This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Internet-based health programs have been shown to be effective in reducing risk for cardiovascular disease. However, their rates of enrollment and engagement remain low. It is currently unclear whether rewards from established loyalty programs can serve as a conditioned stimulus to improve the use of a freely available Internet-based program.
The objectives of the study were to (1) examine enrollment rates and levels of engagement with the My Health eSupport program between a Conditioned Reward group and a Control group, and (2) investigate the influence of loyalty rewards and participant characteristics on levels of enrollment and program engagement.
The study sample (n=142,726) consisted of individuals who were offered enrollment in an Internet-based health intervention (My Health eSupport) after completing the Heart&Stroke Risk Assessment on the Heart and Stroke Foundation website. My Health eSupport programs provided encouragement and tips for lifestyle change. This is a free, self-guided, fully automated program that proactively delivers tailored email messages at 2-week intervals based on the participant’s stage of motivational “readiness” and priority for lifestyle change. Participants in the Conditioned Reward group were offered a single exposure of 20 loyalty reward points from the Air Miles loyalty program for completing the Heart&Stroke Risk Assessment (10 reward points) and enrolling in the Internet-based program (10 reward points). Meanwhile, no rewards were given to the Control group participants. All data were collected between February 1, 2011 and February 10, 2012.
In total, 51.38% (73,327/142,726) of individuals in the Conditioned Reward group and 48.62% (69,399/142,726) of individuals in the Control group completed the Heart&Stroke Risk Assessment. Subsequently, significantly more individuals from the Conditioned Reward group (52.96%, 38,835/73,327) enrolled in the My Health eSupport program than Controls (4.07%, 2826/69,399). Regression analyses indicated that individuals were 27.9 times (95% CI 26.4-29.4;
Our findings suggest that a single exposure of loyalty rewards may be used to encourage individuals to enroll in an Internet-based preventative health program, but additional strategies are required to maintain engagement level. Future studies need to examine the schedules of loyalty reward reinforcement on the long-term engagement level of Internet-based health programs.
Cardiovascular diseases (CVDs) are the number one cause of death globally. It is estimated that the number of people who die from CVDs will reach 23.3 million by 2030 [
The wide adoption of Internet usage presents an incredible opportunity for delivering preventive health initiatives at a population level. In 2012, approximately 83% of Canadians had personal access to the Internet: 85% of whom live in metropolitan areas and 75% in rural areas [
External rewards such as financial and loyalty rewards have been used as a strategy to increase the enrollment and engagement of behavioral interventions [
This was an observational study and the data was collected by the Heart and Stroke Foundation of Canada (HSFC) between February 1, 2011 and February 10, 2012. The HSFC is a non-profit public organization that aims to empower Canadians to live healthy lives free of heart disease and stroke by raising awareness of the key risk factors and encouraging and supporting them to play an active role in managing their health. Study participants were 18 years or older and provided consent for the study. All personal identifiers were removed prior to retrieving the records from the HSFC database. This study was approved by the University Health Network Ethics Board.
The study sample consisted of individuals who were invited to participate in an Internet-based heart health program (My Health eSupport) after completing their Heart&Stroke Risk Assessment on the HSFC website. A total of 142,726 individuals completed the Heart&Stroke Risk Assessment and were included in our analyses. This sample was comprised of two groups: Conditioned Reward vs Control. Participants in the Conditioned Reward group were recruited from the Air Miles loyalty program using a standardized recruitment email. The email described the opportunity to receive a non-cash, uniform reward of up to 20 Air Miles reward points for enrolling in the My Health eSupport HSFC program. Specifically, participants were assured of immediately receiving 10 Air Miles reward points for completing the Heart&Stroke Risk Assessment and another 10 Air Miles for enrolling in My Health eSupport. No additional rewards were given after enrollment. The Heart&Stroke Risk Assessment was a free e-tool on the HSFC website that first enabled individuals to assess their lifestyle in regard to CVD risk. It then offered enrollment to the My Health eSupport program. Enrollment in the My Health eSupport program required the participants to first complete the Heart&Stroke Risk Assessment and create a log-in ID on the HSFC website. Air Miles participants accessed the Heart&Stroke Risk Assessment and claimed the reward using a unique Web link embedded in the recruitment email. Air Miles participants who completed the Heart&Stroke Risk Assessment were assigned to the Conditioned Reward group. Participants in the Control group accessed the Heart&Stroke Risk Assessment on the HSFC website without using the unique Web link provided by the Air Miles loyalty program. Control participants did not receive any rewards for completing either program.
The My Health eSupport program was a free, self-guided, fully automated healthy lifestyle program that proactively delivered tailored email messages at 2-week intervals. The emails contained information on heart healthy living with links to the HSFC website. The initial email from My Health eSupport was delivered to the participants following enrollment. The email guided participants to report their stage of motivational “readiness” to adhere to Health Canada guidelines for diet (daily intake of fruit, vegetables, and restriction of dietary fat and salt), exercise (planned exercise and daily activity), and smoke-free living. Readiness for change was operationally defined for each behavior according to the conventional algorithm of Prochaska’s Transtheoretical Model [
This paper evaluated enrollment rates and levels of engagement in the My Health eSupport program in both the Conditioned Reward and Control groups. Enrollment rate was calculated as the number of individuals who enrolled in the My Health eSupport program divided by the number of participants who completed the Heart&Stroke Risk Assessment. Two measures of engagement level with the program were calculated. The initial engagement was defined as the proportion of individuals who completed the assessment of readiness and priority for lifestyle change during the first email. Ongoing engagement was defined as completion of the assessment of readiness and priority for lifestyle the second time at 6 weeks following initial engagement. Participants completed the second assessment of readiness and priority for lifestyle change within 8 weeks from receiving the re-assessment email. Characteristics of participants were extracted from self-reported data in the Heart&Stroke Risk Assessment, which is an online self-assessment questionnaire. Characteristics included age, gender, education level, ethnicity, employment, medical condition(s), and body mass index (calculated from height and weight). Modifiable CVD risk factors were defined as the following: physical activity level (whether participants achieved 30-60 minutes of moderate exercise, 4 times per week), smoking status (Yes or No), excess alcohol consumption (Male: consumed >2 drinks a day or >14 drinks a week; Female: consumed >2 drinks a day, >9 drinks a week), and whether participants consumed high fat foods (3 or more times per week), fruits and vegetable (5 or more servings per day), and food with high salt content (3 or more times per week).
Chi-square and independent
A total of 73,327 individuals in the Conditioned Reward group and 69,399 individuals in the Control group completed the Heart&Stroke Risk Assessment. Baseline characteristics of the participants are presented in
Baseline participant characteristics.
Characteristics | Conditioned Rewardn=73,327 n (%) | Controln=69,399 n (%) |
|
Effect size | |
Age (years), mean (SD) |
|
50.4 (14.2) | 50.5 (14.4) | .54 | 0.003 |
Female |
|
49,778 (67.88%) | 47,954 (69.11%) | <.001 | 0.01 |
Completed university education |
|
43,783 (60.74%) | 41,208 (60.53%) | .40 | 0.002 |
|
<.001 | 0.13 | |||
|
Caucasian | 62,658 (85.48%) | 56,581 (81.62%) |
|
|
|
South Asian | 1688 (2.30%) | 1911 (2.76%) |
|
|
|
Chinese | 3040 (4.15%) | 1236 (1.78%) |
|
|
|
Aboriginal | 1038 (1.42%) | 1253 (1.81%) |
|
|
|
African/black | 683 (0.93%) | 1160 (1.67%) |
|
|
|
<.001 | 0.03 | |||
|
Married | 42,396 (57.85%) | 40365 (58.18%) |
|
|
|
Widowed, divorced | 10,177 (13.89%) | 8423 (12.14%) |
|
|
|
Single | 10,787 (14.72%) | 10,397 (14.99%) |
|
|
|
|||||
|
Full or part-time | 41,279 (56.33%) | 40,345 (58.16%) | <.001 | 0.07 |
|
Self-employed | 6675 (9.11%) | 6726 (9.70%) |
|
|
|
Retired | 14,644 (19.98%) | 11,654 (16.80%) |
|
|
|
Stay-at-home parent | 3049 (4.16%) | 2350 (3.39%) |
|
|
|
<.001 | 0.08 | |||
|
Management/white collar | 45,660 (63.00%) | 43,364 (68.81%) |
|
|
|
Sales or services | 9491 (13.09%) | 6757 (10.72%) |
|
|
|
Trades | 6408 (8.84%) | 6082 (9.65%) |
|
|
|
<.001 | 0.13 | |||
|
Underweight (<18.5) | 1464 (2.00%) | 1109 (1.60%) |
|
|
|
Normal weight (18.5-24.9) | 21,942 (29.93%) | 19,384 (27.94%) |
|
|
|
Overweight (25-29.9) | 21,240 (28.97%) | 22,356 (32.21%) |
|
|
|
Obesity (≥30) | 16,459 (22.45%) | 20,039 (28.88%) |
|
|
|
|
0.54 (0.93) | 0.51 (0.89) | <.001 | 0.03 |
|
Diabetes | 5298 (7.23%) | 3072 (4.43%) | <.001 | 0.06 |
|
Dyslipidemia | 13,984 (19.07%) | 11,503 (16.58%) | <.001 | 0.03 |
|
Heart disease | 2795 (3.81%) | 2662 (3.84%) | .81 | 0.001 |
|
Hypertension | 15,520 (21.17%) | 16,494 (23.77%) | <.001 | 0.11 |
|
Stroke | 1254 (1.71%) | 1327 (1.91%) | .004 | 0.01 |
|
Mood disorder | 11,906 (16.24%) | 8857 (12.76%) | <.001 | 0.05 |
|
Sleep apnea | 3988 (5.44%) | 3065 (4.42%) | <.001 | 0.02 |
Baseline modifiable risk factors.
Modifiable risk factors | Conditioned Reward n=73,327 n (%) | Control n=69,399 n (%) |
|
Effect size |
Physical inactivity | 30,972 (42.24%) | 31,970 (46.07%) | <.001 | 0.04 |
Smoker | 5704 (8.22%) | 8270 (11.30%) | <.001 | 0.05 |
Excess alcohol | 17,095 (23.31%) | 16,615 (23.94%) | .005 | 0.01 |
Fatty foods | 8537 (11.69%) | 10,663 (15.42%) | <.001 | 0.06 |
Infrequent fruit and vegetable | 30,601 (41.82%) | 31,227 (45.05%) | <.001 | 0.03 |
Salt | 19,738 (26.92%) | 14,205 (20.47%) | <.001 | 0.08 |
Mean number of modifiable risk factors (SD) | 2.51 (1.37) | 2.58 (1.37) | <.001 | 0.05 |
Enrollment in the My Health eSupport program was higher in the Conditioned Reward group (52.96%, 38,835/73,327) than the Controls (4.07%, 2826/69,399). Loyalty rewards provision was the strongest predictor for enrollment in the My Health eSupport program (OR 27.9, 95% CI 26.4-29.4;
Prediction of enrollment and engagement.
Factors | Enrollment | Initial engagement | Ongoing engagement | |||||||
OR | (95% CI) |
|
OR | (95% CI) |
|
OR | (95% CI) |
|
||
Loyalty rewards (Conditioned Reward vs Control) | 27.9 | (26.4-29.4) | <.001 | 0.79 | (0.66-0.92) | .004 | 0.93 | (0.68-1.26) | .662 | |
Gender (female vs male) |
|
0.96 | (0.93-0.99) | .007 | 1.26 | (1.17-1.36) | <.001 | 1.27 | (1.09-1.46) | .002 |
|
||||||||||
|
≤29 years | 1.98 | (1.82-2.15) | <.001 | 1.37 | (0.95-1.97) | .085 | 1.49 | (0.62-3.55) | .371 |
|
30-39 years | 1.69 | (1.56-1.82) | <.001 | 2.32 | (1.64-3.27) | <.001 | 1.99 | (0.86-4.57) | .104 |
|
40-49 years | 1.80 | (1.66-1.94) | <.001 | 4.41 | (3.14-6.17) | <.001 | 5.86 | (2.63-13.0) | <.001 |
|
50-59 years | 1.80 | (1.66-1.94) | <.001 | 6.81 | (4.87-9.49) | <.001 | 9.45 | (4.27-20.8) | <.001 |
|
≥60 years | 2.12 | (1.96-2.30) | <.001 | 8.40 | (6.00-11.7) | <.001 | 12.56 | (5.66-27.8) | <.001 |
Ethnicity (Caucasian vs other) | 1.20 | (1.15-1.25) | <.001 | 1.13 | (1.01-1.24) | .025 | 1.12 | (0.89-1.38) | .318 | |
Total number of modifiable risk factors | 0.96 | (0.95-0.97) | <.001 | 1.37 | (1.33-1.40) | <.001 | 1.38 | (1.31-1.45) | <.001 | |
Employment (employed vs not employed) | 1.02 | (0.99-1.06) | .201 | 0.89 | (0.82-0.95) | <.001 | 1.03 | (0.88-1.20) | .676 | |
University education (completed vs not completed) | 1.05 | (1.01-1.08) | .004 | 0.86 | (0.79-0.91) | <.001 | 0.88 | (0.76-1.01) | .068 |
After participants enrolled in the My Health eSupport program, only 12.43% (4829/38,835) of individuals in the Conditioned Reward group and 8.49% (240/2826) of the Controls assessed their readiness for change and selected a priority area for lifestyle change. Out of these participants, 20.98% (1013/4829) in the Conditioned Reward group and 24.17% (58/240) in the Control group completed the second assessment at week 6. In our regression analyses, loyalty rewards strategy was negatively associated with initial engagement (OR 0.79, 95% CI 0.66-0.92;
The main finding of this study was that a single exposure of loyalty rewards significantly influenced enrollment for the My Health eSupport program. Individuals were 27.9 times more likely to enroll when presented with loyalty rewards. However, contrary to our hypothesis, ongoing engagement was not influenced by loyalty rewards. Individuals were more likely to engage with the program if they were greater than 60 years of age, female, or had one or more modifiable risk factors. These findings suggest that a single exposure of loyalty rewards may be used to encourage individuals to sign up for an Internet-based preventative health initiative, but supplemental strategies are required to maintain engagement.
Our findings on the effects of loyalty rewards with online enrollment were consistent with other study findings using financial rewards [
Engagement level with Internet-based interventions has always been a challenge as high dropout and losses to follow-up are common. This phenomenon is described by Eysenbach as “the law of attrition” [
A single exposure of loyalty reward may have also contributed to the low level of engagement with the My Health eSupport program. Since rewards were presented at the point of enrollment, some individuals in the Conditioned Reward group may have enrolled for the sole purpose of obtaining this reward. This may explain the negative association found between loyalty rewards and initial engagement level. Charness et al [
Some researchers have cautioned against the use of any external incentives as they may have the undesired effects of inhibiting intrinsic motivation [
Several limitations of the present study should be noted. The accuracy of self-reported data is open to challenge for validity. Self-report bias such as lifestyle behaviors of exercise, diet, and smoking could have influenced the accuracy of our results. Engagement with the system was defined when participants re-assessed their readiness for change and selected another priority area for lifestyle change. It is possible that individuals maintained engagement with the program but never completed the re-assessment, which may underestimate the levels of engagement with the My Health eSupport Program. There are other variables that may be associated with levels of enrollment and engagement which were not assessed, such as income, anxiety, and depression. Finally, there may be selection bias in our study as the Conditioned Reward group consisted of participants from the Air Miles loyalty reward program. This may limit our ability to generalize our findings beyond Air Miles members. A strength of this study includes the large sample size. This is one of the first “real-world” (eg, population-based) studies that has been conducted to examine the effects of loyalty rewards on enrollment and engagement with an Internet-based intervention.
Internet-based interventions hold great potential for delivering preventive health initiatives at a population level. A single exposure of loyalty rewards can increase enrollment but additional strategies are required to maintain engagement level. This study has significant design implications for incorporating loyalty rewards as an effective enrollment strategy for future Internet-based interventions. More research is needed that explores the long-term effects of using loyalty rewards to reinforce engagement level and the associated efficacy of Internet-based health programs.
cardiovascular disease
Heart and Stroke Foundation of Canada
None declared.