Depressive symptom complexes of community-dwelling older adults: a latent network model.

Late-life depression has multiple, heterogeneous clinical presentations. The aim of the study was to identify higher-order homogeneous clinical features (symptom complexes), while accounting for their potential causal interactions within the network approach to psychopathology. We analyzed cross-sectional data from community-dwelling adults aged 65-85 years recruited by the European MentDis_ICF65+ study (n = 2623, mean age 74, 49% females). The severity of 33 depressive symptoms was derived from the age-adapted Composite International Diagnostic Interview. Symptom complexes were identified using multiple detection algorithms for symptom networks, and their fit to data was assessed with latent network models (LNMs) in exploratory and confirmatory analyses. Sensitivity analyses included the Partial Correlation Likelihood Test (PCLT) to investigate the data-generating structure. Depressive symptoms were organized by the Walktrap algorithm into eight symptom complexes, namely sadness/hopelessness, anhedonia/lack of energy, anxiety/irritability, self-reproach, disturbed sleep, agitation/increased appetite, concentration/decision making, and thoughts of death. An LNM adequately fit the distribution of individual symptoms' data in the population. The model suggested the presence of reciprocal interactions between the symptom complexes of sadness and anxiety, concentration and self-reproach and between self-reproach and thoughts of death. Results of the PCLT confirmed that symptom complex data were more likely generated by a network, rather than a latent-variable structure. In conclusion, late-life depressive symptoms are organized into eight interacting symptom complexes. Identification of the symptom complexes of late-life depression may streamline clinical assessment, provide targets for personalization of treatment, and aid the search for biomarkers and for predictors of outcomes of late-life depression.