Depression May Drive Disordered Eating in Schizophrenia Through Lost Insight, Study Finds

More than one in five long-term hospitalized patients with schizophrenia shows disordered eating attitudes, according to a new cross-sectional study of 300 patients in Guangdong, China, and the research points to an unexpected psychological mechanism that may connect low mood to abnormal eating: the loss of insight into one’s own illness. The findings, published in BMC Psychiatry, suggest that the emotional and cognitive dimensions of schizophrenia are more tightly intertwined with eating behavior than many clinical teams assume, and that screening for eating pathology in psychiatric wards could identify a hidden burden that currently goes largely untreated.

The research team, led by Gang Zeng of Guangzhou Medical University together with co-first author Xiezhen Chen and colleagues at multiple institutions across South China, recruited patients from psychiatric hospitals between October 2022 and May 2023 using convenience sampling. Each participant completed a battery of standardized instruments: the Eating Attitudes Test-26, a widely used 26-item self-report measure of disordered eating attitudes and behaviors; the Insight and Treatment Attitudes Questionnaire, which quantifies a patient’s awareness of their illness and need for treatment; and the Hamilton Rating Scale for Depression and the Hamilton Rating Scale for Anxiety, clinician-administered measures of mood symptom severity. Sociodemographic details, smoking status, exercise frequency, and diabetic status were also recorded.

The headline number is stark. Using the conventional EAT-26 cutoff of 20 or above, the researchers found that 22 percent of hospitalized schizophrenia patients scored in the disordered eating range. That prevalence is striking for a population in which eating pathology is rarely the focus of clinical attention, and it aligns with a growing body of evidence that people with severe mental illness carry a disproportionate burden of abnormal attitudes toward food, weight, and body shape. Because the EAT-26 captures attitudes rather than diagnosed eating disorders, the figure reflects a spectrum of concern, from preoccupation with thinness and dieting to bulimic behaviors, rather than formal diagnoses.

To understand what drives these attitudes, the team first ran univariate analyses and then a multiple linear regression. Four variables emerged as independently associated with eating attitudes in the univariate stage: gender, smoking, weekly exercise frequency, and diabetic status. In the final regression model, smoking remained positively associated with EAT-26 scores, while insight, exercise frequency, and diabetes status were negatively related to the score, meaning that lower insight, less frequent exercise, and the absence of diabetes each tracked with more disordered eating attitudes. Depression was also significantly correlated with rising EAT-26 scores. Together, the model explained 33.7 percent of the variance in eating attitudes, a substantial share for a psychological outcome measured in a heterogeneous clinical population.

Because EAT-26 scores are typically positively skewed, with most respondents clustering at low values and a minority reporting high scores, ordinary least squares regression can be vulnerable to violations of distributional assumptions. The authors therefore performed a sensitivity analysis using a generalized linear model with a normal distribution, an identity link function, and a robust covariance estimator, a combination that provides more reliable standard errors when residual distributions are imperfect. The key variables, smoking, insight, exercise frequency, and diabetes status, remained statistically significant under this alternative approach, giving the primary findings a degree of statistical resilience that many single-model studies lack.

The most provocative result, however, is exploratory. Framed within the Common Sense Model of Self-Regulation, a theoretical framework in which individuals’ perceptions of their illness shape how they cope with emotional and physical threats, the researchers tested whether insight statistically mediates the relationship between depressive symptoms and eating attitudes. Using a bootstrap mediation procedure, they found a significant total association between depression and eating attitudes, with a standardized total effect of 0.35. This decomposed into a direct effect of 0.21 and an indirect effect of 0.15 that passed through insight. In other words, roughly 41.4 percent of the statistical association between depression and disordered eating attitudes flowed through patients’ level of insight into their illness.

The interpretation is delicate, and the authors are explicit about its limits. Because the data are cross-sectional, with all measurements taken at a single point in time, the analysis cannot establish that depression causes disordered eating, that insight lies on a causal pathway, or that improving insight would improve eating attitudes. Mediation in cross-sectional data is a statistical decomposition of association, not a demonstration of mechanism. Still, the pattern is theoretically coherent: patients with more depressive symptoms may, through the lens of the Common Sense Model, appraise their illness in ways that erode or distort their insight, and diminished insight may in turn leave eating attitudes unexamined and unregulated. Alternatively, low insight and depression may share upstream causes, and the mediation may reflect that shared structure rather than a genuine chain of influence.

The other associated factors deserve clinical attention in their own right. The negative association between exercise frequency and disordered eating attitudes suggests that physical activity, already valued in psychiatric rehabilitation for its metabolic and mood benefits, may also correlate with healthier attitudes toward food. The link with diabetes status is more complex: diabetes was negatively related to EAT-26 scores in the regression, a finding that could reflect dietary education and structured eating routines among diabetic patients, though the cross-sectional design forbids firm conclusions about direction. Smoking, which was positively associated with disordered eating attitudes, is highly prevalent in schizophrenia populations and is itself entangled with appetite regulation, metabolic health, and nicotine’s effects on reward circuitry.

Why should this matter beyond the ward? People with schizophrenia die years earlier than the general population, and cardiometabolic disease is a leading contributor to that gap. Antipsychotic medications frequently cause weight gain and metabolic disturbance, and disordered eating attitudes can compound these risks by shaping food choices and eating patterns over decades of illness. If depressive symptoms and impaired insight jointly mark a subgroup at elevated risk of abnormal eating attitudes, then routine screening with a brief instrument such as the EAT-26, combined with attention to mood and insight during clinical reviews, could flag patients who would benefit from targeted nutritional and psychological support before somatic consequences accumulate.

The study’s limitations frame how far its conclusions can travel. The sample of 300 patients was recruited by convenience sampling in one region of South China, so the prevalence estimate may not generalize to community-dwelling patients, to other regions, or to outpatient populations. The EAT-26 measures attitudes rather than clinical eating disorders, and self-report instruments in patients with variable insight carry inherent measurement challenges, a point that cuts to the heart of a study in which insight is itself a key variable. The mediation analysis is explicitly exploratory and cannot support causal claims. Yet the robustness checks, the theoretically grounded framework, and the size of the association, with insight accounting for over 40 percent of the depression-eating attitude link, make this a study that should prompt psychiatric services to look more closely at the plate as well as the mind. Integrating emotional, cognitive, and behavioral assessment, the authors conclude, may be essential to the clinical management of schizophrenia, and eating attitudes may be one of the most overlooked places where those domains converge.

Subject of Research: Disordered eating attitudes and the mediating role of insight in hospitalized schizophrenia patients

Article Title: Disordered eating attitudes in hospitalized schizophrenia patients of Guangdong, China: prevalence, associated factors, and the exploratory mediating role of insight in the depression-eating attitude link

Article References: Zeng, G., Chen, X., Niu, Z., Wu, F., Chen, X., Liu, C., Shi, X., Huang, Y., Zhou, W., Cao, W., Wang, W., & Zhang, J. (2026). Disordered eating attitudes in hospitalized schizophrenia patients of Guangdong, China: prevalence, associated factors, and the exploratory mediating role of insight in the depression-eating attitude link. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08574-0

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08574-0

Keywords: schizophrenia, disordered eating attitudes, EAT-26, insight, depression, mediation analysis, BMC Psychiatry, Guangdong, cross-sectional study, psychiatric hospitalization, Common Sense Model, exercise