Innovative Self-Regulation Strategies Reduce Weight Gain in Young Adults: The Study of Novel Approaches to Weight Gain Prevention (SNAP) Randomized Controlled Trial
Importance
Weight gain occurs commonly in young adults and has adverse effects on health. Reducing weight gain in young adults would have significant public health impact.
Objective
To compare two self-regulation interventions versus control in reducing weight gain in young adults over an average follow-up of 3 years.
Design
Randomized controlled trial
Setting
Two academic settings
Participants
599 participants (22% men; 27% minority), age 18–35 (27.7 (4.4)), BMI 21–30 kg/m (25.4 (2.6)), recruited via mailings and e-mails between August 2010 and February 2012.
Interventions
Participants were randomly assigned to: Control, Self-Regulation plus Small Changes or Self-Regulation plus Large Changes. The interventions focused on frequent self-weighing and using this information to know when behavior changes were needed. “Small Changes” taught participants to reduce intake and increase activity, both by approximately 100 calories per day. “Large Changes” focused on losing 5–10 pounds initially to buffer against expected weight gain. Interventions were delivered via 10 in-person group sessions during months 1–4 and then primarily online. Assessments occurred at randomization, 4 months, and then annually for 2–4 years (depending on time of recruitment).
Main outcome
Changes in weight from baseline over average follow-up of 3 years. Secondary outcomes: proportion gaining ≥1 pound from baseline, proportion developing obesity (BMI ≥30 kg/m), weight change baseline to 2 years (last assessment reached by all cohorts)
Results
Mean (SE) weight changes over an average follow-up of 3 years were +0.26 kg (0.22), −0.56 kg (0.22), and −2.37 kg (0.22) in Control, Small Changes and Large Changes, respectively (p=0.001). Differences among all three groups were significant (Large Changes vs Control, p<.001; Small Changes vs Control, p<.02 and Large Changes vs Small Changes, p<.001). Both interventions significantly reduced incidence of obesity (p<.05) relative to Control (8.6%, 7.9% and 16.9% in Large, Small and Control, respectively); Large Changes reduced the proportion that gained ≥1 pound (p<.05).
Conclusions and Relevance
Self-regulation with Large or Small Changes both reduced weight gain in young adults over three years relative to Control, but Large Changes was most effective. These interventions could easily be implemented on a large scale and may impact public health.
Trial registration
ClinicalTrials.gov, {"type":"clinical-trial","attrs":{"text":"NCT01183689","term_id":"NCT01183689"}}NCT01183689
Introduction
Weight gain occurs commonly throughout adulthood and is associated with many adverse health outcomes. 1–4 Young adults are the age group that is gaining weight the fastest; 5–7 moreover, weight gain in those age 18 – 35 years has stronger negative associations with critical outcomes such as cancer risk and mortality than weight gain at later ages.8,9 Developing effective approaches to reduce weight gain in young adults is thus an important public health priority.
Previous efforts to prevent weight gain during adulthood, particularly in young adults, have had limited success. 1,10 Although positive effects are often observed initially, few studies have followed participants for two to three years 11–13 and long-term differences between intervention and control are rarely significant. Moreover, many of these programs have been implemented in college courses, limiting generalizability.14
The present study examined two novel interventions for weight gain prevention compared to a control condition on the magnitude of weight change across an average planned follow-up of 3 years. Both interventions were based on a self-regulation model shown previously to be effective in preventing weight regain; 15 both emphasized frequent self-weighing and changes in eating and activity to prevent weight gain. One approach focused on making daily small changes in eating and activity to prevent weight gain. This “small changes” approach has received a great deal of attention through programs such as America on the Move (http://www.anschutzwellness.com/community-wellness-programs/america-the-move and is supported by theoretical papers 16 and small studies 17–20 suggesting that both decreasing intake and increasing activity by approximately 100 calories per day should be sufficient to prevent weight gain. The other approach emphasized larger initial changes in behavior to produce weight loss as a buffer against anticipated weight gains. Evidence favoring the “large changes” approach comes from the Women’s Healthy Lifestyle Project, 21 which showed that producing initial weight losses of 5–15 pounds, even though they were followed by some weight regain, was effective in reducing weight gain in menopausal women. The current study tested whether either or both of these approaches reduced weight gain in young adults.
Methods
Study Design
The Study of Novel Approaches to Weight Gain Prevention (SNAP) is a 3-armed randomized controlled clinical trial, with equal allocation, comparing self-regulation with small daily behavior changes (Small Changes), self-regulation with large periodic behavior changes (Large Changes), and a minimal treatment control condition (protocol available in Supplement).22 The primary outcome is the mean weight gain over an average follow-up of 3 years. Secondary outcomes were the proportion of participants gaining ≥1 pound (chosen to represent a stringent criterion of weight gain over baseline) and the incidence of obesity (BMI≥30 kg/m) over the 3 years. We focused on outcomes over follow-up (rather than at one specific time point) to capture the cumulative effect of the interventions on body weight. We also examined weight change specifically at 2 years, since this is the last time point reached by all participants in SNAP and the outcome in several other current trials on weight gain prevention in young adults.23 The study was funded by the National Heart Lung and Blood Institute and involved 2 clinical sites (Providence, RI and Chapel Hill, NC) and a coordinating center (Winston-Salem, NC) and was approved by each Institutional Review Board. A Data Safety Monitoring Board provided trial oversight.
Study Participants
SNAP targeted an enrollment of 600 participants (25% men and 25% racial / ethnic minorities), age 18–35, with a BMI of 21–30.9 kg/m Both overweight and normal weight individuals were included since young adults in both weight groups (especially those who are overweight) are at greater risk of weight gain than older individuals.5 We included individuals with a BMI of 21 kg/m since epidemiological studies have shown that higher BMI in young adults, even within the normal range, is associated with elevated mortality risk.9 In addition, the earlier the age that the threshold of BMI ≥ 25 kg/m is exceeded, the greater the mortality risk. A weight loss of 5 pounds (as encouraged in one of the interventions), was considered safe in individuals with a BMI of 21 kg/m because their BMI would remain in the normal range. Other eligibility criteria, described previously,22 focused on ability to participate in the program (e.g. internet access, English speaking), safety (no history of eating disorders, ability to walk for activity) and completion of screening and baseline assessment visits. Participants were recruited primarily by mass mailings (38%) and e-mails (23%), using text that sought individuals who were concerned about gaining weight over time.24
Randomization
Randomization assignment used variable block lengths, was stratified by clinical site, gender and ethnicity (non-Hispanic white/other), and was implemented through a web-based data management system.
Study Interventions
The interventions have been described in detail.22 The Control group attended one face-to-face meeting where they were introduced to the issue of weight gain, the concept of self-regulation, and an overview of both Small and Large Change approaches to potentially prevent weight gain.
Small and Large Changes interventions both began with 10 face-to-face group meetings over 4 months. This treatment length was selected to be sufficient to enable the 5 or 10 pound weight loss, but easily disseminated. Subsequently, the interventions were delivered primarily online. Each year, participants were invited to join two 4-week online refresher campaigns reinforcing the behaviors taught during the initial program. All participants received identical quarterly newsletters and personalized feedback reports on their assessment data, including the Control group.
The interventions were both framed in a self-regulation model that forms the basis for several self-control theories 25–28 and has been applied to diabetes 29 and obesity 15 and was used in a pilot study for this trial.30 Self-regulation is based on a negative feedback loop, in which there is a goal, error detector and controlling responses. In SNAP, the goal was to not exceed baseline weight, the error detector was the scale and daily self-weighing, and the controlling responses involved changes in diet and exercise consistent with the Small or Large Change approach. To encourage self-regulation, participants were instructed to weigh themselves daily and submit their weight via the study website, text message or email. They received monthly email feedback on their weight, which was based on a color-coded system15,22 and either reinforced their success, encouraged problem-solving, or recommended additional strategies to help reverse weight gain. Participants who gained above baseline were invited to contact a study interventionist for problem solving assistance via email, phone or face-to-face, but very few requested this assistance.
The specific recommendations related to diet and activity differed for Small Changes vs Large Changes. Participants in Small Changes were taught to make daily small changes (approximately 100 kcal/day) in both diet (e.g. select lower calorie coffee drinks, reduce portion sizes) and physical activity (e.g. park farther from store, use stairs). Participants were given pedometers and instructed to add 2000 steps per day (equivalent to 1 mile) above baseline. If participants in Small Changes experienced weight gains above baseline, they were encouraged to make additional daily small changes.
Large Changes focused on losing weight (5 lbs if normal weight; 10 lbs if overweight) during the initial four-month program to create a buffer against subsequent weight gain. To achieve this, participants were prescribed a calorie goal based on a 500 to 1000 kcal deficit from baseline to use during the initial 8 weeks. They were also encouraged to gradually increase moderate intensity physical activity to a goal of 250 minutes/week, the level recommended for weight loss maintenance,31 and to maintain this over time. If weight exceeded baseline, they were to return to their calorie goal and confirm that they were achieving the activity goal.
Study Assessment
All participants were scheduled to complete assessments at baseline, month 4, year 1 and year 2. Depending on when participants were randomized, some were also scheduled to reach year 3 (N=437) and year 4 (N=106) before pre-specified data close-out on 12/31/14. All assessments were completed by masked staff members, who were centrally trained and certified. Participants received a $50 honorarium for each follow-up assessment.
Weight was measured on a calibrated scale in light clothes, without shoes; height was assessed with a wall-mounted stadiometer. Two measures were taken and averaged. Cellular connected scales (“Smart” scales) were sent to those who had moved or could not attend a clinic visit. These non-protocol smart scale weights constituted 4.7% of all weight data with comparable numbers in the three groups, and were used only in sensitivity analyses.
Statistical approach
The primary hypothesis of SNAP, that the mean weight change across an average planned follow-up of 3 years would differ among the three arms, was assessed by fitting a mixed effects linear model to the changes in measured weight from baseline to 4, 12, 24, 36, and 48 months.32 This is similar to defining the area-under-the curve, and captures the overall exposure to weight. Participants were grouped according to randomization assignment with clinic site as the only pre-specified covariate. Estimated mean differences for each pairwise comparison from linear contrasts were assessed with Wald statistics, Bonferroni adjustment was used to control total Type I error to be 0.05 across the three comparisons. Multiple imputation was used to assess the sensitivity of the primary inference to missing data.33 Secondary aims were to examine group differences over time in the proportion of participants who gained ≥1 pound and the proportion that developed obesity, using generalized estimating equations, and mean weight change from baseline to 2 years, with a linear contrast. All measured weights were included in analyses, except those during or within 6 months after pregnancies. To assess heterogeneity in intervention responses, three subgroup comparisons were pre-specified: baseline BMI (<25 kg/m versus ≥25 kg/m), age (<25 years versus ≥ 25 years), and gender.
The targeted sample of N=600 was projected to provide ≥90% power to detect an average difference between groups of 1.36 kg weight change over time while accommodating loss to follow-up of 7.5% month 4, additional 7.5% at year 1, and 5% per year thereafter.
Study Design
The Study of Novel Approaches to Weight Gain Prevention (SNAP) is a 3-armed randomized controlled clinical trial, with equal allocation, comparing self-regulation with small daily behavior changes (Small Changes), self-regulation with large periodic behavior changes (Large Changes), and a minimal treatment control condition (protocol available in Supplement).22 The primary outcome is the mean weight gain over an average follow-up of 3 years. Secondary outcomes were the proportion of participants gaining ≥1 pound (chosen to represent a stringent criterion of weight gain over baseline) and the incidence of obesity (BMI≥30 kg/m) over the 3 years. We focused on outcomes over follow-up (rather than at one specific time point) to capture the cumulative effect of the interventions on body weight. We also examined weight change specifically at 2 years, since this is the last time point reached by all participants in SNAP and the outcome in several other current trials on weight gain prevention in young adults.23 The study was funded by the National Heart Lung and Blood Institute and involved 2 clinical sites (Providence, RI and Chapel Hill, NC) and a coordinating center (Winston-Salem, NC) and was approved by each Institutional Review Board. A Data Safety Monitoring Board provided trial oversight.
Study Participants
SNAP targeted an enrollment of 600 participants (25% men and 25% racial / ethnic minorities), age 18–35, with a BMI of 21–30.9 kg/m Both overweight and normal weight individuals were included since young adults in both weight groups (especially those who are overweight) are at greater risk of weight gain than older individuals.5 We included individuals with a BMI of 21 kg/m since epidemiological studies have shown that higher BMI in young adults, even within the normal range, is associated with elevated mortality risk.9 In addition, the earlier the age that the threshold of BMI ≥ 25 kg/m is exceeded, the greater the mortality risk. A weight loss of 5 pounds (as encouraged in one of the interventions), was considered safe in individuals with a BMI of 21 kg/m because their BMI would remain in the normal range. Other eligibility criteria, described previously,22 focused on ability to participate in the program (e.g. internet access, English speaking), safety (no history of eating disorders, ability to walk for activity) and completion of screening and baseline assessment visits. Participants were recruited primarily by mass mailings (38%) and e-mails (23%), using text that sought individuals who were concerned about gaining weight over time.24
Randomization
Randomization assignment used variable block lengths, was stratified by clinical site, gender and ethnicity (non-Hispanic white/other), and was implemented through a web-based data management system.
Study Interventions
The interventions have been described in detail.22 The Control group attended one face-to-face meeting where they were introduced to the issue of weight gain, the concept of self-regulation, and an overview of both Small and Large Change approaches to potentially prevent weight gain.
Small and Large Changes interventions both began with 10 face-to-face group meetings over 4 months. This treatment length was selected to be sufficient to enable the 5 or 10 pound weight loss, but easily disseminated. Subsequently, the interventions were delivered primarily online. Each year, participants were invited to join two 4-week online refresher campaigns reinforcing the behaviors taught during the initial program. All participants received identical quarterly newsletters and personalized feedback reports on their assessment data, including the Control group.
The interventions were both framed in a self-regulation model that forms the basis for several self-control theories 25–28 and has been applied to diabetes 29 and obesity 15 and was used in a pilot study for this trial.30 Self-regulation is based on a negative feedback loop, in which there is a goal, error detector and controlling responses. In SNAP, the goal was to not exceed baseline weight, the error detector was the scale and daily self-weighing, and the controlling responses involved changes in diet and exercise consistent with the Small or Large Change approach. To encourage self-regulation, participants were instructed to weigh themselves daily and submit their weight via the study website, text message or email. They received monthly email feedback on their weight, which was based on a color-coded system15,22 and either reinforced their success, encouraged problem-solving, or recommended additional strategies to help reverse weight gain. Participants who gained above baseline were invited to contact a study interventionist for problem solving assistance via email, phone or face-to-face, but very few requested this assistance.
The specific recommendations related to diet and activity differed for Small Changes vs Large Changes. Participants in Small Changes were taught to make daily small changes (approximately 100 kcal/day) in both diet (e.g. select lower calorie coffee drinks, reduce portion sizes) and physical activity (e.g. park farther from store, use stairs). Participants were given pedometers and instructed to add 2000 steps per day (equivalent to 1 mile) above baseline. If participants in Small Changes experienced weight gains above baseline, they were encouraged to make additional daily small changes.
Large Changes focused on losing weight (5 lbs if normal weight; 10 lbs if overweight) during the initial four-month program to create a buffer against subsequent weight gain. To achieve this, participants were prescribed a calorie goal based on a 500 to 1000 kcal deficit from baseline to use during the initial 8 weeks. They were also encouraged to gradually increase moderate intensity physical activity to a goal of 250 minutes/week, the level recommended for weight loss maintenance,31 and to maintain this over time. If weight exceeded baseline, they were to return to their calorie goal and confirm that they were achieving the activity goal.
Study Assessment
All participants were scheduled to complete assessments at baseline, month 4, year 1 and year 2. Depending on when participants were randomized, some were also scheduled to reach year 3 (N=437) and year 4 (N=106) before pre-specified data close-out on 12/31/14. All assessments were completed by masked staff members, who were centrally trained and certified. Participants received a $50 honorarium for each follow-up assessment.
Weight was measured on a calibrated scale in light clothes, without shoes; height was assessed with a wall-mounted stadiometer. Two measures were taken and averaged. Cellular connected scales (“Smart” scales) were sent to those who had moved or could not attend a clinic visit. These non-protocol smart scale weights constituted 4.7% of all weight data with comparable numbers in the three groups, and were used only in sensitivity analyses.
Statistical approach
The primary hypothesis of SNAP, that the mean weight change across an average planned follow-up of 3 years would differ among the three arms, was assessed by fitting a mixed effects linear model to the changes in measured weight from baseline to 4, 12, 24, 36, and 48 months.32 This is similar to defining the area-under-the curve, and captures the overall exposure to weight. Participants were grouped according to randomization assignment with clinic site as the only pre-specified covariate. Estimated mean differences for each pairwise comparison from linear contrasts were assessed with Wald statistics, Bonferroni adjustment was used to control total Type I error to be 0.05 across the three comparisons. Multiple imputation was used to assess the sensitivity of the primary inference to missing data.33 Secondary aims were to examine group differences over time in the proportion of participants who gained ≥1 pound and the proportion that developed obesity, using generalized estimating equations, and mean weight change from baseline to 2 years, with a linear contrast. All measured weights were included in analyses, except those during or within 6 months after pregnancies. To assess heterogeneity in intervention responses, three subgroup comparisons were pre-specified: baseline BMI (<25 kg/m versus ≥25 kg/m), age (<25 years versus ≥ 25 years), and gender.
The targeted sample of N=600 was projected to provide ≥90% power to detect an average difference between groups of 1.36 kg weight change over time while accommodating loss to follow-up of 7.5% month 4, additional 7.5% at year 1, and 5% per year thereafter.
Results
Study Participants
A total of 599 adults were enrolled (292 in Providence RI and 307 in Chapel Hill, NC) between August 2010 and February 2012 and randomly assigned to one of the three groups. Participants in the three groups were similar at baseline (Table 1). The study sample included 22% males and 27% from minority groups. Participants were [Mean (SD)] 28.2 (4.4) years of age and over 70% were above age 25; average BMI was 25.4 (2.6) kg/m and approximately 50% were normal weight. The majority (63%) were employed full-time. Figure 1 presents the CONSORT diagram showing retention at each of the follow-up assessments. Retention did not differ among groups.
Table 1
Characteristics of the SNAP participants at enrollment by intervention assignment: Mean (Standard Deviation) or Frequency (Percent).
| Baseline Characteristics | Intervention Assignment | ||
|---|---|---|---|
| Control N=202 | Small Changes N=200 | Large Changes N=197 | |
| Gender | |||
| Female | 158 (78.2%) | 157 (78.5%) | 154 (78.2%) |
| Male | 44 (21.8%) | 43 (21.5%) | 43 (21.8%) |
| Age, years | |||
| 18–24.9 | 53 (26.2%) | 60 (30.0%) | 56 (28.4%) |
| 25–35 | 149 (73.8%) | 140 (70.0%) | 141 (71.6%) |
| BMI, kg/m2 | |||
| <25 | 97 (48.0%) | 84 (42.0%) | 96 (48.7%) |
| ≥25 | 105 (52.0%) | 116 (58.0%) | 101 (51.3%) |
| Weight, kg | 71.4 (10.2) | 71.9 (11.0) | 70.8 (11.0) |
| Race/ethnicity | |||
| African-American | 19 (9.4%) | 25 (12.5%) | 22 (11.2%) |
| Non-Hispanic White | 148 (73.3%) | 146 (73.0%) | 144 (73.1%) |
| Asian/Pacific Islander | 7 (3.5%) | 11 (5.5%) | 7 (3.5%) |
| Hispanic | 15 (7.4%) | 10 (5.0%) | 21 (10.6%) |
| Multiple/Refusal | 13 (6.4%) | 8 (4.0%) | 3 (1.5%) |
| Education | |||
| Not college graduate | 39 (19.3%) | 46 (23.0%) | 35 (17.8%) |
| College graduate | 163 (80.7%) | 154 (77.0%) | 162 (82.2%) |
| Employment Status | |||
| Employed full time | 127 (62.9%) | 125 (62.5%) | 125 (63.5%) |
| Student full time | 59 (24.3%) | 61 (30.5%) | 55 (27.9%) |
| Other | 26 (12.8%) | 14 (7.0%) | 17 (8.6%) |
Intervention Delivery
Participants in Large and Small Changes attended an average of 87.4% and 86.0% of the 10 intervention sessions; 100% of the control group participants attended their one session. Intervention fidelity, determined by masked raters evaluating a randomly selected sample of 20% of recorded group sessions, was excellent, with 100% accuracy for distinguishing Large from Small Change sessions and for presentation of the appropriate behavioral content. Self-weighing, a cornerstone of self-regulation interventions, was increased in the two interventions. Whereas at baseline, 11–13% of each group reported daily self-weighing, at four months, daily self-weighing was reported by 75%, 72%, and 30% of Large Changes, Small Changes, and Control participants respectively (p<0.001).
Use of the prescribed behavioral strategies also differed significantly (p<0.05) by randomization group; for example at 4 months, 64% of Large Changes participants (and 10% and 11% in Small Changes and Control) reported reducing calories by 500–1000 kcal/day at least “much of the time” and 75% of Small Changes (and 28% and 24% in Large and Control) reported making small changes to diet every day. Weight losses during the first 4 months also differed significantly, with weight losses (mean [SE]) of −0.64 (0.22), −1.48 (0.23) and −3.60 (0.22) kg for Control, Small Changes and Large Changes respectively (all pairwise comparisons p<0.05).
Weight Changes
Figure 2 presents the weight changes for the three groups at each assessment. The primary outcome -- mean (SE) weight change averaged across all the follow-up assessments —was 0.26 (0.22), −0.56 (0.22), and −2.37 (0.22) kg in Control, Small Changes and Large Changes, respectively. All three pair-wise comparisons were significant, with less weight gain (and even some weight loss) in Small Changes (p=0.018) and Large Changes (p<0.001) relative to Control and in Large Changes relative to Small Changes (p<0.001).
Sensitivity analyses (Supplement Figure 1a–1c) including Smart-scale weights, using percent weight change (rather than kg) or censoring the data at 2 years all confirmed that both Large and Small Changes differed significantly from Control and Large Changes differed from Small Changes. Inference generated by multiple imputation of missing weight changes yielded comparable results.
Secondary Weight Outcomes
Results for the two dichotomous secondary outcomes are presented in Figure 3. The proportion of participants who gained ≥1 lb over baseline and the proportion that developed obesity increased steadily over time, with the greatest increases seen in the Control group. Across the 3 years of follow-up (Table 2), weight gains ≥1 lb were less common in Large Changes (23.6%) than in either Small Changes or Control (32.5% and 40.8% respectively). The incidence of obesity was significantly greater in the Control group (16.9%) than in either intervention group (7.9% in Small and 8.6% in Large Changes, respectively, p<0.05). Weight changes between baseline and 2 years (Table 2)) were greater in the Control group than in either Small or Large Changes (p<0.05), which did not differ from each other. Safety alerts occurred very infrequently (Supplement Table 1).
Proportion of participants in the Control, Small Changes and Large Changes groups who at 4, 12, 24, or 36 months had gained ≥1 lb from baseline (Figure 3a) or developed obesity (BMI ≥ 30 kg/m) (Figure 3b). (Figure 3b excludes 16 participants who had a BMI of 30 – 30.9 at baseline)
Table 2
Summary of primary and secondary results: mean (standard errors) from generalized linear models.
| Control | Small | Large | Pairwise comparisons* | ||||
|---|---|---|---|---|---|---|---|
| Overall | C vs S | C vs L | S vs L | ||||
| Primary Outcome | |||||||
| Change in weight (kg) over average of 3 years follow-up | 0.26 (0.22) | −0.56 (0.22) | −2.37 (0.22) | <0.001 | Mean=0.82 95% CI[0.23, 1.41] P=0.018 | Mean=2.64 [2.05, 3.22] <0.001 | Mean=1.81 [1.22, 2.41] <0.001 |
| Secondary Outcomes | |||||||
| % gaining ≥1 lb over average of 3 years follow-up | 40.8 (4.4) | 32.5 (3.8) | 23.6 (2.8) | <0.001 | OR=1.41 95% CI [1.02, 1.98] P = 0.09 | OR=2.28 [1.64, 3.19] <0.001 | OR=1.62 [1.14, 2.31] 0.025 |
| % developing obesity at least once during follow-up a | 16.9 (2.7) | 7.9 (2.0) | 8.6 (2.0) | 0.008 | OR=2.36 95% CI [1.23,4.52] P = 0.002 | OR=2.13 [1.12, 4.10] 0.02 | OR=0.92 [0.44, 1.91] 0.27 |
| Change in weight (kg) from baseline to 2 years | 0.54 (0.33) | −0.77 (0.33) | −1.50 (0.34) | <0.001 | 1.31 95% CI [0.39, 2.24] P = 0.016 | 2.04 [1.11, 2.98] <0.001 | 0.74 [−0.20, 1.66] 0.33 |
Subgroup Effects
Pre-specified interaction tests based on percent weight loss (to control for differences in baseline weights) revealed no significant differences among intervention effects across subgroups based on gender, age, and baseline weight. (Supplement Figure 2). In addition, no differences among race/ethnicity subgroups were evident (p > 0.10).
Study Participants
A total of 599 adults were enrolled (292 in Providence RI and 307 in Chapel Hill, NC) between August 2010 and February 2012 and randomly assigned to one of the three groups. Participants in the three groups were similar at baseline (Table 1). The study sample included 22% males and 27% from minority groups. Participants were [Mean (SD)] 28.2 (4.4) years of age and over 70% were above age 25; average BMI was 25.4 (2.6) kg/m and approximately 50% were normal weight. The majority (63%) were employed full-time. Figure 1 presents the CONSORT diagram showing retention at each of the follow-up assessments. Retention did not differ among groups.
Consort Diagram
Table 1
Characteristics of the SNAP participants at enrollment by intervention assignment: Mean (Standard Deviation) or Frequency (Percent).
| Baseline Characteristics | Intervention Assignment | ||
|---|---|---|---|
| Control N=202 | Small Changes N=200 | Large Changes N=197 | |
| Gender | |||
| Female | 158 (78.2%) | 157 (78.5%) | 154 (78.2%) |
| Male | 44 (21.8%) | 43 (21.5%) | 43 (21.8%) |
| Age, years | |||
| 18–24.9 | 53 (26.2%) | 60 (30.0%) | 56 (28.4%) |
| 25–35 | 149 (73.8%) | 140 (70.0%) | 141 (71.6%) |
| BMI, kg/m2 | |||
| <25 | 97 (48.0%) | 84 (42.0%) | 96 (48.7%) |
| ≥25 | 105 (52.0%) | 116 (58.0%) | 101 (51.3%) |
| Weight, kg | 71.4 (10.2) | 71.9 (11.0) | 70.8 (11.0) |
| Race/ethnicity | |||
| African-American | 19 (9.4%) | 25 (12.5%) | 22 (11.2%) |
| Non-Hispanic White | 148 (73.3%) | 146 (73.0%) | 144 (73.1%) |
| Asian/Pacific Islander | 7 (3.5%) | 11 (5.5%) | 7 (3.5%) |
| Hispanic | 15 (7.4%) | 10 (5.0%) | 21 (10.6%) |
| Multiple/Refusal | 13 (6.4%) | 8 (4.0%) | 3 (1.5%) |
| Education | |||
| Not college graduate | 39 (19.3%) | 46 (23.0%) | 35 (17.8%) |
| College graduate | 163 (80.7%) | 154 (77.0%) | 162 (82.2%) |
| Employment Status | |||
| Employed full time | 127 (62.9%) | 125 (62.5%) | 125 (63.5%) |
| Student full time | 59 (24.3%) | 61 (30.5%) | 55 (27.9%) |
| Other | 26 (12.8%) | 14 (7.0%) | 17 (8.6%) |
Intervention Delivery
Participants in Large and Small Changes attended an average of 87.4% and 86.0% of the 10 intervention sessions; 100% of the control group participants attended their one session. Intervention fidelity, determined by masked raters evaluating a randomly selected sample of 20% of recorded group sessions, was excellent, with 100% accuracy for distinguishing Large from Small Change sessions and for presentation of the appropriate behavioral content. Self-weighing, a cornerstone of self-regulation interventions, was increased in the two interventions. Whereas at baseline, 11–13% of each group reported daily self-weighing, at four months, daily self-weighing was reported by 75%, 72%, and 30% of Large Changes, Small Changes, and Control participants respectively (p<0.001).
Use of the prescribed behavioral strategies also differed significantly (p<0.05) by randomization group; for example at 4 months, 64% of Large Changes participants (and 10% and 11% in Small Changes and Control) reported reducing calories by 500–1000 kcal/day at least “much of the time” and 75% of Small Changes (and 28% and 24% in Large and Control) reported making small changes to diet every day. Weight losses during the first 4 months also differed significantly, with weight losses (mean [SE]) of −0.64 (0.22), −1.48 (0.23) and −3.60 (0.22) kg for Control, Small Changes and Large Changes respectively (all pairwise comparisons p<0.05).
Weight Changes
Figure 2 presents the weight changes for the three groups at each assessment. The primary outcome -- mean (SE) weight change averaged across all the follow-up assessments —was 0.26 (0.22), −0.56 (0.22), and −2.37 (0.22) kg in Control, Small Changes and Large Changes, respectively. All three pair-wise comparisons were significant, with less weight gain (and even some weight loss) in Small Changes (p=0.018) and Large Changes (p<0.001) relative to Control and in Large Changes relative to Small Changes (p<0.001).
Mean (± 1 standard error) weight changes over time by intervention assignment.
Sensitivity analyses (Supplement Figure 1a–1c) including Smart-scale weights, using percent weight change (rather than kg) or censoring the data at 2 years all confirmed that both Large and Small Changes differed significantly from Control and Large Changes differed from Small Changes. Inference generated by multiple imputation of missing weight changes yielded comparable results.
Secondary Weight Outcomes
Results for the two dichotomous secondary outcomes are presented in Figure 3. The proportion of participants who gained ≥1 lb over baseline and the proportion that developed obesity increased steadily over time, with the greatest increases seen in the Control group. Across the 3 years of follow-up (Table 2), weight gains ≥1 lb were less common in Large Changes (23.6%) than in either Small Changes or Control (32.5% and 40.8% respectively). The incidence of obesity was significantly greater in the Control group (16.9%) than in either intervention group (7.9% in Small and 8.6% in Large Changes, respectively, p<0.05). Weight changes between baseline and 2 years (Table 2)) were greater in the Control group than in either Small or Large Changes (p<0.05), which did not differ from each other. Safety alerts occurred very infrequently (Supplement Table 1).
Proportion of participants in the Control, Small Changes and Large Changes groups who at 4, 12, 24, or 36 months had gained ≥1 lb from baseline (Figure 3a) or developed obesity (BMI ≥ 30 kg/m) (Figure 3b). (Figure 3b excludes 16 participants who had a BMI of 30 – 30.9 at baseline)
Table 2
Summary of primary and secondary results: mean (standard errors) from generalized linear models.
| Control | Small | Large | Pairwise comparisons* | ||||
|---|---|---|---|---|---|---|---|
| Overall | C vs S | C vs L | S vs L | ||||
| Primary Outcome | |||||||
| Change in weight (kg) over average of 3 years follow-up | 0.26 (0.22) | −0.56 (0.22) | −2.37 (0.22) | <0.001 | Mean=0.82 95% CI[0.23, 1.41] P=0.018 | Mean=2.64 [2.05, 3.22] <0.001 | Mean=1.81 [1.22, 2.41] <0.001 |
| Secondary Outcomes | |||||||
| % gaining ≥1 lb over average of 3 years follow-up | 40.8 (4.4) | 32.5 (3.8) | 23.6 (2.8) | <0.001 | OR=1.41 95% CI [1.02, 1.98] P = 0.09 | OR=2.28 [1.64, 3.19] <0.001 | OR=1.62 [1.14, 2.31] 0.025 |
| % developing obesity at least once during follow-up a | 16.9 (2.7) | 7.9 (2.0) | 8.6 (2.0) | 0.008 | OR=2.36 95% CI [1.23,4.52] P = 0.002 | OR=2.13 [1.12, 4.10] 0.02 | OR=0.92 [0.44, 1.91] 0.27 |
| Change in weight (kg) from baseline to 2 years | 0.54 (0.33) | −0.77 (0.33) | −1.50 (0.34) | <0.001 | 1.31 95% CI [0.39, 2.24] P = 0.016 | 2.04 [1.11, 2.98] <0.001 | 0.74 [−0.20, 1.66] 0.33 |
Subgroup Effects
Pre-specified interaction tests based on percent weight loss (to control for differences in baseline weights) revealed no significant differences among intervention effects across subgroups based on gender, age, and baseline weight. (Supplement Figure 2). In addition, no differences among race/ethnicity subgroups were evident (p > 0.10).
Discussion
Previous studies have failed to identify interventions with long-term effects on weight gain in young adults, who are at high risk for weight gain.10,12,13 In our study, we found that self-regulation interventions involving small or large changes were both effective in reducing the mean weight gain (and producing small weight loss) relative to control over an average follow-up of 3 years. In addition, Large Changes was more effective than Small Changes. These results were confirmed in several sensitivity analyses. The Large Changes intervention also successfully decreased the proportion of participants who experienced weight gains ≥1 pound over the follow-up, and both interventions reduced the incidence of obesity during follow-up by almost 50% relative to Control, representing a clinically significant reduction in risk of developing obesity.
This is the first study to test two different behavior change approaches to weight gain prevention and the first large study of the small changes approach.16 Although the theoretical basis of small changes has recently been questioned,34 we found that the Small Changes approach had long-term efficacy in preventing weight gain relative to Control. Large Changes was more effective over the follow-up because of the significant weight losses produced at 4 months. This weight loss was followed by gradual regain, as seen in other weight loss 35 and weight gain prevention trials 15 and by 2 years, Large Change participants had regained 2.1 kg (58% of their initial weight loss). In contrast, Small Changes lost less initially (1.48 kg at 4 months), but had a more stable trajectory, gaining only 0.7 kg between months 4 and year 2. At two years, weight change in both Small and Large Changes differed significantly from the Control group, but not from each other. Because the difference between Large and Small Change approaches diminished over time, further follow-up is clearly needed. In addition, it would be important to determine whether success at weight gain prevention would be maximized by periodically repeating the initial 4 month program to reinstate the buffer and/or re-engage participants in the small changes approach.
These interventions could easily be disseminated through community organizations and eHealth or mHealth approaches. The format used in SNAP, with initial face-to-face group meetings followed by Internet delivered maintenance approaches, may have improved outcomes as face-to-face approaches appear superior to Internet delivered weight loss interventions36 and purely eHealth interventions have not been effective in preventing weight gain.37 Likewise, weight loss and maintenance programs, which have included face-to-face or telephone counseling,15,38,39 appear more effective than internet only programs. Further research to determine how best to combine face-to-face and eHealth approaches and which individuals respond better to the Large versus Small Change approach is needed.
Observational studies of weight gain suggest that young adults gain about .6 – .8 kg per year, 4–7 whereas our control group gained .54 kg at 2 years. The smaller weight gains in our control group may reflect the fact that SNAP was a weight gain prevention trial (vs an observational study) and the Control group received some, albeit modest, intervention.
Strengths of this trial include the large sample size, recruitment beyond college campuses, objectively measured outcome by blinded personnel, and the successful implementation and comparison of two different approaches to weight gain prevention. Other strengths include the fact that participants were followed over an average of 3 years and retention rates remained high. The main limitation of this efficacy trial is the generalizability of the results; participants in the trial were disproportionately female, non-Hispanic white, and college graduates. All participants were interested in being in a weight gain prevention trial and the screening process likely led to the selection of a highly motivated sample. Future studies should examine the effectiveness of these two interventions in other individuals more representative of the general population of young adults. The findings from this trial suggest that self-regulation approaches that include frequent self-weighing have clinically significant beneficial effects on reducing weight gain and risk of obesity in young adults and indicate that Large Changes may be particularly effective in reducing average weight gain over three years of follow-up. Further follow-up is clearly needed to determine whether effects are maintained over time and whether Large or Small Changes produce the best long-term outcomes. Since both small and large change interventions reduced weight gain relative to control, it may be important to consider individual preferences in selecting which approach to recommend. Given the success of both approaches in reducing incidence of obesity in the current sample of high-risk young adults, dissemination of these approaches could help to combat the epidemic of obesity.
Supplementary Material
Acknowledgments
Funding Support:
NIH/NHLBI Grant Numbers: U01HL090864 and U01HL090875
Abstract
Importance
Weight gain occurs commonly in young adults and has adverse effects on health. Reducing weight gain in young adults would have significant public health impact.
Objective
To compare two self-regulation interventions versus control in reducing weight gain in young adults over an average follow-up of 3 years.
Design
Randomized controlled trial
Setting
Two academic settings
Participants
599 participants (22% men; 27% minority), age 18–35 (27.7 (4.4)), BMI 21–30 kg/m (25.4 (2.6)), recruited via mailings and e-mails between August 2010 and February 2012.
Interventions
Participants were randomly assigned to: Control, Self-Regulation plus Small Changes or Self-Regulation plus Large Changes. The interventions focused on frequent self-weighing and using this information to know when behavior changes were needed. “Small Changes” taught participants to reduce intake and increase activity, both by approximately 100 calories per day. “Large Changes” focused on losing 5–10 pounds initially to buffer against expected weight gain. Interventions were delivered via 10 in-person group sessions during months 1–4 and then primarily online. Assessments occurred at randomization, 4 months, and then annually for 2–4 years (depending on time of recruitment).
Main outcome
Changes in weight from baseline over average follow-up of 3 years. Secondary outcomes: proportion gaining ≥1 pound from baseline, proportion developing obesity (BMI ≥30 kg/m), weight change baseline to 2 years (last assessment reached by all cohorts)
Results
Mean (SE) weight changes over an average follow-up of 3 years were +0.26 kg (0.22), −0.56 kg (0.22), and −2.37 kg (0.22) in Control, Small Changes and Large Changes, respectively (p=0.001). Differences among all three groups were significant (Large Changes vs Control, p<.001; Small Changes vs Control, p<.02 and Large Changes vs Small Changes, p<.001). Both interventions significantly reduced incidence of obesity (p<.05) relative to Control (8.6%, 7.9% and 16.9% in Large, Small and Control, respectively); Large Changes reduced the proportion that gained ≥1 pound (p<.05).
Conclusions and Relevance
Self-regulation with Large or Small Changes both reduced weight gain in young adults over three years relative to Control, but Large Changes was most effective. These interventions could easily be implemented on a large scale and may impact public health.
Trial registration
ClinicalTrials.gov, {"type":"clinical-trial","attrs":{"text":"NCT01183689","term_id":"NCT01183689"}}NCT01183689
Footnotes
Author Contributions:
Study concept and design: Wing, Tate, Espeland, Lewis, Gokee LaRose, GorinAcquisition and management of data: Wing, Tate, Espeland, Hatley, Ferguson, Bahnson, Perdue
Statistical analysis: Espeland, Garcia, Lang
Drafting of manuscript: Wing, Tate, Espeland
Critical review of manuscript: Lewis, Gokee LaRose, Gorin, Bahnson, Perdue, Hatley, Ferguson, Garcia, Lang
.Financial Disclosures:
Dr. Tate reports being a member of the Scientific Advisory Board for Weight Watcher’s International.
Role of Sponsor: The sponsor approved study design and interpretation of the data and coordinated this trial with others in the EARLY Consortium.
Additional Contributions: We wish to thank the SNAP Data and Safety Monitoring Board and the participants who made this study possible.
References
- 1. James WPT, Gill TP. Prevention of Obesity. In: Bray G, Bouchard C, editors. Handbook of Obesity: ClinicalApplications. 3. New York, NY: Informa Healthcare USA, Inc; 2008. pp. 157–175. [PubMed]
- 2. Colditz GA, Willett WC, Rotnitzky A, Manson JEWeight gain as a risk factor for clinical diabetes mellitus in women. Ann Intern Med. 1995;122:481–486.[PubMed][Google Scholar]
- 3. Stevens VL, Jacobs EJ, Patel AV, Sun J, Gapstur SM, McCullough MLBody weight in early adulthood, adult weight gain, and risk of endometrial cancer in women not using postmenopausal hormones. Cancer Causes Control. 2014 Mar;25(3):321–328.[PubMed][Google Scholar]
- 4. Truesdale KP, Stevens J, Lewis CE, Schreiner PJ, Loria CM, Cai JChanges in risk factors for cardiovascular disease by baseline weight status in young adults who maintain or gain weight over 15 years: the CARDIA study. Int J Obes. 2006 Sep;30(9):1397–1407.[Google Scholar]
- 5. Williamson DF, Kahn HS, Remington PL, Anda RFThe 10-year incidence of overweight and major weight gain in US adults. Arch Intern Med. 1990;150:665–672.[PubMed][Google Scholar]
- 6. Ball K, Crawford D, Ireland P, Hodge APatterns and demographic predictors of 5-year weight change in a multi-ethnic cohort of men and women in Australia. Public Health Nutr. 2003 May;6(3):269–281.[PubMed][Google Scholar]
- 7. Reas DL, Nygard JF, Svensson E, Sorensen T, Sandanger IChanges in body mass index by age, gender, and socio-economic status among a cohort of Norwegian men and women (1990–2001) BMC Public Health. 2007;7:269.[Google Scholar]
- 8. Renehan AG, Flood A, Adams KF, et al Body mass index at different adult ages, weight change, and colorectal cancer risk in the National Institutes of Health-AARP Cohort. Am J Epidemiol. 2012 Dec 15;176(12):1130–1140.[Google Scholar]
- 9. Adams KF, Leitzmann MF, Ballard-Barbash R, et al Body mass and weight change in adults in relation to mortality risk. Am J Epidemiol. 2014 Jan 15;179(2):135–144.[Google Scholar]
- 10. Hebden L, Chey T, Allman-Farinelli MLifestyle intervention for preventing weight gain in young adults: a systematic review and meta-analysis of RCTs. Obes Rev. 2012 Aug;13(8):692–710.[PubMed][Google Scholar]
- 11. Hivert MF, Langlois MF, Berard P, Cuerrier JP, Carpentier ACPrevention of weight gain in young adults through a seminar-based intervention program. Int J Obes (2005) 2007 Aug;31(8):1262–1269.[PubMed][Google Scholar]
- 12. Jeffery R, French SPreventing weight gain in adults: the pound of prevention study. Am J Public Health. 1999;89(5):747–751.[Google Scholar]
- 13. Levine MD, Klem ML, Kalarchian MA, et al Weight gain prevention among women. Obesity (Silver Spring) 2007 May;15(5):1267–1277.[Google Scholar]
- 14. Partridge SR, Juan SJ, McGeechan K, Bauman A, Allman-Farinelli MPoor quality of external validity reporting limits generalizability of overweight and/or obesity lifestyle prevention interventions in young adults: a systematic review. Obes Rev. 2015 Jan;16(1):13–31.[PubMed][Google Scholar]
- 15. Wing RR, Tate DF, Gorin AA, Raynor HA, Fava JLA self-regulation program for maintenance of weight loss. The New England journal of medicine. 2006 Oct 12;355(15):1563–1571.[PubMed][Google Scholar]
- 16. Hill JO, Wyatt HR, Reed GW, Peters JCObesity and the environment: Where do we go from here? Science. 2003;299:853–855.[PubMed][Google Scholar]
- 17. Rodearmel SJ, Wyatt HR, Stroebele N, Smith SM, Ogden LG, Hill JOSmall changes in dietary sugar and physical activity as an approach to preventing excessive weight gain: the America on the Move family study. Pediatrics. 2007 Oct;120(4):e869–879.[PubMed][Google Scholar]
- 18. Lutes LD, Daiss SR, Barger SD, Read M, Steinbaugh E, Winett RASmall changes approach promotes initial and continued weight loss with a phone-based follow-up: nine-month outcomes from ASPIRES II. Am J Health Promot. 2012 Mar-Apr;26(4):235–238.[PubMed][Google Scholar]
- 19. Rodearmel SJ, Wyatt HR, Barry MJ, et al A family-based approach to preventing excessive weight gain. Obesity (Silver Spring) 2006 Aug;14(8):1392–1401.[PubMed][Google Scholar]
- 20. Damschroder LJ, Lutes LD, Goodrich DE, Gillon L, Lowery JCA small-change approach delivered via telephone promotes weight loss in veterans: results from the ASPIRE-VA pilot study. Patient Educ Couns. 2010 May;79(2):262–266.[PubMed][Google Scholar]
- 21. Kuller LH, Simkin-Silverman LR, Wing RR, Meilahn EN, Ives DGWomen’s Healthy Lifestyle Project: A randomized clinical trial: results at 54 months. Circulation. 2001 Jan 2;103(1):32–37.[PubMed][Google Scholar]
- 22. Wing RR, Tate D, Espeland M, et al Weight gain prevention in young adults: design of the study of novel approaches to weight gain prevention (SNAP) randomized controlled trial. BMC Public Health. 2013 Apr 4;13(1):300.[Google Scholar]
- 23. Lytle LA, Svetkey LP, Patrick K, et al The EARLY trials: a consortium of studies targeting weight control in young adults. Transl Behav Med. 2014 Sep;4(3):304–313.[Google Scholar]
- 24. Tate DF, LaRose JG, Griffin LP, et al Recruitment of young adults into a randomized controlled trial of weight gain prevention: message development, methods, and cost. Trials. 2014;15:326.[Google Scholar]
- 25. Kanfer FH, Goldstein AP Helping people change. New York: Pergamon Press, Inc; 1975. [PubMed][Google Scholar]
- 26. Carver CS, Scheier MF. Principles of Feedback Control. In: Carver CS, Scheier MF, editors. On the Self-Regulation of Behavior. Cambridge: Cambridge University Press; 1998. [PubMed]
- 27. Carver CS, Scheier MFControl theory: a useful conceptual framework for personality-social, clinical, and health psychology. Psychol Bull. 1982 Jul;92(1):111–135.[PubMed][Google Scholar]
- 28. Carver CS, Scheier MFOrigins and functions of positive and negative affect: A control-process view. Psychol Rev. 1990;97:19–35.[PubMed][Google Scholar]
- 29. Wing RR, Epstein LH, Nowalk MP, Lamparski DMBehavioral self-regulation in the treatment of patients with diabetes mellitus. Psychol Bull. 1986;99:78–89.[PubMed][Google Scholar]
- 30. Gokee LaRose J, Tate DF, Gorin AA, Wing RRPreventing weight gain in young adults: a randomized controlled pilot study. Am J Prev Med. 2010 Jul;39(1):63–68.[Google Scholar]
- 31. Donnelly JE, Blair SN, Jakicic JM, Manore MM, Rankin JW, Smith BK American College of Sports Medicine Position StandAppropriate physical activity intervention strategies for weight loss and prevention of weight regain for adults. Med Sci Sports Exerc. 2009 Feb;41(2):459–471.[PubMed][Google Scholar]
- 32. SAS System for Mixed Models [computer program] Cary, NC: SAS Institute Inc; 1996. [PubMed]
- 33. Yuan YMultiple imputation using SAS software. J Statistic Software. 2011;45:1–25.[PubMed][Google Scholar]
- 34. Hall KD, Heymsfield SB, Kemnitz JW, Klein S, Schoeller DA, Speakman JREnergy balance and its components: implications for body weight regulation. Am J Clin Nutr. 2012 Apr;95(4):989–994.[Google Scholar]
- 35. Wing RR. Behavioral approaches to the treatment of obesity. In: Bray G, Bouchard C, editors. Handbook of Obesity: Clinical Applications. 3. New York: Informa Health Care USA, Inc; 2008. [PubMed]
- 36. Harvey-Berino J, West D, Krukowski R, et al Internet delivered behavioral obesity treatment. Prev Med. 2010 Aug;51(2):123–128.[Google Scholar]
- 37. Hutchesson MJ, Rollo ME, Krukowski R, et al eHealth interventions for the prevention and treatment of overweight and obesity in adults: a systematic review with meta-analysis. Obes Rev. 2015 May;16(5):376–392.[PubMed][Google Scholar]
- 38. Svetkey LP, Stevens VJ, Brantley PJ, et al Comparison of strategies for sustaining weight loss: the weight loss maintenance randomized controlled trial. JAMA. 2008 Mar 12;299(10):1139–1148.[PubMed][Google Scholar]
- 39. Appel LJ, Clark JM, Yeh HC, et al Comparative effectiveness of weight-loss interventions in clinical practice. N Engl J Med. 2011 Nov 24;365(21):1959–1968.[Google Scholar]