Effects of financial incentives on motivating physical activity among older adults: results from a discrete choice experiment

There is extensive evidence that physical activity exerts a strong protective effect on the health of older adults. This includes reducing the risk and progression of non-communicable diseases and other disabling conditions [1‐4], improving cardiorespiratory and muscle fitness [5], health-related quality of life [6, 7], psychological well-being [1] and longevity. It also has the potential to contain rising medical costs [8, 9]. Despite these benefits, current data reveal low levels of physical activity levels among older adults [10, 11] in nearly every developed country.

Walking is an accessible form of physical activity with proven health benefits [12] that can be conducted across a variety of settings without specialized equipment. It is also the most common form of physical activity among older adults who engage in regular activity [13]. It is for such reasons that walking programs have been promoted as a means to increase activity levels among sedentary older adults. Yet, in many communities, participation in structured programs targeting older adults remains low [14, 15].

By increasing the benefits of participation, economic theory suggests that the use of incentives could increase participation rates in structured walking programs. Incentives have been shown to be effective in positively shaping various health behaviors and health outcomes [16‐24] although there is limited data on its suitability and efficacy among older adults [25]. Moreover, even if effective, incentives come with a cost. One way to offset the cost is to include an enrolment fee. It is possible that the fee may deter some potential participants, but the net effect of incentives and an enrolment fee in affecting uptake by older adults in structured programs remains an empirical question. Therefore, the goal of this study was to conduct a conjoint analysis survey aimed at identifying the extent to which stated uptake of walking programs could be increased via the use of incentives, either with or without a corresponding enrolment fee.

In this stated preference conjoint study we estimate the net effect of several types/magnitudes of incentives and enrolment fees for candidate walking programs among a sample of older, multi-ethnic adults in Singapore. We hypothesize that incentives will increase the stated likelihood of participation in the programs regardless of the presence of an enrolment fee. Conjoint analysis is well suited for this task because it allows for generating predictions of program effects in a low-cost and efficient manner for several candidate programs, the most promising of which can then be tested in a randomized controlled trial to quantify the effects on both uptake and outcomes.

Table 2 shows the demographic characteristics of our sample. The median age was 59.5 years. Over half the respondents were female and had less than primary school education. Consistent with the population in Singapore, the vast majority were Chinese. Slightly under half were employed yet nearly 80% did not meet activity guidelines.

Regression results from the conjoint analysis are presented in Additional file 1: Table S1. Figure 2 graphically presents these results after centering the preference weights to 5 and rescaling to between 0 and 10 for ease of exposition. Attribute levels with higher preference weights were preferred to those with lower weights. The most preferred walking program was a convenient, free (in terms of travel cost and enrolment fee) program with one 45-minute session per week and a $450 cash incentive. When statistically significant, the attribute levels were in the expected direction. None of the levels for travel cost were significantly different from one another nor was there a difference between the $20 and $50 enrolment fee. In terms of incentive type, cash payments were the most preferred incentive but not significantly different from NTUC supermarket vouchers. Both were preferred to Medisave (health savings) credits and sporting goods vouchers were least preferred.

The relative value of various types of incentives can be seen in Table 3. One dollar in cash is roughly equivalent to $1.09 (95% CI: $1.01 to $1.18) in supermarket vouchers, $1.73 ($1.52 to $1.98) in Medisave vouchers, and $3.89 ($2.99 to $5.54) in sporting good vouchers. All of the differences are statistically significant.

Figure 3 shows estimated uptake for the sample program shown in Table 4 with 1) no enrolment fee and 2) a $50 enrolment fee. As expected, the free program results in higher predicted uptake at all incentive levels, and greater incentives generate greater stated levels of enrolment. However, including a $50 enrolment fee only decreases enrolment by an average of 2%. Based on these results, a free program with $500 worth of incentives would generate enrolment rates of 58.5%; charging $50 to enroll only lowers this figure to 55.7%.

Figure 4 shows how uptake varies between those who state they are meeting physical activity guidelines (”Active”) and those who state they are not (“Inactive”). Almost half of Actives claim they would participate in the sample program even if they were not offered any incentives. The percentage of actives increases to as much as 69% for the highest incentive amount. In contrast, only 38% of Inactives claim they would join a free program. However, when incentives are provided, their stated participation increases at a faster rate than for Actives. At $500, over 55% of Inactives state they would join the program. Because they are the majority of the population, regardless of the incentive amount, Inactives represent at least 72% of participants, and this number increases with higher incentive levels.

The objective of this study was to identify a promising walking program for older, multi-ethnic adults in Singapore, and then to test the effects on stated enrolment when incentives and/or an enrolment fee are offered. Not surprisingly, results showed that respondents most prefer the least onerous program with the greatest benefits. This is one involving minimal travel time and cost, comprising the least number of weekly activity sessions (1 session per week), offering a 450SGD cash incentive and without any enrolment fee. Although minimal, one brisk walking session per week would account for 30% of recommended weekly activity if participants walk at the recommend pace. Moreover, this level of activity is still associated with some health benefits, especially among sedentary adults [29‐31]. Therefore, it may be worth offering such a program, even with incentives, if it could encourage sustained increases in activity levels.

A key finding of this study is that financial incentives increase predicted uptake for all programs offered and more so among the least active older adults. For the hypothetical six month program offered, providing SGD$150 for meeting participation goals (SGD$25 ≈ USD20 monthly over 6 months assuming 1SGD = 1.25USD) increases predicted program uptake by 8.5% (assuming no enrolment fee). Raising this to SGD$450 (SGD$75 ≈ USD 60 monthly over 6 months) increases uptake by 13%. These results demonstrate the potential for even modest incentives to increase uptake. Moreover, although cash was clearly the most preferred option, supermarket vouchers, which could potentially be purchased in bulk at a discount, were a close alternative. Health savings account credits and sporting goods vouchers/equipment are not recommended as respondents place a lower value on these rewards and thus they are less of a motivator.

Although we show that incentives increase program uptake and have been shown to be successful in improving outcomes in prior activity studies [25, 32], they are not without cost. One way to offset those costs is via an enrolment fee. These results suggest that an enrolment fee is a viable option as it has a minimal impact on participation. Moreover, the largest impact occurs when going from a free program to a program with a nominal fee. Increases beyond this level can be set so as to recover the cost of the incentives, and will likely have only a modest additional loss in participation; less than 2% based on our estimates when fees are within the $SGD20 to $SGD50 range. Moreover, if the enrolment fee is seen as a deposit whose return is dependent upon program completion, it is likely to have the added benefit of reducing attrition and increasing program effectiveness [21].

When incentives are provided, there is always the concern of adverse selection. In this case the concern is that the incentives go to those who would have been doing the required amounts of activity even without the incentives. Although it is unclear whether these results would generalize to other older adult populations, we found that incentives have a greater stated effect on enrolment for inactive participants. This result may not be surprising given that active participants may have their routines well established and may be less interested in these types of programs. A financial incentive may also be less of a motivation to them, as they have already made the decision to be active. Nevertheless, active participants may benefit through accruing more health benefits to the extent that incentives induce additional physical activity [33‐35]. Monetary incentives are likely to have the greatest effect on participation and outcomes for those who are least active.

Although the first of its kind in a multi-ethnic sample of older adults, this study adds to growing evidence of the potential for financial incentives to motivate health behavior change [24, 25, 36, 37], specifically in promoting activity among older adults in the US and other countries. These results are corroborated by Finkelstein et al. [25] who observed significant, large effects in an incentive-based physical activity trial among older adults aged 50+ where participants assigned to the intervention arm logged an average of 4.1 hours of activity a week relative to their counterparts in the control arm who logged only 2.3 hours over 4 weeks. However, several limitations must be noted. The primary limitation is that the results are based on a survey, and not a real world experiment. The extent to which these surveys serve as a good basis for successful program implementation remains unknown. This should be an area of future research. As such, the results should be interpreted with caution. Yet this is both a limitation and a strength. This study used a low cost survey elicitation technique to generate hypotheses about what might happen were these programs to be offered; future studies should test these predictions in the real world. Moreover, as with any study, even if accurate for Singapore, the generalizability of these results beyond older adults here is unknown. Future studies will need to be targeted to the population of interest. Besides, field studies involving older adults would also need to account for the unique cognitive and psychosocial characteristics of this sub-population [38]. Also, real world cost-effectiveness studies would be a necessary next step to evaluate the long-term feasibility of such programs in specific settings. Finally, incentives, even if effective at increasing uptake, may not be sustainable as they may not translate into sustained behavior change. However, in at least one study, increased activity persisted even after the removal of incentives [39, 40].

The use of incentives has raised concerns about a “crowding out effect”. This would result if the incentives undermine intrinsic motivation. Yet, although theory suggests this could be the case, there is no empirical evidence to support it [41]. Several studies show that incentives increase activity levels of both children and adults. As a result, this appears not to be a legitimate concern [25, 32, 40]. Also, providing incentives may not be affordable in the long term. Although we show that an enrolment fee is one strategy to overcome this concern, other strategies will be required to engage sedentary individuals for whom incentives are ineffective. Identifying these strategies should be a topic of future research. Regardless, these results suggest that incentives may be part of a comprehensive solution aimed at increasing activity levels of older adults.

Study findings add to the growing literature that incentives are a feasible and potentially effective strategy to increase program uptake among older adults. Although cash is the most preferred incentive type, supermarket vouchers are a close alternative. To offset associated costs, enrolment fees are a viable option. Future research should assess the affordability of incentive use and sustainability of model programs in addition to identifying other strategies to encourage sustained activity of older adults.