Discussion

This study examined the longitudinal relationships between symptoms of body dissatisfaction and disordered eating, self-harm and suicidal ideation in two cohorts, TEDS and COPING, revealing distinct patterns of stability and cross-domain associations over time. In the TEDS cohort, symptoms of disordered eating, self-harm and suicidal ideation demonstrated little continuity across adolescence and young adulthood, with peak severity observed around age 21 years and decline by age 26. Cross-domain associations indicated that self-harm and suicidal ideation at age 26 years were more strongly associated with symptoms of disordered eating at age 21 than vice versa. In contrast, the COPING cohort showed greater stability in symptoms over time but minimal evidence of cross-domain effects. In the TEDS cohort, symptoms of disordered eating were observed to peak at age 21 years, with a notable decline by age 26. This pattern suggests that, while disordered eating may be most pronounced in young adulthood, there is a tendency for symptoms to decrease as individuals move into the later stages of adulthood. Similar developmental trends were observed for symptoms of self-harm and suicidal ideation, which aligns with previous research indicating that self-harm may fluctuate or diminish as individuals age, ^{Reference Marzecki, Ahmadzadeh, Oginni, Pingault, McAdams and Zavos33} perhaps in part because of the decrease in symptoms of disordered eating. However, in TEDS, at age 16 years, the items primarily assessed body dissatisfaction, at age 21 they included a combination of body dissatisfaction and disordered eating symptoms and at age 26 they began to align more closely with DSM-5 criteria for eating disorders. As such, the peak in symptoms at age 21 years should be interpreted in light of the specific symptom profile measured at that time, rather than as evidence of a definitive peak in symptoms of disordered eating.

The finding that self-harm and suicidal ideation at age 21 years were much stronger predictors of disordered eating at age 26 than the reverse suggests a directional relationship in which self-injurious thoughts and behaviours play a more central role in the development or persistence of disordered eating, rather than vice versa. This could indicate that self-harm and suicidal ideation represent earlier manifestations of underlying psychological distress, which later evolve into disordered eating symptoms as individuals seek alternative methods of emotional regulation, stress reduction or self-punishment. ^{Reference Foye, Hazlett and Irving34,Reference Foye, Hazlett and Irving35} The markedly stronger association in this direction may also reflect the way in which self-harm functions as a coping mechanism for distress, with disordered eating behaviours emerging subsequently as either an extension of these maladaptive strategies or a more socially acceptable form of self-injurious behaviour. ^{Reference Foye, Kakar, McNamara, Musial, Jewell and Griffiths36} In contrast, the weaker effect of disordered eating on later self-harm or suicidal ideation suggests that, while these conditions often co-occur, eating pathology alone may not be as potent a risk factor for self-harm as self-harm is for disordered eating. Our findings align with the literature’s mixed results regarding the bidirectional relationship between disordered eating and suicidality. The observation that self-harm and suicidal ideation at age 21 years are strongly associated with disordered eating at age 26 emphasises the pathway where self-harm precedes disordered eating. ^{Reference Forrest, Zuromski, Dodd and Smith8,Reference Trujillo, Forrest, Claypool and Smith9}

The observed directional relationship, in which self-harm and suicidal ideation at age 21 years are linked to disordered eating symptoms at age 26, aligns with a hypothesis of the overlap between symptoms of disordered eating and self-injury. ^{Reference Washburn, Soto, Osorio and Slesinger37} This theory suggests that, for some individuals, disordered eating behaviours – such as restricting, bingeing, purging or excessive exercise – may serve as a method of self-harm rather than solely being driven by weight or shape concerns. ^{Reference Washburn, Soto, Osorio and Slesinger37,Reference Fox, Wang, Boccagno, Haynos, Kleiman and Hooley38} Additionally, research indicating that nearly a third of patients in intensive treatment for self-harm engage in disordered eating behaviours, as a form of self-injury, reinforces the hypothesis that self-harm can manifest in multiple ways. ^{Reference Pérez, Marco and Cañabate39} The higher clinical severity at baseline among individuals with self-injurious disordered eating behaviours mirrors findings from the TEDS cohort, where self-harm and suicidality at an earlier stage were associated with later eating disorder symptoms, suggesting that self-injurious behaviours may evolve over time into different forms of psychological distress. ^{Reference Washburn, Soto, Osorio and Slesinger37} The literature addressing the relationship between eating disorders and suicidality has been characterised by a lack of longitudinal studies. ^{Reference Smith, Zuromski and Dodd6} Most research in this area, including meta-analyses, has relied on cross-sectional and retrospective data, which cannot determine the directionality of the relationship between these two constructs. The only meta-analysis that has focused solely on longitudinal studies with suicide outcomes highlighted a substantial gap in this research, with only 14 longitudinal studies in the past 50 years exploring whether symptoms and diagnoses of eating disorders predict suicide attempts or death. ^{Reference Smith, Velkoff, Ribeiro and Franklin7} Notably, no studies have addressed suicidal ideation. One meta-analysis found that eating disorders were weak but significant predictors of suicide attempts, but not death. ^{Reference Smith, Velkoff, Ribeiro and Franklin7}

By contrast, in the COPING cohort, eating disorder symptoms demonstrated greater stability over time. This stability was evident in restrictive behaviours, which showed the strongest temporal associations, compared with purging and bingeing. Unlike the TEDS sample, cross-domain effects in COPING were generally weak although self-harm and suicidal ideation exhibited strong reciprocal associations over time. This difference could be attributed to the unique clinical and demographic characteristics of the COPING sample. TEDS, as a representative population-based sample, captures a broad range of eating disorder symptoms, self-harm and suicidal ideation. By contrast, COPING, as a clinical cohort, comprises individuals with a history of mood and anxiety disorders. This suggests that, in a clinical population with heightened emotional distress, self-harm and suicidal ideation may operate as distinct processes while disordered eating may function as an independent maladaptive coping mechanism.

Other key factors contributing to the differential outcomes between the TEDS and COPING samples are the age of participants and the study design. TEDS follows a developmental trajectory with the current data collected from adolescence into early adulthood, a critical period for the onset and fluctuation of eating disorders, self-harm and suicidal ideation. The dynamic interplay between these behaviours in TEDS suggests that adolescence and early adulthood may be particularly sensitive windows for their co-occurrence. In contrast, COPING is a cross-sectional study with a much older mean age (late forties), capturing a population in which these behaviours may have stabilised over time. The greater temporal stability in COPING of eating disorder symptoms, particularly restrictive behaviours, could reflect the chronic nature of these disorders in individuals with a history of mood and anxiety disorders, and who may have long-established maladaptive coping mechanisms. Moreover, the stronger associations between self-harm and suicidal ideation in COPING suggest that, in an older clinical population, these behaviours may be more ingrained and operate as distinct but interconnected constructs, whereas in younger, general-population samples like TEDS they may still be in flux and influenced more by other psychopathological processes, including disordered eating.

Another crucial difference between the TEDS and COPING samples is the timing of data collection. The COPING data were gathered during the COVID-19 pandemic and associated lockdowns, a period marked by heightened psychological distress, social isolation and disruptions to healthcare access. These factors may have contributed to the strong stability of symptoms, because individuals with pre-existing mental health conditions might have experienced exacerbations of their symptoms due to pandemic-related stressors. In contrast, the TEDS data were collected across a broader developmental span, outside the immediate context of the pandemic, allowing for a more naturalistic examination of how these symptoms evolve from adolescence to early adulthood. The lack of significant cross-domain associations in COPING may reflect the unique psychosocial conditions of the pandemic, where distress was widespread but may not have followed typical trajectories of mental health co-occurrence. Conversely, the TEDS findings suggest that, in a general-population sample, disordered eating, self-harm and suicidal ideation may be potentially influenced by developmental and environmental factors beyond acute stressors. These differences highlight the need to establish broader contextual influences, such as global crises, when interpreting findings on mental health trajectories.

The clinical emphasis on early intervention in self-harm aligns with recommendations by NICE, ¹³ which stress the importance of addressing co-occurring psychiatric conditions such as self-harm to mitigate the risks of both suicidality and eating disorders. Intervening at the stage of self-harm may not only reduce the immediate risk of suicide but also prevent the later onset or escalation of disordered eating. Because self-harm is a well-documented precursor to suicide in the general population, ^{Reference Cliffe, Seyedsalehi, Vardavoulia, Bittar, Velupillai and Shetty11,Reference Foster, Gillespie and McClelland40} identifying and treating self-injurious behaviours early, and addressing deficits in emotional regulation and maladaptive coping strategies, could help break the cycle of distress and maladaptive coping before it leads to more severe psychopathology. Additionally, given the emerging evidence that self-harm and suicidal ideation may be stronger predictors of later disordered eating than the reverse, addressing these behaviours proactively could interrupt the trajectory leading to disordered eating. Because the evidence from the TEDS cohort suggests that adolescence and early adulthood represent critical risk windows for the emergence of both disordered eating and self-harm, research is necessary to determine the precise nature of this relationship and whether early intervention in teenage populations could mitigate long-term risk trajectories. ^{Reference Solmi, Radua, Olivola, Croce, Soardo and Salazar de Pablo41} From a clinical standpoint, the findings suggest that treatment strategies for individuals engaging in self-harm should incorporate screening for emerging disordered eating symptoms. This could include routine screening for symptoms of body dissatisfaction and disordered eating in individuals presenting with suicidal thoughts and behaviours. Research suggests that individuals with both eating disorders and a history of self-harm experience less reduction in body dissatisfaction and drive for thinness following in-patient treatment, but this effect becomes non-significant when controlling for negative affect. ^{Reference Olatunji, Cox, Ebesutani and Wall42} This suggests that self-harm may not independently hinder treatment outcomes but, instead, operate through broader emotional distress encompassing emotion dysregulation, impulsivity and distress intolerance. ^{Reference Lavender, Wonderlich, Engel, Gordon, Kaye and Mitchell43,Reference Slee, Spinhoven, Garnefski and Arensman44} Because negative affect plays a central role in the persistence of both self-harm and disordered eating, interventions should prioritise emotional regulation strategies to improve treatment outcomes and reduce the likelihood of long-term psychopathology. ^{Reference Olatunji, Cox, Ebesutani and Wall42,Reference Solano, Fernández-Aranda, Aitken, López and Vallejo45} Although the present findings suggest that self-harm and suicidal ideation in adolescence may be key risk factors for disordered eating in early adulthood, lack of data on self-harm and suicidality at age 16 years limits our understanding of whether these behaviours in mid-adolescence predict later eating disorder onset, making this a priority for future research.