Can network analysis transform psychopathology?

doi:10.1016/j.brat.2016.06.006

Behaviour Research and Therapy

Volume 86, November 2016, Pages 95-104

https://doi.org/10.1016/j.brat.2016.06.006 Get rights and content

Highlights

•
Latent variable approaches to mental disorder are conceptually flawed.
•
Network analysis views disorders as causal systems of interacting symptoms.
•
Network analysis has clinical implications.

Abstract

Experimental psychopathology has been the primary path to gaining causal knowledge about variables maintaining mental disorders. Yet a radically different approach to conceptualizing psychopathology promises to advance our understanding, thereby complementing traditional laboratory experiments. In contrast to viewing symptoms as reflective of underlying, latent categories or dimensions, network analysis conceptualizes symptoms as constitutive of mental disorders, not reflective of them. Disorders emerge from the causal interactions among symptoms themselves, and intervening on central symptoms in disorder networks promises to foster rapid recovery. One purpose of this article is to contrast network analysis with traditional approaches, and consider its strengths and limitations. A second purpose is to review novel computational methods that may enable researchers to discern the causal structure of disorders (e.g., Bayesian networks). I close by sketching exciting new developments in methods that have direct implications for treatment.

Section snippets

Can network analysis transform psychopathology?

Three years ago the American Psychiatric Association (APA) released the fifth edition of its Diagnostic and Statistical Manual of Mental Disorders (DSM-5; APA, 2013) amidst a storm of controversy. Immediately prior to the unveiling of the new manual, the director of the National Institute of Mental Health (NIMH) announced that the DSM would no longer furnish the requisite framework for grant proposals submitted to NIMH (Insel, 2013). The institute, he said, would be “re-orienting its research

The network approach to psychopathology

The psychometrician, Denny Borsboom, and his colleagues have proposed a radically different explanation for syndromic coherence (e.g., Borsboom and Cramer, 2013, Borsboom, 2008, Borsboom et al., 2011a, Borsboom et al., 2011b, Cramer et al., 2010a, Schmittmann et al., 2013). According to their network¹

Key concepts in network analysis

Networks consist of nodes and edges. Nodes represent the objects of study, and edges represent the connections between them. In psychopathology networks, nodes represent symptoms, and edges represent associations between symptoms.

Networks can consist of either weighted edges or unweighted edges. An unweighted edge merely signifies that two symptoms are connected, whereas a weighted edge signifies the magnitude of the connection (e.g., a Pearson correlation coefficient), represented by thickness

Node centrality metrics

Traditional categorical approaches to psychiatric diagnosis emphasize hallmark symptoms that are strongly associated with a single disorder, but seldom associated with other disorders. Some nosologists have proposed that we purify diagnostic criteria sets of nonspecific symptoms appearing in many disorders, leaving only those strongly associated with the syndrome (e.g., Spitzer, First, & Wakefield, 2007). This recommendation was especially an issue for specialists struggling to make sense of

Types of networks

Psychopathologists have computed several types of networks, most concerning cross-sectional, observational symptom data. Although cross-sectional data cannot alone confirm causality among symptoms, network analysts have devised methods that can bring us closer to characterizing mental disorders as causal systems (McNally, 2012).

Strengths and limitations of network analysis?

A potentially fatal objection to latent variable approaches to psychopathology, whether construed categorically or dimensionally, is their failure to satisfy the axiom of local independence requisite for justifying an inference to an underlying entity as the common cause of symptom emergence and covariance (Borsboom and Cramer, 2013, Borsboom, 2008). Indeed, it seems obvious that causal connections abound between symptoms (e.g., sleep loss causing fatigue; phobic fear causing avoidance

Future directions

The aim of research in abnormal psychology is to discover the causes of mental health problems, thereby enhancing the efficacy of prevention and treatment. The field of experimental psychopathology remains at the forefront of these efforts (van den Hout, Engelhard, & McNally, in press). However, for obvious ethical reasons, many questions concerning causality are unanswerable with experimental methods. Complementing experimental approaches, network analysis aims to elucidate the causal

References (86)

D.M. Clark
A cognitive approach to panic
Behaviour Research and Therapy
(1986)
M.G. Everett et al.
Networks containing negative ties
Social Networks
(2014)
L.C. Freeman
Centrality in social networks
Social Networks
(1978/1979)
A.M. Hayes et al.
Network destabilization and transition in depression: New methods for studying the dynamics of therapeutic change
Clinical Psychology Review
(2015)
F. Hickey et al.
Medical update for children with Down syndrome for the pediatrician and family practitioner
Advances in Pediatrics
(2012)
D. van Kampen
The SSQ model of schizophrenic prodromal unfolding revised: An analysis of its causal chains based on the language of directed graphs
European Psychiatry
(2014)
S. Reiss et al.
Anxiety sensitivity, anxiety frequency and the prediction of fearfulness
Behaviour Research and Therapy
(1986)
M. Rhemtulla et al.
Network analysis of substance abuse and dependence symptoms
Drug and Alcohol Dependence
(2016)
V.D. Schmittmann et al.
Deconstructing the construct: A network perspective on psychological phenomena
New Ideas in Psychology
(2013)
R.L. Spitzer et al.
Saving PTSD from itself in DSM-V
Journal of Anxiety Disorders
(2007)

American Psychiatric Association

Diagnostic and statistical manual of mental disorders

(1994)

American Psychiatric Association

Diagnostic and statistical manual of mental disorders

(2013)

M.C. Ashton et al.

On models of personality structure

European Journal of Personality

(2012)

K. Bollen et al.

Conventional wisdom on measurement: A structural equation perspective

Psychological Bulletin

(1991)

C.D. van Borkulo et al.

Association of symptom network structure with the course of longitudinal depression

JAMA Psychiatry

(2015)

D. Borsboom

Psychometric perspectives on diagnostic systems

Journal of Clinical Psychology

(2008)

D. Borsboom et al.

Network analysis: An integrative approach to the structure of psychopathology

Annual Review of Clinical Psychology

(2013)

D. Borsboom et al.

The small world of psychopathology

PLoS ONE

(2011)

D. Borsboom et al.

Why g is not an adaptation: A comment on Kanazawa (2004)

Psychological Review

(2006)

D. Borsboom et al.

Transdiagnostic networks: Commentary on Nolen-Hoeksema and Watkins (2011)

Perspectives on Psychological Science

(2011)

D. Borsboom et al.

The two disciplines of psychology, or: The disunity of psychology as a working hypothesis

D. Borsboom et al.

The theoretical status of latent variables

Psychological Review

(2003)

L. Boschloo et al.

The network structure of psychopathology in a community sample of preadolescents

Journal of Abnormal Psychology

(2016, March)

L. Boschloo et al.

A prospective study of how symptoms in a network predict the onset of depression

Psychotherapy and Psychosomatics

(2016)

R. Boyd

Realism, anti-foundationalism and the enthusiasm for natural kinds

Philosophical Studies

(1991)

L.F. Bringmann et al.

Revealing the dynamic network structure of the Beck Depression Inventory II

Psychological Medicine

(2015)

L.F. Bringmann et al.

A network approach to psychopathology: New insights into clinical longitudinal data

PLoS One

(2013)

A.O.J. Cramer et al.

The pathoplasticity of dysphoric episodes: Differential impact of stressful life events on the pattern of depressive symptom inter-correlations

Psychological Medicine

(2012)

A.O.J. Cramer et al.

Measurable like temperature or mereological like flocking? On the nature of personality traits

European Journal of Personality

(2012)

A.O.J. Cramer et al.

Comorbidity: A network perspective

Behavioral and Brain Sciences

(2010)

A.O.J. Cramer et al.

Complex realities require complex theories: Refining and extending the network approach to mental disorders

Behavioral and Brain Sciences

(2010)

J.R. Edwards et al.

On the nature and direction of relationships between constructs and measures

Psychological Methods

(2000)

S. Epskamp et al.

qgraph: Network visualization of relationships in psychometric data

Journal of Statistical Software

(2012)

A. Frances

ICD, DSM and the Tower of Babel

Australian & New Zealand Journal of Psychiatry

(2014)

A. Frances et al.

ICD-11 should not repeat the mistakes made by DSM-5

British Journal of Psychiatry

(2013)

E.I. Fried

Problematic assumptions have slowed down depression research: Why symptoms, not syndromes are the way forward

Frontiers in Psychology

(2015)

E.I. Fried et al.

From loss to loneliness: The relationship between bereavement and depressive symptoms

Journal of Abnormal Psychology

(2015)

E.I. Fried et al.

Commentary: “Consistent superiority of selective serotonin reuptake inhibitors over placebo in reducing depressed mood in patients with major depression.”

Frontiers in Psychiatry

(2015)

J. Friedman et al.

Regularization paths for generalized linear models via coordinate descent

Journal of Statistical Software

(2010)

E.I. Fried et al.

The impact of individual depressive symptoms on impairment of psychosocial functioning

PLoS ONE

(2014)

E.I. Fried et al.

Depression is more than the sum score of its parts: Individual DSM symptoms have different risk factors

Psychological Medicine

(2014)

T.M.J. Fruchterman et al.

Graph drawing by force-directed placement

Software - Practice and Experience

(1991)

U. Grömping

Relative importance for linear regression in R: The package relaimpo

Journal of Statistical Software

(2006)

Cited by (569)

Network analysis of social anxiety and problematic mobile phone use in Chinese adolescents: A longitudinal study
2024, Addictive Behaviors
Social anxiety is a common comorbid factor of problematic mobile phone use, but the relationship between them from the inside structure has been underexplored. This two-wave longitudinal study aims to identify the central symptoms of problematic mobile phone use and social anxiety, respectively, as well as the bridge symptoms between the two disorders. Furthermore, this study examined the time and gender differences. A total of 3,294 Chinese adolescents were recruited during two waves (T1: January 2023; T2: August 2023), and completed the questionnaires of social anxiety and problematic mobile phone use. Through network analysis, the results showed that 1) the central symptom of problematic mobile phone use was “loss of control” both at T1 and T2, 2) the central symptom of social anxiety was “fear of negative evaluation” both at T1 and T2, and 3) the bridge symptom between social anxiety and problematic mobile phone use was “withdrawal” both at T1 and T2. Moreover, the network comparison test indicated that the nodes of social anxiety were more strongly associated with “craving”, “loss of control”, and “negative life consequences“ of problematic mobile phone use at T2 than at T1. The nodes of social anxiety were more strongly associated with “negative life consequences” of problematic mobile phone use for boys than for girls, while the “fear of negative evaluation” was more strongly associated with “loss of control” for girls than for boys. These findings provide new scientific basis for interventions targeting adolescents who are susceptible to social anxiety and problematic mobile phone use.
Exploring the complex interrelation between depressive symptoms, risk, and protective factors: A comprehensive network approach
2024, Journal of Affective Disorders
Depressive symptoms seem to be interrelated in a complex and self-reinforcing way. To gain a better understanding of this complexity, the inclusion of theoretically relevant constructs (such as risk and protective factors) offers a comprehensive view into the complex mechanisms underlying depression.
Cross-sectional data from individuals diagnosed with a major depressive disorder (N = 986) and healthy controls (N = 1049) were analyzed. Participants self-reported their depressive symptoms, as well as several risk factors and protective factors. Regularized partial correlation networks were estimated for each group and compared using a network comparison test.
Symptoms of depression were more strongly connected in the network of depressed patients than in healthy controls. Among the risk factors, perceived stress, the experience of negative life events, emotional neglect, and emotional abuse were the most centrally embedded in both networks. However, the centrality of risk factors did not significantly differ between the two groups. Among the protective factors, social support, personal competence, and acceptance were the most central in both networks, where the latter was significantly more strongly associated with the symptom of self-hate in depressed patients.
The network analysis revealed that key symptoms of depression were more strongly connected for depressed patients than for healthy controls, and that risk and protective factors play an important role, particularly perceived stress in both groups and an accepting attitude for depressed patients. However, the purpose of this study is hypothesis generating and assisting in the potential selection of non-symptom nodes for future research.
Trauma behind the keyboard: Exploring disparities in child sexual abuse materials exposure and mental health factors among investigators and forensic examiners – A network analysis
2024, Child Abuse and Neglect
Despite acknowledging the detrimental impact of child sexual abuse material (CSAM) exposure on the mental and physical well-being of investigators and forensic examiners, there is a need for comprehensive exploration into the complex relationship between CSAM exposure, its various dimensions, mental health (i.e., anxiety, depression, and PTSD), and burnout, as well as the presence of positive attitudes towards the job.
To understand how CSAM exposure, mental health and burnout interconnect and cluster within distinct networks of police investigators and forensic examiners.
Police investigators and forensic examiners from across the United States who were exposed to CSAM as part of their professions (N = 470).
Participants, recruited through connections with the National Criminal Justice Training Center, completed an anonymous online survey.
The network analysis revealed differences in centrality between investigators and forensic examiners, particularly in their associations with exposure factors and mental health variables. Edges invariance tests showed differences in the strength of these associations, with some factors being more strongly linked to PTSD avoidance symptoms among investigators and others among forensic examiners. Stability analyses suggested potentially greater heterogeneity among investigators, while both groups displayed high stability in other centrality indices.
This study contributes to our comprehension of the distinct experiences and challenges faced by CSAM investigators and forensic examiners, and specifically the nuanced disparities between CSAM investigators and forensic examiners in terms of their exposure to CSAM content and the associated mental health factors. These insights highlight the imperative need for tailored support mechanisms and interventions that can effectively address the unique challenges encountered by individuals working tirelessly in this critical field.
Contemporaneous and temporal network analysis of complex Posttraumatic stress disorder among Chinese college students with Childhood adversity: A longitudinal study
2024, Psychiatry Research
To enhance understanding of the longitudinal progression of complex posttraumatic stress disorder (CPTSD) symptoms, this longitudinal study examined how CPTSD symptoms interact over time in Chinese college students with childhood trauma.
From 18,933 college students who took part in two surveys 12 months apart, 4,006 participants who reported adverse childhood experiences were screened. Cross-sectional network comparisons and cross-lagged panel network (CLPN) analysis characterized interactions among CPTSD symptoms.
In the cross-sectional networks, feeling like a failure and avoid activities reminiscent of the trauma were the central symptoms. Takes long time to calm down and exaggerated startle are important bridge symptoms in the two networks respectively. The comparison of cross-sectional networks indicates that the global network strength was stable. The findings of the CLPN model reveal that feel worthless and feel like a failure had the highest "out" expected influence; exaggerated startle and avoid thoughts and feelings about the trauma had the highest "in" expected influence.
By conducting cross-sectional network analyses, the study illuminated the attributes of CPTSD networks across various time points. Additionally, the CLPN analysis uncovered the longitudinal patterns of CPTSD symptoms.
Eating disorder symptom dimensions and protective factors: A structural network analysis study
2024, Appetite
Eating Disorders (EDs) and related symptoms pose a substantial public health concern due to their widespread prevalence among both genders and associated negative outcomes, underscoring the need for effective preventive interventions. In this context, deepening our understanding of the interplay between ED symptoms and related protective factors appears crucial. Therefore, this study employed a structural network analysis approach considering both ED symptom dimensions (i.e., drive for thinness, bulimic symptoms, and body dissatisfaction) and related protective factors (i.e., body and functionality appreciation, intuitive eating, and self-esteem) to shed light on how these factors are interrelated. A community sample of 1391 individuals (34.4% men; M_age = 26.4 years) completed a socio-demographic schedule and self-report questionnaires. The network showed that the nodes with the highest positive expected influence were body and functionality appreciation, while those with the highest negative expected influence were eating for physical rather than emotional reasons and unconditional permission to eat (i.e., two components of intuitive eating). Crucially, the most relevant bridges between the conceptual communities “ED symptom dimensions” and “Protective factors” were the negative relations between (a) eating for physical rather than emotional reasons and bulimic symptoms, (b) unconditional permission to eat and drive for thinness, and (c) body appreciation and body dissatisfaction. Finally, age, gender, and body mass index did not moderate any edge in the network. The practical implications of these findings are discussed, especially in terms of preventive interventions for ED symptoms.
Hazardous drinking in young adults with co-occurring PTSD and psychosis symptoms: A network analysis
2024, Journal of Affective Disorders
Existing literature suggests the co-occurrence of post-traumatic stress disorder (PTSD) and psychosis among young adults is related to hazardous drinking. However, the influencing mechanisms among these co-occurrences are inconclusive. Thus, this study aimed to investigate the symptomatic associations between PTSD, psychosis, and hazardous drinking.
This study included 96,218 young Chinese adults, divided into three groups (PTSD, Psychosis, and co-occurring PTSD-Psychosis). PTSD, psychosis, and hazardous drinking were measured by the ten-item Trauma Screening Questionnaire, the seven-item Psychosis Screener Scale, and the four-item Alcohol Use Disorders Identification Test, respectively. Network analysis was utilized to explore and compare the symptomatic correlation between PTSD, psychosis, and hazardous drinking.
In this study, the most crucial symptom (both central and bridge) was "delusion of control" among the three networks. Hazardous drinking was another main bridge symptom. Compared to the Psychosis group and the co-occurring PTSD-Psychosis group, “Delusion of reference or persecution” to “Grandiose delusion” was the strongest edge in "the network structure of the PTSD group".
The cross-sectional study cannot determine the causal relationship. Applying self-reporting questionnaires may cause inherent bias. Young adult participants limited the generalization of the results to other groups.
Among the three network structures, delusion of control was the most crucial symptom, and hazardous drinking was another bridge symptom; the edge of delusion of reference or persecution and grandiose delusion was strongest in the PTSD group's network. Efforts should be taken to develop diverse targeted interventions for these core symptoms to relieve PTSD, psychosis, and hazardous drinking in young adults.

View all citing articles on Scopus

^☆: I thank two anonymous reviewers for their thoughtful, excellent, and very helpful comments. The author has no funding sources to report.

View full text

Can network analysis transform psychopathology?☆

Highlights

Abstract

Section snippets

Can network analysis transform psychopathology?

The network approach to psychopathology

Key concepts in network analysis

Node centrality metrics

Types of networks

Strengths and limitations of network analysis?

Future directions

Behaviour Research and Therapy

Social Networks

Social Networks

Clinical Psychology Review

Advances in Pediatrics

European Psychiatry

Behaviour Research and Therapy

Drug and Alcohol Dependence

New Ideas in Psychology

Journal of Anxiety Disorders

Diagnostic and statistical manual of mental disorders

Diagnostic and statistical manual of mental disorders

On models of personality structure

European Journal of Personality

Conventional wisdom on measurement: A structural equation perspective

Psychological Bulletin

Association of symptom network structure with the course of longitudinal depression

JAMA Psychiatry

Psychometric perspectives on diagnostic systems

Journal of Clinical Psychology

Network analysis: An integrative approach to the structure of psychopathology

Annual Review of Clinical Psychology

The small world of psychopathology

PLoS ONE

Why g is not an adaptation: A comment on Kanazawa (2004)

Psychological Review

Transdiagnostic networks: Commentary on Nolen-Hoeksema and Watkins (2011)

Perspectives on Psychological Science

The two disciplines of psychology, or: The disunity of psychology as a working hypothesis

The theoretical status of latent variables

Psychological Review

The network structure of psychopathology in a community sample of preadolescents

Journal of Abnormal Psychology

A prospective study of how symptoms in a network predict the onset of depression

Psychotherapy and Psychosomatics

Realism, anti-foundationalism and the enthusiasm for natural kinds

Philosophical Studies

Revealing the dynamic network structure of the Beck Depression Inventory II

Psychological Medicine

A network approach to psychopathology: New insights into clinical longitudinal data

PLoS One

The pathoplasticity of dysphoric episodes: Differential impact of stressful life events on the pattern of depressive symptom inter-correlations

Psychological Medicine

Measurable like temperature or mereological like flocking? On the nature of personality traits

European Journal of Personality

Comorbidity: A network perspective

Behavioral and Brain Sciences

Complex realities require complex theories: Refining and extending the network approach to mental disorders

Behavioral and Brain Sciences

On the nature and direction of relationships between constructs and measures

Psychological Methods

qgraph: Network visualization of relationships in psychometric data

Journal of Statistical Software

ICD, DSM and the Tower of Babel

Australian & New Zealand Journal of Psychiatry

ICD-11 should not repeat the mistakes made by DSM-5

British Journal of Psychiatry

Problematic assumptions have slowed down depression research: Why symptoms, not syndromes are the way forward

Frontiers in Psychology

From loss to loneliness: The relationship between bereavement and depressive symptoms

Journal of Abnormal Psychology

Commentary: “Consistent superiority of selective serotonin reuptake inhibitors over placebo in reducing depressed mood in patients with major depression.”

Frontiers in Psychiatry

Regularization paths for generalized linear models via coordinate descent

Journal of Statistical Software

The impact of individual depressive symptoms on impairment of psychosocial functioning

PLoS ONE

Depression is more than the sum score of its parts: Individual DSM symptoms have different risk factors

Psychological Medicine