse off. Although this could be welcomed on equity grounds, politically such a move is likely to face considerable difficulties.

Abandon Incentives

The most radical approach argues that if there are no incentive programs, no inequity can arise from them. Abandoning all incentive programs could mean that resources required for reimbursements and their administration could instead be used for other interventions, such as public health programs aiming to improve health at the population rather than the individual level. However, an important objection to this approach is that practically all population-level interventions also face equity issues, as is well-known from, for example, the uptake of health literacy campaigns, food-labeling requirements, and vaccination programs.

It is also not clear that incentives could not be structured to promote equity. Human psychology is notoriously complex, and incentives, complementing public health efforts at the level of the social determinants of health, may well be effective tools for some people to initiate and sustain behavior change. The rejection argument would therefore be strongest if it was able to demonstrate which equally or more effective alternatives to individual-level incentives are available, without similar drawbacks.

CONCLUSIONS

A wide range of activities can be the subject of process and outcome incentive programs. Incentives can be implemented as carrots or as sticks, with varying levels of financial benefit or disadvantage. Clearly, each proposed program requires an individual assessment, ideally based on empirical data, to assess acceptability and effectiveness. In addition to equity issues, other ethical concepts and principles also need to be considered. It is therefore difficult to come to general conclusions about the permissibility of incentive programs. However, in planning, conducting, and evaluating incentive programs, the needs of groups who will not automatically benefit (the I'll-do-it-tomorrow group or the unlucky ones) or who will resist all inducements (the leave-me-alone group) must not be forgotten. Fairness requires us to assess the impact of policies on all affected groups, and explicit justification is required where the likelihood of uptake is significantly lower for some groups than for others, especially where the levels of incentives are substantial and already disadvantaged groups have lower chances of benefiting.¹⁵

At the same time, the perfect must not be the enemy of the good. Judgments must be made, and incentive programs that will be accepted by all users are likely to remain elusive. As Table 1 shows, only 1 of the 4 policy options (offer incentives universally, with modifications) has the potential to benefit all groups with an interest in an incentive program's objectives, and even then it will still be opposed by those who object to the principal mechanism. Some policymakers may decide that the respective sizes of the groups matter most, and that an acceptable outcome would be to implement the policy that benefits the largest number of people. They should nonetheless be asked for explicit justification and should not assume that all who are offered programs have the ability to use them.

Empirical evidence from monitoring and evaluating people's responses at the offering stage, and assessing whether their attitudes change (for better or worse) over time as they participate, is of crucial importance in determining how many people belong to which group and whether adjustments to policies are required to maximize equality of opportunity. A comprehensive reporting structure of key data would be especially helpful, both on fairness grounds and to ensure that best practice for identifying successful approaches to behavior change can be documented and implemented more broadly. In the absence of such structures, we must trust that employers and their interest groups will independently generate and use these data appropriately.