Collaborative authorship, listed alphabetically by surname.

Abstract

Social media and photo editing tools are becoming more prevalent. These tools have become helpful for connecting; however, they may also be linked to the development of anorexia nervosa (AN). This review analyses how social media and photo editing tools are associated with increased risk of anorexia nervosa in adolescent females between the ages of 10 and 19. Our review primarily looks at research from PubMed, which was then supplemented by PMC and Dergipark to aid our initial findings. We utilised the advanced search on all of these websites, inputting keywords like ‘anorexia nervosa’, ‘females’, ‘age between 10 and 19’, ‘social media’ and ‘photo editing’. While searching PubMed, we realised there were not many studies about our research topic, so we decided to pivot to other websites. In our search, we found that both social media and photo editing apps were consistently associated with AN. Results showed that female adolescents with excessive exposure to social media had poor eating habits, body discontent, depression and poor sleep. Photo editing on social media showed females having exacerbated body dissatisfaction and altered eating behaviours. Video content that shows unrealistic body types on social media was associated with AN and greater use of photo editing apps, which was linked to greater symptom severity. Because of potential recruitment bias and location limitations, as the studies were performed in Western Europe, future research that takes place outside Western Europe and with unbiased recruitment is recommended.

Introduction

Adolescence is a period of psychological, social and neurological development during which beliefs of identity, appearance and self-worth are continually shaped. Social media platforms such as TikTok, Instagram and Snapchat, which are image-based, have become an integral part of teenage life. They are the primary space where social development happens, and they provide continuous exposure to images of peers, influencers and idealised appearances. These media also enable adolescents to actively change their own appearances through filters and photo editing tools. Consequently, adolescents are exposed not only to idealised representations of others but also to opportunities to digitally alter their own appearance before posting it. A recent review of examining photo manipulation and anorexia nervosa (AN) in children and adolescents found associations between frequent selfie sharing and photo manipulation, lower weight satisfaction and eating disorder symptoms (Bozzola et al., 2024).

This issue is specifically significant during adolescence because the brain is undergoing substantial structural and functional development, and this could therefore affect the rest of their adult life. Research on adolescent AN suggests that reward and punishment processes may play a role in the development and maintenance of eating disorder psychopathology (Timko et al., 2019). Therefore, repeated exposure to appearance-focused content and social feedback may interact with development, especially in adolescence, where teens are particularly responsive to social influences.

Social media may influence eating disorder risk through several interconnected mechanisms. Exposure to idealised or digitally altered appearances can promote appearance comparison and integrate unrealistic body standards, and this can contribute to body dissatisfaction and disordered eating cognitions. Photo editing tools may intensify this process by allowing users to alter their own appearance and potentially increase self-surveillance and concern about how they are seen by others (Bozzola et al., 2024). Furthermore, social media exposure is not necessarily passive. Algorithmic recommendation systems can personalise the content users encounter based on their earlier interactions. Recent research analysing more than one million TikTok videos found that algorithms belonging to users with eating disorders showed more appearance-orientated, dieting, exercise and eating disorder-related content than those of healthy controls. Greater algorithmic bias towards such content was also associated with greater eating disorder symptoms (Griffiths et al., 2024).

Importantly, this relationship should not be interpreted as evidence that social media directly causes AN. AN is a multifactorial psychiatric disorder involving biological, psychological and environmental factors. Evidence from adolescent girls diagnosed with AN shows that factors such as perceived loss of control, emotional difficulties, body satisfaction and patterns of social media engagement may interact with eating disorders. Further research proves that adolescents with eating disorders may experience social media content differently depending on their existing vulnerabilities (Sjostrom et al., 2024). This suggests that social media may function not as a single independent cause but as an environmental factor capable of reinforcing existing concerns of disordered eating cognitions.

The public health significance of this issue is considerable because AN is a serious psychiatric disorder that often appears during adolescence and can produce considerable psychological, neurological and physiological consequences. Understanding potentially modifiable environmental influences during this developmental period may therefore contribute to earlier identification and prevention. This paper, therefore, investigates the question: how does exposure to and use of image-based social media platforms, specifically photo editing and filter features on Instagram, TikTok and Snapchat, influence body dissatisfaction, disordered eating cognitions and anorexia nervosa or symptom severity in adolescents aged 10-19?

Methodology

This review restricts “literature” to peer-reviewed, published research, including primary empirical studies (cross-sectional, qualitative and clinical) and existing reviews that report on the relationship between image-based social media use and anorexia nervosa in adolescent females.

This review identifies “social media” as web-based platforms that allow users to create profiles, build networks and share content (Tao et al., 2016). This review focuses on image-based social media platforms centred around photographic and/or video content (Instagram, TikTok and Snapchat) where the main focus is on appearance-editing features (filters, photo manipulation and selfie sharing). This restriction is operationally identified across the literature as distinct from general platform use (Bozzola et al., 2024; Praet et al., 2024).

Anorexia nervosa is defined per DSM-5 criteria (American Psychiatric Association, 2013) as the restriction of energy intake leading to significantly low body weight, intense fear of weight gain and disturbed self-perception of body weight or shape. 

The term “adolescence” is defined by the World Health Organisation (WHO) as the period between ages 10 and 19 years. This review adopted that standard definition directly. All the studies included fell within this range. The term “Western” refers to the sociocultural association with Western media and beauty ideals, focusing on adolescents affected by similar media influences across continents. Under this definition, Turkey meets the inclusion criteria: its adolescents’ social media use is centred around the same image-based platforms as the review’s other literature, and the adolescents’ increased appearance consciousness and body image patterns resemble those observed in the review’s literature on Western European populations (Duygu et al., 2026). Countries with substantially different media environments and culturally specific beauty standards not aligned with Western beauty norms (e.g., East Asian or Sub-Saharan African contexts) were excluded from this broadened definition. 

Literature was identified through the search engines PubMed and Google Scholar. Publications were restricted to 2019-2026, corresponding to the period of rapid social media growth. The PubMed search string combined MeSH and free-text terms (see Supplementary File for the full search string used), filtered using PubMed’s built-in limits to: publication dates 2019-2026; age range 10-19 years; female sex; free full-text availability; English language; and article types restricted to clinical trial, meta-analysis, randomised controlled trial, review and systematic review. This resulted in one piece of literature that met the inclusion criteria. The remaining eight studies were identified through Google Scholar using key terms to capture core concepts (see Supplementary File for key terms).  

Across both databases, 7277 records were identified. After removing duplicates, 978 were screened by title/abstract, and 31 full-text articles were assessed against the inclusion criteria (Table 1). Eight studies met the inclusion criteria and were retained for synthesis in this review (see Supplementary File for names of studies). Following a team discussion, Logrieco et al. (2021) were included despite not meeting the initial inclusion criteria, as the literature demonstrated algorithmic exposure to ostensibly protective content not captured in previously reviewed literature. A ninth study was retained separately despite not meeting the exposure criterion for introductory/epidemiological content only.

Table 1. Inclusion and Exclusion Criteria.

Results and Synthesis

Table 2. Image-Based Social Media, Photo Editing/Filters and Anorexia Nervosa in Adolescent Females.

Across the ten included studies, social media was consistently associated with dimensions of adolescent anorexia nervosa, but the evidence did not support a simple relationship between overall use and illness. Instead, findings converged around four interacting features of social media engagement: appearance-focused practices; patterns of use; algorithmically mediated content exposure; and psychological vulnerability. This distinction is important because the potential relevance of social media appeared to depend less on exposure itself than on what adolescents did online, the content subsequently encountered and the psychological characteristics with which they engaged with it.

Photo Editing, Filters and Appearance-Focused Engagement

The most consistent behavioural pattern concerned appearance management and visual self-presentation. Bozzola et al. (2024) identified selfie-sharing and photo manipulation as behaviours associated with increased eating-disorder risk, while Praet et al. (2024) reported that adolescent girls with eating disorders experienced photo-manipulation tools as exacerbating body dissatisfaction and altering eating behaviours. Although these studies differ in design and evidential scope, their convergence is notable because both implicate the active construction and evaluation of appearance rather than social media exposure in the abstract. The clinical-profile study provides a complementary perspective, linking prolonged, predominantly passive social network use with poorer eating, body-image, depressive and sleep symptoms among adolescent girls with AN. Read together, these findings suggest that active and passive engagement may represent distinct but potentially complementary pathways. Image modification may intensify appearance monitoring, whereas passive consumption may facilitate recurrent comparison with idealised representations. Nevertheless, the evidence remains predominantly observational, precluding conclusions about whether these behaviours precede or follow the development of AN.

Platform-Specific Effects and Algorithmic Exposure

The TikTok studies extend this behavioural account by demonstrating that the content adolescents encounter may itself be algorithmically patterned. Griffiths et al. (2024) found substantially greater delivery of appearance, dieting and pro-ana content to eating-disorder accounts than to control accounts, indicating that recommendation systems can generate markedly different content environments for users. This finding is qualified by Sjöström et al. (2024), whose interviews with adolescents recovering from AN revealed that TikTok could provide both social connection and harmful exposure. The coexistence of these experiences suggests that platform effects cannot be reduced to a uniformly detrimental or beneficial influence.

Logrieco et al. (2021) further complicate the distinction between harmful and protective content. In a single adolescent case, exposure to anti-pro-ana material was followed by imitation of behaviours depicted in the videos. Although this case cannot establish a general effect, it illustrates how ostensibly preventive content may be interpreted differently by vulnerable users. Taken together, the TikTok evidence shifts attention from content categories alone toward the interaction between algorithmic delivery, content interpretation, engagement and individual susceptibility.

Social Media Use, Clinical Severity and Psychological Vulnerability

Evidence linking social media use with clinical severity similarly challenges a simple dose-response interpretation. In the study of 131 adolescent girls with AN, prolonged use, particularly five or more hours daily, was associated with poorer eating, body-image, depressive and sleep outcomes. Jarvers et al. (2023) likewise identified increased social media use during the COVID-19 pandemic as a contextual factor in adolescents with AN. However, their finding that perceived loss of control and alexithymia explained much of the variance in symptom severity suggests that the quality and regulation of engagement may be more clinically informative than duration alone.

Duygu et al. (2026) strengthen this interpretation by identifying higher social media use, appearance consciousness and fear of dislike among adolescents with AN compared with controls, alongside poorer body image in the presence of comorbid social anxiety disorder and major depressive disorder. Considered alongside Jarvers et al. (2023), these findings position social media within a broader constellation of interpersonal and psychological vulnerabilities. They do not establish that such vulnerabilities are caused by social media; rather, they indicate that individual characteristics may shape adolescents’ responses to online environments and may help account for heterogeneity in clinical outcomes.

Overall Synthesis

Taken together, the evidence suggests that social media exposure may not independently produce AN. However, particular forms of engagement may coincide with or reinforce existing appearance concerns and eating-disorder-related difficulties. Appearance-focused self-presentation, passive or prolonged consumption, algorithmically amplified dieting and appearance content, and psychological vulnerabilities appear to converge around processes of appearance evaluation, comparison and behavioural reinforcement.

The epidemiological findings of van Eeden et al. (2023), documenting a marked increase in AN incidence among girls aged 10-14, provide important population-level context but cannot attribute this increase to social media, and its proposed role remains a hypothesis. Interpretation is also constrained by the methodological characteristics of the included studies, including small samples, cross-sectional designs, qualitative evidence and case-level data, which limit causal inference and raise the possibility of reverse or bidirectional associations. Nevertheless, the convergence across methodologies supports a cautious but coherent conclusion. Social media is best understood as a potentially influential environment within the broader ecology of adolescent AN, whose significance depends on the interaction between platform characteristics, patterns of engagement and pre-existing psychological vulnerability.

Discussion

This paper explored how body-image-based content on various social media platforms influences the development of AN in adolescents. Across the included studies, a consistent pattern emerges wherein social media did not operate as a uniform determinant but as a milieu whose influence varies based on how adolescents engaged with it, what content they encountered and the psychological vulnerabilities they brought to that engagement. This demarcation reframes the central question of discussion: rather than isolating the manner in which social media affects the development of AN in adolescents, the evidence raises the question of which specific patterns of engagement interact with an already sensitive developmental period to sustain or intensify illness. 

This foundational vulnerability of AN may explain why active, appearance-editing behaviours carried a more consistent signal than passive exposure alone, extending the distinction established above. Bozzola et al. (2024) linked selfie-sharing and photo manipulation directly to elevated eating-disorder risk, which is consistent with Praet et al. (2024), who found that adolescent girls with eating disorders described photo-editing tools as actively worsening body dissatisfaction, with participants noting that seeing an edited image did not immediately register as unnatural: “You don’t immediately think, ‘That could also be photoshopped,'” (Praet et al., 2024). This suggests the mechanism is not simple exposure but a breakdown in reality-testing around appearance, occurring precisely when adolescents are forming a stable sense of self. Notably, Bozzola et al. (2024) also found a higher-risk subgroup among male adolescents, an unexpected finding that falls outside this review’s female-specific scope but signals a population that current research, including this review, may be underserving.

The evidence from TikTok extends this picture from individual behaviour to platform design. Griffiths et al. (2024) demonstrated that recommendation algorithms delivered considerably more appearance, dieting and pro-ana content to accounts associated with eating disorders, regardless of active user engagement. Praet et al.’s (2024) participant’s description of this dynamic: “I keep looking for the positive, but then the negative always comes back”. A feedback loop emerges, such as adolescent engagement shaping algorithmic delivery and algorithmic delivery reinforcing engagement. Clinically, this reframes the online environment as a potential maintaining factor rather than only a triggering one, a distinction reinforced by Sjöström et al. (2024), who found that TikTok use during AN recovery carried both protective and harmful effects depending on individual regulation. Logrieco et al. (2021) added a further complication: in a single documented case, exposure to ostensibly protective “anti-pro-ana” content was followed by imitation of the harmful behaviour it depicted, suggesting that content-based interventions cannot assume awareness content is inherently safe for a vulnerable viewer.

This evidence base carries several interpretive limits, beginning with its geographic and cultural composition. The included studies were conducted almost entirely within Western European contexts – Italy (Bozzola et al., 2024; Logrieco et al., 2021); France (Sauve et al., 2024); Sweden (Sjöström et al., 2024); Germany (Jarvers et al., 2023); Belgium (Praet et al., 2024); and the Netherlands (van Eeden et al., 2023) — with Duygu et al. (2026) in Turkey offering the only partial departure from this pattern. Because appearance ideals and social media norms are themselves culturally shaped, findings grounded in Western thin-ideal standards may not generalise to adolescents in non-Western cultural contexts. Notably, the age bracket implicated across every one of these studies remains consistent (roughly 11 to 18 years, with van Eeden et al., 2023, specifically identifying ages 10 to 14 as the point of sharpest increase in AN incidence), suggesting that whatever developmental vulnerability underlies these patterns may not be purely culture-bound, even if the current evidence cannot confirm this beyond Western European regions.

Several further limitations qualify these findings. Most included studies that draw from small, clinical or treatment-seeking samples (Griffiths et al., 2024, n = 42; Duygu et al., 2026, n = 30; Praet et al., 2024, n = 7; Sjöström et al., 2024, n = 14; Logrieco et al., 2021, n = 1) recruited through existing eating-disorder services, a convenience-sampling pattern that likely excludes undiagnosed or untreated adolescents who may differ in severity or family support; several studies do not explicitly report their sampling method, which limits confident interpretation. Because most studies relied on self-report surveys or interviews, findings are also vulnerable to social desirability bias, as participants may have minimised or reframed disclosures about disordered eating. Compounding this, the cross-sectional design used by Jarvers et al. (2023), Duygu et al. (2026) and Sauve et al. (2024) means direction of effect cannot be established; it remains equally plausible that greater AN severity drives increased or more passive social media engagement, rather than the reverse. 

Collectively, these findings are directionally consistent with this review’s objective that appearance-editing features of image-based platforms represent a distinct and identifiable risk pathway, operating through mechanisms of distorted self-comparison and algorithmic reinforcement. Given the observational and predominantly cross-sectional nature of the current  literature, this suggests a plausible and coherent pattern rather than an established causal claim.

Future Directions and Public Health Recommendations

Our research includes multiple limitations, including the location where the studies took place, as well as potential bias in the recruitment process. To ensure a more accurate representation of AN in adolescents, future research should include studies that go beyond Western European regions (Bozzola et al., 2024; Sauvé et al., 2024; Sjöström et al., 2024; Jarvers et al., 2023; Praet et al., 2024; Logrieco et al., 2021; van Eeden et al., 2023) to conclude whether AN is prominent everywhere due to differences in cultural body standards. Consequently, another limitation can be observed in the sampling process wherein, in certain studies, participants are aware of their participation, which stands as a leading cause for social desirability bias. Particularly in interview-based studies (Praet et al., 2024; Sjöström et al., 2024) and self-report surveys (Duygu et al., 2026; Jarvers et al., 2023; Sauvé et al., 2024). Participants may be inclined to exclude the severity of their condition, further limiting the integrity of our research.

Public health responses should extend beyond individual awareness to platform-level accountability, given evidence that recommendation algorithms deliver disproportionately harmful content regardless of user engagement (Griffiths et al., 2024). School-based social media literacy education and parental guidance are recommended to help adolescents recognise triggering content (Bozzola et al., 2024; Praet et al., 2024). Social media use patterns should be integrated into routine adolescent AN clinical assessment rather than treated as peripheral (Bozzola et al., 2024; Duygu et al., 2026; Sauvé et al., 2024; Sjöström et al., 2024).

Conclusion

This review concludes that image-based content on social media platforms may be an influence on anorexia nervosa risk and symptom severity among adolescents. Across the studies, appearance-focused practices such as photo manipulation and selfie-sharing, passive or prolonged use, and algorithmically delivered appearance and dieting content were associated with body dissatisfaction, disordered eating behaviours and poorer clinical outcomes, particularly among psychologically vulnerable adolescents. However, the evidence does not establish that social media causes anorexia nervosa, as most studies were observational, qualitative or based on small samples. How adolescents engage with social media appears more important than usage alone, highlighting the need for longitudinal studies.

Bozzola, E., Scarpato, E., Caruso, C., Russo, R., Aversa, T., & Agostiniani, R. (2024). Photo editing and the risk of anorexia nervosa among children and adolescents. Italian Journal of Pediatrics, 50, 229.

Duygu, O. A., Tetik, K., Turan, S., Eray Camli, S., Ergun Ozdel, Z. G., & Mutlu, C. (2026). The relationship between social media use and body image: A comparative analysis on adolescents with anorexia nervosa. Journal of Uludağ University Medical Faculty, 52.

Griffiths, S., Harris, E. A., Whitehead, G., Angelopoulos, F., Stone, B., Grey, W., & Dennis, S. (2024). Does TikTok contribute to eating disorders? A comparison of the TikTok algorithms belonging to individuals with eating disorders versus healthy controls. Body Image, 51, 101807.

Jarvers, I., Ecker, A., Schleicher, D., Kandsperger, S., Otto, A., & Brunner, R. (2023). Anorexia nervosa during COVID-19: Loss of personal control and alexithymia as important contributors to symptomatology in adolescent girls. Journal of Eating Disorders, 11, 170.

Logrieco, G., Marchili, M. R., Roversi, M., & Villani, A. (2021). The paradox of Tik Tok anti-pro-anorexia videos: How social media can promote non-suicidal self-injury and anorexia. International Journal of Environmental Research and Public Health, 18(3), 1041.

Praet, N., Stevens, J., Casteels, K., & Toelen, J. (2024). The association of social media use and eating behaviour of Belgian adolescent girls diagnosed with anorexia nervosa—A qualitative approach. Children, 11(7), 822

Sauve, A., Silva, C., Costa, D., Da Fonseca, D., Givaudan, M., Charvin, I., & Bat-Pitault, F. (2024). Adolescent girls with anorexia nervosa: Clinical profiles in relation to their social networks use. European Eating Disorders Review, 33(3).

Sjöström, D. K., de Mendonça Lindström, T., Kapetanovic, S., & Claesdotter-Knutsson, E. (2024). Helpful or not? A qualitative study on female adolescents’ experience of TikTok when recovering from anorexia nervosa. International Journal of Eating Disorders, 57(11), pp. 2217–2227.

van Eeden, A. E., van Hoeken, D., Hendriksen, J. M. T., & Hoek, H. W. (2023). Increase in incidence of anorexia nervosa among 10- to 14-year-old girls: A nationwide study in the Netherlands over four decades. International Journal of Eating Disorders, 56(12), pp. 2295–2303.

PubMed search strings:

1. ((“Anorexia Nervosa”[Mesh] OR “Feeding and Eating Disorders”[Mesh] OR “body dissatisfaction”[tiab] OR “disordered eating”[tiab]) AND (“Social Media”[Mesh] OR “Instagram”[tiab] OR “TikTok”[tiab] OR “Snapchat”[tiab] OR “photo editing”[tiab] OR “photo manipulation”[tiab] OR “selfie*”[tiab] OR “filter*”[tiab]) AND (“Adolescent”[Mesh] OR adolescen*[tiab] OR teen*[tiab])).
2. ((“Anorexia Nervosa”[Mesh] OR “body image”[Mesh] OR “body dissatisfaction”[tiab]) AND (“Social Media”[Mesh] OR “photo editing”[tiab] OR “photo manipulation”[tiab] OR “Instagram”[tiab] OR “TikTok”[tiab] OR “filter*”[tiab] OR “selfie*”[tiab]) AND (Adolescent[Mesh] OR adolescen*[tiab])).

Google Scholar key terms: 

1. “Anorexia nervosa” (“social media” OR Instagram OR TikTok OR Snapchat OR “photo editing” OR “photo manipulation” OR selfie OR filters) (adolescen* OR teen* OR girls) (risk OR increase OR decrease OR predictor).
2. “Anorexia nervosa” “social media” adolescent (review OR “systematic review” OR “scoping review” OR meta-analysis).

The Eight Studies that met inclusion criteria: 

1. (Bozzola et al., 2024; Sauve et al., 2024; Sjöström et al., 2024; Jarvers et al., 2023; Praet et al., 2024; Griffiths et al., 2024; Duygu et al., 2026; Logrieco et al., 2021).