Eating disorder psychopathology dimensions based on individual co-occurrence patterns of symptoms over time: a dynamic time warp analysis in a large naturalistic patient cohort.

Purpose: Most of the network approaches in eating disorders found the highest degree of centrality for symptoms related to weight and shape concerns. However, longitudinal analyses are scarce and may increase our insight of the complex characteristics and dynamics over time. In the current study, an alternative non-linear method to perform longitudinal network analyses, the dynamic time warp approach, was used to examine whether robust dimensions of eating disorder psychopathology symptoms could be found based on the individual dynamic interplay of eating disorder symptoms co-occurrence patterns in time.

Methods: The study sample included a naturalistic cohort of patients (N = 255) with all eating disorder subtypes who were assessed with the eating disorder examination questionnaire (EDE-Q) at a minimum of four times during treatment. Dynamic time warp analyses yielded distance matrices within each individual patient, which were subsequently aggregated into symptom networks and dimensions at the group level.

Results: Aggregation of the individual distance matrices at the group level yielded four robust symptom dimensions: 1. restraint/rules, 2. secret eating/fasting, 3. worries/preoccupation, and 4. weight and shape concern. The items 'fear of weight gain' and 'guilt' were bridge symptoms between the dimensions 1, 3 and 4.

Conclusion: Dynamic time warp could capture the within-person dynamics of eating disorder symptoms. Sumscores of the four dimensions could be used to follow patients over time. This approach could be applied in the future to visualize eating disorder symptom dynamics and signal the central symptoms within an individual and groups of patients.

Level Of Evidence: Level III: evidence obtained from well-designed cohort or case-control analytic studies. .