Introduction

Extant theoretical models and empirical evidence indicate that body dissatisfaction (BD), the negative subjective evaluation of one’s shape [1], plays a key role in the onset and maintenance of binge eating (BE) pathology. BD has historically been conceptualized and treated as relatively stable, dispositional tendency of individuals with BE pathology to negatively evaluate their body shape. This typical approach to characterizing BD can be defined as trait BD (defined as a stable and unchanging characteristic that is transferable across a wide range of contexts [2]). Consequently, associations between BD and BE have mostly been tested using trait-like measures of BD and longer term assessments of BE (e.g., the Eating Disorder Examination [3]).

Emerging data from ecological momentary assessment (EMA) studies have shown that BD may be better conceptualized as a fluid construct that varies across time and contexts (i.e., state BD) [3, 4]. Studies have found that individuals often experience variability in state BD characterized by changing levels of BD within a single day in response to different contexts [3,4,5]. For instance, a recent study that assessed state BD over 2 weeks of EMA in a non-clinical sample using a 11-point Likert scale found a notable average within-person variation of 4 points between any two consecutive time points [6]. Variability in BD has been shown to arise due to a variety of internal (e.g., stress levels) and contextual factors (e.g., social settings) [7]. Although evidence suggests that state BD is variable in non-clinical samples, it is currently unknown whether individuals with eating disorders (EDs), who have elevated trait BD compared to the general population, also experience variability in BD. If BD is variable within an ED population, this could change how we conceptualize the role of BD in the etiology and maintenance of BE pathology. While trait BD is commonly viewed as a distal risk factor for BE pathology broadly, state variations in BD may be a proximal risk factor for BE episodes.

Day-to-day experiences of individuals with ED (such as seeing oneself in mirrors, exposure to thin-body glorifying media, company of a romantic partner) may cause variability in BD. For instance, in an experimental study, when healthy individuals were asked to focus their attention specifically on the parts of their body with which they were dissatisfied, individuals experienced temporary worsening of BD which was sustained for a brief period before BD returned to its baseline level [8]. Day-to-day body-related experiences may increase discrepancy between one’s actual self and ideal self among individuals with EDs [9] and may promote variability in BD [7]. If individuals with EDs experience variability in BD, we may expect that variability in BD could lead to changes in momentary ED symptoms. For example, variability in BD may be lower on days when BE occurs (BE days) due to either steady or slightly increasing high level of BD over the course of a day compared to days without BEs (non-BE days) when state BD may be more likely to fluctuate in reaction to several factors throughout the day. For instance, an individual who engages in body checking may do so more frequently on some days (e.g., on days off from work) which may lead to rising BD levels throughout the day and ultimately increase risk for a binge episode occurring as a way to cope with the distress of high BD. This may be in contrast with days (e.g., days at work) when there are fewer opportunities for body checking and BD may fluctuate throughout the day in reaction to various stimuli. Similarly, an elevation in BD from one time point to the next could predict the occurrence of a BE episode later on any given day. Studies have found that elevations in state BD prospectively predicts binge episode [10] and engagement in dietary restriction [11] in subclinical ED samples, but no study has tested this relationship within individuals with a full-threshold ED. If momentary increases in BD predict ED symptoms in full diagnostic treatment seeking samples, we may be able to identify high risk times for BE episodes and design interventions to cope with variability in BD.

In the current study, we had three aims: (1) to characterize whether individuals with BE pathology experience variability in BD, (2) to test whether the degree to which individuals with BE experience variability in BD differs on BE and non-BE days, and (3) to test whether momentary increases in BD uniquely predict BE episodes after controlling for other known momentary predictors of BE (i.e., negative affect [NA] and a BE episode that may have occurred at previous times during the day [prior BE episode]). We controlled for a prior BE episode as the occurrence of a binge episode may increase the likelihood of a subsequent BE episode later in the same day. For example, an individual may attempt to compensate for a BE episode by engaging in strict dietary restriction, which increases the risk of a subsequent BE episode [12]. Additionally, an individual may perceive a BE episode as a lapse from rigid dietary rules and experience NA (e.g., guilt), which may predict subsequent BE episodes to cope with negative emotions [12, 13]. We defined within-in person variability in state BD as an individual’s average difference in state BD from one EMA survey/time point to the next, and defined a clinically significant momentary increases in BD as whenever an individual’s state BD is higher than her average variability in BD. We hypothesized that (1) BD ratings will show variability across 2 weeks of EMA assessment, (2) variability in pre-BE ratings of BD will be lower on BE days compared to non-BE days (due to heightened BD on BE days), and (3) momentary elevations in BD will prospectively predict BE occurrence at the next time point.

Methods

Participants

The sample included 12 adult women (Mage = 34.06 ± 12.96 years, range = 19–57 years; MBMI = 32.28 ± 9.60 kg/m2, range = 20.40–58.20 kg/m2; 66.66% Caucasian) with a primary diagnosis of bulimia nervosa (BN) (56.25%, n = 7) or binge eating disorder (BED) (43.75%, n = 5) [14]. While the sample size is small, it is typical within a range of EMA studies that have shown effects [15,16,17]. Because of the analytic approach of EMA studies, the number of observations of the predictors (in this case level of BD) and outcome behavior (in this case, BEs) is a key aspect in determining whether there is a sufficient sample size. We had a total of 55 BE episodes to predict in the current study and a total of 366 surveys that included BD ratings.

Procedure

Participants were recruited through distribution of flyers in the community and posting in online forums as well as referrals from ED treatment clinics. Participants completed a phone screen to determine initial study eligibility. Eligible participants provided informed consent and completed a diagnostic interview using EDE 17.0. EMA data were collected by both device-prompted (signal contingent) and participant-initiated (event and interval contingent). Participants were given written and verbal instructions on how to use the app and how to identify a BE episode.

Materials

Diagnostic interview

The Eating Disorder Examination Interview 17.0 [3] was administered to assess for DSM 5 BN and BED diagnosis.

Body dissatisfaction (BD)

At each prompt, participants reported how much they agreed to the statement, ‘I am dissatisfied with my body shape and/or weight right now’ on a 5-point Likert scale from 0 (not at all) to 5 (completely).

Negative affect (NA)

Participants reported their momentary NA on a 5‐point Likert scale from 1 (not at all) to 5 (extremely) on 11 items of the Negative Affect (NA) scale (sad, distressed, angry, irritable, ashamed, lonely, disgusted, bored, nervous, anxious, and guilty) (α = 0.92) from the adapted version of Positive and Negative Affect Schedule (PANAS) [18].

EMA-measured binge eating

Participants initiated an EMA survey following the occurrence of BE behaviors (i.e., binge eating). Participants received EMA training in which they were asked to record binge eating which was defined as, ‘consuming an amount of food that you consider excessive or an amount of food that other people would consider excessive, along with feelings of loss of control over eating.’ Study staff also gave several examples of excessive amounts of food drawn from participant’s own binge episodes they described during EDE interview to ensure they accurately identify BE episodes.

Statistical analysis

All analyses were conducted using IBM SPSS 24.00. Variability in BD was calculated using mean square successive difference (MSSD), which measures the average an individual’s state BD changes from one time point to the next [19] and has been used in EMA studies to assess variability in BD [20]. Generalized estimating equations (GEE) with gamma with log link models (because variability was not identically distributed across individuals) were used to examine differences in MSSD between BE and non-BE days as has been employed in other EMA studies [6, 20].

GEE models on the basis of a negative binomial distribution with a logit link function and a first‐order autoregressive matrix structure were used to examine whether momentary elevations in BD predicted BE occurrence at the next survey. A model containing only BD was used to examine the effect of momentary elevations in BD on BE occurrence within the same day. All other models included between‐subject effects (i.e., participants' mean level across the 2 weeks) for each potential trigger (i.e. between-subject BD, NA and a prior BE episode). We controlled for a prior BE episode to avoid confounding the relationship between state BD levels and consequent BE occurrence (yes/no) [21]. All between‐subject variables were grand mean centered. Within-subject effects were centered within person.

Results

EMA descriptives

Mean compliance with prompted EMA surveys was 81.6% (SD = 17.2). A total of 645 baseline EMA recordings, representing 168 participant days were obtained. Within 2 weeks, on average, participants reported a total of 3.00 ± 1.00 BE episodes (range 1–7). Roughly, 30% of the days (n = 46 days) included at least one binge episode. Mean BD across the 2 weeks was 3.87 ± 0.97 (range 1–5).

Variability in BD across days

Consistent with our hypothesis, participants reported variability in BD. On average, individuals’ state BD changed by 1.22 (out of 5) points from its level at a preceding time point (MSSD = 1.22, S.D. = 0.43), representing a typical change of 24% of the available scale range.

Difference in variability in BD between BE and non-BE days

Consistent with our second hypothesis, variability in pre-BE BD was significantly lower on BE days compared to variability in BD during the same time period on non-BE days (p = 0.02, OR = 0.92) (see Table 1).

Table 1 Table showing comparative analysis of variability in BD on BE days and non-BE days
Full size table

Association between momentary elevations in BD and occurrence of BE

Consistent with our hypothesis, elevations in BD were positively associated with the occurrence of a BE episode at the next survey after controlling for one’s overall level of body dissatisfaction relative to other participants, NA and a prior BE episode (β = 0.69, χ2 = 5.11, p = 0.02, OR = 0.49) (see Table 2, Fig. 1).

Table 2 Generalized estimating equations examining momentary elevations in BD as a predictor of occurrence of BE at next EMA survey controlling for NA and a BE episode that had occurred at an earlier time during the same day
Full size table
Fig. 1
figure 1

Figure showing odds of occurrence of BE by state BD. BD body dissatisfaction, BE binge eating. *p < .05. Red line indicates level at which the odds of occurrence of BE would be constant regardless of state BD

Full size image

Discussion

This is the first study to examine (1) whether individuals with BE demonstrate variability in BD, and (2) whether momentary elevations in BD prospectively predict the occurrence of a BE episode. As hypothesized, BD ratings were found to vary across 2 weeks of EMA. On average, participants had a 24% change between EMA surveys (1.22 on a 5-point scale) in their level of BD which is only slightly less variability in BD than was reported in a non-clinical sample that used a different scale (i.e., an average change of 36%, or 4 out of a scale of 11, in state BD from its preceding level) [6]. Although future research is needed to compare ED and non-ED samples on variability in BD directly, one possible reason for the slightly lower variability observed in our sample may be that individuals with BE pathology have higher trait levels of BD, and thus may be somewhat less likely to vary towards the low BD end of the scale. Nevertheless, the presence of variability in BD in our ED sample suggests that BD in ED populations remains reactive to varying internal and contextual factors even if the trait BD is elevated. The current study’s methodology did not allow for the determination of the possible reasons for variability in BD and future studies are needed to understand which internal and contextual factors uniquely predict variability in BD in individuals with EDs. If predictable internal and contextual factors for variability in BD are identified, these may be useful intervention targets.

Consistent with our second hypothesis, there was a significant association between variability in BD and days characterized by occurrence of BE. On BE days, the pre-BE BD ratings fluctuated less compared to non-BE days. Of note, on BE days, the pre-BE BD ratings demonstrated gradual upward increase throughout the day. Prior evidence in BN research suggests a slightly increasing pattern in NA across hours of the day leading up to a binge [22]. Since it is well known that there is a strong association between BD and NA, it might be possible that pre-BE increases in NA reciprocally influences BD throughout the day, which could explain a similar upwards trend in BD ratings on BE days. Future research is needed to test this hypothesis. Additionally, future research is needed to (1) identify factors that differentially account for variability in BD across the two types of days, and (2) assess whether the trajectory of BD on two types of days is predictive of BE episodes.

The third aim of our study was to examine whether momentary elevations in BD add unique variance to the prediction of BE episodes. Consistent with our hypothesis, momentary elevations in BD predicted the occurrence of a BE episode at the next survey after controlling for other factors known to precipitate BE episodes (e.g. NA and a prior BE episode). Specifically, we found that an increase of 1 point in state BD on the 0–5-points scale predicted participants being 0.49 times likely to report an episode of BE at the next time point. Most research have supported the link between momentary increases in NA and occurrence of BE [23]; however, our finding adds to the literature and suggests that momentary increase in BD itself may confer additional risk for occurrence of BE in individuals with BE pathology. Although we are limited in our methodology to explain the association between momentary elevation in BD and binge episodes, future studies could explore what factors account for momentary elevations in BD which predisposes individuals to binge.

The present findings are not without limitations. First, our sample was relatively small which limits our ability to generalize these findings. Second, we were unable to assess for other ED behaviors (e.g., body checking) that may have led to variations in BD. Third, it is possible that the process of participants providing repeated reports of BD and behavior in their natural environment impacted these experiences (i.e., reactivity), although existing evidence suggests that EMA reactivity is limited [6, 24]. Fourth, our sample included participants from a wide range of BMI which could have impacted overall BD and variability in state BD levels (e.g., individuals with high BMI might experience lower variability in BD compared to individuals with lower BMI) and make it difficult to interpret the findings. However, the correlation between BMI and BD in this sample was small (r = 0.29), indicating that BD and variability in BD is not influenced by BMI. Study strengths include use of pre-BE BD ratings in the analysis to examine variability in BD and association between variations in BD ratings and BE without the confound of a prior BE episode to impact BD ratings.

Although our findings are preliminary and warrant replication with larger samples, the initial results suggest that BD may be best conceptualized and targeted in treatment from a state rather than trait perspective. Our findings suggest that individuals with BE pathology experience consider variability in BD within and across days. On BE days, average BD is both higher and trends upwards compared to non-BE days before a binge episode occurs, and momentary elevations in BD from one survey to the next confer risk for BEs. Collectively, our findings suggest that BD might be a fluid construct, and that it might be reactive to a range of internal and external factors. Additionally, our findings suggest that BD often worsens prior to a BE rather than after a BE, suggesting that the BE itself does not produce worsening of BD. If this pattern is confirmed, our results suggest a clear need to understand what factors are contributing to the rising BD prior to a BE episode to better understand how we can prevent and address worsening of BD in the moment before risk of BE occurs. These findings highlight the need to additional investigation of variability in body dissatisfaction, and its association with ED behaviors.

What is already known on this subject?

Extant research has established that trait BD is an important factor in the development and maintenance of EDs [1], yet little is known about state BD in ED populations. In non-clinical samples, state BD has been found to be variable, and these fluctuations are related to ED behaviors [10, 11]. To our knowledge, our study was the first to examine if variability in state BD exists in a clinical ED population and if variations in BD are related to ED behaviors.

What does this study add?

The current study’s findings suggest that BD varies in an ED population, such that BD is elevated and trends upwards before a binge episode occurs. Variability in BD was also found to be predictive of subsequent BE episodes. Future research should further examine factors that contribute to rising BD as BD variability may be a relevant risk factor for BE in ED populations.