Hierarchical models of psychopathology: empirical support, implications, and remaining issues

Abstract

There is an ongoing revolution in psychology and psychiatry that will likely change how we conceptualize, study and treat psychological problems.­ Many theorists now support viewing psychopathology as consisting of continuous dimensions rather than discrete diagnostic categories. Indeed, recent papers have proposed comprehensive taxonomies of psychopathology dimensions to replace the DSM and ICD taxonomies of categories. The proposed dimensional taxonomies, which portray psychopathology as hierarchically organized correlated dimensions, are now well supported at phenotypic levels. Multiple studies show that both a general factor of psychopathology at the top of the hierarchy and specific factors at lower levels predict different functional outcomes. Our analyses of data on a large representative sample of child and adolescent twins suggested the causal hypothesis that phenotypic correlations among dimensions of psychopathology are the result of many familial influences being pleiotropic. That is, most genetic variants and shared environmental factors are hypothesized to non‐specifically influence risk for multiple rather than individual dimensions of psychopathology. In contrast, person‐specific experiences tend to be related to individual dimensions. This hierarchical causal hypothesis has been supported by both large‐scale family and molecular genetic studies. Current research focuses on three issues. First, the field has not settled on a preferred statistical model for studying the hierarchy of causes and phenotypes. Second, in spite of encouraging progress, the neurobiological correlates of the hierarchy of dimensions of psychopathology are only partially described. Third, although there are potentially important clinical implications of the hierarchical model, insufficient research has been conducted to date to rec­ommend evidence‐based clinical practices.

Neuroticism vulnerability factors of anxiety symptoms in adolescents and early adults: an analysis using the bi-factor model and multi-wave longitudinal model

“…Past research has shown that neuroticism remains relatively stable after emerging in childhood ( Barlow et al, 2014 ; Eaton, Krueger & Oltmanns, 2011 ). In addition, we chose the bi-factor model as a measure of neurotic structure mainly because it is a direct hierarchical model and assumes that higher-order factors and special factors are orthogonal ( Beaujean, 2015 ; Lahey et al, 2021 ; Reise, 2012 ). Each item affects the general factors directly and sometimes influence a particular factor simultaneously.…”

“…Each item affects the general factors directly and sometimes influence a particular factor simultaneously. This attribute gives the bi-factor model an outstanding advantage in clearly explaining the formation mechanism of factors, especially for the general factors of neuroticism ( Lahey et al, 2021 ; Reise, 2012 ). For other high-order models, they assume that high-order factors and low-order factors are statistically correlated, so it is difficult to explain the formation mechanism of high-order factors of neuroticism independently ( Lahey et al, 2021 ; Mansolf & Reise, 2017 ).…”

Going “Up” to Move Forward: S-1 Bifactor Models and the Study of Neurocognitive Abilities in Psychopathology

“…In recent years, there has been a body of research that has moved away from the study of any single disorder (see [ 1 ]). This work is driven by issues of high comorbidity and low diagnostic stability within the traditional nosological approach to diagnosis (see [ 2 , 3 , 4 ]).…”

“…This work is driven by issues of high comorbidity and low diagnostic stability within the traditional nosological approach to diagnosis (see [ 2 , 3 , 4 ]). In particular, the rise of dimensional structural models of psychopathology has led to explorations of the commonalities of disorders, as well as what may underpin these commonalities [ 1 ]. One of the most prominent structural models of psychopathology, the bifactor model, revealed that a significant amount of variance from symptoms of a range of disorders could be accounted for by a single general factor at the “top” of the model, while specific disorder variance could be largely accounted for by a group of lower-order, or “specific” factors, such as externalising, internalising and thought disorder (e.g., [ 5 ]).…”

“…For example, the p -factor has been claimed to reflect neuroticism [ 6 ], disordered thought [ 7 ], functional impairment [ 8 ], and impulsive responsivity to emotion [ 9 ]. Furthermore, constructs, such as neurocognitive abilities, due to their reliable correlation with the general factor, have been claimed to be a key driver of the p -factor [ 1 ].…”