Female Athlete Body Project Intervention with Professional Dancers: A Pilot Trial

Abstract

As aesthetic athletes, professional dancers have increased vulnerability for eating disorders (EDs), with three times higher risk than non-dancers. Among ballet dancers, generalized risk for EDs associated with internalization of western cultural female beauty ideals is compounded by idealization of a ballet-specific body ideal, a combination that confers unique vulnerability for eating pathology. Empirical support has been established for an athlete-specific intervention promoting body acceptance and reduced eating pathology among general populations of young-adult women and female collegiate athletes; the current study adapted this intervention for pilot implementation among professional ballet dancers. Participants from two elite ballet companies (N = 19) were randomized to a control and intervention condition. All participants self-reported eating pathology and related variables pre- and post-intervention, and at six-week follow-up. Post-intervention, participants receiving the intervention demonstrated reductions in body dissatisfaction, p = .005, r = −.63, dietary restraint, p = .008, r = −.59, and eating pathology, p = .007, r = −.60, as compared to control group counterparts; significant differences were retained at follow-up. Results provide preliminary evidence that this intervention has potential to provide a feasible and acceptable means of ED prevention in female professional ballet dancers. Barriers to feasibility are identified and discussed.

Female dancers report elevated eating pathology as compared to non-dancers (Arcelus, Witcomb & Mitchell, 2014) and, as a population, have unique vulnerability for developing eating disorders (EDs) (Ringham et al., 2006). Specifically, as aesthetic athletes (i.e., those for whom maintaining an objectively low weight might be preferred by viewers, and systemically sanctioned as necessary to success in a competitive field), ballet dancers are at increased risk for EDs (Bratland-Sanda & Sundgot-Borgen, 2013). Further, dancers may face specific ED risk factors within their dance environment, including pressure to attain and maintain thinness in body shape (Thomas, Keel & Heatherton, 2005). Indeed, ballet dancers report higher levels of ‘drive for thinness’ and body dissatisfaction (BD) than non-dancers (Arcelus et al., 2014). Weighing an average of 10–12% under their ideal body weight, ballet dancers also report engaging in dieting behavior to maintain this weight status (Kaufman et al., 2002). While some female dancers may restrict intake in an effort to lose or maintain weight, others may unintentionally consume inadequate calories, given their immensely active lifestyle. Whether decreased caloric intake is concomitant with or independent of ED pathology, dancers with lower energy availability are at elevated risk for injuries and/or a syndrome referred to as Relative Energy Deficiency in Sport (RED-S), characterized by complications related to menstrual function, decreased bone density, metabolic rate, immunity, and cardiovascular health, among others (Manore, Kam & Loucks, 2007; Mountjoy et al., 2014).

Despite this panoply of health risks, there is only one published study (with a companion longitudinal follow-up) of an ED prevention intervention with dancers, and that intervention specifically targeted ballet students (Bar, Cassin & Dionne, 2017; Piran, 1999). To the best of our knowledge, there have been no interventions implemented for female dancers at the professional level, or any interventions specifically tailored for aesthetic athletes.

Given that discontent with weight and shape is a salient risk factor for ED (Stice & Shaw, 2002) and may be exacerbated among elite aesthetic athletes (Sundgot-Borgen & Torstveit, 2004), it is reasonable to hypothesize that an intervention based upon improving body acceptance will mitigate risk for ED pathology in elite ballet dancers. Prior work has established that a healthy weight intervention (HWI) successfully reduced symptoms of ED in collegiate female athletes (Becker, McDaniel, Bull, Powell & McIntyre, 2012). This intervention is based on the premise that encouraging healthy lifestyle choices and pursuit of a healthy ideal body shape can help females to improve body acceptance (Stice, Marti, Spoor, Presnell & Shaw, 2008). For the current study, this intervention was chosen over a similar intervention that highlights cognitive dissonance (Stice, Chase, Stormer, & Appel, 2001) because of previous research suggesting that the HWI was received better by athletic groups (Becker et al., 2012). Although it has been termed HWI previously, the content of the intervention is not weight focused. Given shifts in recent years regarding weight-related terminology in the ED field and society more generally, we refer to it here as a healthy living intervention (HLI). It is unknown whether the benefits of HLI will extend to professional ballet dancers (i.e., elite aesthetic athletes).

1.1. Current study

The current study implemented an adaptation of a HLI that was previously tailored to female athletes (Female Athlete Body Project Intervention [FAB]; for more information, see Becker et al., 2012 and Stewart et al., 2017). The intervention was modified for appropriate implementation in a population of female professional ballet dancers. Study aims were two-fold. First, the current study assessed the acceptability and feasibility of this intervention within a sample of elite, professional ballerinas. Second, this study evaluated the preliminary effectiveness of an adapted FAB in reducing BD, dietary restraint, and ED symptoms both immediately post-intervention, as well as at six-week follow-up. It was hypothesized that the intervention would be both feasible and acceptable as implemented in this specific population. We also expected that within condition, dancers who were randomized to the intervention group would report greater reductions in BD, dietary restraint and ED symptoms at the conclusion of the intervention, which would be retained at the 6-week follow-up assessment.

2. Method

This study was completed with approval by a university Institutional Review Board.

2.1. Participants

This study recruited professional dancers from two elite ballet companies in the United States. Dancers were eligible if gender identified as female and a minimum of 18 years of age. Approximately 30–40 female dancers were eligible from each of the companies; those eligible and in attendance at a recruitment meeting were approximately 80% for Company 1 and 30% for Company 2; differences reflected whether the meeting was scheduled partially within rehearsal time versus voluntary. In total, 13 dancers consented from Company 1 and six dancers consented from Company 2. Participants self-identified as White (n = 15; 79 %), White/Asian (n = 2; 10.5 %), and Native Hawaiian or other Pacific Islander (n = 2; 10.5 %). Baseline descriptives are detailed in Table 1.

Table 1.

Combined Sample Descriptive Statistics at Baseline

Variable	Range	M (SD)	Skewness (SE)
Age	18 – 34	23.21 (5.75)	1.09 (.52)
Body Mass Index	16.83 – 21.97	19.25 (1.46)	.02 (.52)
Years Pro	0 – 16	5.26 (5.46)	1.02 (.52)
EDE-Q Global	.57 – 4.40	2.50 (1.12)	− .05 (.52)
EDE-Q BD	.64 – 5.39	3.27 (1.32)	− .46 (.52)
Dietary Restraint	2.30 – 4.10	3.26 (.54)	− .06 (.52)

Note. Years Pro = years dancing professionally; EDE-Q Global = Eating Disorders Examination Questionnaire – average global scores; EDE-Q BD = Eating Disorders Examination Questionnaire – Body Dissatisfaction subscale scores

2.2. Procedure

Initial intervention sessions occurred in early fall at the start of the conventional dance employment season. Following explanation of study details, consented dancers were randomly assigned to an intervention group or control group. The control condition in the current study consisted of dancers who did not receive any intervention; these dancers were not matched or assigned to a wait-list control group. The intervention includes peer assistance, which aids in delivery as well as in continued practice of the main tenets of the material. Accordingly, one dancer was selected from each intervention group to help facilitate implementation of the intervention. In each case and in line with prior research (e.g., Stewart et al., 2017), this participant was selected by the researcher based on her perceived status as a ‘local champion’ within the group (i.e., of more senior rank and perceived role model standing within the company). In each company, this dancer was asked merely to assist the group leader (i.e., principal investigator) and provided with minimal direction prior to the first session (i.e., approximately 15 minutes) on how to help to introduce and facilitate discussion among her peers.

At consent, all dancers completed questionnaires assessing variables of interest and demographic information. The intervention was provided in three 1.5-hour sessions, over three consecutive weeks. The intervention was conducted similarly in both companies, with the exception of timing; Company 1 received the intervention approximately 1 month prior to Company 2. All participants (including those randomized to the control condition) completed questionnaires assessing the same constructs measured at baseline (T1), at the conclusion of the three-week intervention schedule (T2), and again at six weeks post-intervention (T3).

Focus Groups.

At T3, dancers from the intervention group voluntarily attended a one-hour long focus group led by the researcher to discuss the acceptability and feasibility of this program. With the dancers’ verbal consent, groups were audio recorded.

2.3. The Female Athlete Body Project Intervention

The FAB intervention is led by one group leader, along with one to two peer assistants, and closely adheres to a manual. The FAB manual used in the current study was an adaptation of one used in a prior athlete-modified HWI (Becker et al., 2012). Sections within the manual systematically focus on defining healthy body ideals, increasing nutritional density (i.e., eating foods that yield increased nutrients relative to calories), and on factors such as sleep, appropriate rest time, and other behavioral health domains that might influence performance, other than modification of weight.

Session 1:

Participants defined the traditional thin-ideal (e.g., smaller waist with comparatively larger breasts and hips), ballet-specific thin-ideal (e.g., minimal curves), and ballet-specific healthy-ideal (e.g., lean but with muscular strength and flexibility), and compared the three. They discussed benefits of pursuing the ballet-specific healthy-ideal, the energy intake and output balances in their daily nutrition, and learned about RED-S. Participants were asked to log health behaviors with specific, self-generated goals. They were also asked to complete a homework assignment, and to complete a mirror exposure exercise, where they would view themselves in front of a mirror and describe personality, functionality, and physical qualities.

Session 2:

Participants compared healthy and unhealthy dietary restriction, discussed environmental influence on food choices, identified ways to make their diets healthier, reviewed benefits of physical activity (i.e., not only specific to their profession), and talked about the importance of sleep. Participants were asked to write (but not send) a letter to a hypothetical young dancer who they thought might be at risk for ED or struggling with body image.

Session 3:

Participants created a list of top ten reasons to pursue the ballet-specific healthy-ideal, identified barriers to pursuing the healthy-ideal and ways to overcome those barriers, and discussed ways to promote the ballet-specific healthy-ideal as a dance company. Dancers also talked about ways in which they might address and counter “fat talk” within their company culture.

The current study manual was tailored by the researcher (i.e., a former professional ballet dancer) to align with the language and needs of a dance population. For example, in session material that provides typical ‘fat talk’ for athletics (e.g. “This uniform makes me look so fat”), text was reworded to be more salient within the ballet environment (e.g., “This leotard makes me look so fat”). Other manual material was appended with language to reflect the dancers’ schedule (i.e., referencing layoff weeks vs. an athlete’s ‘off-season’) and dance company staff (i.e., ballet masters/mistresses vs. athletic coaches).

2.4. Measures

Demographics.

Participants completed a demographic questionnaire, assessing items such as age, racial background and self-reported height and weight, used to calculate body mass index (BMI, in kg/m²).

Dutch Restrained Eating Scale (DRES; Van Strien, Frijters, Van Staveren & Defares, 1986).

Dietary restraint was measured with the DRES, a 10-item instrument measuring how often (1 = never, 5 = always, scale range: 1–5) participants report engaging in restrained eating behaviors, where elevated scores represent increased dietary restraint. Prior investigation has determined good internal consistency (α = .95) and predictive validity of the DRES (Stice & Agras, 1998; Van Strien et al., 1986). In the current study internal consistency was questionable at T1 (α =.68), and good for T2 (α =.85) and T3 (α =.85).

Eating Disorders Examination Questionnaire (EDE-Q; Fairburn & Beglin, 1994).

The EDE-Q is a self-report questionnaire assessing disordered eating attitudes and behaviors over the previous 28 days. Responses on 28 items are on a seven-point scale, anchored by 0 (no days) and 6 (every day); higher scores reflect greater eating-related pathology. The measure provides a global score and four subscale scores: Restraint, Eating Concern, Shape Concern, and Weight Concern; the current study used the global score as a general measure of ED psychopathology. Study of the EDE-Q within athletes has demonstrated good validity across review of female samples (α = .78 - .85; Pope, Gao, Bolter & Pritchard, 2015) and in the current sample (T1 α = .86). Internal consistency for the Global scores in the current study were good (T1 α = .88, T2 α = .88, T3 α = .83). Given prior work suggesting the shape and weight concern scales load onto a single factor (Barnes, Prescott & Muncer, 2012), these subscales were combined to yield a measure of body dissatisfaction, demonstrating good internal consistency in the current sample (T1, α = .88, T2 α = .89, T3, α = .90).

2.5. Statistical Analyses

Analyses were conducted with IBM SPSS 21 with several non-parametric statistical methods; non-parametric methods were selected due to their robustness in determining effects in small samples. Mann-Whitney tests were used to compare differences across baseline values of variables of interest, between companies and between conditions. Collapsing across companies, Wilcoxon signed rank tests were conducted separately for each condition to assess the effects of the intervention; due to sample size and anticipated low power to detect effects, an interaction effect (i.e., time by intervention) was not tested. As a non-parametric approach to a repeated measures design, Friedman’s analyses of variance (ANOVAs) were used to examine the change in outcome variables post-intervention, and at follow-up. Given recommendations based on nonparametric calculations conducted with a small sample size (Mundry & Fischer, 1998), all measures of significance were calculated as two-tailed exact and corrected for ties rather than reported as asymptotic.

To evaluate whether the magnitude of change in outcome measures was statistically reliable within the intervention group, a Reliable Change Index (RCI) was calculated (Jacobson and Truax, 1991). RCI was computed by dividing the difference between the pre-intervention and post-intervention scores by the standard error of measurement, referencing SE_meas based upon prior work with the variables of interest among female collegiate athletes (Anderson, Petrie, & Neumann, 2012; Darcy, Hardy, Crosby, Lock, & Peebles, 2013). RCI is a statistic that determines the magnitude of change score necessary of a given self-report measure to be considered statistically reliable, and permits evaluation of meaningful changes in test scores over time.

Criteria for Feasibility and Acceptability:

Specific criteria were identified as benchmarks with which to determine the feasibility of the intervention in the current study: 1) > 50% recruitment of possible participants 2) ability to conduct the intervention within a professional dance employment setting 3) no adverse events. The intervention would be considered acceptable if 1) > 50% of those queried found the intervention to be a positive experience and 2) no participant found the intervention harmful or specifically unhelpful. Based on prior work with the ‘healthy weight’ approach among athletes (Becker et al., 2012) upon which this study was based, it was hypothesized that this intervention would be largely acceptable. However, this population is unique in its cultural milieu (i.e., discussing body image and/or eating behavior with one’s co-worker in a ballet company might be vastly different than with a teammate in the context of college athletics). Therefore, in the current study a conservative approach (i.e., > 50%) was implemented.

3. Results

3.1. Preliminary Analyses

A synopsis of participant flow is included in Figure 1; attendance at all intervention sessions, across both companies was 100%. Due to an injury and scheduling issues precluding assessment at T3, there was a small loss to follow-up across conditions. T1 DRES internal consistency was questionable; all other measures demonstrated acceptable internal consistency and there were no observed violations of normality in variables of interest at any of the three time points.

Figure 1.

CONSORT Flow Diagram

Participant characteristics.

Full-sample descriptive statistics are available in Table 1. We further examined participant characteristics at T1 by condition, as well as by company. Mann-Whitney tests did not indicate significant differences between conditions in BMI, age, or years of professional employment. Additionally, there were no differences between conditions at T1 in EDE-Q global (EDE-Qavg), body dissatisfaction (EDE-Qbd), or DRES scores.

Between companies, while the medians were equivalent, there were significant differences in the time of professional employment, with Company 1 dancers on average reporting more years of employment, (Mdn = 18), U = 15.00, p = .034, r = .49. T1 differences in BMI, age, EDE-Qavg, and EDE-Qbd and DRES were non-significant between companies. Median values for variables of interest are shown for each company across measurement time points in Figure 2; based on similarities at T1, further analyses were collapsed across company.

Figure 2.

Median Values for Variables of Interest Relative to Company

Note. T1 = pre-intervention; T2 = post-intervention; EDE-Qavg = Eating Disorders Examination Questionnaire – average global scores; EDE-Qbd = Eating Disorders Examination Questionnaire – Body Dissatisfaction subscale scores; DRES = Restrained Eating Scale scores

3.2. Intervention effects

Wilcoxon signed rank tests conducted separately for each condition revealed significant reductions in EDE-Qavg, and EDE-Qbd and DRES scores between T1 and T2 for the intervention group, but not for the control group (Table 2). Results indicated that for those in the intervention group, there were significant reductions in global eating pathology, body dissatisfaction and dietary restraint; pre- to post-intervention decreases in these scores were non-significant among those in the control condition.

Table 2.

Wilcoxon Signed-rank Tests for Intervention Effects

	T1 Median (Range)	T2 Median (Range)	T	Z	p	r
Control
EDE-Qavg	2.38 (.57–3.29)	1.77 (.58–2.68)	13.00	−.70	.48	−.16
EDE-Qbd	3.33 (.64–4.78)	2.52 (.64–3.46)	3.00	−1.86	.06	−.44
DRES	3.10 (2.30–4.10)	3.05 92.00–3.90)	12.50	−.25	.80	−.06
Intervention
EDE-Qavg	2.87 (.74–4.40)	2.05 (.29–3.39)	1.00	−2.70	.007	−.60
EDE-Qbd	3.71 (1.19–5.39)	2.64 (.39–3.88)	.00	−2.80	.005	−.63
DRES	3.45 (2.80–4.00)	3.15 (1.70–3.80)	1.50	−2.66	.008	−.59

Note. T1= pre-intervention; T2= post-intervention; EDE-Qavg= Eating Disorders Examination Questionnaire – average global scores; EDE-Qbd = Eating Disorders Examination Questionnaire – Body Dissatisfaction subscale scores; DRES = Restrained Eating Scale scores; bold = p < .01

3.3. Intervention effects sustained over time

A Friedman’s ANOVA was conducted for each dependent variable, assessing change over measurement time points, within condition (Table 3). In the intervention condition, EDE-Qavg scores, EDE-Qbd scores, and DRES scores showed a significant decrease over time; significant change was not evidenced in the control group.

Table 3.

Friedman’s ANOVA Between Conditions

	X2(2 df)	p
Control
EDE-Qavg	2.00	.49
EDE-Qbd	3.19	.22
DRES	.96	.68
Intervention
EDE-Qavg	9.75	.005
EDE-Qbd	13.00	.000*
DRES	9.87	.005

Note. df = degrees of freedom; EDE-Qavg= Eating Disorders Examination Questionnaire – average global scores; EDE-Qbd = Eating Disorders Examination Questionnaire – Body Dissatisfaction subscale scores; DRES = Restrained Eating Scale scores; bold = p < .01, asterisk = p < .001

Wilcoxon signed-rank tests further assessed significant findings within the intervention condition; for all follow-up tests, a Bonferroni correction was applied with effects reported at a .0167 level of significance (Table 4). For all three variables of interest, results indicated a significant relation between T1 and T2, and between T1 and T3; in each case, scores on the variables of interest were significantly reduced from pre- to post-intervention, and from pre-intervention to 6 weeks post-intervention. No significant differences in EDE-Qavg, EDE-Qbd or DRES scores were demonstrated from T2 to T3.

Table 4.

Wilcoxon Signed-rank Tests for Intervention Effects Over Time

	T	Z	p	r
Control
T1 – T2 EDEQavg	13.00	−.70	.55	−.16
T1 - T3 EDEQavg	8.00	−1.01	.38	−.27
T2 – T3 EDEQavg	11.00	−.51	.69	−.14
T1 – T2 EDEQbd	3.00	−1.86	.08	−.44
T1 – T3 EDEQbd	8.00	−1.01	.38	−.27
T2 – T3 EDEQbd	11.00	−.51	.69	−.14
T1 – T2 DRES	12.50	−.25	.84	−.06
T1 – T3 DRES	3.50	−1.78	.09	−.48
T2 – T3 DRES	6.50	−1.27	.23	−.34
Intervention
T1 – T2 EDEQavg	1.00	−2.70	.004	−.60
T1 – T3 EDEQavg	.00	−2.52	.008	−.63
T2–T3 EDEQavg	11.00	−.98	.38	−.25
T1 – T2 EDEbd	.00	−2.80	.002	−.63
T1 – T3 EDEbd	.00	−2.52	.008	−.63
T2 – T3 EDEbd	5.00	−1.82	.08	−.46
T1 – T2 DRES	1.50	−2.66	.006	−.59
T1 – T 3 DRES	.00	−2.53	.008	−.63
T2 – T3 DRES	9.50	−.21	.91	−.05

Note. bold = p < .01; df = degrees of freedom; EDEQavg= Eating Disorders Examination Questionnaire – average global scores; EDEQbd = Eating Disorders Examination Questionnaire – Body Dissatisfaction subscale scores; DRES = Restrained Eating Scale scores

3.4. Reliable Change Indices

In the current study, RCI calculations indicated meaningful reduction in BD (i.e., if RCI > 1.96) amongst HLI dancers both post-intervention (ΔT1–T2: 70% of individuals), and at follow-up (ΔT1–T3: 88% of individuals). Similar RCI results were indicated for meaningful improvement in ED pathology (ΔT1–T2: 80%; ΔT1–T3: 75% of individuals); RCI changes in dietary restraint were less robust (ΔT1–T2: 40%; ΔT1–T3: 50% of individuals).

3.5. Qualitative Results

In a focus group, dancers (Company 1: n = 4; Company 2: n = 3) were prompted to speak about what they liked and disliked about the intervention, what they would like to see modified in future iterations, and for their perspective about recruitment. Two trained research assistants transcribed audio recordings, and core themes were extracted based on a review of the transcriptions; any discrepancies were discussed and resolved by the assistants and principal investigator. Although preliminary, three themes appeared to emerge: recruitment, response to the intervention, and suggested improvements for future implementation (for specific exemplars, see Table 5).

Table 5.

Focus Group Themes and Exemplars

Theme	Exemplars
Recruitment
Why do you think people chose not to participate?
Lack of openness to change	“Maybe you are afraid of other people telling you that you are doing something wrong and you have to change it.” “I think anything where you see nutrition and eating disorders on a piece of paper, people see that and instantly are like “I don’t want to do that,” “I don’t want to address it,” “I don’t want to look into what I am eating.”
Sensitivity of material	“Talking about our vulnerabilities in this way is difficult, and very personal – and I think a lot of people don’t want to get that personal with [colleagues].” “I think some people may not want to be seen as vulnerable.”
Skepticism/Futility of efforts	“It’s kind of grim but I feel like the ballet world is never going to change…. And it’s something that I am not happy about and I wish that it would change but I just kind of have this feeling that it is not going to.” “I don’t know if it is ever going to fully change because […] it’s really hard to change people’s mind-sets on body image.”
Response to Intervention
What did you like about this program?
Program material	“For me it was talking and realizing that everyone has something that they want to improve on and I guess feeling less like I’m the only person in the room with that kind of issue.” “I think I was expecting for it to be more of finding what diet regimen would be best to keep your body.”
Scheduling	“Maybe if we did it over a longer period of time, like shorter meetings but maybe still an hour to an hour and a half meetings but maybe over like four or five weeks.” “I would say like once a week for three weeks.”
Appropriateness of manual adaptation	“[I liked] all the information, it’s so much, it was information that felt tailored to dancers, it wasn’t, it didn’t feel like the generic.” “It was helpful to keep it in the front of your mind and remember the things you said like “I want to work on this” and then throughout the week to have the homework there to kind of remember it.”
Suggested Improvement
What would you change about the program?
Order of programming	“Maybe starting with something not so body image specific…” “Start with sleep.” “[Body image is] a very in your face subject to go at. Because not a lot of people, especially dancers that look at themselves in the mirror everyday like to talk about their own body image.”
Participation	“[Making the intervention mandatory] would be amazing because then everyone would be here…because people who have an issue who, say from the outside, we know have an issue but they maybe themselves are like ‘I’m fine.’” “Making this mandatory actually might feel like there’s more support from artistic … so everyone might feel a little bit more inclined to share and participate fully.”
Including staff/others	“I think [an intervention beyond just for females] would be interesting to do.” “I feel like as if we were all in a room and if I heard someone from artistic say “listen we’re a little bit concerned by some things we’ve seen” I think that would make me feel better, because sometimes it feels that it [i.e., being too thin] is actually encouraged …”

Recruitment.

One proposed recruitment barrier was a perceived lack of openness to change amongst colleagues, further qualified as avoidance of judgment by others. Another barrier cited was the sensitivity of the topic (i.e., EDs) and either denial of having a problem with eating behavior, and/or a lack of willingness to share this information with co-workers. As a third recruitment barrier, dancers voiced skepticism in the belief that despite efforts on their individual parts (i.e., participation in this intervention), systemic issues that contribute to ED risk within the dance environment would not change.

Response to Intervention.

Company 1 dancers expressed enthusiasm for the intervention, self-monitoring, and homework assignments. Likely as a result of scheduling difficulty, Company 2 dancers reported feeling rushed through the intervention (they wanted more!). Dancers reported enjoyment in identification with others in regards to intervention targets (e.g., body image; struggles with dietary choices). Many expressed a desire for more explicit and personalized nutrition counseling. All dancers reported that the manual and delivery were appropriately tailored to a dancer population.

Suggested Improvements.

Both companies recommended that the manual begin with a more innocuous topic than body image (e.g., sleep hygiene), thereby increasing initial comfort within group members. All dancers expressed a desire to have greater participation amongst their peers, with the hope that benefits of the intervention would be more powerful and lasting if disseminated throughout the company. Dancers also posited that those most in need of intervention for eating pathology would be less likely to volunteer to participate. Finally, across companies, there was a call for a companion module of this intervention to be delivered to artistic and other administrative staff in hopes that resulting systemic change would be impactful and sustainable.

3.6. Feasibility and Acceptability Criteria.

Based upon the specific criteria identified to determine the feasibility of the intervention, the current study met two of the three benchmarks identified. The criterion not met, and potentially a barrier to feasibility, was the low recruitment rate, which is discussed further in the Discussion. Based on identified criteria, this intervention met standards of acceptability in the current sample.

4. Discussion

The current pilot study evaluated the acceptability and feasibility, as well as preliminary effects of an aesthetic athlete-modified healthy weight intervention within a sample of female professional dancers. Dancers in the intervention program demonstrated significant decreases in reported BD, dietary restraint, and a global measure of ED pathology following the intervention; no significant changes were evidenced for those in the control condition. Further, the significant reductions in the variables of interest in the intervention group were maintained at six-week follow-up.

While indication of significant improvement in key variables for intervention participants was encouraging, an additional aim of the study was to characterize the feasibility and acceptability of this pilot intervention along with associated barriers in an aesthetic athlete population. Focus groups with intervention participants confirmed a positive response, with minimal suggestions offered with which to improve future iterations. More importantly, all dancers emphatically voiced a desire to have more of their peers participate in this program, as they felt that it would not just be beneficial, but genuinely called for, to improve the current health status of the environment in which they worked. Overall, the acceptability of this program appears to be robust, although it should be noted that this was a self-selected group of individuals who chose to participate.

Those unwilling to participate based on recruitment efforts presented a barrier to meeting feasibility criteria, an issue that must be addressed in continuing work with this intervention with dancers. First, adaptations to recruitment must reflect the issues identified within qualitative results. Specifically, it may be necessary to implement this intervention as standard programming and offer the assessment as optional for consent. While acceptability appeared consistent and positive among those who self-selected into participation, future implementation of this intervention more broadly within a given company (i.e., as standard programming for all employees) may both offset barriers to recruitment, as well as yield a different view of acceptability. Secondly, and related to this consideration, dancers emphasized a call for increased involvement from administrative staff, both as a means to model and support the priority placed on health and well-being within the employment environment, as well as to increase the likelihood of broader, more systemic change. Within this elite dance population, sharing symptoms or ED concerns with administrative staff could have the potential to affect employment, a notable difference from amateur athletes. Given this fear of stigma and discrimination based on ED, increased awareness and education for more senior staff about ED and associated risk factors appears indicated. It should also be noted that the current study did not involve or train peer leaders to the extent that prior work with this intervention has implemented (Stewart et al., 2017); while this was a decision made in light of the current pilot sample, larger and more adequately powered replication of this work may benefit from more peer involvement. Taken together, but also considered individually, these modifications have the potential to increase the feasibility of this intervention.

For further consideration of broader application of this intervention, the relatively low body weight and potentially elevated eating pathology scores within this population should be noted. For a slight majority of dancers, BMI values were within the healthy range, with 7 (37 %) dancers below healthy range (18.5 – 24.9; National Institute of Health, 1998). The majority of participants reported ED symptom levels in the non-clinical range across assessment time points but 2 individuals (10.5 %) reported EDE-Q global scores equal to or exceeding 4.0 at baseline, indicating clinical status (Mond, Hay, Rodgers & Owen, 2006). Body weight and ED symptoms may differentially impact outcomes from this intervention across different athlete populations. For example, an athlete population that is less leanness-focused or aesthetic in nature may respond differently to material in the manual related to ‘fat talk.’ For non-aesthetic athletes such as those playing soccer or tennis, physical enhancement might actually reflect coveted strength, and therefore being labeled as ‘skinny’ within the employment environment might be more negatively salient. In the current study, low body weight and clinical ED status amongst some participants underscores the importance of replication and increasing the scope of implementation for this intervention.

In summary, the intervention demonstrated significant reduction in target variables - BD, dietary restraint, and a global measure of eating pathology - for those who participated, and not for those in the control group. These findings were maintained at six-week follow-up assessment, suggesting that the potential beneficial effects of the intervention may be sustainable over time. Reduction of these specific domains is important, as they are indicated as strongly associated for risk and maintenance of eating pathology. As such, the current study shows preliminary evidence that this dancer-modified HLI may serve as an important means of mitigating ED risk among professional dancers. The intervention was well received by dancers across companies, and with the efficient three-session model of delivery, does not suggest high participant burden. This study is the first of its kind to address the pertinent problem of ED prevention in professional dancers, who comprise a vulnerable population within elite athletics. The acceptability among those who participated, and suggested effectiveness of the current study intervention indicate promise for its benefit to the field, not only in professional dance, but more broadly within aesthetic athletes. Barriers to recruitment are of paramount importance in improving implementation efforts.

Limitations

The current pilot study had several limitations to note. Due to constraints of time and dance company budget, recruitment meetings were organized on the dancers’ own time, during hours when they might otherwise be able to rest or go home at the end of a long day; as such, future recruitment would benefit from being held during paid time. As another logistical limitation, during the intervention weeks, across both companies, scheduling necessitated holding a half session in the morning, completed later in the afternoon. This disruption might have influenced the comprehensive learning from each individual session.

Another limitation was the questionable reliability of the T1 assessment for dietary restraint, which impacts conclusions that can be drawn from this measurement. While reliability improved at later assessment, future study might investigate the construct validity of this particular measure of dietary restraint within a dance population and consider the use of an alternate measure. Further, it is not typical for dietary restraint to decrease with this intervention, suggesting that it is possible this particular measure of restraint may yield different results among aesthetic athletes. It is also notable that while not significant in the control group, both intervention and control groups demonstrated reductions in the three main outcome measures. Given the sample size in this pilot trial, a more robust test for time by intervention interaction was not possible, but is warranted as a direction for future research. The timing of this study (i.e., starting at the beginning of the employment season) suggests that dancers entered the study following a layoff, or period of relatively less activity. We might expect that among aesthetic athletes, transitioning from relative inactivity to two months of rigorous daily activity would result in improved body image, and decreases in eating pathology (e.g., desire to restrict calories). Despite this observed trend in both groups, RCI calculations indicated that significant changes in ED pathology and BD over time in the intervention group were robust; these calculations were less robust for changes in dietary restraint, suggesting measurement error within the DRES. RCIs depend upon reference populations and were calculated for this study with standard deviations and reliability indices from non-dancer reference groups (i.e., female collegiate athletes). Future work should consider the outcome scores in the current study as a valid reference group from which to ascertain meaningful change in future longitudinal study within professional aesthetic athletes.

Dancers who participated in the focus groups may have responded with greater social desirability to queries, given that the groups were led by the researcher. Acceptability was also based upon a conservative metric (i.e., > 50% of dancers found it acceptable) and was not quantified with an empirically supported measure of acceptability. Further, the current study did not implement a means of assessing ‘spill-over effects,’ or the degree to which the intervention, and information about it, was reserved only for those receiving it, and did not provide a rigorous wait-list control. Future work would benefit from all of these modifications.

The current intervention is limited by its specificity for implementation within female samples. An important component of this program is promoting healthy attitudes within the company culture as a whole, and future iterations of this intervention should consider including material that is appropriate for use with men. Finally, as a pilot endeavor, the sample size of this study can only yield preliminary data and therefore lacks statistical power, as well as generalizability. Future work will benefit from replication with an increased sample size.

Conclusions

The current study intervention demonstrated significant change (sustained over time) for participants in ED pathology, BD and dietary restraint. Participants initiated a call for the intervention to be implemented systemically, including dissemination to a broader conglomerate of staff. Preliminary evidence suggests this intervention appeared to be acceptable as it was currently implemented in this (aesthetic athlete) population and may be an effective program to mitigate ED pathology in female professional dancers. Future efforts must prioritize barriers to feasibility, with a specific focus on recruitment.

Clinical Implications.

• In professional ballet dancers, a healthy lifestyle intervention was acceptable

• In ballet dancers, intervention reduced eating pathology and body dissatisfaction

• Psychosocial intervention may mitigate eating disorder risk in ballet dancers

• Barriers to feasibility should be addressed in intervention with ballet dancers