Introduction
Orthorexia nervosa (ON) is a proposed eating disorder characterized by an intense preoccupation with healthy eating and a fixation on food quality and purity.Reference Dunn and Bratman 1 Individuals with ON follow rigid dietary rules, often eliminating entire food groups deemed unhealthy from their diets and spending an excessive amount of time planning, obtaining, and preparing their food.Reference Donini, Barrada and Barthels 2 This level of restriction results in marked distress and psychosocial or medical impairments that disrupt overall functioning.Reference Donini, Barrada and Barthels 2 For instance, individuals with ON may experience significant anxiety in social situations that involve food and thus opt to avoid these events altogether.Reference Şentürk, Güler Şentürk, Erus, Geniş and Coşar 3 They may also distance themselves from family and friends who do not adhere to their self-imposed dietary rules, leading to further isolation.Reference Koven and Abry 4 , Reference McComb and Mills 5 While ON is gaining recognition in the literature and media, it has yet to be established as a disorder in the Diagnostic and Statistical Manual of Mental Disorders (DSM) or the International Classification of Diseases (ICD).
The occurrence of ON varies across populations: in the general population, the rate of ON may range from 1 to 57.6%,Reference Carpita, Nardi and Bonelli 6 , Reference Dunn, Gibbs, Whitney and Starosta 7 but in high-risk groups, such as athletes or health professionals, estimates range from 35 to 57.8%.Reference Varga, Dukay-Szabó, Túry and van Furth Eric 8 However, it must be noted that some of these studies measured ON symptoms with the ORTO-15, an instrument that has been increasingly criticized for its poor reliability and validity.Reference Barrada and Meule 9 , Reference Mitrofanova, Pummell, Martinelli and Petróczi 10 This measure fails to distinguish between pathological and non-pathological healthful eatingReference Dunn and Bratman 1 but continues to be used in ON research, resulting in questionable results. This issue, coupled with the overall lack of consensus on diagnostic criteria for ON, makes it difficult to determine the prevalence of this complex behavior.
One aspect that researchers can agree on regarding ON is that the internet may play a significant role in its development and/or maintenance.Reference Horovitz and Argyrides 11 Problematic usage of the internet (PUI)—also known as internet addiction—is an umbrella term referring to excessive engagement in online activities that results in significant psychosocial and occupational impairments.Reference Beard and Wolf 12 PUI encompasses a variety of behaviors, including gaming, shopping, gambling, social networking, engaging in cyberchondria (the repeated seeking of medical information to alleviate anxiety or distress) or pornography, streaming media, and cyberbullying.Reference Fineberg, Demetrovics and Stein 13 Internet addiction has been linked to obsessive-compulsiveReference Moretta and Buodo 14 and eating disorder symptoms,Reference Hinojo-Lucena, Aznar-Díaz, Cáceres-Reche, Trujillo-Torres and Romero-Rodríguez 15 , 16 which have significant theoretical overlap with ON.Reference Yılmaz, Karakuş, Tamam, Demirkol, Namlı and Yeşiloğlu 17 As such, PUI may also be associated with an ON risk.
Several studies have connected a high level of involvement with nutrition and fitness accounts on visual-based social media platforms (particularly Instagram) with ON symptoms.Reference Scheiber, Diehl and Karmasin 18 –Reference Yurtdaş-Depboylu, Kaner and Özçakal 20 An analysis of “healthy” recipes posted on Instagram found that most of them were unbalanced and did not meet nutritional guidelines.Reference Pozo, Ezan, Moubassat and Déchelotte 21 The link between social media use and ON may be mediated by the drive for thinness and muscularity,Reference Scheiber, Diehl and Karmasin 18 attitudes toward eating,Reference Unal and Kocatepe Avcı 22 and obsessive-compulsive symptoms.Reference Kuş, Tan, Doğan and Bas 23 Additionally, social media addiction has been positively correlated with ON symptoms.Reference Yurtdaş-Depboylu, Kaner and Özçakal 20 , Reference Yılmazel 24
There is a growing concern, however, about how other forms of PUI (outside of social media use) may affect the development and maintenance of eating disorders. Indeed, previous research has connected compulsive shopping,Reference Claes, Bijttebier, Mitchell, de Zwaan and Mueller 25 , Reference Harnish, Gump, Bridges, Slack and Rottschaefer 26 excessive pornography use,Reference Gewirtz-Meydan and Spivak-Lavi 27 , Reference Spivak-Lavi and Gewirtz-Meydan 28 and cyberbullying perpetrationReference Cheng, Chu and Ganson 29 with eating disorder symptoms. In the context of ON, 2 studies have examined its relationship with cyberchondria. Atsizata and Cangöl SögütReference Atsizata and Sögüt 30 found that among nurses, cyberchondria increased their risk of ON. However, in a sample of pregnant women, cyberchondria and ON levels were unrelated.Reference Yıldız, Gökmen, Güler and Demirhan 31 This relationship in a community sample is unclear.
Additionally, no studies have examined the associations between “global” internet use problems or other online activities and ON. These gaps in the literature are important to fill because 96% of US adults use the internet, 32 and various facets of internet usage overlap. For instance, a recent review supported the interaction of several digital hazards, including cyberbullying, exposure to pro-eating disorder content on social media, and online gaming, with preexisting risk factors (eg, thin-ideal internalization) to predict eating disorder symptoms and clinical manifestations of eating disorders.Reference Ioannidis and Chamberlain 33 In the case of ON, someone may excessively seek nutritional information on social media and engage with fitness accounts that post oversexualized content while leaving harmful comments or bullying other users. These aspects of PUI may have distinctive influences on orthorexic symptoms. As such, the current study aimed to determine the association between PUI (including gaming, shopping, social networking, cyberchondria, pornography, and cyberbullying) and ON tendency in a large, non-clinical sample of US adults. We also investigated demographic correlates of ON to add to the growing ON literature. Based on existing research, we hypothesized that PUI would predict ON risk, and a high engagement in social networking (particularly health-related content) and cyberchondria would be more strongly associated with ON tendency than other facets of PUI.
Methods
Participants
A total of 300 adults were recruited to complete an online survey about online behaviors and mental health via Prolific. Subjects had to be 18 years of age or older, currently residing in the United States, able to understand English, and capable of providing informed consent to participate. Participants were excluded if they were unable to understand or undertake study procedures.
Procedures
Subjects completed an online survey via REDCap. They were first required to review the Institutional Review Board (IRB)-approved informed consent, in which they could accept or deny participation in the survey. The survey asserted that all responses would be kept confidential and that no personally identifying information would be collected. Participants were compensated $12 for their participation. Data was collected on 01/09/2024.
Measures
The online survey collected demographic information (age, gender, race, etc.) and mental health history. Subjects were asked if they had ever been diagnosed with an eating disorder (including anorexia nervosa, bulimia nervosa, or binge eating disorder) or obsessive-compulsive disorder, as ON has several similarities to these conditions.Reference Yılmaz, Karakuş, Tamam, Demirkol, Namlı and Yeşiloğlu 17 Data on active substance and psychotropic medication use were also collected, with current nicotine or alcohol use considered as any use within the past 30 days.
To assess ON symptoms, all participants completed the ORTO-R. The ORTO-R is a revised version of the ORTO-15 that consists of 6 questions (selected from the original scale) that best reflect thoughts and behaviors associated with ON.Reference Rogoza and Donini 34 The new version of the scale was created based on a reassessment of the original data used to develop the ORTO-15, and it appears to be a more reliable toolReference Rogoza and Donini 34 that has been validated in several studies of diverse populations.Reference Rogoza and Donini 34 –Reference Vaccari, Cutino and Luisi 37 The 6 items assess the core features of ON, including an excessive preoccupation with healthy eating, a significant effort to eat healthily, food-related emotional distress, and impairments in social functioning.Reference Rogoza and Donini 34 The ORTO-R measures ON tendency in a dimensional manner, eliminating a categorical diagnosis. Subjects were asked to respond to each question on a scale of 1 = Never to 5 = Always (eg, Are your rigid and restrictive dietary choices conditioned by your worry about your health status?). Higher scores indicate more severe ON symptoms. The ORTO-R showed good reliability in the present study (α = 0.809).
Participants also completed the Internet Severity and Activities Addiction Questionnaire (ISAAQ-10), which consists of 2 components that evaluate the severity of internet use and engagement in various online activities, respectively. In the first component, subjects were asked to respond to each question on a scale of 0 = Not at all to 5 = All the time (eg, How often do you find yourself losing track of time while engaging in an internet-related activity?). Higher scores indicate greater problematic internet use, with Ioannidis et al.Reference Ioannidis, Tiego and Lutz 38 proposing a cutoff score of 30 out of 50 as indicating “high PUI.” The ISAAQ-10 demonstrated excellent reliability in the present study (α = 0.903). The second component (part B) prompted participants to indicate how often they engaged in certain online activities during the last 6 mo. These activities included browsing or surfing the web, online gaming with other players, online shopping, social networking, researching medical information (cyberchondria), engaging in cybersex or viewing pornography, streaming music or videos on a platform, and cyberbullying others. The same scale (0 = Not at all, 1 = Rarely, 2 = Occasionally, 3 = Frequently, 4 = Very often, 5 = All the time) was used. Histograms revealed that subjects’ responses to most of the questions in part B clustered around the middle values of the scale. Based on the normal curves, participants were divided into 2 groups (those who chose values greater than or less than the means) to ensure that adequate sample sizes were used in the analyses.
To explore potential expressions of ON-like behaviors on the internet, subjects were asked the following yes or no questions:
-
1. Do you often search for information about healthy foods (eg, research nutritional information of the food you eat, look for healthy recipes) on the internet?
-
2. Do you often eat while engaging in an internet-related behavior?
Statistical analysis
The data were analyzed using IBM SPSS Version 29.0. Between-group differences in continuous variables were tested using independent samples t-tests, analyses of variance (ANOVAs), or analyses of covariance (ANCOVAs) while controlling for covariates. Effect sizes were calculated in the form of Cohen’s d for t tests (0.2 = small, 0.5 = medium, 0.8 = large) and eta squared and partial eta squared for ANOVAs and ANCOVAs, respectively (0.01 = small, 0.06 = medium, 0.14 = large). Pearson correlation analyses were performed to assess the association between continuous variables. The robustness of significant correlations was probed by a multiple linear regression. A two-sided p-value of <0.05 was considered statistically significant in all analyses.
Ethics
The Institutional Review Board of the University of Chicago approved the study and the consent statement. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Results
A total of 300 adults participated in the study. Ten subjects did not complete the ORTO-R and were thus excluded from the analyses. As such, the final sample consisted of 290 adults (Mage = 37.40, SD = 12.61), including 155 women, 120 men, 9 transgender or non-binary individuals, and 6 people who did not report their gender.
Descriptive statistics of the sample are presented in Table 1. Most participants identified as white or Caucasian (69.2%), had a college degree (47.0%), were single (36.6%) or married or engaged (36.9%), identified as heterosexual (78.5%), and did not report active nicotine (82.0%) or psychotropic medication (76.0%) use. Respondents reported an average of 1.22 (SD = 1.49) diagnosed psychiatric conditions. The mean score on the ISAAQ-10 was 18.48 (SD = 9.29), and 12.8% of the sample met criteria for PUI.
Table 1. Descriptive Statistics for a Sample of 290 Adults
Subjects scored an average of 14.52 (SD = 4.95) on the ORTO-R, with women scoring higher than men (M = 15.10, SD = 4.93 versus M = 13.71, SD = 4.90, respectively; t(273) = −2.324, p = .021). Those who reported taking a psychotropic medication had higher ORTO-R scores than those in the comparison group (M = 15.86, SD = 5.47 versus M = 14.06, SD = 4.69, respectively; t(286) = −2.664, p = .008). Additionally, participants who had been diagnosed with an eating disorder (M = 20.83, SD = 3.97), scored higher on the ORTO-R than those who did not report such a diagnosis (M = 14.38, SD = 4.89; t(288) = −3.209, p = .001). ORTO-R scores did not differ between subjects based on all other demographic characteristics, including race, education, relationship status, sexual orientation, nicotine use, alcohol use, or a history of obsessive-compulsive disorder (see Table 2).
Table 2. Differences in Adults’ Orthorexia Nervosa Tendency According to Their Descriptive Statistics
Bolded characteristics indicate a statistically significant difference in variable distribution between groups (p < .05).
ORTO-R scores were higher in participants who did (M = 17.11, SD = 4.77) versus those who did not meet criteria for PUI (M = 14.16, SD = 4.91; F(1, 270) = 7.819, p = .006). Moreover, there were several significant findings in relation to online activities. First, subjects who shopped online “frequently to all the time” (M = 15.44, SD = 4.94) had higher ORTO-R scores than those who shopped online “occasionally to not at all” (M = 14.10, SD = 4.91; F(1, 278) = 5.210, p = .023). The same results were found regarding cyberchondria (M = 15.48, SD = 14.35 versus M = 14.35, SD = 5.06, respectively; F(1, 278) = 4.181, p = .042). Participants who often researched healthy food choices on the internet (M = 17.61, SD = 4.50) had higher ORTO-R scores than those in the comparison group (M = 13.04, SD = 4.47; F(1, 276) = 55.988, p < .001). This variable had the largest effect size (ηpReference Donini, Barrada and Barthels 2 = .169) on the dependent variable. Lastly, subjects who cyberbullied others to any extent (M = 17.05, SD = 4.31) scored higher on the ORTO-R than those who reported no cyberbullying (M = 14.37, SD = 4.95; F(1, 278) = 8.724, p = .003). There was no difference in ON tendency among participants with different rates of other online behaviors, including social networking (see Table 3).
Table 3. Differences in Adults’ Orthorexia Nervosa Tendency According to Online Behaviors While Controlling for Age and Gender
Bolded characteristics indicate a statistically significant difference in variable distribution between groups (p < .05).
Abbreviation: PUI, problematic usage of the internet
Age was negatively correlated with ISAAQ-10 scores (r = −.352, p < .001) and ORTO-R scores (r = −.155, p = .009), and ISAAQ-10 scores were positively correlated with ORTO-R scores (r = .411, p < .001). After controlling for age, gender (0 = male, 1 = female), and an eating disorder diagnosis (0 = no, 1 = yes) in a multiple linear regression, ISAAQ-10 scores were still associated with ORTO-R scores (t = 6.336, p < .001). While the overall utility of the model was significant (F(4, 260) = 16.505, p < .001, R2 = .203), age and gender did not significantly contribute to the model (see Table 4).
Table 4. Multiple Linear Regression Evaluating Predictors of Orthorexia Nervosa Tendency
Bolded characteristics are statistically significant predictors of the outcome variable (p < .05).
Abbreviation: ISAAQ-10, Internet Severity and Activities Addiction Questionnaire.
Discussion
This is the first study to investigate the relationship between PUI and various online activities and ON tendency. In our sample, the average ORTO-R score was 14.52. This is consistent with similar studies: Dell’Osso et al.Reference Dell’Osso, Cremone and Chiarantini 39 reported a mean ORTO-R score of 12.726 in a sample of college students, and in a combined clinical and non-clinical sample, Vaccari et al.Reference Vaccari, Cutino and Luisi 37 found an average ORTO-R score of 16.6. Additionally, we found that women had a higher ON tendency than men. While some studies did not detect gender differences in orthorexic tendencies,Reference Anastasiades and Argyrides 40 , Reference Bundros, Clifford, Silliman and Neyman Morris 41 our results are corroborated by other studies.Reference Atsizata and Sögüt 30 , Reference Dell’Osso, Cremone and Chiarantini 39 , Reference Hanras, Dorard and Boujut 42 The increased risk of ON among women may be explained by the similarities between anorexia and orthorexia nervosa,Reference Dell’Osso, Carpita and Muti 43 as women are more likely to be diagnosed with anorexia nervosa than men.Reference van Eeden, van Hoeken and Hoek 44 Indeed, we found that subjects with an eating disorder and those on psychotropic medications had higher ORTO-R scores than those in the comparison groups. These findings support the growing literature that ON may contain more features of an eating disorder than obsessive-compulsive disorderReference Cosh, Olson and Tully 45 , Reference Hallit, Azzi, Malaeb and Obeid 46 and may be comorbid with conditions that warrant medication usage.
While the mean ISAAQ-10 score in our sample (18.48) is slightly lower than what has been reported in the literature,Reference Ioannidis, Tiego and Lutz 38 the rate of PUI among our participants (12.8%) is the same as that in the global population.Reference Meng, Cheng and Li 47 Additionally, we found that as subjects’ ages increased, their risk of internet addiction and ON decreased. This is consistent with previous research.Reference Bóna, Erdész and Túry 48 , Reference Devine, Ogletree, Shah and Katz 49 ON tendency (measured via ORTO-R scores) was significantly higher in those with than those without PUI, even after controlling for age and gender. Moreover, while considering age, gender, and an eating disorder diagnosis as covariates, a multiple linear regression revealed that ISAAQ-10 scores were significant positive predictors of ORTO-R scores. The mechanisms of this association are unclear, but perhaps anxiety, compulsivity, or a need for control play a significant role, as one may control their diet to cope with other areas of their life that feel uncontrollable (eg, difficulties with regulating internet use). Nevertheless, our finding demonstrates that “global” internet use problems are associated with ON tendency, even after adjusting for potential confounders. This is a valuable contribution to the literature, which has predominantly focused on social media usage in relation to ON.
Interestingly, we did not detect a difference in ON symptoms between subjects who used social media at different frequencies. This supports the idea that ON risk may be more impacted by the type of content viewed on social media than the duration of use.Reference Usta Ulutaş and Okan Bakır 50 Indeed, we found that among all the online activities examined in our study, those who often used the internet to research the nutritional information of their food or look for healthy recipes had the highest ORTO-R scores. The effect size of performing this activity often (ηp Reference Donini, Barrada and Barthels 2 = .169) on ORTO-R scores was large, suggesting a strong association. However, the directionality of this relationship is unclear. Does exposure to clean eating-related content increase one’s risk of ON, or do people with ON symptoms seek out this information online? Future research adopting a longitudinal design is needed to get at this question.
In a similar vein, participants who engaged in high levels of cyberchondria had more ON symptoms than those in the comparison group. This result replicates that of Atsizata and Cangöl Sögüt,Reference Atsizata and Sögüt 30 who found that cyberchondria increased nurses’ risk of ON, in a large community sample of adults. This is concerning given that around 60% of U.S. adults use the internet to search for medical information,Reference Wang and Cohen 51 but most of it (including nutritional information) is incorrect and of poor quality.Reference Daraz, Morrow and Ponce 52 , Reference Denniss, Lindberg and McNaughton 53 As more and more people turn to the internet, which has a considerable amount of misinformation, for guidance about their health, the risk of ON may increase. While again, the directionality of this association is uncertain, it underscores a need for more research on cyberchondria and ON in this digital age.
Apart from excessive engagement with health-related content, our study found that subjects who shopped online “frequently or more” and those who cyberbullied others had higher ORTO-R scores than those in the comparison groups. This is in line with previous research that has suggested overlap between compulsive shoppingReference Claes, Bijttebier, Mitchell, de Zwaan and Mueller 25 , Reference Harnish, Gump, Bridges, Slack and Rottschaefer 26 and cyberbullying perpetrationReference Cheng, Chu and Ganson 29 with eating disorders. Indeed, one study found that low self-esteem, which predicts greater cyberbullying,Reference Brewer and Kerslake 54 also plays a role in the relationship between compulsive shopping and disordered eating.Reference De Pasquale, Morando and Platania 55 While low self-esteem is a theorized risk factor of ON,Reference Mahfoud, Pardini and Mróz 56 future research should investigate the mechanisms of the association between various facets of PUI and ON.
This study has some limitations. First, the ORTO-R measures ON symptom severity in a dimensional manner, so there is no cutoff score available to indicate ON. Thus, most of our analyses were limited to correlations, and the clinical significance of our results is unclear. Yet, since there are still no fixed diagnostic criteria for ON, our findings signify associations between internet addiction and ON within the current scope of research. Next, the cross-sectional design of this study limited our ability to make causal connections. In a previous longitudinal study, PUI was cross-sectionally correlated with but not longitudinally predictive of eating disorder symptoms when considering socio-contextual variables,Reference Ioannidis, Hook and Wiedemann 57 so we emphasize that our results do not suggest that PUI may cause ON. Additionally, all data were self-reported, which may be inaccurate due to social desirability and recall biases. Data regarding psychiatric diagnoses and PUI, in particular, would have been best verified by a clinician-administered diagnostic interview. Lastly, while Prolific has been shown to yield high-quality data,Reference Douglas, Ewell and Brauer 58 , Reference Eyal, David, Andrew, Zak and Ekaterina 59 participants on Prolific may be unrepresentative of the population. In our sample, we had an approximately equal division of genders, but the majority of our participants were Caucasian.
Conclusion
Our results indicate a higher prevalence of ON symptoms among women, individuals who have been diagnosed with an eating disorder, and those who take at least one psychotropic medication. Additionally, we reveal an association between broad internet use problems (including but not limited to social media use) and ON tendency. PUI was still a significant predictor of ON tendency after controlling for age, gender, and a history of an eating disorder. PUI in the form of excessive engagement with health-related content may drive this association; however, other online activities, including cyberbullying and compulsive shopping, may also be involved and warrant more attention in the literature. Further research is needed to establish the directionality and mechanisms of these associations.
Author contribution
Conceptualization: S.B.; Funding acquisition: J.E.G.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Disclosures
Dr. Grant has received research grants from Janssen and Biohaven Pharmaceuticals. Dr. Grant receives yearly compensation from Springer Publishing for acting as an Editor-in-Chief of the Journal of Gambling Studies and has received royalties from Oxford University Press, American Psychiatric Publishing, Inc., Norton Press, and McGraw Hill. Ms. Boutouis reports no conflicts.
Open access
