Exploring weight-related self-monitoring as a potential risk factor for eating disorder symptoms in adults – A systematic review and meta-analysis

Objective: Weight-related self-monitoring (WRSM), which involves the intentional tracking of body weight metrics, has been considered a potential risk factor for eating disorders. Therefore, the aim of this study was to systematically summarize the current state of the literature and to quantify the possible association between WRSM and eating disorder symptoms in adults. Method: Preregistration was carried out using PROSPERO (ID CRD42022366133). The PubMed, PsycInfo, and Web of Science databases were searched until December 21, 2023. A study had to be 1) be available in English or German, 2) be peer-reviewed and quantitative, 3) include adult participants (age ≥ 18 years) from the general population, 4) assess eating disorder symptoms via at least one of the following questionnaires: EDI, EAT, FEV, TFEQ, DEBQ, EDE-Q, Munich ED-Quest or IEG, and 5) include WRSM. Data analyses included descriptive analyses and three-level meta-analysis, corrected for correlations, for the global score and the different subscales of the eating disorder questionnaires. Results: A total of 28 studies ( n = 17,370 participants), with an overall fair methodological quality, were included in the systematic review. Out of these studies, nine studies with n = 13,507 participants were ultimately analyzed in the meta-analysis. The three-level meta-analysis did not reveal a significant association between WRSM and the eating disorder global score ( r = 0.13, 95% CI [-0.02, 0.28]; p = 0.08), with this pattern also being evident in the subgroup analysis (diet monitoring). Discussion: WRSM alone does not generally translate into an increased risk of disordered eating symptoms in the general population. We assume that individual factors are likely to determine whether the use of WRSM could lead to eating disorder symptoms. These factors should be accounted for in future research.


Introduction
Everyday life is hard to imagine without digital devices, such as smartphones and tablets.Unsurprisingly, the global number of smartphone users has increased since 2016 (Statista, 2021), and with it the use of health monitoring technologies.Self-monitoring technologies range from ordinary scales and wearable devices to various smartphone-based health apps (Hahn, Sonneville, Kaciroti, Eisenberg, & Bauer, 2021).Becoming increasingly popular in daily life (Mertens, 2016), health apps are used to monitor general health (e.g., sleep quality), fitness, as well as body composition (Hahn, Sonneville, et al., 2021;Jospe et al., 2018;Levinson, Fewell, & Brosof, 2017), with apps for monitoring physical activity, energy and macronutrient intake, and weight being the most popular and widespread across populations (Fox & Duggan, 2012).In particular, smartphone apps like MyFitnessPal (e.g., for monitoring energy expenditure and macronutrient intake) and wearable fitness trackers (activity monitoring) have emerged as powerful tools to help reach weight-related goals.
Intents to track body-related metrics have been recently summarized broadly as weight-related self-monitoring (WRSM; Hahn, Bauer, et al., 2021) and include both monitoring weight through regular weighing and behaviors that can impact weight, such as monitoring dietary intake or physical activity.WRSM can be further categorized into physical tracking (e.g., using calorie-counting apps) and cognitive tracking (e.g., counting calories in one's head; Hahn, Bauer, et al., 2021).Compared to manual tracking (e.g., records on paper), apps on the smartphone are available at any time and do calculations within seconds.This enables constant engagement with nutrition and fitness content (Eikey, 2021).
While (technology-based) WRSM can rapidly raise awareness of one's own behavior (Hahn, Sonneville, et al., 2021), which in turn initiates behavioral changes that result in the desired weight and body composition, the utilization of diverse WRSM applications can potentially harbor risks.For instance, many apps focus on weight loss (Eikey, 2021) aiming for quick results, which can inadvertently normalize severe energy deficits and cultivate unhealthy practices such as imposing rigid daily dietary goals.Considering that strict dieting behaviors and inflexible weight control methods can potentially lead to eating disorders (Eikey, 2021), particular attention needs to be paid to the adoption and development of such practices, especially as self-monitoring opportunities become more accessible (Hahn, Sonneville, et al., 2021;Mertens, 2016).
In recent years, a significant increase in the prevalence of eating disorders has been reported (Galmiche, Déchelotte, Lambert, & Tavolacci, 2019).The development of eating disorders is multifactorial, with WRSM suggested as a potential risk factor (Hahn, Bauer, et al., 2021).It has been hypothesized that the increased attention to weight and diet as well as the perception of controllability of one's weight, both brought about by WRSM, may increase the risk of eating disorders (Berry, Rodgers, & Campagna, 2020;Hahn, Kaciroti, et al., 2021).Several studies showed that the use of WRSM may be associated with exacerbated eating disorder symptoms (Eikey, 2021;Hahn, Bauer, et al., 2021;Hahn, Sonneville, et al., 2021;Honary, Bell, Clinch, Wild, & McNaney, 2019;Levinson et al., 2017;Linardon & Messer, 2019;Neumark-Sztainer, van den Berg, Hannan, & Story, 2006;Simpson & Mazzeo, 2017), albeit uncertainties exist (Jospe et al., 2018;Neumark-Sztainer et al., 2006).In a large cross-sectional study involving over 10,000 participants, Hahn, Bauer and colleagues (2021) revealed that women who engaged in at least one form of WRSM exhibited more pronounced symptoms of eating disorders.Contrarily, men only exhibited increased eating disorder symptoms if they used several forms of WRSM.Simpson and Mazzeo (2017) reported that individuals who used calorie counters demonstrated significantly higher levels of food worry and dietary restriction; fitness tracker usage was also associated with augmented eating disorder symptomatology.Similarly, results of Eikey's (2021) study revealed that the usage of diet and fitness apps can trigger and exacerbate eating disorder symptoms.Also, the literature review of Pacanowski, Linde, and Neumark-Sztainer (2015) showed that regular self-weighing is negatively associated with various psychological factors, such as self-esteem.While Neumark-Sztainer et al. (2006) reported that frequent self-weighing predicted a higher prevalence of eating disorders in women, this relationship did not show up in male participants.In support of that, Jospe et al. (2018) also could not find any differences in eating disorder symptomatology between individuals who used WRSM and individuals who did not.
While the consequences of an eating disorder require early prevention and treatment and are of importance for public health (Hahn, Sonneville, et al., 2021), no systematic review or meta-analysis has been conducted that relates self-monitoring measures and eating disorder symptoms.In addition, the various existing studies do not provide a consistent picture of the relationship between WRSM and eating disorder symptoms.The individual studies or previous literature reviews mainly looked at certain forms of WRSM (e.g.self-weighing; Pacanowski et al., 2015).To our knowledge, there is not yet a comprehensive overview of the influence of WRSM in general, which is why no clear conclusions can be drawn about the influence of WRSM on eating disorder symptoms.At times when WRSM is widely used to control one's health and weight management, it is important to understand potential risks, especially on eating behavior.A meta-analysis could make an important contribution to creating a more balanced understanding of the impact of WRSM on eating disorder symptoms and ultimately help to promote a healthier use of these technologies.Therefore, the aim of this study was to systematically summarize the current state of the literature and to quantify the possible association between WRSM and eating disorder symptoms in the general population.For this purpose, a three-level meta-analysis was conducted to analyze the relationship between WRSM and global eating disorder score.In addition, the various symptoms of eating disorders (such as eating or weight concern) were analyzed in detail by conducting subgroup analyses.Considering the inconclusive findings from existing studies, conducting a meta-analysis enhances the statistical power beyond what can be achieved by examining individual studies.It enables the quantification of an overall effect within a context that is transparent, objective, and reproducible (Borenstein, Hedges, Higgins, & Rothstein, 2009).Moreover, this study aimed to appraise the current state of literature to highlight the strength of evidence, pitfalls, knowledge gaps, and fruitful areas for future research.Based on previous research (e.g., Eikey, 2021;Hahn, Bauer, et al., 2021;Simpson & Mazzeo, 2017), we hypothesized that WRSM is associated with increased eating disorder symptomatology.

Protocol and preregistration
The study was prepared in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement, with its recent extension for systematic reviews (Moher, Liberati, Tetzlaff, Altman, & ThePRISMA Group, 2009;Page et al., 2021).The completed PRISMA checklist can be found in Appendix A. Preregistration was carried out using PROSPERO (ID CRD42022366133).

Search strategy
The systematic literature search was conducted in accordance with the PRISMA Statement (Moher et al., 2009;Page et al., 2021).The PubMed, PsycInfo, and Web of Science databases were searched until December 2022 with no restrictions.Additional literature search took place on December 21, 2023.The following keywords, and combinations of these, were used for the literature search: "eating disorder", "eating disorders", "disordered eating", "eating behavior", "disordered eating behavior", "self-tracking", "self-monitoring", "fitness-tracking", "health-tracking", "calorie-tracking", "calorie-counting", "MyFitnessPal" and "diet and fitness app".The detailed search strategy is shown in Appendix B. After completing the search, Google Scholar was searched for additional articles.Lastly, author names and the reference lists of the included papers were reviewed to find other potential matching papers (Greenhalgh & Peacock, 2005).

Study selection, data extraction and coding of studies
Titles and abstracts were screened for possible fit by two independent reviewers (JRG and AMR) to reduce selection bias.The remaining studies were then reviewed in full length with respect to inclusion criteria.In case of discrepancies, the inclusion of the study was discussed until a consensus was reached.
Data extraction was also performed by two independent reviewers (JRG and AMR).General information about the study (authors, country, as well as year of publication) was extracted, which was followed by information on participants, WRSM, eating disorder symptoms, outcomes, and reliability measures (internal consistency, Cronbach's α).

Participants
Regarding the participants, information on the studies inclusion criteria, sample size, sample composition (i.e., age, sex, and body mass index [BMI]), as well as group allocation were extracted.

Weight-related self-monitoring
Additionally, information was obtained on the form of WRSM (e.g., diet or activity monitoring) as well as the method of recording (digital or manual).Also, information on the frequency of self-monitoring (days/ week) was extracted.If multiple forms of WRSM were considered in an included study, information on the different forms were collected in a differentiated manner.The different forms of WRSM were coded as "nothing", "weight", "physical activity", "body composition", "diet", "combined", "knowledge only", and "body checking".Studies that looked at different forms of WRSM, e.g.activity and calorie tracking, were categorized under the category "combined"."Knowledge only" involves knowledge about calories one's eaten without using an app or diary for it."Body checking" was defined as checking one's own appearance, e.g. by regularly checking at the mirror.

Eating disorder symptomatology
Regarding the eating disorder symptomatology, the questionnaire information (name of the questionnaire, number of scales/subscales, number of levels) was extracted, along with how it was implemented (digitally or manually) throughout the different studies.Reliability (Cronbach's α) and descriptive values (means and standard deviations, change scores) were obtained for both the overall questionnaire score and the various scales/subscales.

Outcomes
The correlation measures between WRSM and the questionnaires were extracted.If different forms of WRSM as well as multiple scales were implemented in the included studies, all correlation measures between the different WRSM forms and the scales/subscales were extracted in each case.If multiple questionnaires regarding eating disorder symptomatology were used, data were extracted for all questionnaires.

Risk of bias assessment
To assess the risk of bias in the primary outcomes of the individual studies, different tools developed by the National Heart, Lung, and Blood Institute of the National Institutes of Health (NIH) were used depending on the study type (NHLBI, 2021): For cross-sectional analyses, the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies was selected.The tool consists of a total of 14 questions that can be answered with "yes", "no", "cannot be determined", "not applicable", or "not reported".Except "yes", all responses indicate a risk of bias (Linares-Gonzalez et al., 2021).Due to study design, cross-sectional studies are automatically rated "not applicable" for criteria 6, 7, 10, 12, and 13 (Linares-Gonzalez et al., 2021;Musshafen et al., 2021).Furthermore, the NHLBI Quality Assessment of Controlled Intervention Studies tool was used to assess the quality of intervention studies (14 questions), alongside the NHLBI Quality Assessment Tool for Case Series Studies (9 questions).The response schemes are identical between the tools.
The assessment of the methodological quality of the included studies was performed by two independent reviewers (JRG and AMR).In case of disagreement between the reviewers, a discussion was held to reach consensus.

Data analyses
Data analyses included descriptive analyses and three-level randomeffects meta-analyses for the global score (assessed by EAT-26, EDE-Q, and EDE-QS) and the different subscales of the eating disorder questionnaires, as per Harrer, Cuijpers, Furukawa, and Ebert (2021).A three-level meta-analysis is a statistical method that combines data from different studies, taking into account the hierarchy of studies, groups within studies, and individual participants within groups to provide a more precise analysis of variations between studies (Harrer et al., 2021).Data preparation was performed with Microsoft Excel (Version 16.69.1,Microsoft 365, Redmond, WA); data analysis with R (R Core Team, 2022), using the R package metafor (Viechtbauer, 2010).In a first step, the total number of participants was determined, along with the weighted average age, weighted average BMI, form of WRSM, and the gender distribution in the total sample, again separated for meta-analysis and systematic review.Afterwards, a three-level meta-analysis (Harrer et al., 2021), was employed to estimate the association of WRSM and disordered eating behavior.For the global score and the different scales, all forms of WRSM were combined.Subgroup analysis for the different forms of WRSM was applied once at least five studies could be included (Borenstein et al., 2009).Level of significance was α = 0.05.In order to compare the correlations of the studies, they were Fisher z-transformed for analysis.If correlation measures or metrics of the relevant variables were not reported, the corresponding authors were contacted with an inquiry for the required data.If no response was received after 14 days, a reminder message was sent.Herein, the second authors were carbon copied.If still no response was received within two weeks, a final reminder was sent asking for the data within the next 14 days.Studies for which the authors provided no feedback were only included in the systematic review.If data were retrieved, correlation measures (Pearson's r, Spearman's ρ, and η-coefficient) were calculated depending on the study design and scaling level.
Given that the extracted correlations are based on various questionnaires, the effects may not be perfectly comparable between studies (Zhang, 2022).Similarly, even with the same questionnaire, reliability estimates may differ across studies, potentially leading to an underestimation of the true correlation (Harrer et al., 2021).Therefore, the Hunter and Schmidt method was used to correct correlations (Hunter & Schmidt, 2004).Herein, the observed correlation coefficients, which contain measurement errors, were corrected for the measurement reliability of the individual variables.In cases where no reliability measures were reported in the studies, internal consistency measures from similar studies with similar populations were used (Gleaves, Pearson, Ambwani, & Morey, 2014;Löffler et al., 2015;Rose, Vaewsorn, Rosselli-Navarra, Wilson, & Weissman, 2013).
To determine how much of the heterogeneity is due to differences within (level 2, τ 2 Level 2 ) and between (level 3, τ 2 Level 3 ) studies, a multilevel version of I 2 was calculated (Harrer et al., 2021).

Study search, selection, and inclusion
The systematic literature search (11/20/2022) provided a total of 3658 results with one additional study identified by searching google scholar.An additional systematic literature search (12/21/2023) yielded 599 results (Fig. 1).In sum, 284 studies were deemed potentially relevant after screening their titles and abstracts.After reading the full texts of these studies, a total of 28 studies were found to meet the predefined inclusion criteria.The detailed reasons for excluding a study from the analysis are provided in Appendix C. The search of the references of the included studies (01/06/2024) did not reveal any yet unknown studies.

Characteristics of included studies and participants
A total of 28 studies with 17,370 participants (n = 11,807 female, 68%) was included in the systematic review (Table 1).The overall weighted mean age was 27.43 years; weighted mean BMI was 25.54 kg/ m 2 .In total, 10 studies used self-weighing, eight studies used physicalactivity monitoring, 16 studies used diet monitoring, and one study used knowledge of nutrition/calories as forms of WRSM.Also, eight studies used a combination of different forms of WRSM in their analyses, explaining why the number of WRSM forms exceeds the number of studies included.Various questionnaires examining eating disorder symptoms were used within the studies: TFEQ (eight studies), FEV (one study), a version of EDI (two studies), EAT-26 (three studies), a version of the EDE-Q (14 studies), and DEBQ (one study).Only the subscales restraint (EDE-Q, TFEQ), eating concern (EDE-Q), body dissatisfaction/ shape concern (EDE-Q, EDI), weight concern (EDE-Q), uncontrolled eating/hunger (TFEQ), and emotional eating/disinhibition (TFEQ) were interpreted (≥5 studies).
Out of these studies, nine studies with 13,507 participants (n = 9,808 female, 73%) were ultimately included in the meta-analysis.The overall weighted mean age was 24.84 years; weighted mean BMI was 24.41 kg/ m 2 .Out of the nine studies, five used self-weighing, three used physical activity monitoring, five used diet monitoring and one used knowledge of nutrition/calorie facts as forms of WRSM.Two studies used a combination of different WRSM forms in their analyses.In total, seven studies used a version of the EDE-Q; both the TFEQ and EDI-3 were implemented once.Thus, associations between WRSM and eating disorder symptomatology were estimated using the following scales: 1) global score of eating disorder symptomatology, 2) restraint, 3) eating concern, 4) shape concern, and 5) weight concern.

Risk of bias and publication bias
None of the included studies were rated with a strong methodological quality (Appendix D).Five studies were rated with a good and 16 studies with a fair methodological quality.Seven studies were rated with a poor methodological quality.Overall, the methodological quality of the included studies was fair (55%).Note. 1 Some studies had to be excluded based on multiple criteria, therefore, the sum of the excluded and included studies does not equal the number of records assessed for eligibility.A detailed overview of the excluded studies is shown in Appendix C; 2 Given that the same sample was used in two studies (Jospe et al., 2017(Jospe et al., , 2018)), it was included only once in the analyses to avoid bias; ED: eating disorders, WRSM: weight-related self-monitoring.Due to power concerns in detecting true asymmetry, it is recommended to test for funnel asymmetry only if the meta-analysis includes at least 10 studies (Harrer et al., 2021).Given that a maximum of k = 6 studies were included in the analysis of any specific outcome, this asymmetry was not tested here.
Twelve studies considered diet monitoring as a form of WRSM (k = 5 included in the meta-analysis).The subgroup analysis revealed a pooled correlation of r = 0.08 (95% CI [− 0.08, 0.24]; p = 0.25; Fig. 3) between diet monitoring and the global score of eating disorder symptoms.The estimated between-cluster variance was τ 2 Level 3 = 0.02; within-cluster variance was τ 2 Level 2 = 0.00.Therefore, I 2 Level 3 = 80.97% of the total variation can be attributed to between-cluster, and I 2 Level 2 = 0.00% to within-cluster heterogeneity.Out of the 11 studies included in the systematic review, six (54.55%;Chappell et al., 2021;Chung, Law, Fong, & Chung, 2014;Jospe et al., 2018;Linardon & Messer, 2019;Messer et al., 2021;Simpson & Mazzeo, 2017) with 2,205 participants reported a descriptively higher global score of eating disorder symptomatology among individuals who used diet monitoring as a form of WRSM.One study did not provide the requested data.Given that self-weighing (as another form of WRSM) was only reported in two studies with regards to the global score of eating disorder symptoms, only diet monitoring was analyzed using subgroup analysis.

Restraint
Overall, 15 studies reported restraint as a subscale of eating disorder symptomatology.Out of these, five studies were included in the metaanalysis.Based on the multilevel model applied, the pooled correlation was r = 0.15 (95% CI [− 0.05, 0.35], p = 0.13; Figure E.1) between WRSM (all forms) and restraint as a subscale of eating disorder symptoms.The estimated between-cluster variance was τ 2 Level 3 = 0.05; within-cluster variance was τ 2 Level 2 = 0.00.Therefore, I 2 Level 3 = 85.48% of the total variation can be attributed to between-cluster, and I 2 Level 2 = 1.47% to within-cluster heterogeneity.Out of the 14 studies included in the systematic review, seven studies (50.00%;Batra et al., 2013;Linardon & Messer, 2019;Peos et al., 2021;Simpson & Mazzeo, 2017;Steinberg et al., 2014) with n = 1,937 participants revealed a descriptively higher restraint score among individuals who used at least one form of WRSM.One study did not provide the requested data.

Eating concern
Overall, seven studies reported eating concern as a subscale of eating disorder symptomatology.Out of these, four studies were included in the meta-analysis.Based on the multilevel model applied, the pooled correlation was r = 0.00 (95% CI [− 0.10, 0.10], p = 0.98; Figure E.2).
The estimated between-cluster variance was τ 2 Level 3 = 0.01; withincluster variance was τ 2 Level 2 = 0.00.Therefore, I 2 Level 3 = 47.17% of the As the same participants were used in both studies, only the data from Jospe et al. (2018) are included in the meta-analysis, Definition of WRSM in the included studies was categorized as follows.
d diary.
e Weighing.
f Questionnaire.
g Not reported.
h Indicates that the correlations were calculated retrospectively.
total variation can be attributed to between-cluster, and I 2 Level 2 = 1.58% to within-cluster heterogeneity.Out of the six studies included in the systematic review, three studies (50.00%;Jospe et al., 2018;Linardon & Messer, 2019;Simpson & Mazzeo, 2017) with 784 participants revealed a descriptively higher eating concern score among individuals who used at least one form of WRSM.One study did not provide the requested data.

Shape concern
Overall, nine studies reported shape concern as a subscale of eating disorder symptoms.Out of these, five studies were included in the metaanalysis.Based on the multilevel model applied, the pooled correlation was r = 0.04 (95% CI [− 0.10, 0.17]; p = 0.55; Figure E.3) between WRSM (all forms) and shape concern as a subscale of eating disorder symptoms.The estimated between-cluster variance was τ 2 Level 3 = 0.02; within-cluster variance was τ 2 Level 2 = 0.00.Therefore, I 2 Level 3 = 71.89% of the total variation can be attributed to between-cluster, and I 2 Level 2 = 0.00% to within-cluster heterogeneity.Out of the eight studies included in the systematic review, three studies (37.50%; Klos, Esser, & Kessler, 2012;Linardon & Messer, 2019;Simpson & Mazzeo, 2017) with n = 883 participants revealed descriptively higher shape concern scores among individuals who used at least one form of WRSM.One study did not provide the requested data.

Weight concern
Overall, nine studies reported weight concern as a subscale of eating disorder symptoms.Out of these, five studies were included in the metaanalysis.Based on the multilevel model applied, the pooled correlation was r = 0.05 (95% CI [− 0.08, 0.19]; p = 0.42; Figure E.4) between WRSM (all forms) and weight concern as a subscale of eating disorder symptoms.The estimated between-cluster variance was τ 2 Level 3 = 0.02; within-cluster variance was τ 2 Level 2 = 0.00.Therefore, I 2 Level 3 = 73.08% of the total variation can be attributed to between-cluster, and I 2 Level 2 = 0.00% to within-cluster heterogeneity.Out of the eight studies included in the systematic review, three studies (37.50%; Klos et al., 2012;Linardon & Messer, 2019;Simpson & Mazzeo, 2017) with 883 participants revealed descriptively higher weight concern scores among individuals who used at least one form of WRSM.One study did not provide the requested data.

Other subscales
Out of six studies that captured uncontrolled eating/hunger as a subscale of eating disorder symptoms, none showed descriptively higher scores in individuals who used at least one form of WRSM.Two of them did not provide data.
A total of 10 studies reported emotional eating/disinhibition as a subscale of eating disorder symptoms.Only one (10.00%;Peos et al., 2021) exposed slightly higher values in individuals who used at least one form of WRSM.Two of them did not provide data.
Among the included studies, the scales: drive for thinness (three studies), dieting behavior (one study), bulimia (three studies), external eating (one study), body dissatisfaction (one study) and food preoccupation (one study) were further assessed.Due to the small number of studies however, no analyses were carried out.

Discussion
This is the first study to systematically investigate the relationship between WRSM and eating disorder symptoms in the general population.The findings of this systematic review with three-level meta-analyses with over 17,000 participants were: no significant association 1) between WRSM and the eating disorder global score, 2) between diet monitoring (e.g., calorie counting) and the eating disorder global score, and 3) between WRSM and the subscales of eating disorder symptomatology.
Based on the predefined significance level, the applied multilevel meta-analysis, which was corrected for correlations, did not reveal a statistically significant association (r = 0.13, 95% CI [− 0.02, 0.28]; p = 0.08) between WRSM (all forms) and the eating disorder global score.These results are in accordance with the systematic review, in which only 50% of the included studies exposed higher global scores of eating disorder symptomatology among individuals who used at least one form of WRSM.With that said, WRSM does not appear to influence eating disorder symptomatology directly.Rather, we hypothesize that individual factors may play a key role in the relationship between WRSM and eating disorder symptoms.For people in the general population who are more susceptible to developing an eating disorder the use of WRSM could certainly constitute a risk factor.For instance, Plateau, Bone, Lanning, and Meyer (2018) showed small, although non-significant, negative correlations between diet and physical activity monitoring and the EDE-Q global score.Contrarily, the study of Hahn, Bauer, et al.  (2021) with more than 10,000 participants revealed a moderate effect of WRSM on the severity of eating disorder symptoms.While the development of eating disorder symptoms is multifactorial, complex, and not finally elucidated yet (Skowron, Kurnik-Łucka, Dadański, Bętkowska--Korpała, & Gil, 2020;Werthmann, Svaldi, & Tuschen-Caffier, 2021), several risk factors have already been identified, such as baseline disordered eating severity score (Hahn, Kaciroti, et al., 2021;Plateau et al., 2018), gender (Brown, Howatson, Quin, Redding, & Stevenson, 2017;Elavsky, Smahel, & Machackova, 2017;Klos et al., 2012;Messer et al., 2021), and personality traits (Embacher Martin, McGloin, & Atkin, 2018;Hahn, Kaciroti, et al., 2021).
Since the present study considered the association of WRSM and eating disorder symptomatology in the general population, no purely clinical samples were included.Although we can therefore assume that the included studies were comparable in terms of the underlying processes of WRSM usage, the eating disorder global scores still differed to some degree.Hence, it is conceivable that WRSM might exert a more pronounced influence on individuals already predisposed to disordered eating compared to those exhibiting only mild tendencies.Although this was not meta-analytically assessed, it is plausible that for those individuals already at a heightened risk for developing an eating disorder, an amplified focus on food, exercise, or weight (WRSM), may exacerbate their existing issues related to eating or weight (Hahn, Kaciroti, et al., 2021).In certain instances, participants were recruited from health and fitness websites (Linardon & Messer, 2019), suggesting they had already shown interest in their own eating behaviors and weight management.Participants exhibiting low baseline disordered eating symptomatology (Hahn, Kaciroti, et al., 2021) were found to be less impacted, or even not affected at all, by WRSM and vice versa.This factor could potentially account for the substantial heterogeneity observed between studies in this meta-analysis.
Another aspect that could have caused the heterogeneity between the included studies is that different study designs were compared with each other.For example, if WRSM was only one aspect of the intervention in RCTs, the effect of WRSM cannot be considered in isolation, which may then influence the results.In addition, there could be differences regarding the self-initiated use of WRSM.While in crosssectional studies the deliberate use of WRSM is assumed to be recorded, WRSM was part of the intervention in RCTs and therefore externally motivated, which could also influence the results.
Individuals who are already concerned about their eating behavior or weight, may use WRSM to gain additional control.Rather than a risk factor itself, WRSM may be considered as a byproduct of eating disorder symptomatology in the general population (Hahn, Bauer, et al., 2021).We hypothesize that once an individual crosses a suspected threshold into disordered eating, WRSM may exacerbate the severity of eating disorder symptoms.Taking into account other symptoms of eating disorders such as intense preoccupation with food or concerns about body weight and body shape (Dilling & Freyberger, 2019), it is possible that people who already have an eating disorder symptomatology, may become even more deeply and intensely preoccupied with their own body or eating behavior through WRSM (e.g., by counting calories).This is often reported after multiple failures to regulate one's eating behavior and body weight (Guertin & Pelletier, 2021), whereby stricter behavior (e.g., more frequent use of dietary self-monitoring) appears to worsen symptoms of disordered eating.Given that WRSM and disordered eating probably share a reciprocal relationship, which needs closer examination in future research, it also appears that individuals with eating disorders or higher levels of eating disorder symptomatology are more likely to monitor food intake and exercise behavior.Hence, future studies might need to take into account whether participants are already engaged in self-monitoring or management of their own eating behaviors.
While women are four times more likely to experience disordered eating during lifetime (Galmiche et al., 2019), personality traits may also influence the relationship between WRSM and eating disorder symptoms.For instance, perfectionistic individuals hold themselves to high standards and may become rigid in their behavior (Eikey et al., 2017;McCaig, Elliott, Prnjak, Walasek, & Meyer, 2020), especially when it comes to achieving goals (e.g., related to weight management).This may, in turn, lead to obsessive thinking about food or weight, which is indicative of eating disorder symptomatology (Eikey, 2021;Hahn, Kaciroti, et al., 2021;McCaig et al., 2020).However, given that the studies which were included in this systematic review with meta-analysis did not examine the effects of personality traits on the relationship between WRSM and eating disorder symptoms, personality traits could not be adequately considered in the present study.Thus, this appears to be a fruitful area for future research.
Although the included studies considered different forms of WRSM, only diet monitoring was suitable for subgroup analysis.Following the pattern of the overall effect, our model did not reveal an association between diet monitoring and the eating disorder global score, suggesting that the isolated usage of diet monitoring might not generally have an effect on eating disorder symptoms.However, these findings are in contrast with our systematic review, in which 55% of included studies reported higher global eating disorder symptom scores in individuals who used diet monitoring.The inconclusive results between our metaanalysis and our systematic review may be attributed to different intensities of diet monitoring (e.g., time spent tracking) or to the combination of various forms of WRSM in the included studies.While many studies did not investigate diet monitoring as the only form of WRSM (Elavsky et al., 2017;Hahn, Bauer, et al., 2021;Katterman, Goldstein, Butryn, Forman, & Lowe, 2014), large studies that differentiated between different forms of WRSM revealed the greatest association between WRSM and eating disorder symptomatology when multiple WRSM forms were combined (Hahn, Bauer, et al., 2021).
Finally, various subscales of eating disorder symptoms were recorded.Again, our three-level meta-analyses, did not reveal a significant association between WRSM and the eating disorder subscale scores (restraint, eating concern, shape concern, and weight concern).These findings are in line with the majority of studies included in our systematic review (restraint: 50%; shape concern: 38%; weight concern: 38% and eating concern: 50%).While it's plausible that the divergent results between our meta-analysis and our systematic review could be accounted for by the eating disorder risk factors (such as baseline severity of disordered eating, gender, personality traits), as well as the intensity and scope of WRSM, all of which have been described above, some authors have expressed concerns regarding the validity of the fourfactor structure of the EDE-Q (Jenkins & Rienecke, 2022).Therefore, in-depth interpretation of the data obtained is restricted to the overall score of the EDE-Q as suggested by Jenkins and Rienecke (2022).The results regarding the subscales should thus be viewed and interpreted with caution.

Limitations
While the study has multiple strengths (e.g., sample size and methodological quality), it is not without limitations.Firstly, it was not possible to include sufficient primary studies to 1) examine the relationship between eating disorder symptomatology and all forms of WRSM in a differentiated manner (e.g., self-weighing), 2) conduct moderator analyses, and 3) draw straightforward interpretations, especially for specific symptoms of eating disorders.Secondly, given that the pooled effect size of r = 0.13 (eating disorder global score) would be considered clinically meaningful, the absence of statistically significant effects may appear somewhat surprising.However, given the variability across studies and the fair data quality, remaining uncertainties need to be addressed in further high-quality randomized controlled trials.Thirdly, different study designs were included in our study (crosssectional studies, randomized controlled trials, case series study).Accordingly, the history of, exposure to, and motivation to use WRSM might have differed among the included studies.For example, the context/setting of the RCTs or adherence were not considered in our analyses, although they might have a potential influence on the results.Finally, it should be emphasized that our study examined correlations that do not allow drawing causal conclusions.Longitudinal and interventional studies are particularly needed to establish whether WRSM has an influence on eating disorder symptoms, or vice versa.

Conclusion and outlook
This study was the first to systematically examine the relationship between WRSM and eating disorder symptoms in healthy individuals providing important practical implications for research, therapy, and general health.Based on the multilevel meta-analysis, WRSM does not appear to generally translate into an increased risk of eating disorders.Considerable heterogeneity was detected which could be attributed to several risk factors influencing eating disorder symptoms, such as a general tendency for eating disorders, gender, and personality traits.
The present study may help raise awareness of potential risk factors for eating disorders, particularly in at-risk groups, but also provide evidence that WRSM does not generally appear to influence eating disorder symptomatology in the absence of other risk factors.Considering the hypothesized role of WRSM as a consequence of a disorder influenced by multiple factors, it is vital to incorporate WRSM in future research and practical applications.Future meta-analyses are further recommended to focus on high-quality randomized controlled studies.In these studies, individual forms of WRSM (i.e., self-weighing, physical activity monitoring, diet monitoring) should be analyzed in a differentiated manner.In addition, other effectors of this relationship should be examined in detail.Aside from ethnicity, age, or personality traits, it should be included whether and to which extent participants are focused on their own eating behavior.

Fig. 1 .
Fig. 1.PRISMA flow chart for study search and selection as per Page et al. (2021)Note. 1 Some studies had to be excluded based on multiple criteria, therefore, the sum of the excluded and included studies does not equal the number of records assessed for eligibility.A detailed overview of the excluded studies is shown in Appendix C; 2 Given that the same sample was used in two studies(Jospe et al., 2017(Jospe et al., , 2018)), it was included only once in the analyses to avoid bias; ED: eating disorders, WRSM: weight-related self-monitoring.

Fig. 2 .
Fig. 2. Forest plot illustrating the correlation between WRSM (all forms) and eating disorder symptoms (global score) Note.The numbers behind the first author and publishing year help to distinguish between the different study groups (different WRSM forms); WRSM: weight-related self-monitoring, RE: random effect.

Fig. 3 .
Fig. 3. Forest plot illustrating the correlation between diet monitoring and eating disorder symptoms (global score) Note.The numbers behind the first author and publishing year help to distinguish between the different study groups (different WRSM forms); RE: random effect.

Table 1
Summary of the studies analyzed.
a indicates inclusion in meta-analysis, b Δ was as noted in the studies or was calculated as post-pre scores, c