On Nonregularized Estimation of Psychological Networks

Mapping of the network connection between sleep quality symptoms, depression, generalized anxiety, and burnout in the general population of Peru and El Salvador

BACKGROUND

A meta-analysis of randomized controlled trials has suggested a bidirectional relationship between sleep problems and mental health issues. Despite these findings, there is limited conclusive evidence on the relationship between sleep quality, depression, anxiety, and burnout.

OBJECTIVE

The current study aimed to evaluate the relationships between sleep quality symptoms, anxiety, depression, and burnout in samples of adult individuals from two Latin American countries, Peru and El Salvador, through network analysis and to identify key symptoms that reinforce the correlation and intensify the syndromes.

METHODS

A total of 1012 individuals from El Salvador and Peru participated, with an average age of 26.5 years (SD = 9.1). Symptom networks were constructed for both countries based on data from the Jenkins Sleep Scale, Patient Health Questionnaire-2, General Anxiety Disorder-2, and a single burnout item.

RESULTS

The results indicated that Depressed Mood, Difficulty Falling Asleep, and Nervousness were the most central symptoms in a network in the participating countries. The strongest conditional associations were found between symptoms belonging to the same construct, which were similar in both countries. Thus, there is a relationship between Nervousness and Uncontrollable Worry, Anhedonia and Depressed Mood, and Nighttime Awakenings and Difficulty in Staying Asleep. It was observed that burnout is a bridge symptom between both countries and presents stronger conditional associations with Tiredness on Awakening, Depressed Mood, and Uncontrollable Worry. Other bridge symptoms include a Depressed Mood and Nervousness. The network structure did not differ between the participants from Peru and El Salvador.

CONCLUSION

The networks formed by sleep quality, anxiety, depression, and burnout symptoms play a prominent role in the comorbidity of mental health problems among the general populations of Peru and El Salvador. The symptom-based analytical approach highlights the different diagnostic weights of these symptoms. Treatments or interventions should focus on identifying central and bridge symptoms.

Testing Conditional Independence in Psychometric Networks: An Analysis of Three Bayesian Methods

Network psychometrics uses graphical models to assess the network structure of psychological variables. An important task in their analysis is determining which variables are unrelated in the network, i.e., are independent given the rest of the network variables. This conditional independence structure is a gateway to understanding the causal structure underlying psychological processes. Thus, it is crucial to have an appropriate method for evaluating conditional independence and dependence hypotheses. Bayesian approaches to testing such hypotheses allow researchers to differentiate between absence of evidence and evidence of absence of connections (edges) between pairs of variables in a network. Three Bayesian approaches to assessing conditional independence have been proposed in the network psychometrics literature. We believe that their theoretical foundations are not widely known, and therefore we provide a conceptual review of the proposed methods and highlight their strengths and limitations through a simulation study. We also illustrate the methods using an empirical example with data on Dark Triad Personality. Finally, we provide recommendations on how to choose the optimal method and discuss the current gaps in the literature on this important topic.

Simulation-Based Performance Evaluation of Missing Data Handling in Network Analysis

Network analysis has gained popularity as an approach to investigate psychological constructs. However, there are currently no guidelines for applied researchers when encountering missing values. In this simulation study, we compared the performance of a two-step EM algorithm with separated steps for missing handling and regularization, a combined direct EM algorithm, and pairwise deletion. We investigated conditions with varying network sizes, numbers of observations, missing data mechanisms, and percentages of missing values. These approaches are evaluated with regard to recovering population networks in terms of loss in the precision matrix, edge set identification and network statistics. The simulation showed adequate performance only in conditions with large samples (n ≥ 500 ) or small networks (p = 10). Comparing the missing data approaches, the direct EM appears to be more sensitive and superior in nearly all chosen conditions. The two-step EM yields better results when the ratio of n/p is very large - being less sensitive but more specific. Pairwise deletion failed to converge across numerous conditions and yielded inferior results overall. Overall, direct EM is recommended in most cases, as it is able to mitigate the impact of missing data quite well, while modifications to two-step EM could improve its performance.

Time-Varying Network Models for the Temporal Dynamics of Depressive Symptomatology in Patients With Depressive Disorders: Secondary Analysis of Longitudinal Observational Data

BACKGROUND

As depression is highly heterogenous, an increasing number of studies investigate person-specific associations of depressive symptoms in longitudinal data. However, most studies in this area of research conceptualize symptom interrelations to be static and time invariant, which may lead to important temporal features of the disorder being missed.

OBJECTIVE

To reveal the dynamic nature of depression, we aimed to use a recently developed technique to investigate whether and how associations among depressive symptoms change over time.

METHODS

Using daily data (mean length 274, SD 82 d) of 20 participants with depression, we modeled idiographic associations among depressive symptoms, rumination, sleep, and quantity and quality of social contacts as dynamic networks using time-varying vector autoregressive models.

RESULTS

The resulting models showed marked interindividual and intraindividual differences. For some participants, associations among variables changed in the span of some weeks, whereas they stayed stable over months for others. Our results further indicated nonstationarity in all participants.

CONCLUSIONS

Idiographic symptom networks can provide insights into the temporal course of mental disorders and open new avenues of research for the study of the development and stability of psychopathological processes.

Network analysis of maternal parenting practices and adolescent mental health problems: a longitudinal study

BACKGROUND

An extensive literature has shown a strong connection between maternal parenting practices and adolescent mental health problems. However, it has been difficult for previous research to map a dynamic concurrent and prospective relationships within and between types of parenting practices and adolescent mental health problems. The present study addressed these issues using a network analysis approach and a longitudinal design.

METHODS

This study involved 591 Chinese adolescents (249 males; mean age at T1 = 13.53) and their mothers (mean age at T1 = 39.71) at two time points (T1 and T2) with eighteen months apart. Mothers reported their parenting practices including warmth, monitoring, inductive reasoning, hostility, and harshness, while adolescents reported their mental health problems including anxiety, depression, aggression, and conduct problems. Network analysis was conducted for contemporaneous networks at T1 and T2 and temporal networks from T1 to T2.

RESULTS

The contemporaneous networks revealed the negative association between monitoring and conduct problems served as the main pathway through which parenting practices and adolescent mental health mutually influenced each other, and further, warmth was the most influential parenting practice on adolescent mental health. The temporal network revealed that maternal hostility exerted the most influence on adolescent mental health problems, whereas adolescents' depression was most influenced by maternal parenting practices. Moreover, maternal hostility was most predicted by maternal harshness.

CONCLUSIONS

This study presents a novel perspective to gain a better understanding of the dynamics between and within maternal parenting practices and adolescent mental health problems. Findings highlight maternal harshness and warmth as potential prevention and intervention targets for adolescent mental health problems.

Comparing community detection algorithms in psychometric networks: A Monte Carlo simulation

Identifying the correct number of factors in multivariate data is fundamental to psychological measurement. Factor analysis has a long tradition in the field, but it has been challenged recently by exploratory graph analysis (EGA), an approach based on network psychometrics. EGA first estimates a network and then applies the Walktrap community detection algorithm. Simulation studies have demonstrated that EGA has comparable or better accuracy for recovering the same number of communities as there are factors in the simulated data than factor analytic methods. Despite EGA's effectiveness, there has yet to be an investigation into whether other sparsity induction methods or community detection algorithms could achieve equivalent or better performance. Furthermore, unidimensional structures are fundamental to psychological measurement yet they have been sparsely studied in simulations using community detection algorithms. In the present study, we performed a Monte Carlo simulation using the zero-order correlation matrix, GLASSO, and two variants of a non-regularized partial correlation sparsity induction methods with several community detection algorithms. We examined the performance of these method-algorithm combinations in both continuous and polytomous data across a variety of conditions. The results indicate that the Fast-greedy, Louvain, and Walktrap algorithms paired with the GLASSO method were consistently among the most accurate and least-biased overall.

Improving the Walktrap Algorithm Using K-Means Clustering

The walktrap algorithm is one of the most popular community-detection methods in psychological research. Several simulation studies have shown that it is often effective at determining the correct number of communities and assigning items to their proper community. Nevertheless, it is important to recognize that the walktrap algorithm relies on hierarchical clustering because it was originally developed for networks much larger than those encountered in psychological research. In this paper, we present and demonstrate a computational alternative to the hierarchical algorithm that is conceptually easier to understand. More importantly, we show that better solutions to the sum-of-squares optimization problem that is heuristically tackled by hierarchical clustering in the walktrap algorithm can often be obtained using exact or approximate methods for K-means clustering. Three simulation studies and analyses of empirical networks were completed to assess the impact of better sum-of-squares solutions.

Fantasy Made Flesh - A Network Analysis of the Reciprocal Relationship between Sexual Fantasies, Pornography Usage, and Sexual Behavior

Based on different theories in media research (3AM, catalyst model of violent crime, reinforcing spirals model), we further explore the relationship between pornography use, sexual fantasy, and behavior. We suggest that pornography use appears so persistent across time and culture because it is related to a human universal, the ability to fantasize. Consequently, pornography use seems to be an opportunity to acquire media-mediated sexual fantasies, and we believe that pornography use interacts with sexual fantasies and, to a much weaker extent, with sexual behavior. To assess our assumptions, we conducted a network analysis with a large and diverse sample of N = 1338 hetero- and bisexual participants from Germany. Analyses were done separately for men and women. Our network analysis clustered parts of the psychological processes around the interaction of sexual fantasies, pornography use, and behavior into communities of especially strong interacting items. We detected meaningful communities (orgasm-centered intercourse, BDSM) consisting of sexual fantasies and behavior, with some containing pornography. However, pornography use was not part of communities we perceive to account for mainstream/everyday sexuality. Instead, our results show that non-mainstream behavior (e.g., BDSM) is affected by pornography use. Our study highlights the interaction between sexual fantasies, sexual behavior, and (parts of) pornography use. It advocates for a more interactionist view of human sexuality and media use.

Associations of screen use with physical activity and social capital amid the COVID-19 pandemic: A network analysis of youths in China

Inconsistent correlations of screen use with physical activity (PA) and social capital (SC) in youths have been observed in existing cross-sectional studies. This study aimed to elucidate associations among variables in screen use, PA, and SC domains during COVID-19, to improve the prediction and prevention of suboptimal health status in youths. An online survey based on the nationwide COVID-19 Impact on Lifestyle Change Survey (COINLICS) was conducted in China, and 10,082 youths reported their screen use, PA, and SC in the months immediately before, during, and after the COVID-19 lockdown. Cross-sectional and longitudinal network models were used to identify associations of variables in domains of screen use with PA and SC. Effect modifications of bridges and predictors in the associations were examined. The network models suggested that individual SC was a bridge that strongly connected other types of SC, and domains of screen use and PA before lockdown, while phone use became such a bridge during and after lockdown. More PC/TV use before lockdown predicted less household-related PA during lockdown (β = -0.142); more phone use during lockdown was a predictor for higher levels of household-related PA (β = 0.106), active transport (β = 0.096), and individual SC (β = 0.072) after lockdown. Phone use was negatively associated with PA through PC/TV use in the more phone use subgroup. Relationships among screen use, PA, and SC dynamically changed during COVID-19, and phone use that was identified as a bridge and a predictor may be the potential action point for health intervention in youths during lockdown.

Immunometabolic alterations in older adults with heightened depressive symptom trajectories: a network approach

Objective: To analyse the patterns of relationships between depressive symptoms and immunometabolic markers across longitudinal depression status in older people. Methods: A sample of 3349 older adults (55.21% women; initial age: m = 58.44, sd = 5.21) from the English Longitudinal Study of Ageing was used. Participants were classified according to their longitudinal depression status: minimal depressive symptoms (n = 2736), depressive episode onset (n = 481), or chronic depression (n = 132). Network analysis was used to study the relationships between depression symptoms (CES-D 8 items), inflammatory (white blood cell, C-reactive protein, fibrinogen) and metabolic biomarkers (metabolic syndrome markers). Results: Network structure remained invariant across groups. The minimal symptom group had higher overall strength than both clinical groups (p < .01). Moreover, significant relationships between symptoms and markers were observed across group-specific networks. C-reactive protein and effort symptom were positively connected in the minimal symptom group but not in the other groups. Loneliness and diastolic blood pressure were positively associated only in the chronic depression group. Finally, metabolic markers were identified as central nodes in the clinical status networks. Conclusion: The network analysis constitutes a useful approach to disentangle pathophysiological relationships that may maintain mental disorders in old age.

Unique Variable Analysis: A Network Psychometrics Method to Detect Local Dependence

The local independence assumption states that variables are unrelated after conditioning on a latent variable. Common problems that arise from violations of this assumption include model misspecification, biased model parameters, and inaccurate estimates of internal structure. These problems are not limited to latent variable models but also apply to network psychometrics. This paper proposes a novel network psychometric approach to detect locally dependent pairs of variables using network modeling and a graph theory measure called weighted topological overlap (wTO). Using simulation, this approach is compared to contemporary local dependence detection methods such as exploratory structural equation modeling with standardized expected parameter change and a recently developed approach using partial correlations and a resampling procedure. Different approaches to determine local dependence using statistical significance and cutoff values are also compared. Continuous, polytomous (5-point Likert scale), and dichotomous (binary) data were generated with skew across a variety of conditions. Our results indicate that cutoff values work better than significance approaches. Overall, the network psychometrics approaches using wTO with graphical least absolute shrinkage and selector operator with extended Bayesian information criterion and wTO with Bayesian Gaussian graphical model were the best performing local dependence detection methods overall.

Patterns and correlates of insight among patients with schizophrenia in China：A network perspective

OBJECTIVE

To explore the patterns and correlates of insight among patients with schizophrenia in a large Chinese population.

METHOD

A multi-center cross-sectional study was conducted in Guangdong province, China. Patients with schizophrenia were included. Basic socio-demographic and clinical characteristics were collected in this study. Univariate analyses, multivariate logistic regression, and network analysis were conducted.

RESULTS

A total of 6090 participants (58.8% were male, and 41.2% were female) met the study criteria and completed all the assessments. 63.5% (n = 3869) patients with schizophrenia had impaired insight. Fewer drug sides effect, higher psychological and environment domains scores in quality of life have a positive significant impact on insight in patients with schizophrenia. Younger age, higher BPRS scores have a negative significant impact on insight in patients with schizophrenia. The node ITAQ 8 (strength=1.17) was the most central node within the ITAQ network, while node ITAQ 3 was the least central node (strength=0.69). The edge ITAQ 1-ITAQ 2 was the thickest and most saturated edge in network model.

CONCLUSIONS

Considering patterns and correlation of insight, it is necessary to ensure adherence to medications and engagement with mental health services for patients with schizophrenia, which could also improve their quality of life. Taking medication actively is more central to identify ITAQ and might be the potential targets for future interventions.

The general psychopathology factor (p) from adolescence to adulthood: Exploring the developmental trajectories of p using a multi-method approach

Considerable attention has been directed towards studying co-occurring psychopathology through the lens of a general factor (p-factor). However, the developmental trajectory and stability of the p-factor have yet to be fully understood. The present study examined the explanatory power of dynamic mutualism theory - an alternative framework that suggests the p-factor is a product of lower-level symptom interactions that strengthen throughout development. Data were drawn from a population-based sample of girls (N = 2450) who reported on the severity of internalizing and externalizing problems each year from age 14 to age 21. Predictions of dynamic mutualism were tested using three distinct complementary statistical approaches including: longitudinal bifactor models, random-intercept cross-lagged panel models (RI-CLPMs), and network models. Across methods, study results document preliminary support for mutualistic processes in the development of co-occurring psychopathology (that is captured in p). Findings emphasize the importance of exploring alternative frameworks and methods for better understanding the p-factor and its development.

A comparison of logistic regression methods for Ising model estimation

The Ising model has received significant attention in network psychometrics during the past decade. A popular estimation procedure is IsingFit, which uses nodewise l1-regularized logistic regression along with the extended Bayesian information criterion to establish the edge weights for the network. In this paper, we report the results of a simulation study comparing IsingFit to two alternative approaches: (1) a nonregularized nodewise stepwise logistic regression method, and (2) a recently proposed global l1-regularized logistic regression method that estimates all edge weights in a single stage, thus circumventing the need for nodewise estimation. MATLAB scripts for the methods are provided as supplemental material. The global l1-regularized logistic regression method generally provided greater accuracy and sensitivity than IsingFit, at the expense of lower specificity and much greater computation time. The stepwise approach showed considerable promise. Relative to the l1-regularized approaches, the stepwise method provided better average specificity for all experimental conditions, as well as comparable accuracy and sensitivity at the largest sample size.

On maximization of the modularity index in network psychometrics

The modularity index (Q) is an important criterion for many community detection heuristics used in network psychometrics and its subareas (e.g., exploratory graph analysis). Some heuristics seek to directly maximize Q, whereas others, such as the walktrap algorithm, only use the modularity index post hoc to determine the number of communities. Researchers in network psychometrics have typically not employed methods that are guaranteed to find a partition that maximizes Q, perhaps because of the complexity of the underlying mathematical programming problem. In this paper, for networks of the size commonly encountered in network psychometrics, we explore the utility of finding the partition that maximizes Q via formulation and solution of a clique partitioning problem (CPP). A key benefit of the CPP is that the number of communities is naturally determined by its solution and, therefore, need not be prespecified in advance. The results of two simulation studies comparing maximization of Q to two other methods that seek to maximize modularity (fast greedy and Louvain), as well as one popular method that does not (walktrap algorithm), provide interesting insights as to the relative performances of the methods with respect to identification of the correct number of communities and the recovery of underlying community structure.

Using network models to explore the associations between posttraumatic stress disorder symptoms and subjective cognitive functioning

Several studies have identified relationships between posttraumatic stress disorder (PTSD) and cognitive functioning. Here, we aimed to elucidate the nature of this relationship by investigating cross-sectional associations between subjective cognitive functioning (SCF) and 1) the PTSD sum score, 2) symptom domains, and 3) individual symptoms. We also investigated temporal stability by testing whether results replicated over a 3-year period. We estimated partial correlation networks of DSM-5 PTSD symptoms (at baseline) and SCF (at baseline and follow-up, respectively), using data from the National Health and Resilience in Veterans Study (NHRVS; N = 1484; Mdn = 65 years). The PTSD sum score was negatively associated with SCF. SCF was consistently negatively associated with the PTSD symptom domains 'marked alterations in arousal and reactivity' and 'negative alterations in cognitions and mood', and showed robust relations with the specific symptoms 'having difficulty concentrating' and 'trouble experiencing positive feelings'. Results largely replicated at the 3-year follow-up, suggesting that some PTSD symptoms both temporally precede and are statistically associated with the development or maintenance of reduced SCF. We discuss the importance of examining links between specific PTSD domains and symptoms with SCF-relations obfuscated by focusing on PTSD diagnoses or sum scores-as well as investigating mechanisms underlying these relations. Registration Number: 37069 (https://aspredicted.org/n5sw7.pdf).

The use of psychological network analysis in informal dementia care: an empirical illustration

Objective: Theoretical models in informal dementia care have been developed to understand how risk and protective factors interact to cause caregiver's distress. The development of psychological network analysis provides a rich complement to our current models, as explores how different variables (or nodes) are associated using graph theories. Methods: The present study explored the use of network analysis using data from 125 informal caregivers of their partner with dementia (PwD). The included variables were recipient's dependency, self-efficacy, conflict within the family, dyadic adjustment, and caregiver's distress. Results: The analysis suggests a complex network of interacting variables. The core variable was not the caregiver's distress but rather their dyadic adjustment with their PwD. Variables were associated with caregiver distress through a large array of direct and indirect pathways and were associated with each other in the form of an asymmetric spider's web.Conclusion: The results show the complex interplay of variables in a psychological network. The central role of distress suggests a complex and dynamic role, notably through a bidirectional influence with quality of interactions. In the same way, quality of interactions appeared as one of the strongest nodes, its connectivity suggesting a crucial role to consider in our models and interventions.

Network analysis of frontal lobe alpha asymmetry confirms the neurophysiological basis of four subtypes of depressive behavior

Introduction

Although depression is widespread carries a major disease burden, current treatments remain non-universally effective, arguably due to the heterogeneity of depression, and leading to the consideration of depressive "subtypes" or "depressive behavior subtypes." One such model of depressive behavior (DB) subtypes was investigated for its associations with frontal lobe asymmetry (FLA), using a different data analytic procedure than in previous research in this field.

Methods

100 community volunteers (54 males, 46 females) aged between 18 yr. and 75 years (M = 32.53 yr., SD = 14.13 yr) completed the Zung Self-rating Depression Scale (SDS) and underwent 15 min of eyes closed EEG resting data collection across 10 frontal lobe sites. DB subtypes were defined on the basis of previous research using the SDS, and alpha-wave (8-13 Hz) data produced an index of FLA. Data were examined via network analysis.

Results

Several network analyses were conducted, producing two models of the association between DB subtypes and FLA, confirming unique neurophysiological profiles for each of the four DB subtypes.

Discussion

As well as providing a firm basis for using these DB subtypes in clinical settings, these findings provide a reasonable explanation for the inconsistency in previous FLA-depression research.

Network analysis of anxiety and depressive symptoms one year after traumatic brain injury

We used network analysis to explore interrelationships between anxiety and depressive symptoms after traumatic brain injury (TBI). At one year post-injury, 882 adult civilians who received inpatient rehabilitation for moderate-severe TBI self-reported anxiety and depressive symptoms (Hospital Anxiety and Depression Scale). The severity of TBI was characterized acutely by the duration of post-traumatic amnesia (PTA), and TBI-related functional disability was rated by an examiner at one year post-injury using a structured interview (Glasgow Outcome Scale - Extended). We estimated two cross-sectional, partial correlation networks. In the first network, anxiety and depressive symptoms were densely interconnected yet formed three distinct, data-driven communities: Hyperarousal, Depression, and General Distress. Worrying thoughts and having difficulty relaxing were amongst the most central symptoms, showing strong connections with other symptoms within and between communities. In the second network, TBI severity was directly negatively associated with hyperarousal symptoms but indirectly positively associated with depressive symptoms via greater functional disability. The results highlight the potential utility of simultaneous, transdiagnostic assessment and treatment of anxiety and depressive symptoms after moderate-severe TBI. Worrying thoughts, having difficulty relaxing, and the experience of disability may be important targets for treatment, although future studies examining symptom dynamics within individuals and over time are required.

The Role and Impact of the Use of Information Technologies by Police in Response to Violence against Women

The last decade has witnessed an increased awareness of the crucial need to enhance police response and investigation of crimes related to violence against women (VAW). Although some research has been conducted on police decision-making in response to these crimes, there remains a dearth of knowledge concerning the influence of innovative police technologies on the investigative process and resulting case outcomes. This knowledge gap is particularly concerning given (1) the intricate nature and severity of VAW crimes and (2) the substantial advancements in technology that have transformed how the criminal justice system handles violent crime cases. To address this gap, the current study adopted a multi-method, quasi-experimental design to assess the impact of the Miami Police Department's Real-Time Crime Center (MRTCC) technologies on the case processing and case clearance of sexual assault and domestic violence incidents. The results of this study illuminate the distinctive features associated with this form of violent crime and underscore the necessity of continuously advancing the strategies employed to address these incidents.

A Cross-Country Network Analysis of Resilience Systems in Young Adults

Multisystemic resilience has been conceptualised as involving a constellation of protective factors which operate at different levels to promote adaptation and thriving despite experiences of adversity. We used network modelling to discover how protective factors at two different systemic levels (intrapersonal strengths and social-ecological resources) interrelate, drawing on survey data from 5283 emerging adults (M = 24.53 years; 52% female) in Brazil, China, Indonesia, Russia, Thailand, the US and Vietnam. Results indicated that the level of connectivity within and between protective factor levels was similar between the countries, but that there was substantial variation in the specific interrelations among protective factors (both within and between levels), including the presence of some country-specific negative interrelations between protective factors at different levels. The findings support the importance of cultural context in studies of resilience, with implications for the development of appropriate resilience-building interventions for this age group.

Bayesian Uncertainty Estimation for Gaussian Graphical Models and Centrality Indices

In the network approach to psychopathology, psychological constructs are conceptualized as networks of interacting components (e.g., the symptoms of a disorder). In this network view, interest is on the degree to which symptoms influence each other, both directly and indirectly. These direct and indirect influences are often captured with centrality indices, however, the estimation method often used with these networks, the frequentist graphical LASSO (GLASSO), has difficulty estimating (uncertainty in) these measures. Bayesian estimation might provide a solution, as it is better suited to deal with bias in the sampling distribution of centrality indices. This study therefore compares estimation of symptom networks with Bayesian GLASSO- and Horseshoe priors to estimation using the frequentist GLASSO using extensive simulations. Results showed that the Bayesian GLASSO performed better than the Horseshoe, and that the Bayesian GLASSO outperformed the frequentist GLASSO with respect to bias in edge weights, centrality measures, correlation between estimated and true partial correlations, and specificity. Sensitivity was better for the frequentist GLASSO, but performance of the Bayesian GLASSO is usually close. With respect to uncertainty in the centrality measures, the Bayesian GLASSO shows good coverage for strength and closeness centrality, but uncertainty in betweenness centrality is estimated less well.

A network approach to climate change anxiety and its key related features

Research has pointed to startling worldwide rates of people reporting considerable anxiety vis-à-vis climate change. Yet, uncertainties remain regarding how climate anxiety's cognitive-emotional features and daily life functional impairments interact with one another and with climate change experience, pro-environmental behaviors, and general worry. In this study, we apply network analyses to examine the associations among these variables in an international community sample (n = 874). We computed two network models, a graphical Gaussian model to explore network structure, potential communities, and influential nodes, and a directed acyclic graph to examine the probabilistic dependencies among the variables. Both network models pointed to the cognitive-emotional features of climate anxiety as a potential hub bridging general worry, the experience of climate change, pro-environmental behaviors, and the functional impairments associated with climate anxiety. Our findings offer data-driven clues for the field's larger quest to establish the foundations of climate anxiety.

Network analysis studies in patients with eating disorders: A systematic review and methodological quality assessment

INTRODUCTION

Network psychometrics has been enthusiastically embraced by researchers studying eating disorders (ED), but a rigorous evaluation of the methodological quality of works is still missing. This systematic review aims to assess the methodological quality of cross-sectional network analysis (NA) studies conducted on ED clinical populations.

METHODS

PRISMA and PICOS criteria were used to retrieve NA studies on ED. Methodological quality was evaluated based on five criteria: variable-selection procedure, network estimation method, stability checks, topological overlap checks, and handling of missing data.

RESULTS

Thirty-three cross-sectional NA studies were included. Most studies focused on populations that were female, white and, with an anorexia nervosa (AN) diagnosis. Depending on how many criteria were satisfied, 27.3% of studies (n = 9) were strictly adherent, 30.3% (n = 10) moderately adherent, 33.3% (n = 11) sufficiently adherent, and 9.1% (n = 3) poorly adherent. Missing topological overlap checks and not reporting missing data represented most unreported criteria, lacking, respectively, in 63.6% and 48.5% of studies.

CONCLUSIONS

Almost all reviewed cross-sectional NA studies on ED report those methodological procedures (variable-selection procedure, network estimation method, stability checks) necessary for a network study to provide reliable results. Nonetheless these minimum reporting data require further improvement. Moreover, elements closely related to the validity of an NA study (controls for topological overlap and management of missing data) are lacking in most studies. Recommendations to overcome such methodological weaknesses in future NA studies on ED are discussed together with the need to conduct NA studies with longitudinal design, to address diversity issues in study samples and heterogeneity of assessment tools.

PUBLIC SIGNIFICANCE

The present work aims to evaluate the quality of ED NA studies to support applications of this approach in ED research. Results show that most studies adopted basic procedures to produce reliable results; however, other important procedures linked to NA study validity were mostly neglected. Network methodology in ED is extremely promising, but future studies should consistently include topological overlap control procedures and provide information on missing data.

Generalized network psychometrics of eating-disorder psychopathology

OBJECTIVE

As network models of eating disorder (ED) psychopathology become increasingly popular in modeling symptom interconnectedness and identifying potential treatment targets, it is necessary to contextualize their performance against other methods of modeling ED psychopathology and to evaluate potential ways to optimize and capitalize on their use. To accomplish these goals, we used generalized network psychometrics to estimate and compare latent variable models and network models, as well as hybrid models.

METHOD

We tested the structure of the Eating Pathology Symptoms Inventory (EPSI) and Eating Disorder Examination-Questionnaire (EDE-Q) in Recovery Record, Inc. mobile phone application users (N = 6856).

RESULTS

Although all models fit well, results favored a hybrid latent variable and network framework, which showed that ED symptoms fit best when modeled as higher-order constructs, rather than direct symptom-to-symptom connections, and when the relationships between those constructs are described as a network. Hybrid models in which latent factors were modeled as nodes within a network showed that EPSI Purging, Binge Eating, Cognitive Restraint, Body Dissatisfaction, and Excessive Exercise had high importance in the network. EDE-Q Eating Concern and Shape Concern were also important nodes. Results showed that the EPSI network was highly stable and replicable, whereas the EDE-Q network was not.

DISCUSSION

Integrating latent variable and network model frameworks enables tests of centrality to identify important latent variables, such as purging, that may promote the spread of ED psychopathology throughout a network, allowing for the identification of future treatment targets.

Changes in specific and nonspecific psychopathology network structure after intensive cognitive behavior therapy in patients with anorexia nervosa

OBJECTIVE

This study aimed to compare eating disorder-specific and nonspecific clinical features in patients with anorexia nervosa before and after intensive enhanced cognitive behavior therapy (CBT-E) via network analysis.

METHODS

All consecutive patients admitted to intensive CBT-E were eligible, and the sample comprised patients aged ≥16 years who completed a 20-week intensive CBT-E program. Body mass index (BMI), Eating Disorder Examination Questionnaire and Brief Symptoms Inventory responses were gathered at baseline and end of treatment, and used to generate statistical networks of the connections between symptoms (nodes) and the strength and centrality thereof.

RESULTS

A total of 214 patients were included. Most nodes had relatively similar centrality compared to other nodes in the networks. "Eating concern" and "phobic anxiety" showed the greatest bridge centrality at both time points. No differences were found between baseline and the end of treatment in either global network or individual connection strengths.

CONCLUSION

These findings suggest that some clinical expressions not specific to eating-disorder psychopathology remain strongly connected in the generalized network of patients with anorexia nervosa after CBT-E. Future research should examine whether additional procedures specifically designed to target these symptoms should be integrated into this and other treatments.

Cross-sectional and longitudinal AUD symptom networks: They tell different stories

Modern theoretical models of Alcohol Use Disorder (AUD) highlight the different functional roles played by various mechanisms associated with different symptoms. Symptom network models (SNMs) offer one approach to modeling AUD symptomatology in a way that could reflect these processes and provide important information on the progression and persistence of disorder. However, much of the research conducted using SNMs relies on cross-sectional data, which has raised questions regarding the extent they reflect dynamic processes. The current study aimed to (a) examine symptom networks of AUD and (b) compare the extent to which cross-sectional network models had similar structures and interpretations as longitudinal network models. 17,360 participants from Wave 1 (2001-2002) and Wave 2 (2003-2004) of the National Epidemiological Survey on Alcohol and Related Conditions (NESARC) were used to model cross-sectional and longitudinal AUD symptom networks. The cross-sectional analyses demonstrate high replicability across waves and central symptoms consistent with other cross-sectional studies on addiction networks. The longitudinal network shared much less similarity than the cross-sectional networks and had a substantially different structure. Given the increasing attention given to the network perspective in psychopathology research, the results of this study raise concerns about interpreting cross-sectional symptom networks as representative of temporal changes occurring within a psychological disorder. We conclude that the psychological symptom network literature should be bolstered with additional research on longitudinal network models.

Development and Validation of a New Measure of Work Annoyance Using a Psychometric Network Approach

Existing measures of the impact of job characteristics on workers’ well-being do not directly assess the extent to which such characteristics (e.g., opportunity to learn new skills) are perceived as positive or negative. We developed a measure, the Work Annoyance Scale (WAS), of the level of annoyance that workers feel about certain aspects of the job and evaluated its psychometric properties. Using archival data from two cohorts (n = 2226 and 655) of workers that had undergone an annual medical examination for occupational hazard, we show the usefulness of the network psychometric approach to scale validation and its similarities and differences from a traditional factor analytic approach. The results revealed a two-dimensional structure (working conditions and cognitive demands) that was replicable across cohorts and bootstrapped samples. The two dimensions had adequate structural consistency and discriminant validity with respect to other questionnaires commonly used in organizational assessment, and showed a consistent pattern of association with relevant background variables. Despite the need for more extensive tests of its content and construct validity in light of the organizational changes due to the COVID-19 pandemic and of an evaluation of the generalizability of the results to cultural contexts different from the Italian one, the WAS appears as a psychometrically sound tool for assessment and research in organizational contexts.

Network analysis of impulse dyscontrol in mild cognitive impairment and subjective cognitive decline

OBJECTIVES

To investigate conditional dependence relationships of impulse dyscontrol symptoms in mild cognitive impairment (MCI) and subjective cognitive decline (SCD).

DESIGN

A prospective, observational study.

PARTICIPANTS

Two hundred and thirty-five patients with MCI (n = 159) or SCD (n = 76) from the Prospective Study for Persons with Memory Symptoms dataset.

MEASUREMENTS

Items of the Mild Behavioral Impairment Checklist impulse dyscontrol subscale.

RESULTS

Stubbornness/rigidity, agitation/aggressiveness, and argumentativeness were frequent and the most central symptoms in the network. Impulsivity, the fourth most central symptom in the network, served as the bridge between these common symptoms and less central and rare symptoms.

CONCLUSIONS

Impulse dyscontrol in at-risk states for dementia is characterized by closely connected symptoms of irritability, agitation, and rigidity. Compulsions and difficulties in regulating rewarding behaviors are relatively isolated symptoms.

Youth depression and inflammation: Cross-sectional network analyses of C-Reactive protein, interleukin-6 and symptoms in a population-based sample

BACKGROUND

Inflammation-related proteins constitute a promising avenue in studying biological correlates of major depressive disorder (MDD). However, MDD is a heterogeneous condition - a crucial aspect to be considered in association studies. We examined whether inflammatory proteins are associated with categorical diagnosis, a dimensional total sum-score, and specific depressive symptoms among youths.

METHODS

We analyzed data from the 1993 Pelotas Birth Cohort, a population-based study in Brazil that followed individuals up to age 22 years. Categorical psychiatric diagnoses were derived using adapted modules of the Mini International Neuropsychiatric Interview (MINI). Dimensional symptomatology was assessed using the Brazilian Portuguese version of the Center for Epidemiological Studies-Depression Scale-Revised (CESD-R). We estimated network structures that included individual depressive symptoms as measured by CESD-R items, peripheral inflammatory markers (C-Reactive Protein [CRP] and Interleukin-6 [IL-6]), as well as relevant covariates.

RESULTS

We evaluated 2586 participants (mean age = 22.5[SD = 0.33]) There were no associations between concentrations of inflammatory proteins and categorical diagnosis of MDD or with CESD-R total sum-scores. In symptom-specific analysis, CRP and IL-6 were positively connected to somatic and cognitive items.

DISCUSSION

We found cross-sectional connections of two commonly studied inflammatory proteins and specific depressive symptoms. Conducting symptom-specific analyses in relation to biological markers might advance our understanding of the heterogeneity of MDD.

Coexisting service-related factors preceding suicide: a network analysis

OBJECTIVES

The overall objective was to analyse service-related factors involved in the complex processes that precede suicide in order to identify potential targets for intervention.

DESIGN AND SETTING

Explorative network analysis study of post-suicide root cause analysis data from Swedish primary and secondary healthcare.

PARTICIPANTS

217 suicide cases reported to the Swedish national root cause analysis database between 2012 and 2017.

PRIMARY AND SECONDARY OUTCOME MEASURES

A total of 961 reported incidents were included. Demographic data and frequencies of reported deficiencies were registered. Topology, centrality indices and communities were explored for three networks. All networks have been tested for robustness and accuracy.

RESULTS

Lack of follow-up, evaluations and insufficient documentation issues emerged as central in the network of major themes, as did the contributing factors representing organisational problems, failing procedures and miscommunication. When analysing the subthemes of deficiencies more closely, disrupted treatments and staffing issues emerged as prominent features. The network covering the subthemes of contributing factors also highlighted discontinuity, fragile work structures, inadequate routines, and lack of resources and relevant competence as potential triggers. However, as the correlation stability coefficients for this network were low, the results need further investigation. Four communities were detected covering nodes for follow-up, evaluation, cooperation, and procedures; communication, documentation and organisation; assessments of suicide risk and psychiatric status; and staffing, missed appointments and declined treatment.

CONCLUSION

The results of this study suggest that healthcare providers may improve patient safety in suicide preventive pathways by taking active measures to provide regular follow-ups to patients with elevated suicide risk. In some cases, declined or cancelled appointments could be a warning sign. Tentative results show organisational instability, in terms of work structure, resources and staffing, as a potential target for intervention, although this must be more extensively explored in the future.

Resolving the paradox of posttraumatic growth and event centrality in trauma survivors

When a traumatic experience is central to an individual's identity and worldview, it can result in either severe posttraumatic stress disorder (PTSD) symptoms, perceived posttraumatic growth (PTG), or, paradoxically, both. To resolve this apparent paradox, we used network analytic methods to estimate the relations among components of event centrality (EC), PTSD symptoms, and PTG in 1,136 undergraduates who had experienced trauma. Participants completed surveys on their experiences with traumatic events as well as the degree to which they experienced PTSD symptoms, components of EC, and components of PTG. We performed network analysis to examine EC, PTSD, and PTG and identify which components of EC were most conducive to its associations with PTSD versus those with PTG. We found that the components of EC most associated with PTSD, the extent to which trauma serves as a script for the future, were markedly distinct from the components associated with PTG, the extent to which trauma is seen as a turning point in one's life. The combined findings suggest that EC may be a catalyst for subsequent positive or negative effects contingent upon how an individual interprets the centrality of their traumatic experience.

Possible Futures for Network Psychometrics

This commentary reflects on the articles included in the Psychometrika Special Issue on Network Psychometrics in Action. The contributions to the special issue are related to several possible future paths for research in this area. These include the development of models to analyze and represent interventions, improvement in exploratory and inferential techniques in network psychometrics, the articulation of psychometric theories in addition to psychometric models, and extensions of network modeling to novel data sources. Finally, network psychometrics is part of a larger movement in psychology that revolves around the analysis of human beings as complex systems, and it is timely that psychometricians start extending their rich modeling tradition to improve and extend the analysis of systems in psychology.

Dependence Graphs Based on Association Rules to Explore Delusional Experiences

Methods to estimate dependence graphs among variables, have quickly gained popularity in psychopathology research. To date, multiple methods have been proposed but recent studies report several drawbacks impacting on the validity of the conclusions as it is argued that assumptions and conditions underlying the methods commonly used and the nature of the data is lacking alignment. A particularly important issue is that underlying dynamics potentially present in heterogeneous datasets are disregarded, as the methods focus on the variables but not on individuals. This work also argues that the networks may lack relevant components as current methods ignore connections beyond pairwise interactions between individual symptoms. This study addresses these issues with a novel method for constructing dependence graphs based on applying Association Rules to binary records, which is often the type of records in the psychopathology domain. To demonstrate the benefits, we examine 12 delusional experiences in a sample of 1423 subjects with psychotic disorders. We show that by extracting Association Rules using an algorithm called apriori, in addition to facilitating an intuitive interpretation, previously unseen relevant dependencies are revealed from higher order interactions among psychotic experiences in subgroups of patients.

Meta-analytic Gaussian Network Aggregation

A growing number of publications focus on estimating Gaussian graphical models (GGM, networks of partial correlation coefficients). At the same time, generalizibility and replicability of these highly parameterized models are debated, and sample sizes typically found in datasets may not be sufficient for estimating the underlying network structure. In addition, while recent work emerged that aims to compare networks based on different samples, these studies do not take potential cross-study heterogeneity into account. To this end, this paper introduces methods for estimating GGMs by aggregating over multiple datasets. We first introduce a general maximum likelihood estimation modeling framework in which all discussed models are embedded. This modeling framework is subsequently used to introduce meta-analytic Gaussian network aggregation (MAGNA). We discuss two variants: fixed-effects MAGNA, in which heterogeneity across studies is not taken into account, and random-effects MAGNA, which models sample correlations and takes heterogeneity into account. We assess the performance of MAGNA in large-scale simulation studies. Finally, we exemplify the method using four datasets of post-traumatic stress disorder (PTSD) symptoms, and summarize findings from a larger meta-analysis of PTSD symptom.

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

Using network analysis, we estimated the structure of relations among manic and depressive symptoms, respectively, in 486 patients (59% women; age: M = 37, SD = 12.1) with bipolar disorder prior to their entering a clinical trial. We computed three types of networks: (a) Gaussian graphical models (GGMs) depicting regularized partial correlations, (b) regression-based GGMs depicting nonregularized partial correlations, and (c) directed acyclic graphs (DAGs) via a Bayesian hill-climbing algorithm. Low energy and elevated energy were consistently identified as central nodes in the GGMs and as key parent nodes in the DAGs. Across analyses, pessimism about the future and depressed mood were the symptoms most strongly associated with suicidal thoughts and behavior. These exploratory analyses provide rich information about how bipolar disorder symptoms relate to one another, thereby furnishing a foundation for investigating how bipolar disorder symptoms may operate as a causal system. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Excessive Worrying as a Central Feature of Anxiety during the First COVID-19 Lockdown-Phase in Belgium: Insights from a Network Approach

Since the WHO declared the COVID-19 pandemic on March 11, 2020, the novel coronavirus, SARS-CoV-2, has profoundly impacted public health and the economy worldwide. But there are not the only ones to be hit. The COVID-19 pandemic has also substantially altered mental health, with anxiety symptoms being one of the most frequently reported problems. Especially, the number of people reporting anxiety symptoms increased significantly during the first lockdown-phase compared to similar data collected before the pandemic. Yet, most of these studies relied on a unitary approach to anxiety, wherein its different constitutive features (i.e., symptoms) were tallied into one sum-score, thus ignoring any possibility of interactions between them. Therefore, in this study, we seek to map the associations between the core features of anxiety during the first weeks of the first Belgian COVID-19 lockdown-phase (n = 2,829). To do so, we implemented, in a preregistered fashion, two distinct computational network approaches: a Gaussian graphical model and a Bayesian network modelling approach to estimate a directed acyclic graph. Despite their varying assumptions, constraints, and computational methods to determine nodes (i.e., the variables) and edges (i.e., the relations between them), both approaches pointed to excessive worrying as a node playing an especially influential role in the network system of the anxiety features. Altogether, our findings offer novel data-driven clues for the ongoing field's larger quest to examine, and eventually alleviate, the mental health consequences of the COVID-19 pandemic.

The network of stress-related states and depression and anxiety symptoms during the COVID-19 lockdown

BACKGROUND

The coronavirus (COVID-19) pandemic and the social distancing protocols used to impede the spread of the virus may have severe mental health consequences. The purpose of this study was to investigate the network of components of pandemic-related negative psychological states (i.e., fear of infection, financial worries, loneliness) and symptoms of major depressive disorder (MDD) and generalized anxiety disorder (GAD).

METHODS

Data from 10,061 Norwegian adults recruited through an online survey during a period of strict social distancing protocols were analyzed by cross-sectional network methods.

RESULTS

Of the infection fears, fear of being infected, fear of dying from the coronavirus and fear of significant others dying from it had notable connections to the GAD symptoms anxiety and/or fear of awful events. The financial worry component worry about personal economy was connected to the MDD symptom sleep problems and to the GAD symptom generalized worry. Each of the loneliness components was connected to a specific MDD symptom. Depressed mood, low energy and worthlessness had the highest strength centrality among the MDD symptoms; generalized worry, uncontrollability of worry, and trouble relaxing among the GAD symptoms; fear of dying from the virus among the fear of infection components; and feeling isolated among the loneliness components.

LIMITATIONS

Full random sampling was not conducted, although the sample turned out to be relatively representative of the Norwegian population.

CONCLUSIONS

Some components of the pandemic-related distressing states of fear of infection, financial worry and loneliness seem to be associated with specific symptoms of MDD and GAD.

Change in eating-disorder psychopathology network structure in patients with anorexia nervosa treated with intensive cognitive behavior therapy

OBJECTIVE

This study was designed to compare the change in eating-disorder feature networks in patients with anorexia nervosa after treatment with intensive enhanced cognitive behavior therapy (CBT-E).

METHODS

Patients seeking treatment for anorexia nervosa were consecutively recruited from January 2016 to September 2020. All patients aged ≥16 years who completed a 20-week intensive CBT-E program (13 weeks of inpatient followed by 7 weeks of day-hospital treatment) were included in the study. Body mass index (BMI) was measured, and the Eating Disorder Examination Questionnaire completed for each patient, both at baseline and the end of treatment.

RESULTS

The sample comprised 214 patients with anorexia nervosa. Treated patients showed significant improvements in BMI and eating-disorder psychopathology. Network analysis revealed a significant reduction in the network global and connection strengths at the end of treatment. The most central and highly interconnected nodes in the network at baseline were related to the drive for thinness, but at the end of treatment to body image concerns. Some edge connections were significantly stronger at baseline than at the end of treatment, while others were significantly stronger at the end of treatment than at baseline.

DISCUSSION

CBT-E reduces the psychopathology network connectivity over time in patients with anorexia nervosa. The differences in central nodes and edge connections between baseline and end of treatment, not detected by classical inferential analysis, may be informative for understanding the centrality of symptoms in the psychopathology network, and how a specific treatment may act to reduce symptoms and change their connections over time.

Emotional information-processing correlates of positive mental health in adolescence: a network analysis approach

The combined cognitive bias hypothesis proposes that emotional information-processing biases may conjointly influence mental health. Yet, little is known about the interrelationships amongst cognitive biases, particularly in adolescence. We used data from the CogBIAS longitudinal study (Booth et al., 2017), including 450 adolescents who completed measures of interpretation bias, memory bias, and a validated measure of general mental health in a typically developing population. We used a moderated network modelling approach to examine positive mental health-related moderation of the cognitive bias network. We found that mental health was directly associated with positive and negative memory biases, and positive interpretation biases, but not negative interpretation biases. Further, we observed some mental health-related moderation of the network structure. Network connectivity decreased with higher positive mental health scores. Network approaches allow us to model complex relationships amongst cognitive biases and develop novel hypotheses for future research.

A Network Analysis of DSM-5 Posttraumatic Stress Disorder Symptoms and Event Centrality

The centrality of a traumatic event to one's autobiographical memory has been associated with posttraumatic stress disorder (PTSD) symptom severity. In the present study, we investigated the associations between specific features of event centrality (EC), as measured using the Centrality of Event Scale, and specific symptoms of PTSD. We computed a cross-sectional graphical lasso network of PTSD symptoms and specific features of EC in a sample of trauma-exposed individuals (n = 451), many of whom met the clinical threshold for a PTSD diagnosis. The graphical lasso revealed intrusive memories, negative trauma-related feelings, and the perception that the traumatic event was central to one's identity to be influential nodes. Viewing the future through the lens of one's trauma exposure was the EC feature most strongly linked to PTSD. Among all PTSD symptoms, blaming oneself or others for the traumatic event showed the strongest link to EC. The network was stable, allowing for reliable interpretations. Future longitudinal research is needed to clarify the associations among EC features and PTSD symptoms over time.

Network Analysis of Psychopathology: Controversies and Challenges

Empirical publications inspired by the network approach to psychopathology have increased exponentially in the twenty-first century. The central idea that an episode of mental disorder arises from causal interactions among its symptomatic elements has especially resonated with those clinical scientists whose disenchantment with traditional categorical and dimensional approaches to mental illness has become all too apparent. As the field has matured, conceptual and statistical concerns about the limitations of network approaches to psychopathology have emerged, inspiring the development of novel methods to address these concerns. Rather than reviewing the vast empirical literature, I focus instead on the issues and controversies regarding this approach and sketch directions where the field might go next.

Latent cause inference during extinction learning in trauma-exposed individuals with and without PTSD

BACKGROUND

Problems in learning that sights, sounds, or situations that were once associated with danger have become safe (extinction learning) may explain why some individuals suffer prolonged psychological distress following traumatic experiences. Although simple learning models have been unable to provide a convincing account of why this learning fails, it has recently been proposed that this may be explained by individual differences in beliefs about the causal structure of the environment.

METHODS

Here, we tested two competing hypotheses as to how differences in causal inference might be related to trauma-related psychopathology, using extinction learning data collected from clinically well-characterised individuals with varying degrees of post-traumatic stress (N = 56). Model parameters describing individual differences in causal inference were related to multiple post-traumatic stress disorder (PTSD) and depression symptom dimensions via network analysis.

RESULTS

Individuals with more severe PTSD were more likely to assign observations from conditioning and extinction stages to a single underlying cause. Specifically, greater re-experiencing symptom severity was associated with a lower likelihood of inferring that multiple causes were active in the environment.

CONCLUSIONS

We interpret these results as providing evidence of a primary deficit in discriminative learning in participants with more severe PTSD. Specifically, a tendency to attribute a greater diversity of stimulus configurations to the same underlying cause resulted in greater uncertainty about stimulus-outcome associations, impeding learning both that certain stimuli were safe, and that certain stimuli were no longer dangerous. In the future, better understanding of the role of causal inference in trauma-related psychopathology may help refine cognitive therapies for these disorders.

Inflammatory phenotype of depression symptom structure: A network perspective

BACKGROUND

There has been increasing interest in classifying inflammatory phenotypes of depression. Most investigations into inflammatory phenotypes only have tested whether elevated inflammation is associated with elevated levels of depression symptoms, or risk for a diagnosis. This study expanded the definition of phenotype to include the structure of depression symptoms as a function of inflammation.

METHODS

Network models of depression symptoms were estimated in a sample of 4157 adults (mean age = 47.6, 51% female) from the 2015-2016 National Health and Nutrition Examination Survey (NHANES). Analyses included comparisons of networks between those with elevated (C-reactive protein (CRP) values ≥ 3.0 mg/L; N = 1696) and non-elevated CRP (N = 2841) as well as moderated network models with CRP group status and raw CRP values moderating the associations between depression symptoms.

RESULTS

Differences emerged at all levels of analysis (global, symptom-specific, symptom-symptom associations). Specifically, the elevated CRP group had greater symptom connectivity (stronger total associations between symptoms). Further, difficulty concentrating and psychomotor difficulties had higher expected influence (concordance with other symptoms) in the elevated CRP group. Finally, there was evidence that several symptom-symptom associations were moderated by CRP.

CONCLUSIONS

This study provides consistent evidence that the structure of depression symptoms varies as a function of CRP levels. Greater symptom connectivity might contribute to why elevated CRP is associated with treatment-resistant depression. Additionally, differences in symptom structure might highlight different maintenance mechanisms and treatment targets for individuals with compared to those without elevated CRP. Finally, differences in symptom structure as a function of CRP highlight a potential misalignment of standard depression measures (the structure of which are evaluated on groups unselected for CRP levels) and the presentation of depression symptoms in those with elevated CRP.

On Penalty Parameter Selection for Estimating Network Models

Network models are gaining popularity as a way to estimate direct effects among psychological variables and investigate the structure of constructs. A key feature of network estimation is determining which edges are likely to be non-zero. In psychology, this is commonly achieved through the graphical lasso regularization method that estimates a precision matrix of Gaussian variables using an ℓ1-penalty to push small values to zero. A tuning parameter, λ, controls the sparsity of the network. There are many methods to select λ, which can lead to vastly different graphs. The most common approach in psychological network applications is to minimize the extended Bayesian information criterion, but the consistency of this method for model selection has primarily been examined in high dimensional settings (i.e., n < p) that are uncommon in psychology. Further, there is some evidence that alternative selection methods may have superior performance. Here, using simulation, we compare four different methods for selecting λ, including the stability approach to regularization selection (StARS), K-fold cross-validation, the rotation information criterion (RIC), and the extended Bayesian information criterion (EBIC). Our results demonstrate that penalty parameter selection should be made based on data characteristics and the inferential goal (e.g., to increase sensitivity versus to avoid false positives). We end with recommendations for selecting the penalty parameter when using the graphical lasso.

Unraveling the comorbidity of depression and anxiety in a large inpatient sample: Network analysis to examine bridge symptoms

BACKGROUND

Comorbidities in mental disorders are often understood by assuming a common cause. The network theory of mental disorders offers an alternative to this assumption by understanding comorbidities as mutually reinforced problems. In this study, we used network analysis to examine bridge symptoms between anxiety and depression in a large sample.

METHOD

Using data from a sample of patients diagnosed with both depression and an anxiety disorder before and after inpatient treatment (N = 5,614, mean age: 42.24, 63.59% female, average treatment duration: 48.12 days), network models of depression and anxiety symptoms are estimated. Topology, the centrality of nodes, stability, and changes in network structure are analyzed. Symptoms that drive comorbidity are determined by bridge node analysis. As an alternative to network communities based on categorical diagnosis, we performed a community analysis and propose empirically derived symptom subsets.

RESULTS

The obtained network models are highly stable. Sad mood and the inability to control worry are the most central. Psychomotor agitation or retardation is the strongest bridge node between anxiety and depression, followed by concentration problems and restlessness. Changes in appetite and suicidality were unique to depression. Community analysis revealed four symptom groups.

CONCLUSION

The estimated network structure of depression and anxiety symptoms proves to be highly accurate. Results indicate that some symptoms are considerably more influential than others and that only a small number of predominantly physical symptoms are strong candidates for explaining comorbidity. Future studies should include physiological measures in network models to provide a more accurate understanding.