Direction of Dependence Between Specific Symptoms of Depression: A Non-Gaussian Approach

Abstract

Psychopathology could arise from direct interactions between symptoms. Evidence suggests that the mechanisms underlying somatic and cognitive-affective symptoms of depression are different. The aim of this study was to explore dynamic associations among cognitive-affective depression criteria. We used distribution-based direction of dependence models, which estimate whether the presence of symptom A is more likely to depend on the presence of symptom B than vice versa. We analyzed six large samples of adults from the United States ( N = 34,963) and conducted a simulation study to test the performance of the algorithm with ordinal variables and a second simulation study focusing on Type I error. Our results were consistent with the literature: Depressed mood and anhedonia were reactive to changes in other symptoms, whereas suicidality may reinforce other symptoms or reflect factors doing so. We discuss the results in the context of other empirical findings and theories of depression, reflect on the potential of these methods in psychopathology, and consider some practical implications.

Longitudinal associations between specific symptoms of depression: Network analysis in a prospective cohort study

“…We found that deterioration in suicidal ideation was most likely to induce deterioration in other symptoms over time. This accords with studies using directed network analyses (Bringmann et al, 2015) and direction of dependence methods (García-Velázquez et al, 2020) which suggest that suicidality is likely to reinforce other depressive symptoms.…”

Estimating the symptom structure of bipolar disorder via network analysis: Energy dysregulation as a central symptom.

“…Hence, our findings can only be treated as a provisional sketch of a potentially causal system among bipolar disorder symptoms. As García-Velázquez et al (2020) observed, acyclicity is a simplifying assumption that does suggest interventional hypotheses. For example, clinically targeting parent symptoms should be far more beneficial than only targeting descendants.…”

Depression Symptoms are Linked to Music Use

“…For example, the latent models treat all the symptoms more or less equally central to the disorder (symptom equivalence) because all the symptoms arise out of a single underlying condition (Cramer et al, 2010; Lux & Kendler, 2010; Schmittmann et al, 2013). However, research studies using large population cohorts and longitudinal samples have shown us that the latent variable model of major depression does not fit the observed data and fails to meet unidimensionality expectations (Fried et al, 2016; García-Velázquez et al, 2017, 2019). The underlying premise that depression is a unidimensional construct contributes to the categorisation of sum scores into healthy or depressed based on the assumption that all symptoms of depression are equal contributors to the severity of depression.…”