Introduction

Eating disorders (EDs) such as anorexia nervosa, bulimia nervosa, and binge-eating disorder have a global presence [1], and are clinically recognized terms according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-5). Although not fully agreed upon, orthorexia nervosa (ON) is defined as the pathological obsession with consuming only (biologically) pure foods that may lead to the development of specific ritualistic attitudes towards food preparation and the restriction of specific food groups that are perceived as harmful to one’s health [2, 3]. The first formal proposal for diagnostic criteria appeared in 2015 [4]. ON has been mostly quoted through the internet and social media, with little data on the prevalence of ON and it has largely been neglected in the professional and scientific literature [5]. Nianias [6] suggests that specifically, Instagram users may use various diets (e.g., paleo, vegan) to cover up their need for obsessive control over their pure eating habits. The fixation on food quality is thought to be a desire to maximize physical health and well-being, rather than religious beliefs, sustainable agriculture, environment, and/or animal welfare [7]. Currently, tools do exist that assess ON such as the ORTO-15 questionnaire [8] that focuses on the quality rather than the quantity of food, a key characteristic of ON [3]. However, this questionnaire does not account for the obsessive–compulsive behaviour of the individual, and thus it is often used in conjunction with other assessment tools (e.g., Maudsley Obsessive Compulsive Inventory (MOCI) [7]). Moreover, a recent study has suggested that the ORTO-15 questionnaire may be unable to distinguish pathology and lacks clinical significance [8]. Albeit, being difficult to generalize, as sample sizes and populations have varied across studies, and researchers have deleted items to increase the scale’s internal consistency (rendering different cut off points) [8], research regarding the prevalence rates of ON (using ORTO-15 questionnaire) has been emerging. Studies have reported results ranging from 27.0% (among 864 young Polish adults) [9] to 81.9% (among 392 Brazilian dieticians) [10]. Moreover, a recent US study found a prevalence rate of 71.0% (among 275 college students), however, when further explored for clinical significance, less than one-half of 1% of this sample suffered from ON [8]. Such variability could be due to a cultural issue, as the instrument was heavily influenced by the United States, and/or psychometric problems with the instrument [8].

The categorization of ON as an eating disorder has been debated since it does not include the core symptoms of anorexia nervosa or bulimia nervosa, including the fear of becoming overweight, extreme weight-control behaviours, and/or the overevaluation of body shape [3]. ON does share some common characteristics with other EDs, such as the individual’s diet provides them with an identity and sense of control over one aspect of their lives [11, 12]. Specifically, a recent study with 220 participants suggests that ON shares similarities with anorexia nervosa and bulimia nervosa with regard to perfectionism, body image attitudes, and attachment style [13]. In addition, it has been suggested that ON may serve as a coping strategy in anorexic individuals [14], and may be a comorbid condition or a residual symptom for anorexic and bulimic individuals [15]. Further, obsessive–compulsive disorder, anorexia nervosa, and bulimia nervosa were studied for overlapping symptoms with ON and the strongest commonalities among all three disorders were anxiety, certain food avoidance, perfectionism, and ritualistic behaviours [16]. However, an argument for clinically distinguishing ON from anorexia nervosa and obsessive–compulsive disorder is that people with ON are likely to flaunt their health behaviours [11], such as via social media, because these behaviours are largely socially desirable and celebrated, making it difficult to determine whether their health behaviours are obsessive and/or no longer healthy [16]. This is further supported by a recent study that reported, through the use of vignettes, those with ON are perceived as less distressing, less likely to evoke sympathy, and more acceptable than anorexia nervosa, bulimia nervosa, and binge-eating disorder [18]. Past research has also shown the use of social media to be a setting for online ED communities to create connections, support, and understanding for like-minded individuals that would not have been available in the past [19]. The encouragement of knowledge, attitudes, and behaviours to achieve unhealthy eating beahviours can be validated by the support of online communities on social networking sites (SNSs) and pro-EDs websites (i.e., pro-ana and pro-mia), which may lead individuals to develop high levels of body dissatisfaction and eating disturbances [20, 21].

Although much of the work investigating SNSs and psychosocial outcomes to date has utilized social comparison theory [22] and sociocultural theory [23, 24], the present study utilized social comparison as it in particular suggests that individuals evaluate their opinions, skills, and identities by comparing themselves to others who are similar or dissimilar [25, 26]. SNSs provide a virtually unlimited source of social comparisons. Research suggests that young female social media users make upward comparisons (comparing oneself to someone who is better off in the domain of interest) of their own bodies to images of slim and toned female universalistic targets (distant sources of influence in mass media) [26,27,28]. As such, SNSs have also served as a way for young females to compare their bodies to thin and toned body images of celebrities and fitness models and ‘inspire’ them to replicate these often, unattainable ideals [29,30,31]. SNSs allow users to create public or private online profiles that they can use to build relationships and interact with other individuals on the same website, thus, creating pro-ana and pro-mia groups for social support and encouragement of unhealthy eating behaviours [19, 32]. SNSs, such as Instagram, allows users to search posts of interest using hashtags that function as a descriptor for individual images and themed pages [32, 33]. According to Tanner [33], coded hashtags such as #ana (anorexia), #mia (bulimia), #Ed (eating disorder) and #thinspo (short form of thinspiration, the amalgamation of the words thin and inspiration), have been used to create dangerous pro-eating disorder communities. Holland and Tiggemann [34] also explored the relationship between #fitspiration (the amalgamation of the words fit and inspiration) and EDs and found that women who posted #fitspiration images scored significantly higher on drive for thinness, bulimia nervosa, drive for muscularity and compulsive exercise, with almost a fifth of the sample size (n = 101) at risk for diagnosis of a clinical eating disorder. Through SNS platforms, social comparison to these universalistic targets promote the perceived message of what healthy, fit, and strong looks like, which can result in lower self-esteem and body dissatisfaction [25, 31] and lead to engaging in behaviours that ultimately damage health and well-being [35, 36].

A recent study has suggested that healthy eating community (i.e., social media users following health food accounts) on Instagram has a high prevalence of ON symptoms, with higher Instagram use being linked to increased symptoms [37]. However, to the authors’ knowledge there is currently no research on the specific #orthorexia within social media research, even though there are over 120,000 posts on Instagram (as of July, 2018). However, great overlap exists with hashtag use on Instagram (i.e., individuals use multiple hashtags on a single image) and, therefore, it may be impossible to use #othorexia to capture ON only on Instagram. Thus, as Instagram makes it easy to share images and connect with like-minded individuals/communities (or those with other EDs or in ED recovery), coupled with its use being linked to higher ON symptoms [37], using #orthorexia may create opportunities for individuals to be exposed to others who promote specific body ideals and post perfect images of their dinner plate, which may be causing psychological and physical health issues [38]. Therefore, the purpose of this paper is to investigate the following research questions (RQ):

RQ 1: Descriptively, what is the nature of the larger #orthorexia conversation on Instagram?

RQ 2: Among a random subsample of #orthorexia images, what types of images are being posted?

RQ 3: How do the authors of #orthorexia images describe themselves in their biographies?

Methods

Multiple methods were utilized for this study. Using the Netlytic program [38], a text analysis was conducted; an image content analysis/coding scheme was developed and implemented for the #orthorexia images; and finally, author biographies were analysed.

Data collection (image selection) and Netlytic analysis

Using the open-sourced software Netlytic [39], all media tagged with the #orthorexia hashtag were downloaded in real time from Instagram (i.e., when the picture was tagged, not necessarily when it was posted). The download began on 25th October 2016 (all posts with the tag are captured every hour for 7 days; October 25th–31st October 2016), capturing only publically available data.

The Netlytic program [39] finds emerging themes of discussion based on the text within the posts. For this study, Netlytic [39] was used to identify popular topics in the #orthorexia data set, as measured by word frequency. Word frequency analysis removes ‘filler words’ (i.e., the, a, if) and punctuation, resulting in a final set of meaningful words and hashtags. Netlytic [39] also creates categories of words and phrases to represent broader categories (i.e., positive vs. negative words) based on pre-determined synonyms, and then automatically identifies what entries belong to what category (see Fig. 1 for category names). In addition to the pre-determined categories of Netlytic [39], the authors also added the synonyms “strong” to the good feelings category, “guilty”, “bloated”, and “full” to the bad feelings category, and “healthy” and “fit” to the appearance category to better reflect the context of ON.

Fig. 1
figure 1

Categories of popular words/phrases associated with #orthorexia posts (N = 4533)

Full size image

The Netlytic program [39] created an excel output file that recorded the image link, publication date, author of the hashtag (who wrote #orthorexia, which is not always the same as who posted the picture), the actual comment containing hashtag, geographical location, and to whom the post was directed (if applicable). The excel file (N = 4533) was downloaded and 245 (3%) images were randomly sampled for content analysis and author biographies. The authors chose to sample 3% since manual coding of the photos was performed and this number was deemed reasonably manageable by the authors.

Content analysis and author biographies

Prior to the coding of the images (N = 245), data cleaning procedures involved using list-wise deletions to delete an image that was a deleted account/photo (n = 38) or a duplicate image (n = 60), leaving 145 unique images for analysis. The study used a novel content analysis/coding scheme that involved coding in multiple categories: people, food, other, and lifestyle. The coding scheme was developed through a combination process of both deductive (a priori) and emergent processes. A similar coding scheme to Tiggemann and Zaccardo’s [40] and Santarossa et al. [31] was used in this study, as images were coded in content categories (people, food, or other). Emergent coding often evolves from the data and are different from the a priori codes [41]. After a preliminary review of the images, emergent coding was used to develop the subcategories for the food category, as well as the development of the lifestyle category. Two homogenous coders (upper year Kinesiology undergraduate students), previously screened for ON, were assigned the same list of randomized images. A codebook with examples and instructions was created for the coding process. Due to the inclusion of the lifestyle category and the complexity of coding, once the coders individually assessed each image, the coders came together to discuss each picture and agree on an answer. The negotiated style of coding is beneficial for studies implementing new coding procedures such as the one utilized in this study [42].

Table 1 describes the overall content analysis/coding scheme. If the image included people, the image was coded for (1) gender, (2) activity engaged in, (3) whether or not it was a selfie (i.e., photo of the poster taken by the poster), and (4) the presence of the following objectifying features: a sexually suggestive pose, obstruction of the face/head, and if one body part was specifically the focus. Images of food were coded as (1) type of food [e.g., real food (sandwich, salad)] or supplement(s) (protein bar/shake), (2) not processed, processed, or mixed, and (3) similar to Webb et al. [43], healthy (i.e., stereotypical “healthy” foods including fresh fruits and vegetables, salads, lean meats, smoothies, sandwiches, etc.) or not healthy (i.e., stereotypical “junk” foods including fast food burger, snack chips, French fries, pop, chocolate cake, etc.). For images that included a cartoon, quote, or scenery, we coded it as other. Instead of looking at simply the image (as per people, food, and other categories above), for the lifestyle category, the overall image and poster’s comments and corresponding photo caption were analysed, thus recording the overall tone of the image. Four variables were coded based on the presence of the related terminology (in the form of a hashtag or plain text): (1) clean eating (e.g., organic, raw, eat clean, non-genetically modified organism, plant-powered), (2) diet-specific (e.g., paleo, vegan, gluten-free, dairy-free, ketogenic, Atkins, Weight Watchers, macronutrient distribution), (3) fitness-related (e.g., fitfam, fitspo, girlswholift), and (4) ED-specific (e.g., EDrecovery, eatingdisorderrecovery).

Table 1 Content analysis/coding scheme of a subsample of #orthorexia images (n = 145)
Full size table

Among the 145 images, duplicate authors were deleted (N = 82), leaving 63 unique authors for analysis. Gender (based on the photograph or taken from the biography description), account type (i.e., personal or non-personal/promotional account), and if the account was verified or not was recorded. The authors biographies were then analysed and categorized into 4 categories: (1) ED recovery (fully or partially recovered) and/or a food/fitness focused descriptor (e.g., vegan, plant-powered, yoga lover), (2) health professional (e.g., body coach, dietitian, yoga teacher, personal trainer, midwife, food blogger, or student) and/or a food/fitness focus descriptor, (3) non-personal ED recovery account that served as platforms for ED recovery/support or inspiration to recover from an ED, and (4) other category that did not mention food/fitness, recovery, or health background.

Results

Text analysis

Among the 4533 downloaded records (i.e., description of the picture or in the comment section), there were 48,780 unique, meaningful words/hashtags associated with #orthorexia. The top 30 most commonly used words or hashtags are described in Table 2, with the top words being love (n = 535) and #edrecovery (n = 425). When placed into categories, the majority of posts were good feelings (47%; see Fig. 1).

Table 2 Most commonly used words associated with #orthorexia (N = 4533 records)
Full size table

Content analysis and author biographies

Of the 145 images, 13% had people in them, 68% contained food, and 12% were categorized as other (12 of which were quotes). Table 1 describes a breakdown on the image content. Of the 63 unique authors, the majority could be identified as female (84%, n = 53 vs. 16% male, n = 10). Additionally, majority of the accounts were personal (87%, n = 55) compared to not personal/promotional accounts (13%, n = 8). Among the authors, 41% (n = 26) mentioned ED recovery (fully or partially) and food/fitness focused, 33% (n = 21) were a health professional and food/fitness focus, 13% (n = 8) were a non-personal ED recovery account, and 13% (n = 8) were considered other.

Discussion

The purpose of this study was to understand the nature of (1) the larger #orthorexia conversation on Instagram, and (2) among a random subsample of images, to analyse the types of images and author biographies. The nature of the conversation did not support the hypothesis that #orthorexia on Instagram encourages social support and highlights negative body image ideals and unhealthy eating behaviours [19, 34]. In contrast, the #orthorexia conversation was associated with good feelings (47%) and encouraging hashtags (#edfamily, #edrecovery) that suggest that Instagram may be being used as a supportive and interactive community that focuses on recovery and adopting healthier eating behaviours. This unexpected nature of the #orthorexia conversation could be explained by the fact that online social networks have been found to foster a sense of community [44]. Furthermore, one can use social norms to cultivate social change through social media [45]. Thus, if the social norm for online conversations around ON is positive, encouraging, and driven by ED recovery, this attitude and behaviour becomes further perpetuated in the communities that have or will develop. In addition, the finding of a generally positive conversation and one not focused on the promotion of body dissatisfaction and/or disordered eating supports that literature that suggests ON individuals do not display negative body image attitudes and are not concerned with weight loss [11]. This study also partially supported that the nature of the images and authors’ biographies would highlight food and lifestyle characteristics to promote the effects of ON, specifically focusing on dieting and restrictive eating practices (by following a certain way of eating). The fact social media motivates by a desire to belong [46, 47] may help to explain and support this finding. Perhaps, this suggests that authors may be using their biographies to attract followers and/or reflect whom they follow through descriptors that create a sense of belonging. Overall, results indicate a need for future researchers to continue to explore online social networks and their influence on EDs, such as ON.

The #orthorexia conversation on SNS appears to be a supportive and interactive community that focuses on recovery and adopting healthier eating behaviours. The interactions also had a positive and encouraging theme, as 47% of words associated with #orthorexia were good feelings and emphasized recovery (#edrecovery, 417 messages; #recovery, 340 messages). However, further research into the correlates of ON is needed, as although past research has suggested ON is not associated with body image, a recent study has found conflicting results, suggesting ON is correlated with body image attitudes such as, appearance orientation, overweight preoccupation, and self-classified weight [13]. Perhaps, the most promising fact given the nature of the #orthorexia conversation is that 41% of authors mentioned ED recovery (full or partial) in their Instagram biography. This finding is reflected in literature, which suggests that history of an ED is the most important predictor of ON [13], and that ON may be a coping strategy or residual symptom for other EDs [14, 15]. Therefore, the present results do not support the notion that Instagram is supporting and encouraging the ideals of ON. However, future research in this area may benefit from involving other social media platforms, as past research has shown content variability with the same hashtag on different platforms (e.g., #thispiration on Twitter vs. Pinterest) [30]. In addition, as social media continues to advance, there is greater ability for interconnectivity between platforms. This interconnectivity, may allow ED communities (supporting recovery and not supporting recovery) to stay connected across multiple sites further developing social norms and a deeper sense of community.

As presented, the #orthorexia images and authors’ biographies mainly highlighted food and lifestyle characteristics. Most images that contained food were pictures of real food (89%) that were not processed (39%) and deemed healthy (68%). This was as expected, given the nature of ON being about only consuming (biologically) pure foods [2, 3], however, some of the images depicted processed (31%) or mixed foods (29%), which might suggest that those posting with #orthorexia may be trying to be supportive of non-restrictive eating habits and lessen the perfectionist tendencies associated with ED [30]. Interestingly, 27% of the images were associated with a specific diet suggesting that an emphasis on diet and adhering to strict rules associated with diets such as vegan, paleo, gluten-free could be used as a way for one to explain their restricted eating preferences in social situations without explicitly mentioning their ON tendencies [6]. Therefore, the nature of #orthorexia may be contradictive, as it is associated with both those who are attempting to recover and are seeking support from the #edrecovery community yet may also be used by those who may be engaging in orthorexic behaviours. However, without further research, it is difficult to make such interpretations with this descriptive study.

Several limitations exist within this study, including that it was based on self-reported data, in both authors’ biographies and images posted, suggesting that those who discuss #orthorexia on Instagram might not actually have ON and instead, just want to be a part of a close knit and highly interactive community. In addition, it is difficult to know if someone who is posting images of “healthy” foods is actually recovering or not, as this is dependent on the cessation of the obsessive thoughts they have towards healthy eating [11] which is difficult to identify in the images they post. It can be difficult to differentiate between someone who is eating healthy and someone who suffers from ON [17] and further research is needed that investigates ON and Instagram/SNS use prospectively. Another limitation is that hashtags and words are considered equivalent in the descriptive analysis of captions using the #orthorexia, despite Instagram users using hashtags as they can serve as both a label and searchable keyword [48]. There is limited research within the area of ON, as ON is a relatively new and often debated term. Any research done in the future would, therefore, be formative. Gender also was a limitation in this study, as images were coded based on the binary labels of female and male, despite the multiple identifications associated with gender [49]. Lastly, the image coding and author biography coding was difficult as this is a relatively new methodology. However, elements of the coding procedures were taken from others [28, 31] and adapted to the current area of study, however, replication studies are needed.

This study used the Netlytic program [39] to aid in exploring conversation and author details, as well as types of images that were posted using #orthorexia. Using this multi-method approach is a strength of the research and adds to the literature as it provides an opportunity to view #orthorexia on Instagram from a variety of perspectives, which may aid in understanding ON further. Overall, this study suggests that the #orthorexia conversation on Instagram is associated with good feelings and encouragement of ED recovery. This small, interactive, and supportive online community does not appear to be propagating the more negative ideals of ON, as originally hypothesized. Although the majority of the images were food focused, most of the authors’ biographies reflected that of the online ON conversation and mentioned ED recovery. These encouraging results may suggest that online communities, such as Instagram, can create social norms that will support positive behaviour change and may be used as a combative tool against EDs. Thus, we believe the current findings may help future researchers design studies to better understand how online social support networks (i.e., Instagram) can be leveraged to encourage ED recovery and the adoption of healthy eating behaviours.