Regular self-weighing has been associated with weight loss and maintenance in adults enrolled in a behavioral weight loss intervention; however, few studies have examined the patterns of adherence to a self-weighing protocol. The study aims were to (1) identify patterns of self-weighing behavior; and (2) examine adherence to energy intake and step goals and weight change by self-weighing patterns.
This was a secondary analysis of self-monitoring and assessment weight data from a 12-month behavioral weight loss intervention study. Each participant was given a scale that was Wi-Fi-enabled and transmitted the date-stamped weight data to a central server. Group-based trajectory modeling was used to identify distinct classes of trajectories based on the number of days participants self-weighed over 51 weeks.
The sample (N=148) was 90.5% female, 81.1% non-Hispanic white, with a mean (s.d.) age of 51.3 (10.1) years, had completed an average of 16.4 (2.8) years of education and had mean body mass index of 34.1 (4.6) kg m−2. Three patterns of self-weighing were identified: high/consistent (n=111, 75.0% self-weighed over 6 days per week regularly); moderate/declined (n=24, 16.2% declined from 4–5 to 2 days per week gradually); and minimal/declined (n=13, 8.8% declined from 5–6 to 0 days per week after week 33). The high/consistent group achieved greater weight loss than either the moderate/declined and minimal/declined groups at 6 months (−10.19%±5.78%, −5.45%±4.73% and −2.00%±4.58%) and 12 months (−9.90%±8.16%, −5.62%±6.28% and 0.65%±3.58%), respectively (P<0.001). The high/consistent group had a greater mean number days per week of adherence to calorie intake goal or step goal but not higher than the moderate/declined group.
This is the first study to reveal distinct temporal patterns of self-weighing behavior. The majority of participants were able to sustain a habit of daily self-weighing with regular self-weighing leading to weight loss and maintenance as well as adherence to energy intake and step goals.
Self-monitoring of body weight has been recommended as a component of standard behavioral treatment (SBT) for weight loss.1, 2 Frequent self-weighing may improve individuals’ awareness of their eating and exercise behaviors, provide early detection of subtle weight increases and prevent weight regain after weight loss.3 In recent years, the number of research articles on this topic has increased, which reflects the growing interest in self-weighing as a treatment strategy for weight loss.4, 5, 6, 7 A recent systematic literature review reported that regular self-weighing is associated with successful weight loss, weight maintenance and weight gain prevention in adults seeking behavioral weight loss treatment.8 Another recent literature review also reported that daily self-weighing may be a useful strategy for most adults to prevent weight gain, lose weight or prevent weight regain after loss.9 It is possible that some individuals, such as those who are considered restrained eaters or vulnerable to practicing disordered eating behaviors, may experience negative effects of daily self-weighing. However, the currently available data do not present a cogent argument for daily self-weighing leading to these behaviors.9 There are ongoing studies that continue to provide supportive evidence for daily weighing leading to better weight-related outcomes.10, 11
Technology now permits recording and transmitting weight data in real time. Steinberg et al.12 conducted a behavioral weight loss study focusing on daily weighing. They used a smart scale that displayed the current weight and sent these data directly to a website (www.bodytrace.com) via the wireless cellular network.12 The results demonstrated that the intervention group self-weighed on average±s.d. 6.1±1.1 days per week during the initial 6 months but declined to 4.0±2.3 days per week during the subsequent 3 months. In our previous 18-month behavioral intervention for weight loss trial, we used a scale that stored the data and asked participants to weigh every other day or at least 3 days per week. The results showed that the mean number days of self-weighing per week significantly declined from 2.2±1.1 days per week during the first 6 months to 1.8±1.1 days per week during the second 6 months and to 1.5±1.2 in the final 6-month period.13
Reports of self-weighing interventions are limited by investigators examining barriers for the entire sample to adhere to this strategy over time; however, this approach ignores whether there may be subsamples of individuals with distinct patterns of self-weighing, for example, consistent vs irregular self-weighing. Identifying distinct patterns of self-weighing in real time could provide information on how these patterns affect weight loss and whether interventions can be developed to address deficits in self-weighing behaviors. To date, no study has reported long-term distinct patterns of self-weighing. To address this gap, the aims of this investigation were to: (1) identify the distinct temporal patterns of self-weighing in a sample of adults undergoing standard behavioral treatment for weight loss; and (2) examine the differences in demographic predictors, adherence to lifestyle behaviors and the percentage of weight change by self-weighing patterns.
Subjects and methods
This was a secondary analysis of self-weighing data that were collected using Wi-Fi-enabled scales in the EMPOWER (LE Burke, R01HL107370) study. EMPOWER is a recently completed 12-month, descriptive study where all participants received a standard behavioral intervention for weight loss in overweight and obese adults as the context to observe intentional weight loss and subsequent behaviors that might lead to weight regain. The primary study aim of EMPOWER used ecological momentary assessment methods to assess participants daily in real time in their own environment to determine the triggers for dietary lapses or relapse. The standard behavioral intervention for weight loss included group sessions, self-monitoring of dietary intake and exercise behaviors and provision of daily dietary and weekly exercise goals. The participants were encouraged to attend a total of 24 group treatment sessions that were held across the 12-month study. Participants were instructed to self-monitor their calorie and fat gram intake and minutes of physical activity using a self-monitoring application (Lose It!, FitNow, Inc., Boston, MA, USA) on their smartphone or computer. The self-monitoring data were transmitted to the research server every night with a maximum 24-h lag time in the event the participant did not complete the daily self-monitoring that evening. The study interventionist had access to the self-monitoring data in real time through a study-specific portal and provided feedback to the participant via an email message at the same frequency as the intervention sessions (weekly for the first 3 months followed by bi-weekly for next 3 months, then monthly for the last 6 months). Participants were also given a Withings Wi-Fi scale (Withings, Inc., Issy-les-Moulineaux, France) with instructions to weigh themselves at home daily soon after they arose from bed in the morning or at the same time every day. The scale date-stamped each weighing episode and transmitted the weight data to the LoseIt! server and to the project server in real time; the participants also were able to view their weight data on their smartphone or computer. The University of Pittsburgh Institutional Review Board approved the study protocol and all participants provided written informed consent.
For the EMPOWER study, individuals had to meet the following criteria to be eligible: (1) be ⩾18 years of age, (2) have a body mass index (BMI) between 27 and 44 kg m−2, inclusive, and (3) not have participated in another weight loss program in the past 3 months. Individuals were excluded if they: (1) had the presence of any condition that may confound study findings (for example, diabetes, pregnancy, post-bariatric surgery); (2) planned to become pregnant in next 12 months; (3) planned frequent travel, extended vacations or relocation in next 12 months; (4) were receiving current treatment for a serious mental illness (for example, schizophrenia); (5) reported alcohol intake ⩾4 drinks per day; or (6) were unable or unwilling to use the smartphone.
These data were collected using the self-administered Socio-demographic and Lifestyle Questionnaire. This questionnaire consists of 25 questions that were designed to assess standard socio-demographic and socioeconomic characteristics of the participant, including age, gender, marital status, education, employment status, income and ethnicity/racial background.
Outcome weight was measured semiannually at baseline, 6, 12 and 18 months by a digital scale (Tanita Corporation of America, Inc., Arlington Heights, IL, USA) at the project office. The assessment was performed following an overnight fast with participants wearing light clothing and no shoes. Weight data were analyzed as the percentage of weight change relative to baseline levels (t=0). That is, the percentage of weight change was defined as ((weightt−weight0)/weight0) × 100%, where t=6 and 12 months.
Self-weighing data were transmitted from the Wi-Fi-enabled scale to the research server. Based on the date-stamped information, each day was coded as binary variable indicating whether or not there was any self-weighing. We then calculated number of days of self-weighing for each week for analysis.
The data on energy intake were obtained from the daily dietary recordings from LoseIt!. Adherence to the energy goal was calculated by dividing the total number of calories consumed on a specific day by the daily calorie goal and then multiplied by 100 to express the value as a percentage. For example, if a participant with a daily calorie goal of 1800 reported consuming 1500 total calories in a day, the level of adherence to the energy goal was calculated as 83.3% (1500/1800 × 100%). Based on the calculation of adherence to the energy intake goal, participants were categorized as adherent (reported consuming 85–115% of the daily goals) or non-adherent (reported consuming <85% or >115% of the daily goals) on a daily basis. For days where there were no records retrieved from Lose It! server, adherence to energy intake goals was coded as non-adherent for those days. For analysis, we calculated the number of days that the participant was adherent to the energy intake goal.
We determined adherence to self-monitoring by comparing the recorded calories with the goal calories and defined adherence to self-monitoring for each day as recording ⩾50% of the daily calorie goal. Non-adherence to self-monitoring for each day was defined as recording <50% of the calorie intake goal or no recording of food intake.14 Using these data, the number of days adherent to self-monitoring was calculated for each week for analysis.
Average steps per day were calculated from accelerometer data at baseline, 6 and 12 months. Participants were instructed to wear the accelerometer (ActiGraph GT3x, Tabtronics, Inc., Dayton, OH, USA) for ⩾3 weekdays and 1 weekend day for ⩾10 h day−1 for each assessment period. Adherence to the step goal was defined as ⩾7500 steps per day.
Descriptive analyses were conducted using IBM SPSS Statistics version 23 (IBM Corp., Armonk, NY, USA). Statistical significance was set at 0.05 for two-sided hypothesis testing. Descriptive statistics for continuous variables (for example, the percentage of change in weight, age, education, BMI) were reported as mean±s.d. Categorical variables (for example, gender, race, employment, household income) were described using frequency counts and percentages.
The group-based trajectory modeling15 using PROC TRAJ in SAS version 9.4 (SAS Institute, Cary, NC, USA) was used to identify distinct classes of trajectories of self-weighing over 51 weeks. Group-based trajectory modeling (sometimes called latent class growth analysis or semiparametric finite mixture modeling) is used to identify groups of individuals following similar progressions of some behavior or outcome over age or time.15 Group-based trajectory modeling assumes that there is a certain number of discrete underlying groups in the sample and that each group has its own intercept, slope or shape.15 The predicted trajectory group membership can be used to understand the etiological underpinnings of different developmental trajectories.16
Once the final group-based trajectory model was identified, the resulting predicted group membership was treated as a grouping variable to identify demographic predictors. Chi-square test of independence and general linear modeling were performed to examine the differences in demographic time-invariant categorical and continuous characteristics, respectively, among the levels of the predicted group membership. Linear mixed modeling with time as a categorical factor was used to examine the difference in weight change. To examine the differences in time-dependent variables of adherence to energy intake, steps and self-monitoring of dietary intake among the levels of the predicted group membership, we initially used random coefficient modeling using PROC MIXED in SAS, assuming normal error; however, post-hoc model assessment revealed that the distribution of residuals was not normal. Hence, generalized linear mixed modeling assuming binomial error distribution was performed using PROC GLIMMIX in SAS. Sensitivity analyses were conducted for possible influential cases identified as outliers through graphical methods. The conclusions did not change when outliers were omitted, supporting the robustness of our findings. Hence, the results based on random coefficient models using the full sample were reported.
The total sample for analysis was 148 participants as data for 3 participants (2.0%) were excluded because 2 of them were pregnant during the first 2–3 months of the study and 1 participant withdrew on the first day of the study. The sample was 90.5% female, 81.1% non-Hispanic white, 62.2% married or cohabiting, 72.3% had a household income >$50 000, with a mean age (s.d.) of 51.3 (10.1) years, and had completed average 16.4 (2.8) years of education and had a mean BMI of 34.1 (4.6) kg m−2 at entry.
On average, self-weighing frequency declined from 5.8 to 4.8 days per week from week 1 to the final week of the study. Using group-based trajectory modeling to identify distinct classes of trajectories based on the number of days per week participants self-weighed, three temporal patterns of self-weighing were identified in this sample (Figure 1): high/consistent (75.0% (n=111) self-weighed >6 days per week regularly); moderate/declined (16.2% (n=24) declined from 4–5 to 2 days per week gradually); and minimal/declined (8.8% (n=13) declined from 5–6 to 0 days per week after week 33).
Examination of demographic time-invariant or baseline characteristics revealed a difference in the racial distribution across three self-weighing trajectory groups (P=0.001), with more Asian and white individuals demonstrating a high/consistent self-weighing pattern than black individuals (100.0% (n=3) vs 79.2% (n=95) vs 52.0% (n=13), respectively) and fewer Asian and white individuals following minimal/declined self-weighing pattern than black individuals (0.0% vs 4.2% vs 32.0%). No differences were found among self-weighing trajectory groups in terms of baseline BMI, age, years of education, gender, marital status, employment status and household income level (Table 1).
When we examined the relation between weight change over 12 months and predicted self-weighing trajectory group membership, we found a significant group differences in the percentage of weight change by self-weighing trajectory groups (P<0.001) (Figure 2). The high/consistent group lost on average 10.19%±5.78% at 6 months and 9.90%±8.16% at 12 months. In contrast, the moderate/declined group lost on average only 5.45%±4.73% and 5.62%±6.28% weight at 6 and 12 months, respectively, and the minimal/declined group lost 2.00%±4.58% of baseline weight at 6 months but regained 0.65%±3.58% over their baseline weight at 12 months.
Regarding differences in the change in the adherence variables by self-weighing trajectory group membership, there was a significant interaction effect (self-weighing trajectory group × cubic time) on adherence to calorie intake goal (P=0.009); the high/consistent group had a greater mean number days per week of adherence to calorie intake goal (estimated mean±s.e.: 2.29±0.14) than the minimal/declined groups (1.00±0.40) (t=5.83, P<0.001) but not higher than the moderate/declined group (1.47±0.29) (t=1.71, P=0.09) (Figure 3). Similarly, there was a significant interaction effect (self-weighing trajectory group × cubic time) on adherence to self-monitoring of dietary intake (P<0.001) (Figure 4); the high/consistent group (4.73±0.17) had higher mean number days per week of adherence to self-monitoring of dietary intake than the moderate/declined group (3.45±0.37) (t=3.10, P=0.002) and minimal/declined group (2.12±0.51) (t=4.85, P<0001). For adherence to step goal, only a significant group effect was found (P=0.02), where the high/consistent group had higher probability of being adherent to the step goal than the minimal/declined group (t=2.68, P=0.008) but not higher than the moderate/declined group (t=1.09, P=0.28). As there was no significant group × time effect, we dropped the interaction term and only reported the group and linear time effects for adherence to step goal (Table 2).
This is the first study to identify and report three distinct patterns of self-weighing behavior over 12 months of a behavioral intervention for weight loss, with a majority of participants sustaining a habit of daily self-weighing. The high/consistent group that consistently self-weighed >6 days per week achieved greater weight loss and weight maintenance. The high/consistent self-weighing group also demonstrated greater adherence to calorie intake and step goals as well as adherence to self-monitoring of dietary intake. However, our data demonstrated that one-fourth of the study’s sample was not able to establish a habit of daily self-weighing.
Our findings reveal that the high/consistent self-weighing participants achieved a weight loss larger than what is considered clinically meaningful (for example, a loss of 5% of baseline body weight), which was significantly greater than those who did not establish the daily self-weighing habit. Steinberg et al.12 conducted a behavioral weight loss study focusing on daily weighing and used a smart scale that displayed current weight and transmitted the data directly to a website (www.bodytrace.com) via the wireless cellular network. Their results were similar to ours, although their intervention was less intensive, in that individuals who weighed every day over 6 months achieved significantly greater weight loss than those weighing less often. However, their study was only of 6 months in duration and did not report self-weighing patterns over time. Typically, individuals regain weight after an initial weight loss in a short-term weight loss study, for example, 6 months.17, 18 Another study examined temporal associations between adherence and non-adherence to daily self-weighing and weight changes by analyzing longitudinal self-weighing data in a health-promoting program.19 This study found that weight loss took place during periods of daily self-weighing, whereas self-weighing breaks >1 month posed a risk of weight regain.19 They also found that the more consecutive days without weighing, the larger the weight regain.19 Our study adds to these findings by examining the patterns of self-weighing over 12 months and revealing that individuals who established a daily self-weighing habit had greater weight loss and weight maintenance. We also found that the likely reason for achieving a weight difference might be that high/consistent self-weighers had greater adherence to other lifestyle behaviors (for example, daily calorie intake goal) compared with those who did not establish the habit of daily self-weighing, which is consistent with that reported by Steinberg et al.20 Also, the average percentage of weight loss was higher within the high/consistent self-weighing group in our study compared with trial by Steinberg et al.,20 which might be due to intensive nature of the behavioral intervention in our study.
We observed that 25% of participants did not self-weigh every day over 12 months, including 16.2% who declined to self-weigh only 2 days per week; also 8.8% stopped self-weighing during the study. This finding indicates that it is important to identify barriers to daily weighing among members of these subgroups to help inform the development of interventions to enhance self-weighing for a sustained period. We explored the factors influencing the different patterns of self-weighing. There were no differences in baseline BMI, age, years of education, gender, marital status, employment status or household income level by self-weighing trajectory groups. However, the racial representation was different across the self-weighing trajectory groups. There was a higher percentage of Asian and white individuals following a high/consistent self-weighing pattern than black individuals. No published work has reported on reasons why there is a difference among racial groups in their self-weighing behaviors. Therefore, future work needs to further validate this result as the sample sizes for the racial groups of Asian and blacks in this study were small. If these findings were replicated, another fruitful area of investigation would be to identify possible barriers that interfere with a subgroup of black individuals being able to establish a daily self-weighing behavior.
The main limitation of the study is that the sample was well educated, white and female, which precludes us from generalizing the findings to groups that represent males or less educated women. Additionally, the objective measure of adherence to the step goals was assessed only at three time points; therefore, we could not examine the patterns of change in this behavior by three self-weighing trajectory groups. Strengths of this study include the use of daily prospective data, which allowed us to explore pattern changes of self-weighing behaviors over time. Additionally, this is the first study to identify distinct patterns of self-weighing behavior. Our work demonstrated that not all participants’ self-weighing frequency declined over time, which is distinctly different from previous findings which reported that self-weighing significantly declined over time.12, 13, 19 Also, the findings from the trajectory analysis provided insights into the longitudinal associations between self-weighing patterns and weight changes. Another strength of this study is that self-weighing behaviors were objectively measured and also date-stamped, which eliminated the potential bias of self-reported weight and also documented when the weighing occurred. Moreover, trajectory analysis could be applied to other behavior domains, for example, dietary intake, physical activity or medication adherence.
In conclusion, our work makes a significant and unique contribution to the literature related to self-weighing. Three distinct patterns of self-weighing behavior were identified over 12 months of a behavioral intervention for weight loss. Seventy five percent of participants were able to sustain a habit of daily self-weighing and achieved greater weight loss and weight loss maintenance. However, there was a subgroup of participants who could not establish the daily weighing habit. Thus, it is important to identify the barriers to implementing this strategy in these subgroups; doing so could help inform the development of interventions to enhance self-weighing for a sustained period. Building on the use of the Wi-Fi scales and its delivery of data in real time, future consideration could be given to delivery of feedback and adherence-enhancing messages in real time.
Cooper Z, Fairburn CG . A new cognitive behavioural approach to the treatment of obesity. Behav Res Ther 2001; 39: 499–511.
National Institute of Health. National Heart Lung and Blood Institute. Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults–The Evidence Report. Obes Res 1998; 6 (Suppl 2): 51S–209S.
Wing RR, Tate DF, Gorin AA, Raynor HA, Fava JL . A self-regulation program for maintenance of weight loss. N Engl J Med 2006; 355: 1563–1571.
Madigan CD, Jolly K, Lewis AL, Aveyard P, Daley AJ . A randomised controlled trial of the effectiveness of self-weighing as a weight loss intervention. Int J Behav Nutr Phys Act 2014; 11: 125.
Harrison CL, Teede HJ, Lombard CB . How effective is self-weighing in the setting of a lifestyle intervention to reduce gestational weight gain and postpartum weight retention? Aust N Z J Obstet Gynaecol 2014; 54: 382–385.
Wing RR, Tate D, LaRose JG, Gorin AA, Erickson K, Robichaud EF et al. Frequent self-weighing as part of a constellation of healthy weight control practices in young adults. Obesity (Silver Spring) 2015; 23: 943–949.
Pacanowski CR, Levitsky DA . Frequent self-weighing and visual feedback for weight loss in overweight adults. J Obes 2015. e-pub ahead of print 12 May 2015 doi:10.1155/2015/763680.
Zheng Y, Klem ML, Sereika SM, Danford CA, Ewing LJ, Burke LE . Self-weighing in weight management: a systematic literature review. Obesity (Silver Spring) 2015; 23: 256–265.
Pacanowski CR, Bertz F, Levitsky DA . Daily self-weighing to control body weight in adults: a critical review of the literature. SAGE Open 2014; 4: 1–16.
Madigan CD, Jolly K, Roalfe A, Lewis AL, Webber L, Aveyard P et al. Study protocol: the effectiveness and cost effectiveness of a brief behavioural intervention to promote regular self-weighing to prevent weight regain after weight loss: randomised controlled trial (The LIMIT Study). BMC Public Health 2015; 15: 530.
Linde JA, Jeffery RW, Crow SJ, Brelje KL, Pacanowski CR, Gavin KL 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.
Steinberg DM, Tate DF, Bennett GG, Ennett S, Samuel-Hodge C, Ward DS . The efficacy of a daily self-weighing weight loss intervention using smart scales and email. Obesity (Silver Spring) 2013; 21: 1789–1797.
Zheng Y, Sereika SM, Danford CA, Ewing LJ, Burke LE . Association between self-weighing and percent weight change: mediation effects of adherence to energy intake and expenditure goals. J Acad Nutr Diet 2015; 116: 660–666.
Acharya SD, Elci OU, Sereika SM, Music E, Styn MA, Turk MW et al. Adherence to a behavioral weight loss treatment program enhances weight loss and improvements in biomarkers. Patient Prefer Adherence 2009; 3: 151–160.
Nagin DS . Analyzing developmental trajectories: semi-parametric, group-based approach. Psychol Methods 1999; 4: 139–157.
Jones BL, Nagin DS . Advances in group-based trajectory modeling and an SAS procedure for estimating them. Sociol Methods Res 2007; 35: 542–571.
Butryn ML, Webb V, Wadden TA . Behavioral treatment of obesity. Psychiatr Clin North Am 2011; 34: 841–859.
Katan MB . Weight-loss diets for the prevention and treatment of obesity. N Engl J Med 2009; 360: 923–925.
Helander EE, Vuorinen AL, Wansink B, Korhonen IK . Are breaks in daily self-weighing associated with weight gain? PLoS One 2014; 9: e113164.
Steinberg DM, Bennett GG, Askew S, Tate DF . Weighing every day matters: daily weighing improves weight loss and adoption of weight control behaviors. J Acad Nutr Diet 2015; 115: 511–518.
This work was funded by the NIH grants. The R01HL107370 was awarded to LE Burke at the University of Pittsburgh School of Nursing and the UL1RR024153 was the CTSA grant that was awarded to the CTSI at the University of Pittsburgh, PA, USA.
The authors declare no conflict of interest.
About this article
Cite this article
Zheng, Y., Burke, L., Danford, C. et al. Patterns of self-weighing behavior and weight change in a weight loss trial. Int J Obes 40, 1392–1396 (2016). https://doi.org/10.1038/ijo.2016.68
Nightly Variation in Sleep Influences Self-efficacy for Adhering to a Healthy Lifestyle: A Prospective Study
International Journal of Behavioral Medicine (2021)
Problem-solving, Adherence to Lifestyle Goals, and Weight Loss Among Individuals Participating in a Weight Loss Study
International Journal of Behavioral Medicine (2021)
Self-weighing frequency and the incidence of type 2 diabetes: post hoc analysis of a cluster-randomized controlled trial
BMC Research Notes (2020)
Consistent self-monitoring in a commercial app-based intervention for weight loss: results from a randomized trial
Journal of Behavioral Medicine (2020)
Temporal patterns of self-weighing behavior and weight changes assessed by consumer purchased scales in the Health eHeart Study
Journal of Behavioral Medicine (2019)