Data-driven subtypes of internalizing and functional somatic symptoms: A hybrid mixture modeling approach.

BACKGROUND

The mechanisms underlying the high comorbidity between internalizing disorders (IDs) and functional disorders (FDs) remain unclear. This study aimed to identify data-driven subgroups of major depressive disorder (MDD), generalized anxiety disorder (GAD), myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS), fibromyalgia (FM), and irritable bowel syndrome (IBS) symptoms in the general population, capturing shared symptom patterns while accounting for variation in symptom severity.

METHOD: We analyzed cross-sectional data from 72,919 adults in the Dutch Lifelines Cohort Study, randomly divided into training and validation subsets. Twenty-seven ID and FD criteria symptoms were examined using mixed-measurement item response theory models in the training set, with the optimal model validated in the validation set. Class characteristics were assessed through associations with risk factors, ID/FD diagnosis and comorbidity patterns.

RESULTS

Six classes best described the data: Healthy (57.5%), Pain (13.8%), Tension/Pain (9.1%), Anxiety (9.0%), Cognition/Fatigue (6.4%), and Depression (4.3%). All classes showed a combination of ID and FD symptoms, with the Cognition/Fatigue and Depression classes being most mixed, and the Pain and Anxiety classes most domain-specific. Transdiagnostic symptoms were highly endorsed across all classes. ID-FD comorbidity was highest in the Cognition/Fatigue and Depression classes, with the Depression class showing the greatest functional impairment.

CONCLUSIONS

The identified classes comprised mixed symptoms from multiple disorders, rather than disorder-specific profiles. ID-FD comorbidity may partly result from non-specific defining symptoms lowering the threshold for fulfilling criteria for multiple disorders. Symptom-level evaluation can help identify more homogenous and clinically relevant subgroups. Longitudinal research is needed to clarify underlying mechanisms.