Is Social Media Contributing to an Unhealthy Fixation with Health?

Steven Bratman devised the term ‘Orthorexia Nervosa ’ in order to describe a pathological, unhealthy fixation with eating healthy food ¹. The word orthorexia is derived from the Greek words orthos meaning right and orexis meaning appetite. The wish to consume healthy foods is not a condition in itself, but a single-minded fixation and obsessive preoccupation with eating healthy foods may lead to increased levels of stress and anxiety about food ².

The current social and cultural environment seems to encourage the proliferation of conditions such as ON, as individuals are increasingly preoccupied with health and the appearance of health online. ON references in the media have been extensive, and research regarding ON has grown in the past few years. The research has, however, been restricted by the lack of both clear diagnostic criteria and a valid and reliable measurement tool. It is clear throughout the literature that a good measurement tool cannot be developed without clear criteria being established, despite ORTO-15 having become the most widely used measurement tool for ON ³. The measurements of prevalence, possible links to demographic data, use of social media and causal factors have all therefore been problematic.

The current study intends to enhance our understanding of ON by examining the prevalence of ON in an opportunity sample; assessing whether there is a relationship between ON scores and the use of social media; analysing participants’ understanding of the aforementioned potential link; and exploring the reliability and validity of a new ON measure. Given that most ON research uses a single measure for ON, which has led to varied results, the study employs the ORTO-15 as well as a proposed new measure which attempts to compensate for the limitations of the ORTO-15 and introduces a ‘healthy Orthorexia’ threshold. The Tereul Orthorexia Scale (r = .43) consists of 17 items, which aim to evaluate a healthy preoccupation with diet, and the negative social and emotional influence of rigid adherence to a healthy eating programme ⁴.

The study adopts the ideas proposed in a recent study by Turner and Lefevre in order to build upon earlier systematic data ⁵. The study attempted to examine the associations between social media use and ON indications, with the use of an online survey examining social media use and eating behaviours. The study indicated that a higher level of Instagram use was related to a greater predisposition for ON ⁵.

The study aims to compensate for the shortcomings in quantitative exploration of ON, and the difficulties in measurement, by also using a qualitative approach to offer deeper insight into the potential link between ON and social media use. The study aims to contribute to a greater understanding of the motivators for ‘healthy’ eating and of the current climate contributing to ON, by exploring through the use of probing questions the participants’ understanding of the impact of social media on their eating patterns.

The study used self-reporting online questionnaires to obtain the data. The questionnaires included a question regarding whether the participants had a current diagnosis of an eating disorder or were on a medically prescribed diet, in order to exclude them from the study. The questionnaire also involved questions regarding the participants’ demographic groups; measures from the TOS the ORTO-15 the social media channels used; frequency of access to social media; time spent on social media and content of feed; as well as probing questions regarding the participants’ understanding of the possible effect on their eating patterns of specific social media channels, frequency of access and exposure time to social media, and the content of their social media feeds ^{4, 6}.

The participants’ age ranged from 20 to 70 (M=32.92, SD=9.624). 34% of the sample was male and 66% was female. Most participants had a Master’s degree or equivalent, were White and came from Cyprus or the UK.

The Kruskal-Wallis H test showed that there was a statistically significant difference in ORTO-15 scores between age groups (H(3)=13.16, p=0.004). The test suggested that Generation Z (M=39.75) and Millennials (M=38.75) were more likely to rate highly on ON scores using the ORTO-15 measure than Generation X (M=41.93) and Boomers (M=44.56).

The Mann-Whitney U Test revealed no significant differences in the HeOr scores of Males (M=20.73) and Females (M=22.04), U = 4123, z = -1.57, p = 0.12. It also generated no significant differences in the OrNe scores of Males (M=12.49) and Females (M=13.58), U = 4096.50, z = -1.64, p = 0.10. The Mann-Whitney U test indicated that ORTO-15 scores were significantly greater for Males (M=39.06) than Females (M=38.24), U = 3668, z = 2.11, p = 0.03, suggesting a higher likelihood of ON in Females.

Out of 201 participants, 146 scored below the threshold value of 40 for ON on the ORTO-15 scale. This resulted in a 72.63% prevalence in the sample. There were no thresholds reported for the TOS (Barrada & Roncero, 2018), hence prevalence of HeOr and OrNe in the sample using this measure was not computed.

An internal consistency reliability analysis on the 201 responses resulted in Cronbach’s alpha of 0.86 for all 17 items, Cronbach’s alpha of 0.84 for the nine items identifying HeOr and Cronbach’s alpha of 0.79 for the eight items measuring OrNe. The internal consistency for the ORTO-15 produced a Cronbach’s alpha of 0.76. The Spearman’s rho correlation coefficient indicated that there were significant correlations between HeOr, OrNe and ORTO-15. Findings showed a correlation between HeOr and the ORTO-15 rs=-0.28, p=0.00 and between OrNe and ORTO-15 rs=-0.55, p=0.00. The above findings denote that the TOS is therefore internally consistent, valid and reliable.

The Mann-Whitney U Test revealed a significant difference in the ORTO-15 scores of Instagram users (M=38.77, n=146) and non-Instagram users (M=41.35, n=55), U = 3044, z = –2.64, p = 0.008. Hence there is evidence to suggest that Instagram users are more likely to score higher on ON symptoms on the ORTO-15 than non-Instagram users.

In order to examine the participants’ understanding of the potential link between ON and social media use, the probing questions were analysed using thematic analysis and the six-phase model outlined by Braun & Clarke ⁷. Three main themes and nine subthemes developed from the analysis. The first main theme is ‘The Need to Belong (Inclusivity/Exclusivity)’, which tried to encompass the psychological and basic human need to belong, and feel accepted within a social group ⁸. The theme also attempts to incorporate the contrast in feelings that being a part of something, being included, can generate and the feelings that can manifest when individuals feel excluded, all of which seem to be illustrated on social media.

The first main theme of ‘The Need to Belong (Inclusivity/Exclusivity)’ consists of the sub-themes ‘Stimulation To Do Something’, which describes the use of social media for motivation or inspiration to try something new; ‘Constant Comparisons’, which encompasses the continuous access to social media and hence a perpetual exposure to appearance-related comparisons online; and ‘Meeting Expectations’ which relates to the feelings of pressure to adhere to an explicit set of regulations in order to belong.

The second main theme is ‘Health as Art’, which attempts to encompass the notion that being healthy is projected in a certain way online, and comprises the subthemes ‘What Healthy Bodies Should Look Like’ which describes the beliefs presented on social media regarding what a healthy body looks like visually as well as what a healthy body should be consuming (eating) and doing (exercising); ‘Cult of Social Media’ which encompasses the echo chamber effects of social media use, and how businesses and individuals advertise themselves and their beliefs online; and ‘Illusory Ideal’ which relates to the artistry that goes into the alluring, edited images of food, bodies and health to which participants are exposed on Instagram.

The third main theme is ‘Craving’, which attempts to cover both the literal cravings for food that participants reported and a metaphorical longing, and includes the sub-themes ‘Procrastination’, which entails using social media to avoid doing specific tasks; ‘Escape’, which involves the use of social media to break free of the real world and engross oneself in the online domain; and ‘Food Porn’, which involves the glamourised visual presentation of food.

The following includes direct quotations from the online responses received, some of which included grammatical errors.

The first main theme attempts to include both how belonging can contribute to improved wellbeing and how being excluded can generate negative emotions. Whilst many responses shared the positive effects of social media, and how it had been helpful for them to gain new information and ideas with regard to their eating and health, there were also responses indicating negative effects when participants were unable to meet the expectations raised or to emulate what others appeared to be doing online. The following response outlines an effect of social media use on a participant’s eating, and how seeing others engage in exercise and healthy eating inspires them but leaves them feeling low when they cannot do the same: ‘Yes it does affect my eating. Seeing other people exercise and eating healthy motivates me to do so, however due to the lifestyle and busy schedule I have, I can’t keep up with being healthy which then results in me feeling guilty, stressed and depressed.’

The second main theme of ‘Health as Art’ attempted to encompass the notion that healthy appears in a certain way online, as participants elaborated on how health has a specific appearance, either to make it look appealing to the consumer, like an advertisement, or to adhere to the societal norms of what healthy and attractive food or bodies look like, with an emphasis on food being ‘Instagrammable’. The following response offers evidence of this as it proposes that Instagram has a greater effect on their eating patterns, encouraging them to view content by strangers whom they are more likely to admire rather than that of their friends: ‘Instagram is definitely the biggest one as a large proportion of people I follow on their I am not personally connected to. They are more figures to “idolise” or attempt to emulate as to gain a physique like theirs etc. Facebook on the other hand is basically following your friends who are far less likely to “inspire” or “motivate”.’

Another participant alludes to the counterfeit content they encounter on specific social media channels, and how it adheres to the thin and athletic ideals that have been related to healthism: ‘Social media, such as Instagram & Facebook, that provoke the users to upload photographs of daily activities, such as eating, can sometimes be used to fool others of a false & unhealthy diet but being able to remain skinny & fit.’ Further references to ‘Unrealistic body images’ and ‘Seeing people with abs as being normal’ were included in the responses by participants who show an awareness of deceptive online content, but not necessarily an immunity to the effects of exposure to such content.

Finally the third main theme of ‘Craving’ included both the literal cravings for food that participants reported and a metaphorical longing. Individuals seem to be constantly bombarded with images of food on social media as well as traditional media. Cravings for food can therefore be triggered by these food cues in the environment. Numerous responses offered evidence to support the effect of food cues on social media, by explaining that looking at images of food and videos of food being made or eaten created cravings for that specific food which could be healthy or unhealthy. The actual cravings that participants reported were evident in extracts like the following one, in which the participant reports that social media affects their eating patterns solely when they are exposed to what they deem as desirable food: ‘The only way that it does affect my eating is if I see a picture of some yummy food (a bowl of pasta for example), I then have a craving for that!’ Another participant added that being exposed to food imagery could lead to them thinking about that specific food throughout the day, and sometimes acting on their cravings: ‘Yes, if I see doughnuts on my newsfeed I think of them all day. Sometimes I go and buy some, but not always.’

Previous findings regarding the link between age and ON have been inconclusive, with the majority of studies reporting either a link between ON and younger age, or no relation between the two. More research was needed to understand the link between age and ON, and specifically research employing older adults, as previous research was predominantly focussed on participants between the ages of 20 and 30. The current study benefitted from a wide range age range, between 20 and 70. However the majority of participants were Millennials, aged between 24 and 38. Findings regarding age and its link to ON scores suggested that younger participants, specifically those in Generation Z and Millennials, were more likely to display orthorexic traits than older participants from Generation X and Boomers. Responses to the probing questions also suggested that age might be a protective factor in the effects of social media on eating behaviours, which seems to support the quantitative findings. Participants mentioned an increased sense of self as a protective factor, suggesting that this may develop with age.

Findings suggested a higher likelihood of ON in females. The average scores for both genders were below 40 on the ORTO-15 scale, and therefore lay within the orthorexic range, which may be an indication of the high prevalence found in the study. The variance of ON scores was greater for males than females, suggesting that the male participants in the study had a wider range of ON scores than the females. This could imply that some males are very concerned about healthy eating whereas others have a minimal interest.

Although the prevalence of HeOr and OrNe in the sample was not computed due to the absence of cut-off points, it was apparent that participants’ scores in both categories were relatively high, which matched the 72.63% prevalence found with the ORTO-15 measure. Barrada & Roncero outline a possible range of scores for HeOr (0, 27) and add that the mean score for HeOr in their study was quite low at M=12.52 ⁴. Their participants also tested quite low for OrNe, with M=3.57 and the possible range is (0, 24). In the current study the average HeOr was M=21.59 which was significantly more than that in the TOS study ⁴. This may suggest that the sample had what has been deemed as a healthy preoccupation with healthy eating without the adverse effects of ON. The OrNe in the current study was M=13.21 which again was much higher than the average of OrNe in the TOS study ⁴.

The high prevalence in the current study may be due to various factors. The sample was predominantly made up of Millennials (78.2%), who according to the current study seem to be more prone to ON than other age groups. More than half the sample was made up of females, 66%, who according to findings are again more likely to display orthorexic traits than males. The prevalence may have also been affected by the recruitment method or by response bias, as individuals who agreed to take part in the study may have done so because of an interest in healthy eating.

Other studies recognize that in the current environment ON behaviours can often be seen as socially acceptable and even commendable ^{9, 10}. Healthy lifestyle choices are currently praised and therefore it may be difficult for both the measures to distinguish between following a currently fashionable diet and pathological eating patterns. The high average of HeOr in the sample may suggest the former, but the high ORTO-15 prevalence and high OrNe scores suggest the latter. Brytek-Matera et al. explain that the appeal of healthy eating practices could generate an overestimation of ON behaviours in self-reporting measures, which could further explain the high prevalence in the current study ⁹.

The argument that ORTO-15 over-pathologises what could be deemed normative eating behaviour and detects individuals who are simply on diets does not explain the high scores generated for HeOr and OrNe ^{11, 12}. The TOS scale was developed in order to quantify both non-clinical healthy eating and pathological eating behaviours. Hence, if the sample were saturated with individuals on healthy diets who did not experience any of the detrimental effects of dieting, we would have expected high HeOr scores and low OrNe. Even so, the high prevalence found in the current sample raises the question whether ON could truly be deemed as pathological if it is highly present in a non-clinical population.

The findings from the reliability analysis suggested that the TOS measure has high internal consistency and seems to measure what it is supposed to do. Nevertheless, statistically significant differences between the ON scores of different demographics and between Instagram users and non-Instagram users were only apparent when using the ORTO-15 and not when using the TOS scale. This may be explained by the greater variance in ORTO-15 scores as the questions posed on the measure may have generated more polarised responses than the TOS. This may have been due to the different format of the ORTO-15 and TOS questions.

The high prevalence in the sample could also be explained by the large proportion of social media users, and specifically users of Instagram (29.67%), in the sample, with only 3.46% of the sample not using any social media platform. The content depicted on social media, and described in the probing questions, seems to have a direct effect on participants’ view of themselves, their concern with health, their body image and food intake. Participants reported an abundance of health-promoting content online which encouraged them to eat healthy foods and to exercise in order to adhere to the socio-cultural ideals of health promoted online. Quantitative findings confirmed a link between social media and eating patterns, suggesting that individuals who accessed social media more frequently were more likely to describe their interest in healthy food as an important part of themselves and the way they understand the world.

Further results offered evidence to support Turner and Lefevre’s findings of a link between Instagram use and ON, as Instagram users in the sample were more likely be within the orthorexic range than non-Instagram users ⁵. This was further supported in the qualitative segment, as participants described this specific platform as having a greater influence due to its visual format.

Most studies examining the effect of social media on eating patterns have related adverse eating behaviours to online appearance-based comparisons and body image issues. These were noticeable in the current study as participants described idealised images of unrealistic body types adhering to the slim and toned physique that denotes healthism ¹³. The high prevalence of ON in the sample and the responses of participants who report appearance-based comparisons suggest that weight and shape concerns may in fact be a factor in ON. Whilst Bratman previously highlighted the lack of weight and shape concerns in ON, the latest proposed diagnostic criteria take body image into consideration ¹⁴.

Although many participants referred to idealised images of unrealistic bodies and constant image-based comparisons, this was not the only type of content that affected their eating patterns. Preoccupation with food has been related to eating disorders and ON and the rise of social media use and mobile access has contributed to increased exposure to images of food, recipes and cooking channels ¹. The online popularity of food is clear, as the hashtag ‘#food’ is one of the 25 most popular hashtags on Instagram and was the second most popular content type on participants’ feeds. Studies have indicated that social media could be used to influence actual food selection and participants reported cravings for specific foods when being exposed to this content online ¹⁵. Participants added that they were likely to act upon these cravings, but described these behaviours, especially when the cravings were for what they deemed as unhealthy foods, as negative and guilt-inducing.

Findings also suggest that individuals adhering to strict and restrictive diets in pursuit of health may be more prone to expose themselves to this content, as is evident in other eating disorders. Acting upon the cravings induced by food-related online content could lead to negative emotions. This interpretation of cravings and the outcomes of acting upon them could suggest that ON is similar to other eating disorders, and supports the transdiagnostic nature of eating disorders. It could be suggested that the combination of unrealistic body images and the glamourisation of food on social media, specifically Instagram, generates an environment that promotes detrimental food thoughts and behaviours outlined in new ON criteria.

The study profited from a diverse sample population in regard to the age, gender, level of education and country of residence of participants; this was not present in prior research, as most previous studies regarding ON were limited to specific age groups, professions or student groups and cultures. One of the study’s strengths was the large sample size, which was a result of the online survey employed. An online survey questionnaire was considered as the most effective method, given the study’s methodology and epistemological positioning as well as its time and cost restraints.

However, online surveys are not without their shortcomings. As the setting of this method is uncontrolled, participants are able to complete the measure wherever they wish. Participants may have been able to complete the questionnaire without any interruptions or distractions, or may have filled it out whilst preoccupied with other tasks.

Self-selection bias in opportunity samples is also unavoidable, as participants who are interested in the topic of healthy eating or social media use were more likely to take part in the study. This may have resulted in a sample that was more concerned about healthy eating than the average population, and could explain the high prevalence of ON in the sample.

Participants may have not completed the whole questionnaire as it may have taken them longer to complete than expected, they may have been interrupted or they may not have felt comfortable answering certain questions. In the current study, the TOS measure was presented first, followed by the ORTO-15 scale, followed by a social media use measure and finally the probing questions. Participants had to complete two measures on quite similar content. This repetition may have been off-putting to participants and may have further contributed to order effects. Future studies could control for potential order effect by randomising the order of the measures for each participant.

Responses regarding the probing questions were of particular concern, as many of the themes were related to the thoughts, emotions and behaviours outlined in the ON measures. It is uncertain whether the questions provided theoretical support for the measures or were simply a result of recency and priming effects. The study could possibly have benefitted from fewer probing questions, which may have encouraged participants to elaborate on just one aspect of the link between eating patterns and social media use.

The findings from the current study regarding the link between age and ON suggest that younger participants are more prone to ON than older participants. Further studies could attempt to assess whether these findings reflected an existing inclination for healthy eating amongst younger individuals who tend to be more susceptible to online trends.

The notion that the ORTO-15 detects and over-pathologises individuals who are on diets is refuted by the high HeOr and OrNe scores in the sample ¹². Further studies could employ both measures, in order to successfully distinguish between the proportion of individuals pathologised as ON sufferers by ORTO-15 and individuals who are simply concerned with healthy eating without the detrimental effects of ON. Although results indicated that the TOS is a valid and reliable measure, it needs to be used in more studies in order further to quantify its ability to generate repeatable and consistent outcomes in different samples. Despite the current study’s attempts to make up for the shortcomings of the lack of official diagnostic criteria and an official ON measure by utilising two measures and qualitative analysis, ON research and the current study have been limited due to this.

Furthermore, previous studies and the responses from participants in the current study all seemed to agree with the increased influence of the visual format of Instagram. Further work could explore possible effects of varying Instagram content, in order to provide further insight into whether images, videos, IGTV or live feeds are more influential. Finally, the findings regarding food content and cravings suggest that there is an abundance of food cues online that appear to affect individuals’ real-life food choices. The combination of unrealistic body images and the online glamourisation of food may be contributing to an environment that promotes detrimental food thoughts and behaviours. Studies examining the variety of content online, not solely focussed on fitspiration or the effect on body image, may be beneficial to gain a deeper understanding of social media’s contribution to ON.

1. 1.Bratman S. (1997) Orthorexia Nervosa.Yoga Journal: Retrieved from http://www.orthorexia.com/original-orthorexia-essay/. 42-50.
   Search at Google Scholar

1. 2.Bratman S, Knight D. (2000) Health food junkies.New York, NY:Broadway Books.
   Search at Google Scholar

1. 3.Donini L, Marsili D, Graziani M, Imbriale M, Cannella C. (2004) Orthorexia Nervosa: a preliminary study with a proposal for diagnosis and an attempt to measure the dimension of the phenomenon.Eating and Weight DisordersST 9:. 151-157.
   View article·Search at Google Scholar

1. 4.J R Barrada, Roncero M. (2018) Bidimensional structure of the orthorexia: Development and initial validation of a new instrument.Anales de Psicología. 34(2), 283-291.
   Search at Google Scholar

1. 5.P G Turner, C E Lefevre. (2017) Instagram use is linked to increased symptoms of orthorexia nervosa.Eating and Weight Disorders. 22(2), 277-284.
   View article·Search at Google Scholar

1. 6.Donini L, Marsili D, Graziani M, Imbriale M, Cannella C. (2005) Orthorexia Nervosa: Validation of a diagnosis questionnaire.Eating and Weight Disorders—Studies on Anorexia. , Bulimia and Obesity 10(2), 28-32.
   Search at Google Scholar

1. 7.Braun V, Clarke V. (2006) Using thematic analysis in psychology.Qualitative research in psychology.3. 77-101.
   View article·Search at Google Scholar

1. 8.R F Baumeister, M R Leary. (1995) The need to belong: Desire for interpersonal attachments as a fundamental human motivation.Psychological. , Bulletin 117(3), 497-529.
   Search at Google Scholar

1. 9.Brytek-Matera A, L M Donini, Krupa M, Poggiogalle E, Hay P. (2015) Orthorexia Nervosa and self-attitudinal aspects of body image in female and male university students.Journal of Eating Disorders. 3(2), 1-8.
   View article·Search at Google Scholar

1. 10.Segura-Garcia C, Ramacciotti C, Rania M, Aloi M, Caroleo M et al. (2015) The prevalence of Orthorexia Nervosa among eating disorder patients after treatment.Eating and Weight Disorders. 20, 161-166.
   View article·Search at Google Scholar

1. 11.Missbach B, T M Dunn, J S Koenig. (2016) We need new tools to assess Orthorexia Nervosa. A commentary on “Prevalence of Orthorexia Nervosa among college students based on Bratman's Test and associated tendencies”.Appetite,108. 521-524.
   Search at Google Scholar

1. 12.Roncero M, J R Barrada, Perpiñá C. (2017) Measuring orthorexia nervosa: Psychometric limitations of the ORTO-15. , The Spanish Journal of Psychology 20(41), 1-9.
   View article·PubMed·Search at Google Scholar

1. 13.Crawford R. (2004) Risk ritual and the management of control and anxiety in medical culture.Health:. , An Interdisciplinary Journal for the Social Study of Health, Illness and Medicine 10(4), 401-420.
   PubMed·View article·Search at Google Scholar

1. 14.Dunn T, Bratman S. (2016) On orthorexia nervosa: A review of the literature and proposed diagnostic criteria.Eating. , Behaviours 21, 11-17.
   PubMed·View article·Search at Google Scholar

1. 15.The Hartman Group (2012) Clicks and cravings: the impact of social technology on food culture. http://www.unitedfreshshow. org/files/clicks_and_craving.june_lee.pdf
   Search at Google Scholar