Abstract
Objectives:
Males often lose more weight than females during treatment, and early weight loss predicts weight loss longer-term. Yet, mechanisms for sex differences in early weight loss are unknown and were examined in this study.
Methods:
Adults≥21 years old with overweight or obesity and prediabetes (N=206) participated in a lifestyle intervention and completed baseline psychosocial questionnaires. Percent weight loss, session attendance, and number of days participants self-monitored dietary intake and weight were determined at week 5. Principal components, regression, and mediation analyses were conducted to determine whether weight loss differed by sex and potential mediators of weight change.
Results:
Mean (±SD) weight loss was greater for males (2.59±1.62%) than females (2.05±1.54%; p=.02). Attendance, self-monitoring, and beliefs regarding disease risk were independent predictors of weight loss (all p<.05) but did not explain sex differences. The association between attendance and weight loss was stronger for males than females (p<.05).
Conclusions:
Additional research is needed to identify mechanisms that explain sex differences in early weight loss. However, strengthening risk beliefs, attendance, and self-monitoring may promote greater early weight loss for all participants.
Keywords: obesity, prediabetes, weight loss, lifestyle risk reduction, mediation analysis
Early weight loss success experienced by individuals who are overweight or obese during behavioral weight loss interventions is predictive of longer-term weight loss success. 1 For example, the Look AHEAD trial among people with type 2 diabetes found that participants who failed to achieve a 1- or 2-month weight loss threshold had significantly greater odds of not achieving ≥ 5% or≥10% weight loss at one year.2 Other investigators also found that weight loss during the first month of treatment predicted greater weight loss at treatment end,3–6 and weight loss is the primary factor resulting in reduced incidence for type 2 diabetes.7
A review examining sex differences in weight loss found that, although females lost a significant amount of weight, males lost more weight than females, on average, over the course of treatment. 8 Despite the observed sex differences in weight loss post-treatment, mechanisms to explain these differences were not examined. Psychosocial and behavioral pretreatment predictors of weight control have been investigated (eg, eating and exercise self-efficacy, binge eating), and fewer previous weight loss attempts was the most consistent pretreatment predictor of weight loss in mixed gender samples but with a small effect size (r=0.10; 95% CI: 0.05–0.15) and moderate heterogeneity across 8 studies. 9 A weight loss study conducted among females only found that previous weight loss attempts explained about 6% of 12-month weight change following intervention.10 Despite these findings, limited evidence exists regarding the impact of psychosocial factors on weight change to draw reliable conclusions.9 Thus, there is a need for high-quality studies about the role of pretreatment predictors on weight control behaviors, particularly in understanding sex differences in weight loss.
Engagement with the intervention, defined by behaviors such as attendance and self-monitoring, may differentially impact males and females, and predict early success. In previous research, session attendance was related to greater early weight loss.11 Similarly, more frequent self-monitoring of dietary intake and body weight, common components of behavioral weight loss trials, have been consistently and significantly associated with weight loss.6,12–14 The association between weight loss and engagement with the treatment may differ based on sex; in fact, daily self-weighing mediated the treatment effect of a weight loss intervention for males only.15
Attendance and treatment adherence are behavioral markers of intervention engagement. However, there also are social, cognitive, and emotional factors that reflect engagement with an intervention and motivation for change. The motivational phase of behavioral change encompasses socio-cognitive constructs implicated in intention formation. Once an intention is established, volitional constructs, such as setting specific and moderately challenging goals, lead to greater effort and success.16,17
A few studies have attempted to explain sex differences in weight-loss by examining differential responses to intervention elements. For example, the change in autonomous motivation for diet, self-efficacy regarding dietary intake, and frequency of using weight control strategies that promote self-regulation of eating behaviors previously mediated the intervention effect on weight loss in males.15 In a separate study, higher pretreatment levels of self-efficacy predicted weight change for males but not for females; the change in self-efficacy to resist eating when experiencing negative affect predicted lower body mass index following a residential obesity treatment program in both sexes.18 Greater research is needed to improve understanding of how interventions impact engagement and targeted socio-cognitive constructs to promote weight loss, and how these constructs help explain sex differences in weight loss. Therefore, we examined whether early weight loss (ie, following one month of intervention) differed by sex, and whether markers of engagement with the intervention (ie, session attendance and self-monitoring of dietary intake and body weight) and pretreatment socio-cognitive constructs accounted for any sex differences in weight loss in a sample of adults with prediabetes participating in a diabetes prevention trial.
METHODS
Research Design
An adaptive intervention design was employed at a university worksite where participants attended 16 group sessions occurring weekly during the intensive intervention phase. In person sessions were offered prior to the SARS-CoV-2 pandemic and videoconferencing sessions were offered following pandemic onset. Body weight was obtained prior to each session either in person (prior to the pandemic) or via Bluetooth-enabled scales provided to participants and was used to determine weight change at the beginning of session 5. In our prior research, failure to achieve>2.5% weight loss following the first month of intervention predicted failure to achieve≥5% weight loss at follow-up.3 The current study aimed to adapt treatment for slower weight loss responders at-risk for poorer outcomes. Thus, if participants achieved>2.5% weight loss at the beginning of week 5, they remained in the standard intervention at week 5. If participants achieved≤2.5% weight loss, they were stratified to an augmented intervention designed for slower weight loss responders starting week 5 (not described here). Demographic, weight history, and socio-cognitive characteristics were obtained at baseline prior to intervention initiation. Session attendance was recorded weekly. Participants were encouraged to self-monitor their calorie and dietary fat intake daily following the second intervention session using either paper or electronic diaries (ie, MyFitnessPal) based on individual preference. Self-monitoring records were submitted weekly by participants and reviewed by the health coach who recorded the number of days each person reported their calorie and fat intake (21 days possible). Participants were encouraged to weigh themselves at home at least once a week, preferably on the day of their group session, and report their weight on the self-monitoring record. Study recruitment and intervention sessions have been completed; participants were followed for up to 18 months.
Study Eligibility Criteria and Recruitment
To be eligible, participants were≥21 years old, English-speaking employees of the university with prediabetes. and had a body mass index (BMI)≥25 kg/m2 for non-Asians or≥23 kg/m2 for Asians. Height and weight were assessed at baseline and a fasting capillary blood glucose and A1c were obtained. Individuals with a fasting glucose 100–125 mg/dL or an A1c 5.7–6.4% were potentially eligible19 and screened for any exclusionary criteria. Individuals diagnosed with diabetes, chronically using corticosteroids, participating in a structured weight loss program, preparing for bariatric surgery, planning to leave university employment, or intending to move from the community were ineligible. Potentially eligible individuals completed the Physical Activity Readiness Questionnaire and those who answered positively to more than one question were excluded.20 The Patient Health Questionnaire-8 also was administered to assess the presence of depressive symptoms in the previous 2 weeks; those who scored>9 were ineligible.21 Individuals who scored>26 on the Binge Eating Scale, indicating the potential for severe binge eating, also were excluded.22
Participants were recruited through electronic advertisements on university websites, campus flyers, postcards distributed to employee mailboxes, and employee electronic mail advertisements. A telephone number and e-mail address were provided on recruitment material for interested individuals to contact to receive more information
Month 1 of the Group Lifestyle Balance (GLB) Intervention
All participants received the first month of the GLB intervention, which is a group-based adaptation of the intervention used in the Diabetes Prevention Program.23 Lifestyle goals include losing 7% of body weight by study end, progressively increasing physical activity to 150 minutes/week, and consuming<25% of energy from fat to reduce energy intake. Sessions 1–4 focus on risk factors for type 2 diabetes, benefits of a healthy lifestyle and weight loss, reading nutrition labels, and strategies for reducing energy and fat intake and increasing physical activity. Participants received their individual weight loss goal during session 1 and their fat and calorie goals during session 2. Self-monitoring of food intake began following session 1; recording of number of calories and fat grams consumed began following session 2. We encouraged participants to record minutes of physical activity of≥10 continuous minutes after session 4.
Socio-cognitive Indicators of Behavioral Change
The intervention addressed socio-cognitive factors to promote behavioral change, including goal-setting and efficacy expectations regarding weight loss. Self-regulatory models of behavioral change, such as the Health Action Process Approach (HAPA), differentiate between a motivational phase of behavioral change leading to an intention to change and a volitional phase of behavioral change leading to initiation of the behavior.24 An intention to act is influenced by risk perception, positive outcome expectancies, and action self-efficacy (the perceived capability to perform a desired action). Once an intention is set, a goal is established and planning is needed to bridge the gap between intention and action (ie, action planning), and the identification of strategies to overcome potential obstacles encountered along the way (ie, coping planning) is necessary. During the behavior change process, different tasks are enacted, and different self-efficacy beliefs are required to master those tasks. Action self-efficacy refers to an optimistic belief that success is possible, coping self-efficacy represents the belief regarding one’s capability to overcome obstacles following behavior initiation, and recovery self-efficacy represents the conviction to get back on track after being derailed. In this study, action control also was assessed and described participants’ belief in their ability to control food choices in high-risk situations (eg, when anxious or dining out).
Goals refer to future valued outcomes (eg, weight loss), and high, or challenging, goals are more motivating than easy goals. Challenging goals require the attainment of more to promote satisfaction and thus promote greater effort. Commitment to the goal, which is enhanced by self-efficacy, and perceived difficulty of the goal are moderators of goal setting.17 Finally, self-efficacy is task specific and may vary across goals. Thus, efficacious beliefs regarding weight loss, the primary outcome of the current study, are central to goal attainment in this study.
Measures of Socio-cognitive Outcomes
HAPA outcomes were adapted to assess diet-related behaviors based on previous research.24–26 Response options for outcome expectancies, action control, action and coping planning, and action, coping, and recovery self-efficacy included a 7-point scale. Intention was assessed using a scale from 0% to 100% confident in response to the statement, “How likely is it that you will eat a low-fat diet on most days of the week within the next month?”27
Risk perception was assessed using a questionnaire specific to the risk for type 2 diabetes. Four subscales were included: comparative disease risk (15 items), comparative environmental risk (9 items), optimistic bias (the level of optimism, realism, or pessimism about developing diabetes; 2 items), and personal control regarding the development of diabetes (4 items). Two additional dimensions of disease risk, diabetes risk knowledge (the assessment of 11 risk factors for type 2 diabetes) and worry (2 items) about developing diabetes, also were assessed.28
Because weight loss was the primary study outcome, self-efficacy expectations regarding weight control behaviors were assessed using the Weight Efficacy Lifestyle Questionnaire (WEL).29 The 20-item WEL asks respondents to rate their confidence regarding the ability to resist the desire to eat in various situations. Five subscales, with 4 items each, include: experiencing negative emotions, availability of food, social pressure to eat, experiencing physical discomfort (eg, headache), and engaging in positive activities (eg, watching TV).
Goal commitment to losing weight was assessed using a 7-item questionnaire. Two items were reverse scored so that a higher score indicates greater goal commitment.30 Goal difficulty was assessed with one item asking participants to indicate how easy or difficult it is to lose weight.31
Data Analyses
Measures of central tendency, distribution, and normality were examined for all variables. The Fisher exact test, Pearson chi-square test, or one-way analysis of variance were used to compare between-sex differences in participant demographic and weight-related characteristics at baseline. Mean (±SD) socio-cognitive scale scores at baseline and percent weight loss following session 4 were obtained and compared between sex using the Welch t-test, the Fisher exact test, or the Wilcoxon rank sum test.
We determined session attendance and the number of days participants self-monitored their calorie intake, fat intake, and body weight. Due to skewed distributions, number of sessions attended, and the number of days participants self-monitored and self-weighed, we dichotomized into fully adherent or less than fully adherent to improve model fit and to meet the assumption of multiple linear regression models. We also tested whether the number of pretreatment weight loss attempts predicted intervention session attendance to determine whether the novelty of attending a lifestyle intervention increased attendance. We used stepwise regression models to determine whether there were sex differences in early weight loss. We examined models for outliers, and we produced the final estimates after removing one male outlier.
We then conducted exploratory mediation analyses to improve understanding of the sex differences in weight loss (Figure 1). To reduce the number of tests in these analyses, we created grouped variables with principal components analysis of the socio-cognitive variables that significantly differed by sex at baseline. Three theoretically consistent components with eigenvalues>1.0 were retained. Component 1 (“dietary self-efficacy and difficulty”) reflected participants’ confidence to manage dietary intake in high-risk situations and their belief regarding the difficulty of losing weight. Component 2 (“disease risk”) reflected participants’ beliefs regarding their risk for diabetes and other chronic health conditions (eg, cardiovascular disease, cancer) and environmental risks (eg, air pollution, pesticides). Component 3 (“diabetes risk knowledge and worry”) reflected participants knowledge of risk factors specific for developing type 2 diabetes and their level of worry regarding developing diabetes. The direct and indirect effects of sex on percent weight loss through each combination of the 3 socio-cognitive components and 3 intervention engagement variables were then tested using serial mediation analyses, estimating 95% confidence intervals using bootstrapping procedures. Statistical analyses were completed using JMP version 15 (Carey, NC, 2019) and the lavaan package version 0.632 in R version 4.1.233 for mediation analysis.
Figure 1:
Final Model regarding the Effect of Sex on the Socio-cognitive Variable of Disease Risk, Intervention Engagement of Self-monitoring Dietary Intake, and Percent Weight Loss
*p-value<.05;**p-value<.01
RESULTS
Overall, 232 individuals initiated the intervention and 25 (10.8%) withdrew during the first month. There were no statistically significant differences in demographic characteristics between those who remained in the intervention compared to those who withdrew (all ps>.05). There were no statistically significant differences in percent weight loss following changes in intervention delivery method due to the pandemic (p>.05). Weight loss did not differ between those who attended in-person sessions and those who attended videoconference sessions. There were, however, between-sex differences in educational attainment, occupation, and previous weight loss attempts at baseline for those who remained in the intervention (Table 1).
Table 1.
Baseline Demographic Characteristics and Pretreatment Weight Loss History of Participants by Sex
| Characteristic a | Males (N=52) | Females (N=154) | p-value b |
|---|---|---|---|
|
| |||
| Mean±SD | |||
|
| |||
| Age (years) | 51.10±12.46 | 51.60±9.60 | .7568 |
| Body mass index (kg/m2) | 36.02±7.26 | 36.86±8.07 | .6272 |
| n (%) | |||
| Race | |||
| Non-Hispanic white | 40 (76.92) | 124 (80.52) | |
| Non-Hispanic black | 4 (7.69) | 20 (12.99) | .1391 |
| Asian | 7 (13.46) | 7 (4.55) | |
| Not reported | 1 (1.92) | 1 (0.65) | |
| >1 race | 0 (0.00) | 2 (1.30) | |
| Ethnicity | 1.00 | ||
| Hispanic | 1 (1.92) | 4 (2.60) | |
| Education | |||
| High school or GED education or vocational training | 1 (1.92) | 7 (4.58) | .0031 |
| Some college or bachelor’s degree | 16 (30.77) | 84 (54.90) | |
| Postgraduate training or degree | 35 (67.31) | 62 (40.52) | |
| Marital status | |||
| Married | 39 (75.00) | 97 (63.82) | .1733 |
| Not married | 13 (25.00) | 55 (36.18) | |
| Occupation | |||
| Professional | 38 (77.55) | 78 (52.70) | |
| Clinical | 2 (4.08) | 22 (14.86) | < .0001 |
| Clerical | 1 (2.04) | 38 (26.68) | |
| Technical | 7 (14.29) | 9 (6.08) | |
| Other | 1 (2.04) | 1 (0.68) | |
| Annual household income | |||
| <$30,000 | 1 (2.04) | 3 (2.04) | |
| $30,000–49,999 | 3 (6.12) | 15 (10.20) | |
| $50,000–69,999 | 11 (22.45) | 22 (14.97) | .8093 |
| $70,000–99,999 | 8 (16.33) | 31 (21.09) | |
| $100,000–149,999 | 14 (28.57) | 42 (28.57) | |
| ≥$150,000 | 12 (24.49) | 34 (23.13) | |
| Tried to lose weight previously | 42 (80.77) | 143 (92.86) | .0180 |
| Frequency of self-weighing at home | |||
| >1 time/day | 0 (0.00) | 3 (1.96) | |
| 1 time/day | 6 (11.54) | 17 (11.11) | .8336 |
| 1 time/week | 23 (44.23) | 56 (36.60) | |
| 1 time/month | 15 (28.85) | 53 (34.64) | |
| 1 time/year or less | 8 (15.38) | 24 (15.69) | |
Note.
Values may not sum for the entire sample due to missing data.
Between-sex comparison using Wilcoxon rank sum test for continuous responses and Fisher’s exact test for categorical responses.
Mean (±SD) percent weight loss was significantly greater for males (2.59±1.62%) than females (2.02±1.54%) after session 4 (p=.0225; Table 2). The engagement variables of session attendance, self-monitoring calorie or fat intake, and self-weighing also independently predicted weight loss. Participants who attended all 4 sessions (N=135) lost a mean (±SD) of 2.41±1.53% weight compared to participants (N=70) who attended<4 sessions (1.68±1.56%; p=.0019) but session attendance did not differ by sex (p=1.00). The reported number of pretreatment weight loss attempts did not predict session attendance or weight loss. Participants who self-monitored≥20 days (N=100) lost a mean (±SD) of 2.60 (±1.46)% weight compared to participants (N=104) who self-monitored<20 days (1.70±1.58%; p<.0001). Participants who weighed≥4 days during the first month (N=60) lost a mean (±SD) of 2.57±1.59% weight compared to participants (N=144) who weighed fewer days (1.99±1.55%; p=.0174). Baseline BMI, self-monitoring of dietary intake, attendance, and the interaction between sex and attendance were statistically significant predictors of percent weight loss (R2=18.74%) in regression models. Critically, sex moderated the association between attendance and weight loss such that attendance at all sessions was more strongly associated with weight loss for males than females.
Table 2.
Differences in Weight Loss, Intervention Engagement, and Baseline Scores for Socio-cognitive Variables by Sex
| Component | Males (N=52) | Females (N=154) | p-value a |
|---|---|---|---|
|
| |||
| Outcome or Dependent Variable | |||
| Weight loss following intervention session 4 (mean±SD%) | 2.59±1.62 | 2.02±1.54 | .0225 |
|
| |||
| Intervention Engagement Variables | |||
| Attended 4 group sessions | 65.38% | 65.58% | 1.0000 |
| Self-weighed (≥4 days across weeks 1–4) | 36.54% | 26.80% | .2172 |
| Mean±SD | |||
| Self-monitored calorie intake (days monitored across weeks 2–4) | 15.31±6.48 | 16.27±6.02 | .3126 |
| Self-monitored fat intake (days monitored across weeks 2–4) | 15.80±5.76 | 16.22±6.04 | .4607 |
| Self-monitored calorie or fat intake (days monitored across weeks 2–4) | 15.81±5.80 | 16.40±5.99 | .3531 |
| Self-weighed (days weighed across weeks 1–4) | 4.08±4.42 | 3.35±4.03 | .3205 |
|
| |||
| Socio-Cognitive Variables | |||
|
| |||
| Health Action Process Approach Variables: b | |||
|
| |||
| Intention to eat a lower fat diet c | 54.33±26.02 | 66.76±17.94 | .0014 |
| Outcome expectancies | 5.56±1.08 | 5.84±0.80 | .1733 |
| Action self-efficacy | 4.79±1.09 | 4.80±1.37 | .9655 |
| Action planning | 4.50±1.41 | 4.69±1.39 | .3469 |
| Action control | 4.60±0.92 | 4.20±0.96 | .0096 |
| Coping planning | 3.40±1.39 | 3.36±1.43 | .9484 |
| Coping self-efficacy | 4.87±1.01 | 4.90±1.06 | .5666 |
| Recovery self-efficacy | 4.95±1.08 | 4.97±1.18 | .6743 |
|
| |||
| Risk Perception Variables: | |||
|
| |||
| Diabetes risk knowledge d | 6.35±1.73 | 7.36±1.66 | .0003 |
| Personal control e | 3.34±0.47 | 3.35±0.46 | .6564 |
| Personal disease risk f | 2.24±0.49 | 2.50±0.53 | .0086 |
| Worry f | 2.55±0.58 | 2.89±0.62 | .0002 |
| Optimistic bias f | 2.20±0.58 | 2.04±0.52 | .0308 |
| Comparative environmental risk g | 1.57±0.42 | 1.75±0.44 | .0075 |
|
| |||
| Weight Lifestyle Self-efficacy Variables: h | |||
|
| |||
| Negative emotions | 23.81±7.95 | 19.63±8.24 | .0020 |
| Food availability | 20.71±7.04 | 18.07±6.80 | .0361 |
| Social pressure | 22.60±8.59 | 21.80±7.67 | .4191 |
| Physical discomfort | 26.44±6.22 | 24.22±7.16 | .0949 |
| Positive activities | 26.06±6.16 | 25.39±6.26 | .5937 |
|
| |||
| Goal Setting Variables regarding Weight Loss: | |||
|
| |||
| Perceived goal difficulty i | 5.15±1.50 | 5.78±1.49 | .0057 |
| Goal commitment j | 28.88±3.02 | 28.53±3.31 | .5090 |
Note.
Between-sex comparison using t-test, Fisher exact test, or Wilcoxon rank sum test
Response options ranged from 1=not at all true to 7=exactly true
Response options ranged from 0% to 100% regarding the likelihood that you will eat a low-fat diet on most days of the week within the next month
Responses included the sum of 11 items with response options ranging from 1=increases the risk to 3=decreases the risk
Response options ranged from 1=strongly agree to 4=strongly disagree
Response options ranged from 1=strongly disagree to 4=strongly agree
Response options ranged from 1=almost no risk to 4=high risk
Response options ranged from 0=not confident at all that I can resist the desire to eat to 9=very confident that I can resist the desire to eat
One item ranging from 0=very easy to 8=very difficult
Response options ranged from 1=strongly agree to 5=strongly disagree with higher total scores indicating greater commitment
To explore potential sex differences in mechanisms of weight loss, we conducted mediation analyses (Table 3). In terms of the first path in the mediational models, males had significantly higher Component 1 scores and significantly lower Component 2 scores than females (both ps<.01). In the second path of the models, Components 2 and 3 were significantly related to self-monitoring dietary intake (both ps<.05). However, none of the principal components significantly predicted percent weight loss (all ps>.05) in separate regression models.
Table 3.
First Principal Components Structure for the Socio-cognitive Variables and Mean (±SD) Values for Each Component by Sex
| Factor | Factor Loading | Eigenvalue (Percent Variance Explained) | Mean (±SD) M: Male; F: Female [p-value] a |
|---|---|---|---|
|
| |||
| Principal Components 1: Dietary Self-efficacy & Goal Difficulty regarding Weight Loss | |||
|
| |||
| Action control | 0.5093 | 2.2895 (57.24%) |
M: 0.65 (±1.55) F: −0.22 (±1.45) [.0008] |
| Negative emotions | 0.5498 | ||
| Food availability | 0.5650 | ||
| Perceived goal difficulty | −0.3451 | ||
| Principal Components 2: Disease Risk | |||
| Optimistic bias | −0.5286 | 1.6373 (54.58%) |
M: −0.52 (±1.26) F: 0.17 (±1.24) [.0010] |
| Comparative environmental risk | 0.5541 | ||
| Personal disease risk | 0.6430 | ||
| Principal Components 3: Diabetes Risk Knowledge & Worry | |||
| Diabetes risk knowledge | 0.7071 | 1.0579 (52.90%) |
M: −0.02 (±1.03) F: 0.006 (±1.03) [.8923] |
| Diabetes risk worry | −0.7071 | ||
Note.
Comparison determined using Welch t-test
As noted previously, the direct effect of sex on percent weight loss was statistically significant, (−0.687; p=.006). Of the 9 combinations of socio-cognitive components and intervention engagement variables, only a serial mediation analysis of sex on percent weight loss through Component 2 (ie, perceived disease risk) and dietary self-monitoring included significant effects for each path in the model (ie, sex predicting perceived risk, perceived risk predicting dietary self-monitoring, dietary self-monitoring predicting one-month weight loss; Table 4). However, the overall indirect effect was not significant, indicating that the association between sex and early weight loss was not explained by these variables (Figure 1).
Table 4.
Mediation Analysis of the Direct and Indirect Effects of Sex, Intervention Engagement, and Principal Components 2 (Disease Risk) on Percent Weight Loss Following Intervention Session 4
| Effect | Estimate | SE | Z | p-value | 95% Lower CI | 95% Upper CI |
|---|---|---|---|---|---|---|
|
| ||||||
| Principal component 2 (disease risk) and sex | 0.699 | 0.202 | 3.457 | .001 | 0.300 | 1.098 |
| Self-monitoring dietary intake and sex | 0.014 | 0.084 | 0.166 | .868 | −0.151 | 0.177 |
| Self-monitoring dietary intake and principal component 2 | 0.055 | 0.027 | 2.011 | .044 | 0.000 | 0.108 |
| Percent weight loss and principal component 2 | 0.088 | 0.086 | 1.019 | .308 | −0.080 | 0.255 |
| Percent weight loss and self-monitoring dietary intake | 0.900 | 0.212 | 4.253 | < .001 | 0.478 | 1.309 |
| Percent weight loss and sex | −0.687 | 0.248 | −2.766 | .006 | −1.166 | −0.198 |
| Indirect path | 0.035 | 0.024 | 1.411 | .158 | 0.000 | 0.095 |
| Direct path | −0.687 | 0.248 | −2.766 | .006 | −1.166 | −0.198 |
| Total effect | −0.652 | 0.253 | −2.575 | .010 | −1.141 | −0.149 |
| Mediation through principal component 2 | 0.061 | 0.065 | 0.947 | .343 | −0.056 | 0.202 |
| Mediation through self-monitoring dietary intake | 0.013 | 0.078 | 0.161 | .872 | −0.146 | 0.167 |
DISCUSSION
This is one of the first studies to examine the impact of sex on early weight loss success. Sex was a primary determinant of early weight loss in this study, and no prior studies reported the effect of sex on early weight loss success.1 Critically, this work explored reasons for these sex differences, and found evidence that males and females differentially respond to intervention features. Specifically, session attendance was particularly important for males to achieve early weight loss. Males who attended all 4 sessions had a mean (±SD) of 3.12 (±0.25%) weight loss compared to those who attended fewer sessions. Frequency of contact (ie,>2.7 contacts per month) was related to achieving≥5% weight loss at final assessment in one review of male-only weight loss interventions.34 Thus, regular contact may be important for helping males initiate the behavior change process. In analyses not presented here, we tested the association between early weight loss and previous number of weight loss attempts as a proxy for the novelty of the intervention. However, number of previous attempts did not predict early weight loss for either sex. More research is needed, such as qualitative interviews among males, to determine incentives and barriers to attendance and strategies for promoting session attendance, especially for males with poor early attendance.
Our study is also one of the first to examine whether engagement or socio-cognitive variables designed to promote behavioral change explained early success and to explore whether these mechanisms differed by sex. No sex differences in the mechanisms that accounted for weight loss as a function of socio-cognitive variables were identified. Males had higher pretreatment scores on self-efficacy beliefs for controlling food intake when experiencing negative emotions and when food was available (ie, Component 1) than females. However, these efficacy beliefs predicted neither greater intervention engagement nor weight loss. Similarly, females had greater pretreatment pessimistic beliefs about and personal control for developing diabetes (ie, Component 2) than males, but these risk-related variables did not predict early weight loss directly. The presence of prediabetes was a criterion for study enrollment; thus, participants were aware of their greater risk status compared to the general US population. Only perceived disease risk (ie, Component 2) predicted self-monitoring of dietary intake. Because self-monitoring intake predicted early weight loss success, increasing the saliency of risk beliefs, strengthening participants’ understanding of the positive impact of self-monitoring on weight loss,12 and enhancing understanding that weight loss is the dominant predictor of risk reduction7 may be critical. Moreover, given the significant relation between variables in each path in the mediation model, intervention research needs to strengthen all paths in the model to overcome the direct effect of sex on weight loss. Strengthening each model path may allow females (but not males) to achieve greater early weight loss. However, more research is needed in a larger sample (especially with a greater number of males) with greater statistical power, as we did not observe a statistically significant indirect effect in this study.
In the absence of evidence for socio-cognitive constructs predicting sex differences in early weight-loss, these differences may have a biological basis. Numerous hormones, peptides, and nutrients help regulate body weight. Compensatory changes in regulators of appetite that encourage weight regain after weight loss do not revert to baseline values within 12 months following weight loss.35 In addition, lower energy expenditure persisted in those who maintained a reduced body weight for greater than one year.36 Metabolic adaptation, defined as a lower measured versus predicted resting metabolic rate, was observed in females with overweight who lost weight such that for each 10 kcal/day increase in metabolic adaptation, time to reach the weight loss goal increased by one day.37 For those with the largest magnitude of metabolic adaptation, findings showed they would need to adhere to a diet for 70 additional days compared to those with no adaptation. A greater proportion of females in this study reported pretreatment weight loss attempts compared to males. Thus, biological mechanisms following repeated weight loss attempts may contribute to metabolic adaptation and appetite regulation. However, baseline metabolic rate and hormone levels were not measured in this study, and additional research is needed to determine the impact of these factors on early weight loss by sex.
It is worth noting that in addition to attendance, self-monitoring dietary intake was another indicator of intervention engagement that predicted early weight loss. One review found that frequent self-monitoring was significantly associated with weight loss; however, in all but 2 of the 22 studies reviewed, the samples were predominantly white and female.12 The relationship between self-monitoring dietary intake in all male samples has not been reported. Current findings replicate the importance of self-monitoring dietary intake in a mixed gender sample. Information obtained through self-monitoring is necessary to determine how frequently individual dietary goals are achieved. Thus, encouraging all participants in weight management programs to monitor as often as possible may ensure greater success. Participants in the current study self-monitored using either a study-provided paper self-monitoring log or a digital self-monitoring app based on personal preference. Prior research found no statistically significant difference in weight loss for either paper-based or digital approaches to self-monitoring intake.38 Thus, participants can choose the form of self-monitoring that is most convenient for them as long as they monitor.
Self-weighing, another component of engagement, did not mediate early weight loss for either sex. Prior research found that daily self-weighing was an effective self-monitoring practice and associated with clinically significant weight loss at 6 months.39 The feedback provided from self-weighing is an important component of self-regulation. Participants stratified to the adaptive intervention at week 5 in the current study were encouraged to self-weigh daily following session 5. Thus, additional research will determine whether consistent self-weighing mediates weight loss following the intensive intervention phase at 4 months.
Finally, many institutions and payers reimburse patients for costs associated with enrolling in weight loss programs but make later reimbursement of sessions contingent on early weight loss success.40 For example, the Centers for Medicare and Medicaid Services cover the costs of treatment for weight loss following the first 6 months of treatment if enrollees lose≥3 kg. Insurance companies often follow Centers for Medicare and Medicaid Services reimbursement policies; thus, it is important to optimize the efficacy of “rescue” interventions for early slow responders or non-responders to treatment. Additionally, it may be important to review these reimbursement practices given that females in our study had less early weight loss than males, even when they were equally engaged in and committed to the intervention. Whereas our participants were enrolled in the university health insurance program, federal reimbursement of costs of participation based on weight loss outcomes may create discriminatory practices. Our findings regarding intervention session attendance, self-monitoring intake, and disease risk may be keys to early success.
Limitations
Although this research contributes to the understanding of sex differences in early weight loss, a few limitations should be noted. The sample included more females than males and non-Hispanic white individuals, thereby limiting the ability to generalize the findings to the broader population of males with prediabetes and to individuals who do not identify as white, reducing our power to identify moderators and mediators of the sex difference findings. Also, the sample, recruited from a university community, was well-educated. The number of days participants self-weighed was based on self-report obtained from self-monitoring records. Participants may have weighed more days than they recorded in their monitoring records. The availability of wireless scales to monitor self-weighing objectively can be implemented in future studies to verify self-monitoring practices. Similarly, engagement in physical activity was not assessed during the first month of intervention. Differences in activity levels may help explain weight loss differences. Whereas several socio-cognitive variables based on an integrated theoretical model of self-regulation were measured in this study, other variables than those measured may have greater explanatory power regarding early weight loss. In addition, measures may need to occur more frequently (eg, weekly or daily) to capture the change in socio-cognitive (or other) variables as people progress through the behavior change process to explain mechanisms of change.
Future research is needed to confirm whether the predictors of early weight loss identified in this research are found in other, more diverse populations at risk for type 2 diabetes. It may be critical to discuss with individuals the role of session attendance and self-monitoring in promoting weight loss during the recruitment and screening phase for diabetes prevention programs. An explanation of disease risk and how weight loss reduces risk for type 2 diabetes should be included early in lifestyle interventions for diabetes prevention. Prior research found the implementation of a presession in which motivational interviewing and problem-solving regarding barriers to intervention engagement were addressed significantly improved session attendance, program retention, and weight loss.41
Conclusion
In summary, males experienced greater early weight loss than females following one month of intervention for diabetes prevention. Males benefited more from full intervention session attendance than females. Exploratory analyses failed to find any differences in the mechanisms of weight loss between males and females as a function of socio-cognitive variables and intervention engagement. However, there was some evidence that weight loss may be explained by greater perceived disease risk, especially in females, and increased self-monitoring of dietary intake for both sexes. Future research clarifying how specific intervention components differentially impact males and females are needed to ensure lifestyle interventions for diabetes prevention can address potential sex differences in response to treatment. Moreover, additional research needs to improve understanding of how early weight loss differs between sexes and across racially and ethnically diverse samples to be able to tailor interventions that facilitate greater weight loss overall.
Acknowledgements
The National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health supported this research under award number R01DK112930 and by award number UL1TR002733 from the National Center for Advancing Translational Sciences. The content is solely the responsibility of the authors and does not necessarily represent the official view of the National Center for Advancing Translational Sciences or the National Institutes of Health (Clinical Trial Registration Number: NCT03382873, clinicaltrials.gov). The authors are grateful for the time and support provided by study participants.
Footnotes
Conflict of Interest Disclosure Statement
All authors of this article declare they have no conflict of interest.
Human Subjects Approval Statement
Eligibility and written informed consent were obtained from participants; the study protocol was approved by the Institutional Review Board at the Ohio State University. (Protocol Number: 2016H0392).
Contributor Information
Carla K. Miller, Indiana University School of Public Health-Bloomington, Bloomington, IN, United States..
Haikady N. Nagaraja, The Ohio State University, College of Public Health, Division of Biostatistics, Columbus, OH, United States..
Jennifer S. Cheavens, The Ohio State University, Department of Psychology, Columbus, OH, United States..
Kentaro Fujita, The Ohio State University, Department of Psychology, Columbus, OH, United States..
Sophie A. Lazarus, The Ohio State University, Department of Psychiatry & Behavioral Health, Columbus, OH, United States..
Daniel S. Brunette, The Ohio State University, Department of Psychology, Columbus, OH, United States..
References
- 1.Unick JL, Pellegrini CA, Demos KE, et al. Initial Weight Loss Response as an Indicator for Providing Early Rescue Efforts to Improve Long-term Treatment Outcomes. Curr Diab Rep. 2017;17(9):69. 10.1007/s11892-017-0904-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Unick JL, Hogan PE, Neiberg RH, et al. Evaluation of early weight loss thresholds for identifying nonresponders to an intensive lifestyle intervention. Obesity (Silver Spring). 2014;22(7):1608–16. 10.1002/oby.20777 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Miller CK, Nagaraja HN, Weinhold KR. Early weight-loss success identifies nonresponders after a lifestyle intervention in a worksite diabetes prevention trial. J Acad Nutr Diet. 2015;115(9):1464–71. 10.1016/j.jand.2015.04.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Wadden TA, Foster GD, Wang J, et al. Clinical correlates of short- and long-term weight loss. Am J Clin Nutr. 1992;56(1 Suppl):271s–74s. 10.1093/ajcn/56.1.271s [DOI] [PubMed] [Google Scholar]
- 5.Hadžiabdić MO, Mucalo I, Hrabač P, et al. Factors predictive of drop-out and weight loss success in weight management of obese patients. J Hum Nutr Diet. 2015;28(Suppl 2):24–32. 10.1111/jhn.12270 [DOI] [PubMed] [Google Scholar]
- 6.Patel ML, Hopkins CM, Bennett GG. Early weight loss in a standalone mHealth intervention predicting treatment success. Obes Sci Pract. 2019;5(3):231–37. 10.1002/osp4.329 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hamman RF, Wing RR, Edelstein SL, et al. Effect of weight loss with lifestyle intervention on risk of diabetes. Diabetes Care. 2006;29(9):2102–7. 10.2337/dc06-0560 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Williams RL, Wood LG, Collins CE, et al. Effectiveness of weight loss interventions--is there a difference between men and women: a systematic review. Obes Rev. 2015;16(2):171–86. 10.1111/obr.12241 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Carraça EV, Santos I, Mata J, et al. Psychosocial Pretreatment Predictors of Weight Control: A Systematic Review Update. Obes Facts. 2018;11(1):67–82. 10.1159/000485838 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Jorge R, Santos I, Tomás R, et al. Behavioural and psychological pretreatment predictors of short- and long-term weight loss among women with overweight and obesity. Eat Weight Disord. 2020;25(5):1377–85. 10.1007/s40519-019-00775-9 [DOI] [PubMed] [Google Scholar]
- 11.Nackers LM, Ross KM, Perri MG. The association between rate of initial weight loss and long-term success in obesity treatment: does slow and steady win the race? Int J Behav Med. 2010;17(3):161–7. 10.1007/s12529-010-9092-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Burke LE, Wang J, Sevick MA. Self-monitoring in weight loss: a systematic review of the literature. J Am Diet Assoc. 2011;111(1):92–102. 10.1016/j.jada.2010.10.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Milsom VA, Middleton KM, Perri MG. Successful long-term weight loss maintenance in a rural population. Clin Interv Aging. 2011;6:303–9. 10.2147/cia.s25389 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kong A, Beresford SAA, Alfano CM, et al. Self-monitoring and eating-related behaviors are associated with 12-month weight loss in postmenopausal overweight-to-obese women. J Acad Nutr Diet. 2012;112(9):1428–35. 10.1016/j.jand.2012.05.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Crane MM, Ward DS, Lutes LD, et al. Theoretical and Behavioral Mediators of a Weight Loss Intervention for Men. Ann Behav Med. 2016;50(3):460–70. 10.1007/s12160-016-9774-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zhang CQ, Zhang R, Schwarzer R, et al. A meta-analysis of the health action process approach. Health Psychol. 2019;38(7):623–37. 10.1037/hea0000728 [DOI] [PubMed] [Google Scholar]
- 17.Locke EA, Latham GP. New Directions in Goal-Setting Theory. Curr Dir Psychol Sci. 2006;15(5):265–68. 10.1111/j.1467-8721.2006.00449.x [DOI] [Google Scholar]
- 18.Presnell K, Pells J, Stout A, et al. Sex differences in the relation of weight loss self-efficacy, binge eating, and depressive symptoms to weight loss success in a residential obesity treatment program. Eat Behav. 2008;9(2):170–80. 10.1016/j.eatbeh.2007.07.005 [DOI] [PubMed] [Google Scholar]
- 19.ElSayed NA, Aleppo G, Aroda VR, et al. 2. Classification and Diagnosis of Diabetes: Standards of Care in Diabetes-2023. Diabetes Care. 2023;46(Suppl 1):S19–s40. 10.2337/dc23-s002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Thomas S, Reading J, Shephard RJ. Revision of the Physical Activity Readiness Questionnaire (PAR-Q). Can J Sport Sci. 1992;17(4):338–45. https://pubmed.ncbi.nlm.nih.gov/1330274 [PubMed] [Google Scholar]
- 21.Kroenke K, Strine TW, Spitzer RL, et al. The PHQ-8 as a measure of current depression in the general population. J Affect Disord. 2009;114(1–3):163–73. 10.1016/j.jad.2008.06.026 [DOI] [PubMed] [Google Scholar]
- 22.Gormally J, Black S, Daston S, et al. The assessment of binge eating severity among obese persons. Addict Behav. 1982;7(1):47–55. 10.1016/0306-4603(82)90024-7 [DOI] [PubMed] [Google Scholar]
- 23.The Diabetes Prevention Program Research Group. The Diabetes Prevention Program (DPP): Description of lifestyle intervention. Diabetes Care. 2002;25(12):2165–71. 10.2337/diacare.25.12.2165 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Schwarzer R Modeling health behavior change: How to predict and modify the adoption and maintenance of health behaviors. Appl Psychol. 2008;57(1):1–29. 10.1111/j.1464-0597.2007.00325.x [DOI] [Google Scholar]
- 25.Schwarzer R, Renner B. Social-cognitive predictors of health behavior: action self-efficacy and coping self-efficacy. Health Psychol. 2000;19(5):487–95. 10.1037/0278-6133.19.5.487 [DOI] [PubMed] [Google Scholar]
- 26.Scholz U, Schüz B, Ziegelmann JP, et al. Beyond behavioural intentions: planning mediates between intentions and physical activity. Br J Health Psychol. 2008;13(Pt 3):479–94. [DOI] [PubMed] [Google Scholar]
- 27.Plotnikoff RC, Blanchard C, Hotz SB, et al. Validation of the decisional balance scales in the exercise domain from the transtheoretical model: A longitudinal test. Meas Phys Educ Exerc Sci. 2001;5(4):191–206. 10.1207/S15327841MPEE0504_01 [DOI] [Google Scholar]
- 28.Walker EA, Mertz CK, Kalten MR, et al. Risk perception for developing diabetes: comparative risk judgments of physicians. Diabetes Care. 2003;26(9):2543–8. 10.2337/diacare.26.9.2543 [DOI] [PubMed] [Google Scholar]
- 29.Clark MM, Abrams DB, Niaura RS, et al. Self-efficacy in weight management. J Consult Clin Psychol. 1991;59(5):739–44. 10.1037//0022-006x.59.5.739 [DOI] [PubMed] [Google Scholar]
- 30.Seijts GH, Latham GP. The Construct of Goal Commitment: Measurement And Relationships with Task Performance. In Goffin RD, Helmes E, (Eds). Problems and Solutions in Human Assessment: Honoring Douglas N Jackson at Seventy. Boston, MA: Springer US; 2000:315–32. 10.1007/978-1-4615-4397-8_14 [DOI] [Google Scholar]
- 31.Erez M, Zidon I. Effect of Goal Acceptance on the Relationship. of Goal Difficulty to Performance. J Appl Psychol. 1984;69(1):69–78. 10.1037/0021-9010.69.1.69 [DOI] [Google Scholar]
- 32.Rosseel Y lavaan: An R package for structural equation modeling. J Stat Softw. 2012;48:1–36. 10.18637/jss.v048.i02 [DOI] [Google Scholar]
- 33.R Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria 2021. https://www.r-project.org/index.html
- 34.Young MD, Morgan PJ, Plotnikoff RC, et al. Effectiveness of male-only weight loss and weight loss maintenance interventions: a systematic review with meta-analysis. Obes Rev. 2012;13(5):393–408. 10.1111/j.1467-789x.2011.00967.x [DOI] [PubMed] [Google Scholar]
- 35.Sumithran P, Prendergast LA, Delbridge E, et al. Long-term persistence of hormonal adaptations to weight loss. N Engl J Med. 2011;365(17):1597–604. 10.1056/nejmoa1105816 [DOI] [PubMed] [Google Scholar]
- 36.Rosenbaum M, Hirsch J, Gallagher DA, et al. Long-term persistence of adaptive thermogenesis in subjects who have maintained a reduced body weight. Am J Clin Nutr. 2008;88(4):906–12. 10.1093/ajcn/88.4.906 [DOI] [PubMed] [Google Scholar]
- 37.Martins C, Gower BA, Hunter GR. Metabolic adaptation delays time to reach weight loss goals. Obesity (Silver Spring). 2022;30(2):400–06. 10.1002/oby.23333 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Burke LE, Conroy MB, Sereika SM, et al. The effect of electronic self-monitoring on weight loss and dietary intake: a randomized behavioral weight loss trial. Obesity (Silver Spring). 2011;19(2):338–44. 10.1038/oby.2010.208 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Steinberg DM, Tate DF, Bennett GG, et al. The efficacy of a daily self-weighing weight loss intervention using smart scales and e-mail. Obesity (Silver Spring). 2013;21(9):1789–97. 10.1002/oby.20396 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Unick JL, Ross KM, Wing RR. Factors associated with early non-response within an Internet-based behavioural weight loss program. Obes Sci Pract. 2019;5(4):324–32. 10.1002/osp4.341 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Ritchie ND, Kaufmann PG, Gritz RM, et al. Presessions to the National Diabetes Prevention Program May be a Promising Strategy to Improve Attendance and Weight Loss Outcomes. Am J Health Promot. 2019;33(2):289–92. 10.1177/0890117118786195 [DOI] [PubMed] [Google Scholar]