Abstract
Background
This analysis helps clarify the individual and joint effects of moderate to vigorous physical activity (MVPA) in concert with significant life events (eg, divorce, marriage, job change or loss, pregnancy, etc) on weight following a behavioral weight loss intervention.
Methods
Data from the Tracking Study weight loss trial were utilized to perform a 4-way decomposition of moderation and mediation of life events (≥ 1 vs 0) and MVPA (low <2500 kcal vs high ≥ 2500 kcal) on 24-month weight.
Results
The total effect of life events and MVPA on weight was estimated to be 1.63 kg (95% confidence interval = 0.30 to 2.96; P = .02). The controlled direct effect of life events on 24-month weight suggested that experiencing at least one life event controlling for MVPA was associated with an increase of 2.31 kg (95% confidence interval = 0.29 to 4.33) at 24 months. Other interaction and mediation estimates were not statistically significant.
Conclusions
This analysis offers new potential for examining health behaviors that may act as both mediators and effect modifiers of health. Although more work is needed to understand the interaction of life events and MVPA on weight loss maintenance, findings help rule out mediation. Life events and MVPA should be considered for their unique effects on weight loss maintenance in the future.
Keywords: methods, obesity, interventions
During the maintenance phase following weight loss programs, when intervention support for weight loss behaviors taper off, individuals are at higher risk for weight gain. Predicting and promoting successful maintenance is challenging and few randomized controlled trials have been conducted to examine behavioral factors associated with maintenance.1 Cohort studies have identified regular moderate to vigorous physical activity (MVPA) as a strong correlate of successful weight loss maintenance.2,3 Although the caloric expenditure associated with physical activity is likely the major mechanism through which higher levels of physical activity mitigate weight regain, the role of life stressors in this process has not been examined. Importantly, MVPA has been shown to improve mood and ability to cope with life stressors including daily hassles, divorce, legal issues, job loss or change, and death of close friends or family members.4–6 Although research has focused on specific life events that have triggered individuals to lose weight,7 the role of life stressors in the weight loss maintenance phase, including whether and/or how physical activity may affect the relationship between life events and weight regain has received limited attention.
Obtaining a better understanding of the relationships among physical activity, stressful life events, and weight regain is important given that physical activity may function as a moderator and a mediator of weight regain. For example, MVPA participation may moderate the relationship between stressful life events and associated weight gain. That is, the effect of life events on weight regain may differ depending on how physically active is an individual, with higher levels of MVPA a protective factor. Conversely, MVPA may mediate the relationship between life events and weight regain, such that the occurrence of stressful life events is associated with reductions in MVPA, which in turn contribute to weight regain. Current research has focused on specific life events that have prompted individuals to lose weight and take on behavior change that has been sustained.7 However, examination of the impacts of major life events that may interrupt or negatively impact behavior change progress after weight loss programs, or factors such as physical activity that may modify these impacts, have not received the same attention. This pathway is worthy of exploration given that individuals who have recently adopted or increased MVPA are at higher risk for disengagement and relapse due to stressful life events.8 This risk for behavioral relapse due to the occurrence of life events has the following intervention as a common consideration in behavioral maintenance theory as well.9 However, little empirical evidence exists about the effects of major life events on physical activity in the context of weight loss interventions.
It is important to understand whether physical activity is a moderator and/or a mediator of the relationship between life events and weight loss maintenance following participation in a behavioral weight loss intervention. The aim of this work was to examine the impact of physical activity, in the context of life events, on weight following an active weight loss intervention. Although adherence to high MVPA has been associated with better weight loss maintenance,2 this analysis attempted to clarify the individual and joint effects of significant life events and MVPA on weight out to 24 months. We hypothesized the total effect of greater reporting of life events on weight within a lifestyle weight loss intervention would be influenced by individuals’ physical activity levels during the maintenance phase. Using this decomposition method, we further expected a portion of the effect of reported life events on weight following a behavioral weight loss intervention would be moderated by physical activity level during the maintenance period, such that individuals who participated in higher levels of physical activity would weigh less at 24 months despite life events that were experienced. Finally, a portion of the effect of reported life events on weight following a behavioral weight loss intervention would be mediated by physical activity during the maintenance period, such that the occurrence of significant life events would lead to increased weight through low participation in MVPA.
Methods
Study Design and Population
To assess differences in life events and the association with MVPA and 24-month weight following a year-long weight loss intervention, data from a randomized controlled weight loss trial were utilized. The Tracking Study was a 3-arm randomized trial comparing the efficacy of self-weighing prescriptions during a behavioral weight loss intervention on weight and psychological outcomes over 24 months.10
Men and women residing throughout the Minneapolis-St Paul, MN metropolitan area were enrolled between July 2012 and August 2013. Eligibility criteria required that participants had a body mass index between 25 and 40 kg/m2, access to home wireless internet, no recent weight loss greater than 10 pounds or history of bariatric surgery, and no history of significant physical or mental health concerns, including a binge eating disorder. Women could not be pregnant or breastfeeding or have plans to become pregnant for the 24 months of the study.10 A total of 339 individuals gave consent and were enrolled in the Tracking Study. Both the Tracking Study and this secondary data analysis were approved by the University of Minnesota Internal Review Board.
Individuals were recruited, randomized, and began the intervention in 3 waves staggered 6 months apart beginning in July 2012. Participation in the final wave concluded in September 2015. Detailed information regarding the Tracking Study has been published elsewhere.10 A standard behavioral weight loss intervention was conducted for the first 12 months of the study.10,11 Activities included, dietician-lead in-person group meetings with education and discussion weekly for 6 months, biweekly for 2 months, and monthly for 4 months, followed by a 12-month measurement-only maintenance period. Participants completed questionnaires and anthropometric measures at baseline, 6-, 12-, 18-, and 24-month time points.
Life Events
Individuals reported life events at 24 months, occurring in the prior 12-month maintenance period. An adaptation of the Life Experiences Survey12 assessed different types of life events, including marriage, pregnancy, death of spouse or close family members, divorce, job loss or change, and change of residence, retrospectively, over the 12 months prior to administration. The measure was shortened to avoid undue participant burden during lengthy survey procedures for this study while still capturing key life events; all events assessed are listed in Table 1. Although participant responses tend to fluctuate over time, the measure has shown satisfactory reliability (r = .63–.64) over 6-week test–retest evaluation.12 Individuals were categorized and compared using a classification of no life events (0 life experiences reported) and some life events (≥1 life experiences reported). This definition is consistent with previous work has examined the impact of one major life event on weight.13 To adjust for previous behaviors and experiences during the weight loss intervention the total number of life events reported during the active weight loss phase (measured at 12 mo) was included as a covariate.
Table 1.
Life Events Survey List
| Event |
|---|
| Marriage |
| Detention in jail |
| Death of spouse |
| Major change to sleeping habits (much more or less sleep) |
| Death of a close family member: |
| Mother |
| Father |
| Brother |
| Sister |
| Grandmother |
| Grandfather |
| Other_______ |
| Major change in eating habits (much more or much less food intake) |
| Foreclosure on mortgage or loan |
| Death of a close friend |
| Outstanding personal achievement |
| Minor law violations (traffic tickets, disturbing the peace, etc) |
| Men: wife/girlfriends pregnancy |
| Women: pregnancy |
| Changed work situation (different work responsibility, major change in working conditions, working hours, etc) |
| New job |
| Serious illness or injury of close family member: |
| Mother |
| Father |
| Brother |
| Sister |
| Grandmother |
| Grandfather |
| Other_______ |
| Sexual difficulties |
| Trouble with employer (in danger of losing job, being suspended, demoted, etc) |
| Major change in financial status (a lot better off or a lot worse off) |
| Major change in closeness of family members (increased or decreased closeness) |
| Gaining a new family member (through birth, adoption, family member moving in, etc) |
| Change of residence |
| Marital separation of mate (due to conflict) |
| Major change in religious activities (increased or decreased attendance) |
Physical Activity
Participants self-reported MVPA via the Paffenbarger Activity Questionnaire (PAQ) at baseline, 6, 12, 18, and 24 months.14 This questionnaire assessed respondent’s MVPA in the previous week in both closed and open-form questions. Activity responses were categorized into light, moderate, and vigorous physical activity, using self-reported duration and researcher-coded intensity to estimate calories spent participating in the various physical activity intensities during the previous week. In addition, the amount of walking and stair climbing were specifically assessed in 2 separate questions.14,15 Responses to these questions were incorporated into a total caloric expenditure calculation. Estimated energy expenditure was validated against VO2 maximal output and had acceptable validity (r = .60).15 A binary variable indicating high MVPA at 24 months was defined as low (<2500 kcal) and high (≥2500 kcal). This cut point was defined by previous work into MVPA recommendations required for successful maintenance following weight loss.16–18 Baseline MVPA was included as a covariate to account for physical activity level prior to intervention.
Weight
The primary outcome of interest was weight (in kilograms) measured at 24 months. Trained staff measured weight at all assessment visits. Baseline and 12-month (the end of active intervention) weights were included in the models to account for prior weight change during active intervention.
Covariates
Covariates included self-reported demographic characteristics collected at baseline: gender, race/ethnicity, educational attainment, marital status, and age. Due to the demographic background of the sample, race/ethnicity was collapsed to a binary covariate (white/other) as well as education level (college degree or more/less than a college degree), marital status (married/other), and gender (male/female). Age was treated as a continuous variable as well as weight and PAQ measured physical activity.
Finally, diet measured at 12 months was used as a covariate in all models to account for weight status associations with caloric intake. Dietary measures were collected using a self-reported Dietary History Questionnaire-II19 at baseline, 6-, 12-, 18-, and 24-month time points. This online questionnaire asks participants whether they consumed a list 134 foods and beverages over the past year, followed by average frequency and portion consumed in the past month to estimate caloric average intake. It has been validated using 24-hour dietary recalls (r = .62–.66).20 The 12-month time point was most important as a covariate to correspond with dietary behaviors at the end of the active intervention, prior to the maintenance phase.
Analysis
Recently proposed statistical methods allow for the assessment of mediation and moderation simultaneously.21 This decomposition of mediating and moderating effects, proposed by VanderWeele,21 has thus far been used to examine the effects of genetic and physiologic exposures on health outcomes. However, this method also has the potential to offer a unique and innovative way to explore the effects of health behaviors that may act as both mediators and moderators of health outcomes.
A 4-way decomposition model to assess mediation and moderation, developed by VanderWeele21 was used to examine the proposed hypotheses. This decomposition method resulted in separate 4 components that make up the total effect of life events and high MVPA on weight. Figure 1A–1D illustrates these 4 components and the relationships they represent. The first component, the controlled direct effect (CDE), estimated the portion of the effect of life events on weight controlling for MVPA. This component is the independent effect of life events alone on weight. The second component, the reference interaction (INTref), estimated the effect of life events and high MVPA on weight due only to moderation. The third component, the mediated interaction (INTmed), estimated the portion of the effect of life events and high MVPA on weight that was both mediated and moderated. The fourth and final component, the mediated main effect or pure indirect effect (PIE), accounted for the portion of the effect of life events on weight that is due only to mediation through reported MVPA.21
Figure 1 —
Graphical depictions of single components in decomposition. (A) Controlled direct effect of life events on weight. (B) Reference interactions effect of life events and MVPA on weight. (C) Mediated interaction effect of life events and MVPA on weight. (D) Pure indirect effect of life events on weight mediated through MVPA. MVPA indicates moderate to vigorous physical activity.
Regression models, conditional on covariates were specified using PROC NLMIXED (version 9.4; SAS, Cary, NC). Code provided by VanderWeele21 was used to estimate the CDE, the PIE, and the INTmed and the INTref. Our main hypothesis of interest was the total effect of life events on 24-month weight, computed by combining all 4 components. This method utilized the binary exposure variable (≥1 life events vs 0 life events), the binary mediator (<2500 kcal vs ≥2500 kcal), and the continuous weight variable. All covariates were standardized to the mean and SD prior to regression modeling.
To test the first subhypothesis the percent of the effect due to interaction was computed. Using this method, only additive interactions were assessed. The percent of the total effect that is due to interaction between high MVPA and life events was found by summing the INTref and the INTmed over the TE.21
To test the second subhypothesis, the percent of the total effect that is due to mediation was computed. The percent of the total effect of life events on 24-month weight that is mediated through MVPA was found by summing of the PIE and the INTmed over the TE.21
Results
Summary statistics for participants reporting each category of life events (≥1 life events vs 0 life events) reported at 24 months as well as the baseline sample are provided in Table 2. Of the 339 individuals who were enrolled in the original study, 230 participants contributed complete data for this analysis. Pairwise comparisons examined differences between the baseline characteristics of participants with complete data at 24 months to the sample that was excluded from this analysis for incomplete data. Those who were included in this analysis were older (48.2 [9.5] y vs 43.1 [10.8] y; P < .01), weighed less at baseline (93.9 [15.6] kg vs 98.7 [13.5] kg; P < .01), and had a higher proportion of married (0.72 [0.45] vs 0.60 [0.49]; P = .03) participants than the excluded sample. No statistical differences in education level, race, gender, baseline physical activity, or condition were detected.
Table 2.
Weight, Change in PA, Caloric Intake, Previous Life Events, and Demographics for Levels of Life Events During the Maintenance Phase
| Characteristics | Baseline sample, n = 339 | No life events (reported at 24 mo), n = 52 | ≥1 life events (reported at 24 mo), n = 178 |
|---|---|---|---|
| Baseline weight, mean (SD), kg | 95.5 (15.1) | 94.6 (14.3) | 93.8 (15.9) |
| 12-mo weight | – | 84.5 (14.9) | 85.9 (14.9) |
| 24-mo weight | – | 87.4 (15.3) | 89.5 (15.5) |
| Baseline PA, measured by PAQ, kcal | 1309.9 (1250.9) | 1268.2 (1147.0) | 1305.4 (1191.8) |
| 12-mo PA | – | 2604.9 (2433.7) | 2017.1 (1635.8) |
| 24-mo PA | – | 2618.0 (2158.4) | 1972.2 (1624.0) |
| Life events, number reported during active weight loss (measured at 12 mo), mean (SD) | – | 1.9 (1.6) | 2.9 (2.1) |
| Age, mean years (SD) | 46.5 (10.2) | 48.5 (9.7) | 48.0 (9.5) |
| Women, n (%) | 220 (65) | 35 (67) | 117 (66) |
| White, n (%) | 293 (86) | 50 (96) | 155 (87) |
| College graduate, n (%) | 216 (64) | 31 (60) | 121 (68) |
| Married, n (%) | 230 (68) | 40 (77) | 125 (70) |
| Weighing assignment | |||
| No weighing, n | 116 | 19 | 58 |
| Weekly weighing, n | 109 | 15 | 63 |
| Daily weighing, n | 114 | 18 | 57 |
Abbreviations: PA, physical activity; PAQ, Paffenbarger Physical Activity Questionnaire.
At 24 months, 52 participants reported having experienced no life events and 178 reported at least one life event in the past 12 months. Those who reported experiencing at least one life event during the maintenance phase had previously reported experiencing on average 2.9 (2.1) life events during the active weight loss phase, whereas those who reported no life events during the maintenance phase had reported experiencing on average 1.9 (1.6) life events during the weight loss intervention. Of those participants who reported experiencing no life events during the maintenance phase, most were white (96%) and married (77%). These percentages were smaller in the group that reported experiencing life events (87% white and 70% married). Finally, those reporting no life events reported substantially more MVPA at 12- and 24-month time points compared with those reporting ≥1 life events: 2604.9 (2433.7) kcal versus 2017.1 (1635.8) kcal at 12 months, 2459.1 (2260.6) kcal versus 1881.7 (1720.0) kcal at 18 months, and 2618.0 (2158.4) kcal versus 1972.2 (1624.0) kcal at 24 months (see Table 2).
Prior to decomposition, multivariable linear regression models that assessed physical activity and life events independently showed that both life events (β = 1.42 kg, SE = 0.68, P = .04) and MVPA level (β = 1.32 kg, SE = 0.63, P = .04) were independently associated with weight (Table 3). Models with both life events and MVPA treated continuously examined the association with weight out to 24 months and found that an increase in 1 minute of MVPA was statistically significantly associated with a 0.001 kg (SE = 0.0001; P = .01) decrease in 24-month weight, whereas life events were not statistically significantly associated. Interactions between life events and MVPA at both time points were not statistically significant in the linear models. Therefore, the interaction terms were excluded from subsequent models.
Table 3.
Linear Regression Model of Life Events and MVPA on 24-Month Weight (in kilograms)
| β (SE) | P | ||
|---|---|---|---|
| Adjusted effects at 24-mo PAa | Life events (1 vs 0) | 1.42 (0.68) | .04 |
| PA (<2500 kcal vs ≥2500 kcal) | 1.32 (0.63) | .04 |
Abbreviations: PA, physical activity; MVPA, moderate to vigorous physical activity.
Adjustment variables include age, gender, race, education level, marital status, baseline and 12-month weight, baseline PA, 12-month calorie intake, life-event frequency during active weight loss, and randomization arm.
Results of the full decomposition analysis are found in Table 4. Using 24-month MVPA, the total effect of life events and MVPA on weight was 1.63 kg (95% confidence interval [CI] = 0.30 to 2.96; P = .02). The CDE of life events on weight was β = 2.31 kg (95% CI = 0.29 to 4.33). All other interaction and mediation decomposition estimates (ie, INTref, INTmed, PIE) were not statistically significant (Table 4).
Table 4.
Decomposition Model of Life Events and MVPA on 24-Month Weight (in kilograms)
| β (95% CI) | P | |
|---|---|---|
| Adjusted effects of life events with 24-moa PA | ||
| Total effect | 1.63 (0.30 to 2.96) | .02 |
| CDE | 2.31 (0.29 to 4.33) | .03 |
| INTref | −0.84 (−2.31 to 0.64) | .26 |
| INTmed | −0.24 (−0.72 to 0.25) | .34 |
| PIE | 0.39 (−0.17 to 0.95) | .17 |
| Percent mediation | 9% (−7% to 26%) | .28 |
| Percent interaction | −66% (−199% to 57%) | .31 |
Abbreviations: CDE, controlled direct effect; CI, confidence interval; INTmed, mediated interaction; INTref, reference interaction; MVPA, moderate to vigorous physical activity; PIE, pure indirect effect.
Adjustment variables include age, gender, race, education level, marital status, baseline and 12-month weight, baseline physical activity, 12-month calorie intake, life-event frequency during active weight loss, and randomization arm.
The estimated proportion of the total effect attributed to the interaction (−66%; 95% CI = −199% to 57%) appeared to be larger than the proportion of the total effect attributed to mediation (9%; 95% CI = −7% to 26%, respectively). However, these estimates were very small and not statistically significant (Table 4).
Discussion and Limitations
This exploration is innovative in the use of a novel statistical method to elucidate the relationship between life events, MVPA, and weight maintenance. The decomposition approach used here offers more information about the relationship between life events, MVPA, and weight than the traditional multivariable regression model reported in Table 3. Results of the decomposition support our main hypothesis that the total effect of life events and high MVPA at 24 months (during the maintenance period) was statistically significantly associated with weight. This is consistent with previous work examining the separate effects of life events22 and MVPA2,3 on weight maintenance. Results of the decomposition model highlight the importance of accounting for MVPA when considering the effect of life events on weight loss maintenance.
The subhypotheses assessing moderation and mediation were not supported as the 4 decomposed results were not statistically significant. Though not statistically significant, both interaction estimates (INTref and INTmed) point to a negative interaction or moderation, which is suggestive that those achieving a higher level of MVPA weigh less despite life events experienced. In addition, the CDE, the effect of life events in isolation on weight, controlling for any mediation or moderation by MVPA, was statistically significant and greater than the total effect. This suggests that mediation of the effect of life events on weight gain through MVPA is unlikely. This is somewhat inconsistent with previous work suggesting that major life events often lead to low physical activity.23 However, to our knowledge, this is the first study that has explored the mediation effects of several different potential life events and MVPA on weight in a weight loss sample, suggesting that although MVPA is integral for weight loss maintenance,3 more work is needed to understand the role of life events on this relationship.
Overall, findings suggest that more understanding of how major life events may enable or disrupt behavior changes in weight loss interventions is needed. Although the 24-month follow-up allows exploration of behavior change over a relatively long period of time following an active weight loss intervention, future work should examine whether these relationships change depending on the nature of the life events. For example, although a positive life event such as a work promotion may result in positive feelings and a negative life event of facing unemployment may result in negative feelings, research has suggested that working longer hours are associated with weight gain.24 Therefore, the effects of positive and negative life events on health behaviors, psychosocial factors, and weight should be further explored. Moreover, other factors that might mitigate the effects of life events on weight including social support, education level, and socioeconomic status should be examined. Successful weight loss maintenance is the outcome of a relatively complex set of maintained behaviors beyond MVPA that require further examination to be fully understood.25 Similarly, life events impact many behaviors and factors that may influence weight. Therefore, this study may only offer narrow insights into the relationship between life events and 24-month weight. However, noting the relatively small number of participants who experienced no life events, intervention components to help mitigate the effects of life events will be important for successful maintenance going forward.
There are several limitations that should be noted. First, the identification of the 4 component effects requires several assumptions regarding confounding. After covariate adjustment, this method assumes no other unmeasured confounding of the effects of life events on either 24-month weight or MVPA, as well as no unmeasured confounding of the effects of MVPA on 24-month weight. Finally, it is assumed that all recanting witness variables are included as covariates. That is, all confounders between MVPA and 24-month weight who are affected by life events have been measured and accounted for in the models. Each of these assumptions are described in more detail by VanderWeele21 and Vansteelandt26 and Pearl.27
Although multiple covariates were included in this analysis, the above assumptions are very strong and unmeasured confounding within this complex relationship of life events, MVPA, and weight likely remain. In addition, measuring only weight and height may ignore body composition changes that did not result in weight loss but may affect health, such as by increases in muscle mass and decreased fat mass. However, due to the focus of the curriculum on lowering caloric intake through dietary changes,10 we expect that individuals who had gained weight through muscle mass increase coupled with fat loss to be few and far between.
The potential for measurement error in the self-reporting of life events and MVPA is another limitation. It is plausible, that participants may incorrectly remember the time window of these life events. Furthermore, life events were reported concurrently with MVPA, which has the potential to introduce recall bias and impact the temporality assumptions of the statistical models as well. Examining life events in shorter intervals, and prior to measuring MVPA and weight in future work may help to address this issue.
Self-reporting MVPA may introduce additional measurement error. Previous work has suggested differential MVPA reporting dependent on weight status suggesting that individuals with higher weight may over report MVPA.28 At baseline and 12 months, approximately half of the individuals in this sample were randomized to wear accelerometers for 7 days to assess the validity of this questionnaire. At 12 months, accelerometery measured MVPA was sufficiently correlated with the PAQ measure (r = .56; P < .01). However, future work using objective MVPA measurements will be important to improve this work. Furthermore, the PAQ only assessed MVPA, thus no conclusions can be made regarding the role of light intensity physical activity.
This analysis method required complete data for effect estimation and missing data from participants narrowed the available sample size considerably. To address this issue, imputation models were considered to increase precision. However, these resulted in similar, nonsignificant findings. Thus, only the complete data models were reported in this analysis. In addition, the total effect of life events and MVPA on weight maintenance, though statistically significant, was small. However, the large number of covariate parameters limited the statistical power to detect any small decomposed effects. In the case of the moderated effects, in particular, the decomposed effects were less than 1 kg each, limiting the detectable signal that may have been overpowered by the number of covariates and the limited sample. Future analyses utilizing a larger, more diverse sample may improve statistical power and should be explored.
Finally, this sample had a large proportion of white, married, and college-educated adults who may have social and economic resources that offer protection from the detrimental effects of life events on weight during weight loss maintenance. When examining those who experienced no life events, the number of nonwhite and nonmarried individuals was even smaller. However, being married, having children, and increased job responsibilities entail their own sets of stressors. Therefore, assessing these relationships in a variety of situations is warranted. Specifically, future work should consider examining this relationship of life events, MVPA, and weight loss maintenance in ethnically and socioeconomically diverse populations who may be likely to experience major life events29 more often.
Conclusions
This paper explored the effects of behavior changes on health outcomes within the context of one’s personal environment. Findings suggest that the effects of life events and MVPA on weight loss maintenance should be considered separately when considering weight loss maintenance and designing interventions to prevent weight regain. More work is needed in larger, more diverse samples, with interventions that specifically target weight loss maintenance to fully understand these effects. Further work that explores relationships between life events, MVPA as well as mood and other behaviors affecting weight maintenance remains critical.
Acknowledgments
Funding for the Tracking Study was provided by National Institutes of Health Grant R01DK093586-05 (J.A.L). K.L.G. is supported by National Cancer Institute training grant (T32 CA139139, PIs: Penedo and Spring).
Contributor Information
Kara L. Gavin, Behavioral and Psychosocial Research Training Program in Cancer Prevention and Control, Department of Preventive Medicine, Feinberg School of Medicine, Northwestern University, Chicago, IL.
Julian Wolfson, University of Minnesota, Minneapolis, MN..
Mark Pereira, University of Minnesota, Minneapolis, MN..
Nancy Sherwood, University of Minnesota, Minneapolis, MN..
Jennifer A. Linde, University of Minnesota, Minneapolis, MN.
References
- 1.MacLean PS, Wing RR, Davidson T, et al. NIH working group report: innovative research to improve maintenance of weight loss. Obesity. 2015;23(1):7–15. doi: 10.1002/oby.20967 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Klem ML, Wing RR, McGuire MT, Seagle HM, Hill JO. A descriptive study of individuals successful at long-term maintenance of substantial weight loss. Am J Clin Nutr. 1997;66(2):239–246. PubMed ID: 9250100 doi: 10.1093/ajcn/66.2.239 [DOI] [PubMed] [Google Scholar]
- 3.Catenacci VA, Grunwald GK, Ingebrigtsen JP, et al. Physical activity patterns using accelerometry in the National Weight Control Registry. Obesity. 2011;19(6):1163–1170. doi: 10.1038/oby.2010.264 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kim J-H, McKenzie LA. The impacts of physical exercise on stress coping and well-being in University Students in the context of leisure. Health. 2014;6:2570–2580. doi: 10.4236/health.2014.619296 [DOI] [Google Scholar]
- 5.Norris R, Carroll D, Cochrane R. The effects of physical activity and exercise training on psychological stress and well-being in an adolescent population. J Psychosom Res. 1992;36(1):55–65. PubMed ID: 1538350 doi: 10.1016/0022-3999(92)90114-H [DOI] [PubMed] [Google Scholar]
- 6.Harris AH, Cronkite R, Moos R. Physical activity, exercise coping, and depression in a 10-year cohort study of depressed patients. J Affect Disord. 2006;93(1–3):79–85. PubMed ID: 16545873 doi: 10.1016/j.jad.2006.02.013 [DOI] [PubMed] [Google Scholar]
- 7.Epiphaniou E, Ogden J. Evaluating the role of life events and sustaining conditions in weight loss maintenance. J Obes. 2010;2010: 859413 PubMed ID: 20798851 doi: 10.1155/2010/859413 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Oman RF, King AC. The effect of life events and exercise program format on the adoption and maintenance of exercise behavior. Health Psychol. 2000;19(6):605–612. PubMed ID: 11129364 doi: 10.1037/0278-6133.19.6.605 [DOI] [PubMed] [Google Scholar]
- 9.Kwasnicka D, Dombrowski SU, White M, Sniehotta F. Theoretical explanations for maintenance of behaviour change: a systematic review of behaviour theories. Health Psychol Rev. 2016;10(3):277–296. PubMed ID: 26854092 doi: 10.1080/17437199.2016.1151372 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Linde JA, Jeffery RW, Crow SJ, et al. The tracking study: description of a randomized controlled trial of variations on weight tracking frequency in a behavioral weight loss program. Contemp Clin Trials. 2015;40:199–211. PubMed ID: 25533727 doi: 10.1016/j.cct.2014.12.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Unick JL, Beavers D, Jakicic JM, et al. Effectiveness of lifestyle interventions for individuals with severe obesity and type 2 diabetes: results from the Look AHEAD trial. Diabetes Care. 2011;34(10): 2152–2157. PubMed ID: 21836103 doi: 10.2337/dc11-0874 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sarason IG, Johnson JH, Siegel JM. Assessing the impact of life changes: development of the Life Experiences Survey. J Consult Clin Psychol. 1978;46(5):932–946. PubMed ID: 701572 doi: 10.1037/0022-006X.46.5.932 [DOI] [PubMed] [Google Scholar]
- 13.Ogden J, Stavrinaki M, Stubbs J. Understanding the role of life events in weight loss and weight gain. Psychol Health Med. 2009;14(2):239–249. PubMed ID: 19235083 doi: 10.1080/13548500802512302 [DOI] [PubMed] [Google Scholar]
- 14.Paffenbarger RS Jr, Wing AL, Hyde RT. Physical activity as an index of heart attack risk in college alumni. Am J Epidemiol. 1978;108(3):161–175. PubMed ID: 707484 doi: 10.1093/oxfordjournals.aje.a112608 [DOI] [PubMed] [Google Scholar]
- 15.Ainsworth BE, Leon AS, Richardson MT, Jacobs DR, Paffenbarger RS Jr. Accuracy of the College Alumnus Physical Activity Questionnaire. J Clin Epidemiol. 1993;46(12):1403–1411. PubMed ID: 8263567 doi: 10.1016/0895-4356(93)90140-V [DOI] [PubMed] [Google Scholar]
- 16.Donnelly JE, Blair SN, Jakicic JM, et al. American College of Sports Medicine Position Stand. Appropriate physical activity intervention strategies for weight loss and prevention of weight regain for adults. Med Sci Sports Exerc. 2009;41(2):459–471. PubMed ID: 19127177 doi: 10.1249/MSS.0b013e3181949333 [DOI] [PubMed] [Google Scholar]
- 17.Jeffery RW, Wing RR, Sherwood NE, Tate DF. Physical activity and weight loss: does prescribing higher physical activity goals improve outcome? Am J Clinl Nutr. 2003;78(4):684–689. doi: 10.1093/ajcn/78.4.684 [DOI] [PubMed] [Google Scholar]
- 18.Catenacci VA, Ogden LG, Stuht J, et al. Physical activity patterns in the National Weight Control Registry. Obesity. 2008;16(1):153–161. doi: 10.1038/oby.2007.6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.National Cancer Insitute [Internet] Applied Research CCaPS, National Cancer Institute. Diet History Questionnaire II and Canadian Diet History Questionnaire II (C-DHQII). National Institutes of Health: Washington: c2010. http://appliedresearch.cancer.gov/dhq2/. Accessed December 23, 2012. [Google Scholar]
- 20.Millen AE, Midthune D, Thompson FE, Kipnis V, Subar AF. The National Cancer Institute diet history questionnaire: validation of pyramid food servings. Am J Epidemiol. 2006;163(3):279–288. PubMed ID: 16339051 doi: 10.1093/aje/kwj031 [DOI] [PubMed] [Google Scholar]
- 21.VanderWeele TJ. A unification of mediation and interaction: a 4-way decomposition. Epidemiology. 2014;25(5):749–761. PubMed ID: 25000145 doi: 10.1097/EDE.0000000000000121 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Turk MW, Sereika SM, Yang K, Ewing LJ, Hravnak M, Burke LE. Psychosocial correlates of weight maintenance among black & white adults. Am J Health Behav. 2012;36(3):395–407. PubMed ID: 22370440 doi: 10.5993/AJHB.36.3.10 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Allender S, Hutchinson L, Foster C. Life-change events and participation in physical activity: a systematic review. Health Promot Int. 2008; 23(2):160–172. PubMed ID: 18364364 doi: 10.1093/heapro/dan012 [DOI] [PubMed] [Google Scholar]
- 24.Au N, Hauck K, Hollingsworth B. Employment, work hours and weight gain among middle-aged women. Int J Obes. 2013;37(5): 718–724. doi: 10.1038/ijo.2012.92 [DOI] [PubMed] [Google Scholar]
- 25.Wing RR, Hill JO. Successful weight loss maintenance. Annu Rev Nutr. 2001;21(1):323–341. doi: 10.1146/annurev.nutr.21.1.323 [DOI] [PubMed] [Google Scholar]
- 26.VanderWeele TJ, Vansteelandt S. Conceptual issues concerning mediation, interventions and composition. Stat Interface. 2009;2: 457–468. doi: 10.4310/SII.2009.v2.n4.a7 [DOI] [Google Scholar]
- 27.Pearl J. The causal mediation formula—a guide to the assessment of pathways and mechanisms. Prev Sci. 2012;13(4):426–436. PubMed ID: 22419385 doi: 10.1007/s11121-011-0270-1 [DOI] [PubMed] [Google Scholar]
- 28.Warner ET, Wolin KY, Duncan DT, Heil DP, Askew S, Bennett GG. Differential accuracy of physical activity self-report by weight status. Am J Health Behav. 2012;36(2):168–178. PubMed ID: 22370255 doi: 10.5993/AJHB.36.2.3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lantz PM, House JS, Mero RP, Williams DR. Stress, life events, and socioeconomic disparities in health: results from the Americans’ changing lives study. J Health Soc Behav. 2005;46(3):274–288. PubMed ID: 16259149 doi: 10.1177/002214650504600305 [DOI] [PubMed] [Google Scholar]