Published online by Cambridge University Press: 07 January 2020
In the past decade, network analysis (NA) has been applied to psychopathology to quantify complex symptom relationships. This statistical technique has demonstrated much promise, as it provides researchers the ability to identify relationships across many symptoms in one model and can identify central symptoms that may predict important clinical outcomes. However, network models are highly influenced by node selection, which could limit the generalizability of findings. The current study (N = 6850) tests a comprehensive, cognitive–behavioral model of eating-disorder symptoms using items from two, widely used measures (Eating Disorder Examination Questionnaire and Eating Pathology Symptoms Inventory).
We used NA to identify central symptoms and compared networks across the duration of illness (DOI), as chronicity is one of the only known predictors of poor outcome in eating disorders (EDs).
Our results suggest that eating when not hungry and feeling fat were the most central symptoms across groups. There were no significant differences in network structure across DOI, meaning the connections between symptoms remained relatively consistent. However, differences emerged in central symptoms, such that cognitive symptoms related to overvaluation of weight/shape were central in individuals with shorter DOI, and behavioral central symptoms emerged more in medium and long DOI.
Our results have important implications for the treatment of individuals with enduring EDs, as they may have a different core, maintaining symptoms. Additionally, our findings highlight the importance of using comprehensive, theoretically- or empirically-derived models for NA.