Using network analysis to compare diagnosis-specific and age-specific symptom networks in eating disorders.

Abstract

The network theory of mental disorders conceptualizes eating disorders (EDs) as networks of interacting symptoms. Network analysis studies in EDs mostly have examined transdiagnostic and/or mixed age samples. The aim of our study was to investigate similarities and differences of networks in adolescents and adults with anorexia nervosa (AN) or bulimia nervosa (BN). Participants were 2,535 patients (n=991 adults with AN, n=821 adolescents with AN, n=473 adults with BN, and n=250 adolescents with BN) who completed the Eating Disorder Inventory-2. Twenty-seven items were selected. Cross-sectional networks were estimated via Joint Graphical Lasso. Core symptoms were identified using strength centrality. Spearman correlations and network comparison tests (NCTs) were used to compare groups. Across diagnoses and ages, feeling ineffective, desire to be thinner, worries that feelings will get out of control, guilt after overeating as well as doing things perfectly emerged as most central symptoms. There were moderate to high correlations between symptom profiles (0.62-0.97, mean: 0.78) as well as high correlations between network structures (0.83-0.93, mean: 0.87) and network strengths (0.73-0.95, mean: 0.85). Global strength significantly differed in two of the six NCTs, and 2.5-10% of edges differed between networks. Considerable similarities in network structures and strengths across diagnoses and ages speak in favor of the transdiagnostic approach to EDs. Besides drive for thinness, ineffectiveness, emotion regulation difficulties, and perfectionism might be the most consistent factors in ED networks. These symptoms as well as their symptom connections should be especially focused in treatment regardless of age and diagnosis.