Effect of Escalating Financial Incentive Rewards on Maintenance of Weight Loss

This randomized clinical trial assesses an escalating lottery-based incentive tied to daily self-weighing for weight loss maintenance among adults in an online weight management program.


Introduction
Because of its high prevalence and association with multiple illnesses, obesity has become a leading cause of preventable death in the United States. 1 Identifying effective strategies for treating obesity is a clinical challenge and a public health priority.Although a variety of approaches have been successful in achieving initial weight loss, maintenance of weight loss has proven much more difficult.
Research has shown that the frequency of self-weighing is important for prevention of weight gain and for weight loss. 2,3Financial incentives have been successful at producing behavior change, including weight loss, in a variety of settings and formats, with lottery-based incentives being an effective design.Financial incentives, however, have been studied infrequently in the context of weight loss maintenance.In a previous study enrolling participants who had already lost 5 kg in a commercial weight management program, direct and lottery-based incentives did not enhance weight outcomes beyond the control intervention. 4However, participants who self-weighed at least 6 days each week experienced better weight outcomes than those weighing less frequently.Given the difficulties of sustaining adherence over time during weight control efforts, we hypothesized that a reward that increased over time with persistent adherence to self-weighing at least 6 days each week would improve weight outcomes.Therefore, our primary aim was to assess the effectiveness of escalating lottery rewards, relative to a control condition, on maintenance of weight loss during the 6 months after an initial weight loss of at least 5 kg.Our secondary aim was to assess the durability of this effect after an additional 6 months without rewards.

Overview of Study Design
The study was a 2-phase, 2-arm randomized clinical trial (trial protocol is available in Supplement 1)   in which participants who lost at least 5 kg in a commercial weight management program were randomized in a 1:1 ratio to (1) daily self-weighing and weekly feedback (control group) or (2) daily self-weighing, weekly feedback, and a weekly, escalating, lottery-based financial incentive for 6 months (phase 1) (Figure 1).Participants in both groups were then observed without intervention for an additional 6 months (phase 2).Recruitment for the study began May 23, 2016, and ended November 13, 2016, when target sample size was reached, with the final participant completing the follow-up period on November 13, 2017.The study was approved by the institutional review boards of the University of Pennsylvania, Philadelphia, and Duke University, Durham, North Carolina.All patients provided written informed consent.This report follows the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline for randomized clinical trials. 5

Participants and Setting
Participants were recruited by email from WW (formerly Weight Watchers) Digital to members who had opted in for communications from WW about research studies and met the following inclusion criteria: aged 30 to 80 years, body mass index (calculated as weight in kilograms divided by height in  Participants in the incentive group were informed of the structure of the escalating lottery at the onset of the intervention.Incentive group participants were eligible to win lotteries worth an expected value of $3.98 in week 1 that increased by $0.43 per week for each week that they achieved the goal of weighing 6 of 7 days.The lottery provides infrequent large payoffs (a 1-in-100 chance at a $110 reward in week 1) and frequent small payoffs (an 18-in-100 chance at a $16 reward in week 1) given that individuals are motivated by the future (fixation on large potential winnings) and the past (how often did I win?). 8,9Lotteries also have a theoretical advantage over direct payments in providing variable reinforcement, which has been demonstrated as more effective in reinforcing behavior than consistent reinforcement. 10The lottery was designed with rewards that start small and then escalate over time (as internal motivation begins to wane) and as weight maintenance becomes more difficult.Increasing the reward magnitude, conditional on ongoing engagement, is a way to leverage the endowment effect and loss aversion because participants face an ever-greater lost opportunity if they discontinue engagement. 11If participants missed the target 2 weeks in a row, they returned to the beginning of the escalating rewards schedule and were warned of this outcome the preceding week.If participants missed the target one week but adhered the next week, their potential reward would resume where they left off.When a participant won the lottery but did not meet the self-weighing goal, the participant was notified that he or she would have won because anticipated regret could motivate future adherence.Other than the total number of lottery weeks (24 weeks), there was no limit to the number of times a person could win the lottery.For participants who achieved all their weekly goals for the entire 24 weeks of the intervention, the expected value of the lottery, based on probabilities of winning the small and large lotteries, was approximately $214.
All participants were asked to complete online surveys at months 6 and 12 and received automated emails or text messages from the Way to Health platform notifying them of their compensation on completion.To enhance retention, participants received $50 each time for completing the online survey at the end of 6 and 12 months, independent of self-weighing adherence.All payments were approved by staff electronically from the Way to Health platform, and a text message was sent to participants alerting them of the payment.Payment information was transmitted to a bank to process and sent to participants via check weekly.

Measurements
Weight was measured by each participant using the wireless scale.Physical activity was measured using the short form of the International Physical Activity Questionnaire, 12 and eating habits were assessed with the Three-Factor Eating Questionnaire-R18, designed to assess three cognitive and behavioral domains of eating in obese individuals: cognitive restraint, uncontrolled eating, and emotional eating. 13,14

Safety Monitoring
The study was monitored by an independent data and safety monitoring board.Daily self-weighing data were also used to screen for unsafe weight loss strategies.Study staff contacted participants who lost more than 4.5 kg in 1 week or more than 9.0 kg in 1 month and asked about potential reasons for such weight loss, including unsafe strategies.

Statistical Analysis
Participants were randomized in a 1:1 ratio stratified by sex and amount of weight lost (<13.6 vs Ն13.6 kg) in their first 4 to 6 months of WW using permuted block randomization with variable block sizes.
The study was powered to detect between-group differences in weight change from baseline to the end of phase 1, when the intervention should achieve its maximal effect.Assuming a 5.0-kg SD for change in weight at 6 months based on a previous study 4   Additional secondary analyses included a between-group comparison of the percentage of individuals who maintained their weight (defined as gaining Յ1.36 kg) at 6 and 12 months using a χ 2 test.The percentage of weeks that individuals weighed in at least 6 times was also compared between groups for each phase using an unequal-variance t test.The association between the mean self-weighing frequency and amount of weight lost at the end of the phase was estimated using the Spearman correlation coefficient in a cross-sectional analysis for each phase.The trend over time for the frequency of self-weighing measurements was also compared between groups using a generalized estimating equation with an autoregressive working correlation model; specifically, the mean number of days per week was modeled and compared between study groups after adjusting for study week and a week-by-group interaction.Comparisons were also made of the changes in selfreported physical activity and the 3 domains of eating behaviors (cognitive restraint, uncontrolled eating, and emotional eating) at 6 and 12 months after initial weight loss, in analyses similar to those described above for the change in weight.
All reported hypothesis tests were 2 sided.P < .05 was considered statistically significant.
Analyses were conducted using SAS, version 9.4 (SAS Institute Inc).

Recruitment and Enrollment
A total of 4847 WW Digital members received an invitation email; of these, 421 initiated the enrollment process, and 259 were randomized (Figure 1).After 1 patient in the incentive group was excluded becausue of being diagnosed with cancer, a total of 258 participants were included in the analysis (128 in the incentive group and 130 in the control group).The mean (SD) age of the participants was 48.0 (10.5) years; 225 (87.2%) were women and 33 (12.8%) were men; 235 (91.1%) were white; 149 (57.8%) had at least a college degree; and 102 (39.5%) had an annual income of at least $100 000 (Table 1).The mean (SD) weight loss before study enrollment was 11.6 (4.2) kg or 11.5% (3.7%) of original body weight (Figure 2); the mean (SD) body mass index at randomization was 32.1 (3.9).

Primary Weight Outcome
Participants in both groups successfully maintained their initial weight loss during the 6 months of the intervention (phase 1) and the additional 6 months of follow-up after the intervention (phase 2) (Table 2 and Figure 2).The 6-month weight outcome was observed in 250 (96.9%) of the 258 participants in the analysis cohort; the 12-month weight outcome, in 227 (88.0%).In the intentionto-treat analysis, mean weight changes at the end of phase 1 were −1.1 (95% CI, −2.1 to −0.1) kg in the incentive group and −1.9 (95% CI, −2.9 to −0.   a Percentages have been rounded and may not total 100.
b One participant from the incentive group had physical activity data at baseline excluded because it was considered an outlier by the IPAQ scoring protocol.
c Measured at baseline.
d Measured at baseline using the Three-Factor Eating Questionnaire-R18.Scores range from 18 to 76, with higher scores indicating higher cognitive restraint and emotional and uncontrolled eating.0.7 (95% CI, −0.7 to 2.2) kg (P = .30for comparison).Mean weight changes from baseline to the end of phase 2 (end of study) were 0.2 (95% CI, −1.2 to 1.7) kg in the incentive group and −0.6 (95% CI, −2.0 to 0.8) kg in the control group, with a mean difference of 0.8 (95% CI, −1.2 to 2.8) kg (P = .41for comparison).Analyses adjusted for baseline covariates yielded similar results, with a mean difference between groups in weight change from baseline to the end of phase 1 of 1.2 (95% CI, −0.2 to 2.5) kg (P = .09for comparison) and from baseline to the end of phase 2 of 1.2 (95% CI, −0.8 to 3.2) kg (P = .23for comparison).Using a complete case analysis, results were similar to the intention-to-treat analysis; mean weight changes from baseline to the end of phase 1 were −1.1 (95% CI, −2.2 to −0.1) kg in the incentive group and −1.8 (95% CI, −2.8 to −0.8) kg in the control group; from baseline to end of phase 2, 0.4 (95% CI, −1.0 to 1.8) kg in the incentive group and −0.4 (95% CI, −1.9 to 1.1) kg in the control group.The percentage of incentive vs control participants who maintained their weight

Secondary Outcomes
During the 6-month intervention (phase 1), the percentage of weeks that incentive and control participants achieved mean self-weighing at least 6 times was 85.3% vs 75.8%, respectively (P = .002for comparison).After an additional 6 months of follow-up without intervention (phase 2), the percentages were 37.7% vs 50.2% for the intervention and control groups, respectively (P = .009).The trends in self-weighing frequency over time by group (Figure 3) demonstrated a decrease in frequency immediately after cessation of incentives in that group.requirement in consecutive weeks well enough to step up their reward a mean of 11 (of a possible 23) times before their first nonadherent week, which would translate into $27 in escalated small lottery winning or $385 in escalated large lottery winning.Among incentive group participants, 33 (25.8%) were adherent every week in phase 1, and 76 (80.9%) were adherent the week after a nonadherent week.In contrast, 35 (27.3%) did not self-weigh 6 of 7 days for 2 weeks in a row, returning them to the baseline incentive amount at least once during phase 1.

Excess Weight Loss Events and Other Adverse Events
A total of 71 weight loss alerts, including alerts for 66 participants with a loss of at least 4.5 kg in 1 week and 5 with a loss of at least 9 kg in 1 month, occurred during 12 months of follow-up.No unhealthy weight loss behaviors were reported.The most common reason for an alert was another family member using the scale (15 of 71 alerts [21.1%]) followed by scale calibration error (14 [    b Calculated using the Three-Factor Eating Questionnaire-R18.Scores range from 18 to 76, with higher scores indicating higher cognitive restraint and emotional and uncontrolled eating.
c Comparison between groups are calculated using χ 2 tests.

and 2
scores related to delayed gratification behavior.We also estimated the difference between groups at 6 months in an adjusted analysis in which the change from baseline was regressed in a linear model on study group and the same baseline covariates as in the imputation model.Analyses of the secondary outcome, weight change from baseline to 12 months (phase 2), were conducted in a similar manner as the primary outcome.

JAMA Network Open | Health Policy Effect
6f Escalating Financial Incentive Rewards on Maintenance of Weight Loss of 30.0 to 45.0 before starting WW, documented weight loss of at least 5 kg in the previous 4 to 6 months, and stable health.Participants were enrolled in the study by a clinical research coordinator (C.R.).Potential participants were excluded for the following: substance abuse; bulimia nervosa or related behaviors; pregnancy or breastfeeding; medical contraindications to counseling about diet, physical activity, or weight reduction; unstable mental illness; positive screen findings for pathologic gambling6; or inability to read or complete forms in English.Enrollment, electronic written informed consent, and data collection occurred via the Penn Way to Health portal, a web-based infrastructure used to run behavioral economic intervention studies.

JAMA Network Open | Health Policy
Downloaded From: https://jamanetwork.com/ on 09/16/2023 Data were analyzed from May 23, 2016, through November 13, 2017.The primary analysis was an intention-to-treat, unadjusted, between-group comparison of the mean change in weight from baseline (randomization) to 6 months with 95% CIs, using an unequal-variance t test performed at the significance level of P < .05 and multiple imputation for handling of missing data.The 6-month weight was determined as the first self-weighing measurement that occurred inside a 4-week window starting at week 24.Our multiple imputation strategy used the following baseline covariates to impute missing data for the primary outcome: study group, baseline weight, age, sex, selfreported race, body mass index, educational level, annual income, qualifying weight loss, minutes of moderate and vigorous activity and walking from the International Physical Activity Questionnaire, the 3 domains of eating behaviors (cognitive restraint, uncontrolled eating, and emotional eating), and 20% missingness, there was 90% power to detect at least 2.3 kg more weight loss during 6 months for the intervention group compared with a control group.JAMA Network Open | Health PolicyEffect of Escalating Financial Incentive Rewards on Maintenance of Weight Loss JAMA Network Open.2019;2(11):e1914393. doi:10.1001/jamanetworkopen.2019.14393(Reprinted) November 1, 2019 4/12 Downloaded From: https://jamanetwork.com/ on 09/16/2023

Table 1 .
Baseline Participant Characteristics by Group 8) kg in the control group, with a mean difference of JAMA Network Open | Health Policy Effect of Escalating Financial Incentive Rewards on Maintenance of Weight Loss JAMA Network Open.2019;2(11):e1914393. doi:10.1001/jamanetworkopen.2019.14393(Reprinted) November 1, 2019 5/12 Downloaded From: https://jamanetwork.com/ on 09/16/2023 Figure 2. Mean Percentage Weight Change by Group From Entry Into Weight Management Program Over Time

Table 2 .
Mean Changes in Weight, Physical Activity, Eating Habits, and Self-weighing Frequency by Group