(PS10-9) Exploring Symptom Communities Underlying Internalizing Features: A Network Approach

Disorders marked by fear, distress, avoidance, and other internalizing features frequently co-occur and commonly display symptoms cutting across diagnostic boundaries. The present study employed a network approach to symptom-level data from six clinical questionnaires assessing depression, generalized anxiety disorder (GAD), social anxiety disorder, panic disorder, posttraumatic stress disorder (PTSD), and obsessive-compulsive disorder (OCD). Here, we: (a) construct a Gaussian graphical model to assess overarching symptom structure, (b) employ an exploratory graph analysis to examine data-driven communities’ alignment with traditional diagnostic boundaries, (c) examine the reproducibility and stability of communities unbound by traditional groupings, and (d) utilize proximity-based methods to quantify node-wise similarities among clinical symptoms. In a large sample of youth and adults with varying levels of psychopathology (n=1,423), empirical and bootstrapping exploratory graph analyses identified five communities of interconnected symptoms with moderate modularity (Q = 0.67); groupings reflected GAD/depression, social anxiety, panic, PTSD, and OCD symptoms. Proximity-based results quantified the density and similarities of nodes, most notably revealing a marked interconnectedness of GAD and depressive symptoms as shown by common neighbors and shortest path length. Node-wise properties of GAD symptoms show conceptual distinction from social anxiety and panic, and similarity to depressive symptoms. This lends data-driven evidence to distress-based internalizing symptoms but does not support the prediction that social anxiety and panic would form a fear-based dimension. These results also support the conceptualization of PTSD and OCD as distinct from the other internalizing disorders.