Exploring the role of mindful eating and self-compassion on eating behaviours and orthorexia in people following a plant-based diet

Orthorexia nervosa (ON) is a new concept that is more prevalent in vegan populations. The aim of this study was to explore problematic eating behaviours in a vegan population, and to explore whether mindful eating and self-compassion have an impact on orthorexia nervosa. Three hundred thirteen individuals following a vegan diet completed scales in Orthorexia, Self-Compassion, Mindful, Emotional, External and Restraint Eating. The results indicated that individuals with high levels of orthorexia nervosa display low levels of self-compassion, and high levels of restrained eating. Moreover, the ndings indicated that self-compassion, but not mindful eating, partially mediated the relationship between restrained eating and orthorexia nervosa. The present results contribute to a better understanding of orthorexic eating behaviours in a vegan population, and identies the mediating capacity of self-compassion. Future directions are discussed.


Introduction
In recent years, there has been a signi cant increase in individuals who consume plant-based foods such as vegetarian and vegan diets (Timko, Hormes & Chubski, 2012). Individuals follow a plant-based diet for a number of reasons generally summarised under health, morals, environment and animal welfare (Ruby, 2012). Plant-based diets are often associated with better health outcomes such as lowering risks of cancers, lower body weight and decreased likelihood of cardiovascular disease (Bardone-Cone et al., 2012;Key et al., 2009). Studies have shown that plant-based (compared to omnivore) diets are more effective for weight loss (Moore et al., 2015). Kahleova et al. (2018) showed that individuals who are overweight bene ted more from a plant-based diet showing improvements in their body weight, fat mass and insulin resistance markers. Not consuming any meat, dairy, or any products that come from animal origin bene t individuals on plant-based diets (Radnitz, Beezhold & DiMatteo, 2015;The Vegan Society, 2017), especially since meat availability is highly correlated to prevalence of obesity (You & Henneberg, 2016). Contrary, some research has suggested that individuals who consume a meat-free diet are more likely than others to develop an eating disorder (e.g. Bardone-Cone et al., 2012). Evidence indicated that 50% of patients with anorexia nervosa consumed a vegetarian and vegan diet (Hansson et al., 2011) compared to six percent in the general population (Gilbody, Kirk & Hill, 1999). The mixed ndings in a rapidly developing population mandates further research, as well as potential solutions to problems that may be associated to adopting a plant-based diet.
A rather less explored eating behaviour that is prevalent in plant-based dieting and is bordering to the disordered spectrum of food consumption is orthorexia nervosa (ON) (e.g., Barthels, Meyer & Pietrowsky, 2018). ON is a relatively new concept, and is characterised by a xation with healthy eating (Brytek-Matera, 2019). For a person experiencing ON, manifestations of in exible eating behaviours ranging from a focus on consuming organic, raw foods to a complete avoidance of any foods that they consider being unhealthy, frequently leading to malnutrition (Dunn et al., 2017;Brytek-Matera et al., 2017). Extreme dieting restrictions necessitate lengthy food preparation and often severely reduce the enjoyment of food (Cena et al., 2019), creating further psychological implications that are associated to ON. For example, thoughts about only consuming clean and healthy foods and avoiding unhealthy foods may become persistent and intrusive, and distress at breaking food rules may be extreme (Cena et al., 2019;Brytek-Matera et al., 2019;Sfeir et al., 2021). There are no diagnostic criteria for ON in the current classi cation systems (American Psychiatric Association, 2013), and a lack of wider recognision in clinical and medical elds, mandating a greater need for further research as the prevalence for ON varies from 1-10% in the general population (Alles et al., 2017;Chard et al., 2018;Reynolds, 2018).
Research has showed an association between ON and plant-based diets Dittfeld et al., 2017;Heiss, Co no & Hormes, 2019), where people who adhere to vegetarian and vegan diets could be at greater risk of developing ON than individuals who consume meat. Following a plant-based diet requires eliminating certain foods and adhering to strict nutritional rules, behaviours that are characteristic of ON  obesity (Elfhag & Morey, 2008;Loeber et al., 2018). Restrained eating is characterised by deliberately limiting food intake to control weight (Polivy, Herman & Mills, 2020), and emotional eating is characterised by the consumption of signi cant amounts of food in response to negative emotions (e.g. Frayn, Livshits & Knauper, 2018). Some research has shown that individuals who display signs of ON typically display high levels of restrained eating (Bartherls et al., 2018;Brytek-Matera, 2020b), while no research to date explored potential associations of ON to emotional eating. Two studies have found that stress, anxiety, and negative emotions were signi cant predictors of ON in both men and women (Stutts, 2020;Strahler et al., 2018); factors found to predict emotional eating (Tan & Chow, 2014;Rose et al., 2018), and potentially signposting a relationship between ON and emotional eating. Understanding the role of restrained and emotional eating with ON will allow for the development of prevention and intervention programs for individuals who are displaying higher orthorexia symptoms.
These eating behaviours are problematic as they are characterised by consumption of food due to emotional stimuli rather than hunger and satiety (Alberts, Thwissen & Raes, 2012). A number of studies have shown that use of mindfulness techniques improve problematic eating behaviours and eating disorders (Atkinson & Wade, 2015;Jacobs et al., 2013). Mindful eating refers to eating in a conscious way, focusing on the present moment with the aim of satisfying hunger (Wnuk & Du, 2017), however past de nitions of mindful eating have been critiqued as being vague and unreliable. Mantzios (2020) proposed that "mindful eating behaviour is de ned as the sustained attention on a sensory element of the eating exercise (e.g. the taste), and a non-judgmental (or non-evaluative) awareness of thoughts and feelings that are incongruent to the sensory elements of the present eating experiences." (p.3), which is more speci c to behaviour itself, rather than decision making that is co-occurring when eating mindfully. He further suggested that "mindful eating behaviour (i.e. sensory experience of eating, non-judgmentally)" is separate "from decision making for mindful eating (i.e. 'am I still hungry?' 'I will not multitask while eating')" (p.4), which are different elements that may signify different associations and clinical implications for people who are regulating their eating.
Recent research proposed that mindful eating assists in the gradual change of external to internal motivations when eating, such as hunger, when participating in mindful eating interventions (Mantzios, Egan & Asif, 2019), promoting heathier eating behaviours (Mantzios & Wilson 2014, 2015aMantzios & Giannou, 2014;Zeros et al., 2021), including an increased intake of fruit and vegetables (Dutt et al., 2018;Gilbert & Waltz, 2010), and reductions in high sugar and energy dense food consumption (Mantzios, Skillett & Egan, 2020;Mason et al., 2016). Research has also found a negative association between mindful eating and motivations to eat palatable foods ( (2019) found mindful eating to successfully reduce the intensity of food cravings and promoted control over dietary intake, when used as an intervention. Egan and colleagues (2020) drew further associations where mindful eating related negatively to emotional eating (see also Lattimore, 2020), other research highlighted the negative associations to weight gain (Mantzios, Wilson, Linnell & Morris, 2014), and the impact of portion size and self-regulation (Hussein, Egan, Keyte & Mantzios, 2021); all evidence that could blur the line between health behaviours and the potential promotion of ON. To our knowledge, only one study has looked at the effects of mindfulness, rather than mindful eating, on ON; Strahler (2020) showed that orthorexia has two dimensions; healthy orthorexia and orthorexia nervosa. Participants who were in the healthy orthorexia group displayed higher levels of mindfulness (in comparison to orthorexia nervosa group). Therefore, mindfulness may protect against the development of ON and promote healthier eating, but mindful eating and the attentional and attitudinal aspects directed to the food and the eating experience may propose a different, more relevant outlay to the development of relevant mindfulness-based practices.
Self-compassion -a concept related to mindfulness and mindful eating (Mantzios et al., 2018a, b) -is de ned as the recognition that suffering, inadequacy and failure are part of shared human experience (Neff, 2003). This construct includes three components: self-kindness, common humanity, and mindfulness. In a recent systematic review, Braun et al., (2016) provided evidence that self-compassion acts as a protective factor against body dysmorphia and eating pathology. Adams and Leary (2007) showed that introducing self-compassion interventions to restrictive eaters reduced their distress-related eating. All of the research has been either focused on general or overweight/obese populations, which makes the ndings not generalizable, and di cult to relate to ON. Considering the association of selfcompassion to restrictive eating, and the latter to ON, the potential of supporting individuals experiencing higher symptomatology of ON is explored in the present research.
According to Brytek-Matera (2019), a number of studies include semi-vegetarians and pescatarians as individuals who consume a plant-based diet, which was a barrier to drawing clear conclusions, and therefore, this study focused only on vegans. Furthermore, most of studies on ON focus only on vegetarians, and not vegans, when vegans have more dietary restrictions than vegetarians, signifying the added importance to explore vegans and potential associations to ON. The aim of this study was to explore problematic eating behaviours such as emotional eating, restrictive eating, and external eating in a vegan population. Additionally, to our knowledge only one study has looked at effects of mindfulness and ON, but no research has explored the link between ME and self-compassion to ON. In accordance with previous literature, it is hypothesised that emotional and restrained eating will be positively correlated with ON. Moreover, self-compassion and potentially most aspects of mindful eating will be negatively correlated with ON.

Participants
Participants (n = 313; 287 females and 26 males) who were all adults (18 years or over; M = 37.44, SD = 12.33) with BMI of M = 24.86 and SD = 4.87. Participants were recruited through volunteering sampling by advertising the study on several social media platforms such as Facebook and Instagram. The advertisement on Facebook has been posted in a number of vegan/vegetarian groups requesting individuals to participate in the study. Furthermore, the university's Research Participation Scheme was also utilised. Participants recruited through the scheme were rewarded with research credits upon completion of the study. To participate in the study, the participants had to be over the age of 18, have good knowledge of English language and not be diagnosed with an eating disorder.

Demographic information
Participants were asked to report age, gender, ethnicity, weight and height and to select the form of diet that best describes their eating. There were seven options; vegan, lacto-vegetarian, lacto-ove-vegetarian, pescatarian, semi-vegetarian, occasional omnivore and omnivore.
Düsseldorf Orthorexia Scale (DOS; Chard et al., 2018). This 10-item self-reported questionnaire that measures the Orthorexic eating behaviour of participant. A 4-point Likert scale from 1 (this does not apply to me) to 4 (this applies to me) is used. Higher score indicates a more pronounced Orthorexic behaviour. Sample questions include "I have certain nutrition rules that I adhere to" and "I feel upset after eating unhealthy foods". This scale demonstrates Cronbach's alpha of 0.83. Furthermore, high retest reliability was shown where it ranges between 0.67 and 0.79 (Chard et al., 2018) Mindful Eating Behaviour Scale (MEBS; Winkens et al., 2018). The scale contains 17 items that ask about mindful eating in general as well as attention and awareness of mindful eating. This scale utilised a 5-point Likert scale which ranged from 1 (never) to 5 (very often), the higher the score the more mindful individual. Sample questions include "I trust my body to tell me when to eat" and "I notice avours and Sussex-Oxford Compassion Scales (SOCS; Gu et al., 2019). This scale measures compassion for the self and others. For this study, only the compassion for self was measured. This consists of 20 items and responses range from 1 (not true at all) to 5 (always true), the higher the score the more self-compassion individual has. Sample items include "I connect with my own suffering without judging myself" and "When I'm upset, I can let the emotions be there without feeling overwhelmed". The present study produced Cronbach's alpha of 0.82 for recognising suffering, 0.83 for understanding universality of suffering, 0.83 for feeling for the person suffering, 0.79 for tolerating uncomfortable feelings and 0.83 for acting to alleviate suffering.
Salzburg Emotional Eating Scale (SEES; Meule et al., 2018). This scale was developed to measure the changed in food intake in response to emotional experiences. This scale has 20 items which are scored from 1 (I eat much less than usual) to 5 (I eat much more than usual). SEES assesses four emotional states of happiness, sadness, anger and anxiety. The higher scores indicate that individual eats more when experiencing those emotions. Sample items include "When I feel happy…" and "When I am jealous".
The Cronbach's alpha for the present study was 0.86.
Dutch eating behaviour questionnaire (DEBQ; van Strein et al., 1986). DEBQ consists of 33 items where it measures the eating behaviour of adults and it assesses three dimensions of eating behaviour such as emotional eating, external eating, and restrained eating. It utilises a 5-point Likert scale 1 (never) to 5 (very often). Sample items include "If you have put on weight, do you eat less than you usually do" and "If you see others eating, do you also have the desire to eat". The Cronbach's alpha for the present study was 0.83 for diffuse emotions, 0.89 for clearly labelled emotions, 0.91 for emotional eating, 0.84 for external eating and 0.92 for Restrained eating.

Procedure
Ethical approval was obtained from the ethical committee of an institution based in the midland region of the United Kingdom. The study followed ethical practices fully adhering to the British Psychological Society. Participants were recruited via forums and social media groups (e.g., Facebook and Instagram) where they were presented with a hyperlink to Qualtrics). Participants consented and created a unique code to identify data in case of withdrawal. Participants were then presented with demographic information, DOS, MEBS, SOCS, SEES and DEBQ. Upon completion the participant was presented with a debrief form. Participants only attended one online session, which lasted approximately 20 minutes.

Data Analysis
Data analysis was conducted using SPSS software (version 25.0; IBM Corp., 2017). Means, standard deviations, ranges and internal consistency reliability were calculated for all continuous variables. Additionally, descriptive statistics were used to summarise the sociodemographic of the sample. The relationships between BMI, DEBQ, SEES, MEBS, SCOS, DOS and the vegan population were examined using correlational analysis (see Table 1). A further correlation analysis was conducted to examine the relationships between BMI, DOS and subscales of SOCS (see Table 2). Furthermore, mediation analysis was used to evaluate the indirect effects (via self-compassion) of restrained eating on orthorexia nervosa (see Figure 1). Hayes' (Preacher & Hayes, 2008b) PROCESS macro (v3.3) was installed on SPSS (version 25.0) and was used to conduct mediation analyses (model 4) using 10,000 bootstrapping resamples to generate 95% bias-corrected con dence intervals for the indirect effect (Preacher & Hayes, 2008a). According to speci ed guidelines using mediation analyses, Fritz and MacKinnon (2007) suggested that a sample size of 462 participants would enable research to observe an indirect effect of a medium-small sized alpha pathway coe cient (i.e. predictor to mediator) and a medium-small sized beta pathway coe cient (i.e. mediator to criterion) at 80% power using bias-corrected bootstrapping estimating procedures. Participant's body mass index has been calculated using the information provided by the participant.

Results
A multiple correlation analysis has been used to identify which scales (BMI, SOCS, SEES, DEBQ and MEBS) relate to Orthorexia (DOS).
Inter-correlations between DOS, BMI, SOCS, SEES, DEBQ and MEBS are presented in Table 1. Findings suggest that there is a signi cant negative relationship between ON and SOGS (p = .007). Furthermore, there is also a signi cant positive relationship between ON and restrained eating in DEBQ scale (p < .001). All other correlations were non-signi cant. On the other hand, BMI had all signi cant correlations. BMI was negatively associated with SOCS (p = .006), whereas SEES (p < .001), emotional eating (p < .001), external eating (p = .012), restrained eating (p = .021) and MEBS (p < .001), all had a positive relationship with BMI.
A further correlation analysis has been conducted between the ON, BMI and subscales of SOCS, the ndings are presented in Table 2. The correlations had showed that ON is negatively associated with Recognising suffering subscale (p = .009), understanding the universality of suffering subscale (p = .010) and Feeling for the person suffering subscale (p = .040). The subscale of Tolerating uncomfortable feelings and acting or being motivated to act to alleviate suffering were non-signi cant. Furthermore, BMI is negatively associated with Person suffering subscale (p = .001), uncomfortable feelings subscale (p = .009) and alleviate suffering (p = .010). All other subscales were non-signi cant relating to BMI. The mediational model analyses (see Figure 1) use orthorexia as the dependent variable, restrained eating as independent variable, and self-compassion as potential mediator. The c indicated a signi cant relationship between Restrained eating and orthorexia b = .252, p < .001, 95%CI [.186, .318]. Pathway a showed that restrained eating predicted self-compassion b = -.432, p < .001, 95%CI [-.583, -.282], however for pathway b self-compassion did not predict orthorexia b = -.018, p > .05, 95%CI [-.067, .032]. When selfcompassion was included in the mediation model, it remained signi cant b = .244, p < .001, 95%CI [.175, .313], this therefore suggests that the relationship between restrained eating and orthorexia is partially mediated by self-compassion.

Discussion
The aim of the present study was to explore the role of mindful eating and self-compassion on problematic eating behaviours in a vegan population. As previously mentioned, orthorexia nervosa is a new concept for an eating disorder that has not been extensively researched. Previous research has only looked at restrained eating and orthorexia, where it showed that individuals who display high levels of orthorexia engage in high levels of restrictive eating (Brathels et al., 2018;Sfeir et al., 2020). Our study con rmed that vegans who show high levels of orthorexic behaviours do engage in more restrictive eating. Our ndings are consistent to the observation that individuals who consume a vegan diet already have a lot of restrictions as they do not consume any dairy, meat, or products from animal origin (Larson & Johansson, 2002); however, this does not mean that people who consume a vegan diet do not consume the right macronutrients. Typically, studies have shown that vegan populations do consume the right nutrients (Clarys et al., 2014;Key et al., 2009) and they are more likely to be in better health such as lower weight, lower risk of cancers and cardiovascular diseases (Bardone-Cone et al., 2012;Key et al., 2009). The mean BMI reported for this sample was 24.86, which falls under the healthy weight (Brown, 2006;cf. Lebow, Sim and Kransdorf, 2015), and suggests that vegans are physiologically healthy. However, engaging in restrictive eating can be problematic over time as it may lead to psychosocial di culties, such as not engaging in social events, placing more restrictions on food and severe distress if food rules were broken (Cena et al., 2019;. Therefore, restrictive eating in people who consume a vegan diet should be considered with caution as it can be more problematic if aligned to orthorexic behaviours. In accordance with previous studies and what we know about orthorexia, it was hypothesised that individuals who are high in orthorexic behaviours will also have low levels of mindful eating. Our study has showed that there was no relationship between orthorexia and mindful eating, which was surprising as there is a clear attentional aspect that is more evident in following a speci c diet, and past research indicated that orthorexia nervosa was negatively correlated with mindfulness (Strahler, 2020). Our ndings go against all the studies conducted into mindfulness and mindful eating, and associations to healthier eating (Beshara, Hurchinson & Wilson, 2013;Dutt et al., 2018;Jordan et al., 2014;Mantzios, Egan & Asif, 2019) and protective values against the development of disordered eating (Moore et al., 2014). It was also surprising to nd that BMI and mindful eating were signi cantly positively associated to each other, suggesting that individuals with higher BMIs were more prone to eat mindfully (or vice versa). This is an unexpected nding as previous studies showed that mindful eating was effective in interventions to lose weight and to reduce food intake (Warren, Smith & Ashwell, 2017); however, this was conducted in overweight/obese populations and the present sample was of an average weight. Then again other researchers have found similar ndings with average weight populations , proposing that there may be a differential relationship within vegan populations in the weightregulation when compared to omnivores. Essentially the ndings are indicative that there are signi cant differences in the way mindful eating works for people who consume a vegan diet.
A further explanation vital in understanding the ndings could be the motivations for diet choice (MacInnis & Hodson, 2021). For example, if vegans chose to switch to plant-based diet for ethical/sustainable reasons it could explain the ndings as they will be conscious of consuming sustainable and organic produce and not be motivated by weight-regulation. Essentially external and internal motivations may signal different aspects and levels of adaptability and health, and potential in enhancing the effectiveness of practices when knowing and distinguishing healthy versions of healthy eating. For the present research, this creates a limitation to the present study as motivation for the diet choice were not taken into consideration. Future studies will greatly bene t from the inclusion of motivators for the diet choice, as studies have showed that these motivators can further add to the restrictiveness of diets (Fox & Ward, 2008), and combining all people who consume a vegan diet into one group could lead to research outcomes that are not identifying differences within vegan populations. (Braunsberger & Flamm, 2019).
This is also the rst study that considers self-compassion in relation to orthorexia nervosa. The ndings are novel as they indicate that individuals who experience high levels of orthorexic behaviours display low levels of self-compassion. Past research has demonstrated that self-compassion has been related to a number of positive eating behaviours; individuals with higher levels of self-compassion tend to display lower levels of disordered eating (Webb & Forman, 2013), and showed more intuitive eating where they rely on satiety cues and lower dietary restraint (Schoenefeld & Webb, 2013). Furthermore, it was also shown that high self-compassion was linked to more mindful eating, lower disordered eating, and lower BMI (Mantzios et al., 2018a, b;Taylor, Daiss, & Krietsch, 2015 There are contradictory ndings that shows self-compassion is not related to BMI (Ali et al., 2017;. As explained by Mantzios and Egan (2017) this could be due to self-compassion and selfkindness capturing the eating of unhealthy foods as a method of being kind to oneself. There seems to be a case worth exploring further whether self-compassion is useful for self-regulation of weight and disordered eating, where the potential of looking into aspects of self-kindness for body and mind unlock an element of holistic self-care, as seen in previous literature . Despite the mixed ndings, a future study may use a qualitative approach to explore more clearly the utility of selfcompassion (and mindful eating) for orthorexia and people who follow a vegan diet.
In addition, a mediation analysis was conducted where restrictive eating and orthorexia nervosa was explained through self-compassion as a mediator. A signi cant mediation was observed in the present study, which to date, is a novel nding that has not been previously explored, and is consistent with previous literature (e.g., Adam & Leary, 2007). The present ndings have demonstrated that selfcompassion could be introduced to individuals who are restrictive eaters and display orthorexic behaviours.

Limitations:
There are two limitations with the current study. The rst limitation of this study is that the male sample is too low, and therefore, it cannot be generalised to the whole population. Usually, females are more likely to become vegan, whereas males typically consider meat as a masculine food (Modlinska et al., 2020) and are less likely to change their diet. This could explain recruitment numbers, as less males are following a vegan diet, but also highlights the signi cance in exploring males further, especially when considering the differences in eating behaviours between males and females.
Second, this study did not take into consideration any motivational factors to consume a vegan diet (Braunsberger & Flamm, 2019). This could have an impact on the ndings as becoming a vegan for weight reduction versus ethical and environmental reasons may signify psychological and psychosocial exibility. Future research should establish motivational factors and groupings of vegan populations to allow for results to be clear and more controlled within research paradigms.
It is clear that present study has produced some novel ndings that add to the research on orthorexia nervosa. This study has demonstrated that self-compassion is negatively associated with orthorexia nervosa. Additionally, the ndings also showed that orthorexia is positively associated with restrained eating, and self-compassion mediates the relationship between restrictive eating and orthorexia nervosa. This is bene cial as it suggests that self-compassion could be effective for treatments and interventions in populations that are following a vegan diet and are scoring high on both restrictive eating behaviours and orthorexia.