Eiko I. Fried

Estimating Psychological Networks and their Accuracy : A tutorial paper

The usage of psychological networks that conceptualize behavior as a complex interplay of psychological and other components has gained increasing popularity in various research fields. While prior publications have tackled the topics of estimating and interpreting such networks, little work has been conducted to check how accurate (i.e., prone to sampling variation) networks are estimated, and how stable (i.e., interpretation remains similar with less observations) inferences from the network structure (such as centrality indices) are. In this tutorial paper, we aim to introduce the reader to this field and tackle the problem of accuracy under sampling variation. We first introduce the current state-of-the-art of network estimation. Second, we provide a rationale why researchers should investigate the accuracy of psychological networks. Third, we describe how bootstrap routines can be used to (A) assess the accuracy of estimated network connections, (B) investigate the stability of centrality indices, and (C) test whether network connections and centrality estimates for different variables differ from each other. We introduce two novel statistical methods: for (B) the correlation stability coefficient, and for (C) the bootstrapped difference test for edge-weights and centrality indices. We conducted and present simulation studies to assess the performance of both methods. Finally, we developed the free R-package bootnet that allows for estimating psychological networks in a generalized framework in addition to the proposed bootstrap methods. We showcase bootnet in a tutorial, accompanied by R syntax, in which we analyze a dataset of 359 women with posttraumatic stress disorder available online.

Mental disorders as networks of problems: a review of recent insights

PurposeThe network perspective on psychopathology understands mental disorders as complex networks of interacting symptoms. Despite its recent debut, with conceptual foundations in 2008 and empirical foundations in 2010, the framework has received considerable attention and recognition in the last years.MethodsThis paper provides a review of all empirical network studies published between 2010 and 2016 and discusses them according to three main themes: comorbidity, prediction, and clinical intervention.ResultsPertaining to comorbidity, the network approach provides a powerful new framework to explain why certain disorders may co-occur more often than others. For prediction, studies have consistently found that symptom networks of people with mental disorders show different characteristics than that of healthy individuals, and preliminary evidence suggests that networks of healthy people show early warning signals before shifting into disordered states. For intervention, centrality—a metric that measures how connected and clinically relevant a symptom is in a network—is the most commonly studied topic, and numerous studies have suggested that targeting the most central symptoms may offer novel therapeutic strategies.ConclusionsWe sketch future directions for the network approach pertaining to both clinical and methodological research, and conclude that network analysis has yielded important insights and may provide an important inroad towards personalized medicine by investigating the network structures of individual patients.

Moving Forward: Challenges and Directions for Psychopathological Network Theory and Methodology:

Since the introduction of mental disorders as networks of causally interacting symptoms, this novel framework has received considerable attention. The past years have resulted in over 40 scientific publications and numerous conference symposia and workshops. Now is an excellent moment to take stock of the network approach: What are its most fundamental challenges, and what are potential ways forward in addressing them? After a brief conceptual introduction, we first discuss challenges to network theory: (1) What is the validity of the network approach beyond some commonly investigated disorders such as major depression? (2) How do we best define psychopathological networks and their constituent elements? And (3) how can we gain a better understanding of the causal nature and real-life underpinnings of associations among symptoms? Next, after a short technical introduction to network modeling, we discuss challenges to network methodology: (4) heterogeneity of samples studied with network analytic models, and (5) a lurking replicability crisis in this strongly data-driven and exploratory field. Addressing these challenges may propel the network approach from its adolescence into adulthood and promises advances in understanding psychopathology both at the nomothetic and idiographic level.

The 52 symptoms of major depression: Lack of content overlap among seven common depression scales

BACKGROUNDnDepression severity is assessed in numerous research disciplines, ranging from the social sciences to genetics, and used as a dependent variable, predictor, covariate, or to enroll participants. The routine practice is to assess depression severity with one particular depression scale, and draw conclusions about depression in general, relying on the assumption that scales are interchangeable measures of depression. The present paper investigates to which degree 7 common depression scales differ in their item content and generalizability.nnnMETHODSnA content analysis is carried out to determine symptom overlap among the 7 scales via the Jaccard index (0=no overlap, 1=full overlap). Per scale, rates of idiosyncratic symptoms, and rates of specific vs. compound symptoms, are computed.nnnRESULTSnThe 7 instruments encompass 52 disparate symptoms. Mean overlap among all scales is low (0.36), mean overlap of each scale with all others ranges from 0.27 to 0.40, overlap among individual scales from 0.26 to 0.61. Symptoms feature across a mean of 3 scales, 40% of the symptoms appear in only a single scale, 12% across all instruments. Scales differ regarding their rates of idiosyncratic symptoms (0-33%) and compound symptoms (22-90%).nnnLIMITATIONSnFuture studies analyzing more and different scales will be required to obtain a better estimate of the number of depression symptoms; the present content analysis was carried out conservatively and likely underestimates heterogeneity across the 7 scales.nnnCONCLUSIONnThe substantial heterogeneity of the depressive syndrome and low overlap among scales may lead to research results idiosyncratic to particular scales used, posing a threat to the replicability and generalizability of depression research. Implications and future research opportunities are discussed.

False alarm? A comprehensive reanalysis of "Evidence that psychopathology symptom networks have limited replicability" by Forbes, Wright, Markon, and Krueger (2017)

Forbes, Wright, Markon, and Krueger (2017) stated that “psychopathology networks have limited replicability” (p. 1011) and that “popular network analysis methods produce unreliable results” (p. 1011). These conclusions are based on an assessment of the replicability of four different network models for symptoms of major depression and generalized anxiety across two samples; in addition, Forbes et al. analyzed the stability of the network models within the samples using split-halves. Our reanalysis of the same data with the same methods led to results directly opposed to theirs: All network models replicated very well across the two data sets and across the split-halves. We trace the differences between Forbes et al.’s results and our own to the fact that they did not appear to accurately implement all network models and used debatable metrics to assess replicability. In particular, they deviated from existing estimation routines for relative importance networks, did not acknowledge the fact that the skip structure used in the interviews strongly distorted correlations between symptoms, and incorrectly assumed that network structures and metrics should be the same not only across the different samples but also across the different network models used. In addition to a comprehensive reanalysis of the data, we end with a discussion of best practices concerning future research into the replicability of psychometric networks.

Replicability and Generalizability of Posttraumatic Stress Disorder (PTSD) Networks : A Cross-Cultural Multisite Study of PTSD Symptoms in Four Trauma Patient Samples

The growing literature conceptualizing mental disorders like posttraumatic stress disorder (PTSD) as networks of interacting symptoms faces three key challenges. Prior studies predominantly used (a) small samples with low power for precise estimation, (b) nonclinical samples, and (c) single samples. This renders network structures in clinical data, and the extent to which networks replicate across data sets, unknown. To overcome these limitations, the present cross-cultural multisite study estimated regularized partial correlation networks of 16 PTSD symptoms across four data sets of traumatized patients receiving treatment for PTSD (total N = 2,782). Despite differences in culture, trauma type, and severity of the samples, considerable similarities emerged, with moderate to high correlations between symptom profiles (0.43–0.82), network structures (0.62–0.74), and centrality estimates (0.63–0.75). We discuss the importance of future replicability efforts to improve clinical psychological science and provide code, model output, and correlation matrices to make the results of this article fully reproducible.

Network Structure of Perinatal Depressive Symptoms in Latinas: Relationship to Stress and Reproductive Biomarkers

Based on emerging evidence, mood disorders can be plausibly conceptualized as networks of causally interacting symptoms, rather than as latent variables of which symptoms are passive indicators. In an innovative approach in nursing research, we used network analysis to estimate the network structure of 20 perinatal depressive (PND) symptoms. Then, two proof-of-principle analyses are presented: Incorporating stress and reproductive biomarkers into the network, and comparing the network structure of PND symptoms between non-depressed and depressed women. We analyzed data from a cross-sectional sample of 515 Latina women at the second trimester of pregnancy and estimated networks using regularized partial correlation network models. The main analysis yielded five strong symptom-to-symptom associations (e.g., cry-sadness), and five symptoms of potential clinical importance (i.e., high centrality) in the network. In exploring the relationship of PND symptoms to stress and reproductive biomarkers (proof-of-principle analysis 1), a few weak relationships were found. In a comparison of non-depressed and depressed womens networks (proof-of-principle analysis 2), depressed participants had a more connected network of symptoms overall, but the networks did not differ in types of relationships (the network structures). We hope this first report of PND symptoms as a network of interacting symptoms will encourage future network studies in the realm of PND research, including investigations of symptom-to-biomarker mechanisms and interactions related to PND. Future directions and challenges are discussed.

Assessment of Symptom Network Density as a Prognostic Marker of Treatment Response in Adolescent Depression

One in 4 adolescents with depression does not respond favorably to treatment.1 Prognostic markers to identify this nonresponder group are lacking and urgently needed.2 It has been suggested that the network structure of depressive symptoms (ie, group-level covariance or connectivity between symptoms) may be informative in this regard.3 Intuitively, one may expect that more densely connected networks would be more inclined to result in negative spirals (eg, sleeplessness causes an individual to be too tired to go out, which leads to a lack of friends, resulting in sadness) and therefore more liable to nonresponse. An influential naturalistic study by van Borkulo et al published in JAMA Psychiatry3 reported that adult patients with depression who continue to experience problems in subsequent years have more densely connected networks at baseline than patients who later recover. Here, we performed a conceptual replication of that study in adolescents with depression who participated in a psychological treatment trial. We tested whether network characteristics at baseline were prognostic for long-term outcomes.1

The centrality of DSM and non-DSM depressive symptoms in Han Chinese women with major depression

INTRODUCTIONnWe compared DSM-IV criteria for major depression (MD) with clinically selected non-DSM criteria in their ability to represent clinical features of depression.nnnMETHODnWe conducted network analyses of 19 DSM and non-DSM symptoms of MD assessed at personal interview in 5952 Han Chinese women meeting DSM-IV criteria for recurrent MD. We estimated an Ising model (the state-of-the-art network model for binary data), compared the centrality (interconnectedness) of DSM-IV and non-DSM symptoms, and investigated the community structure (symptoms strongly clustered together).nnnRESULTSnThe DSM and non-DSM criteria were intermingled within the same symptom network. In both the DSM-IV and non-DSM criteria sets, some symptoms were central (highly interconnected) while others were more peripheral. The mean centrality of the DSM and non-DSM criteria sets did not significantly differ. In at least two cases, non-DSM criteria were more central than symptomatically related DSM criteria: lowered libido vs. sleep and appetite changes, and hopelessness versus worthlessness. The overall network had three sub-clusters reflecting neurovegetative/mood symptoms, cognitive changes and anxiety/irritability.nnnLIMITATIONSnThe sample were severely ill Han Chinese females limiting generalizability.nnnCONCLUSIONSnConsistent with prior historical reviews, our results suggest that the DSM-IV criteria for MD reflect one possible sub-set of a larger pool of plausible depressive symptoms and signs. While the DSM criteria on average perform well, they are not unique and may not be optimal in their ability to describe the depressive syndrome.

A reassessment of the relationship between depression and all-cause mortality in 3,604,005 participants from 293 studies

As reported in the February issue of this journal, over three decades of research suggest that depression is associated with an increased risk of all-cause mortality, although some large recent studies have found negative or null associations. To better inform clinical decision making and evidence-based service provision, it is crucial to resolve this discrepancy. Here we summarize the principal findings of the largest ever investigation of the relationship between depression and all-cause mortality, comprising 3,604,005 participants and over 417,901 deaths, based on a reassessment of 293 studies derived from 15 systematic reviews. We observed that several factors moderate the relationship between depression and mortality, and found no evidence of an association when controlling for comorbid mental disorders and health behaviors (see https:// osf.io/svywu/ for the complete report and the extracted data). The purpose of this reassessment was to better understand the features of studies that have sought to address the depression-mortality relationship, to delineate some methodological reasons for heterogeneity between studies (sample size and characteristics, number of deaths and follow-up periods, and adjustment for mental disorders and health behaviors), and to explore whether estimates of the relationship between depression and mortality on the basis of the methodologically most rigorous studies differed from those of previous meta-analyses. The three main results of the study are as follows. First, there was a pronounced publication bias, as indicated by the positive intercept (1.02; 95% CI: 0.72-1.31) of effect estimates on their standard errors favoring imprecise studies with large positive associations. The largest estimates consistently came from studies with small samples, low number of deaths, and brief follow-up periods. Second, only 16 ( 5%) of the included studies adjusted for at least one comorbid mental condition. This is surprising, given that more than half of individuals diagnosed with major depressive disorder suffer from at least one additional comorbid mental disorder in their lifetime. The pooled relative risk (RR) of these 16 estimates (1.08; 95% CI: 0.98-1.18) was smaller than the RR of the 266 estimates that were unadjusted for comorbid mental disorders (1.33; 95% CI: 1.29-1.37). Additionally, there was no evidence of an association between depression and all-cause mortality among the fraction of eight of these estimates that also adjusted for health behaviors (smoking, drinking or physical inactivity) (1.04; 95% CI: 0.87-1.21). Third, apart from sample size, follow-up duration, and lack of adjustment for important variables, other substantial sources of heterogeneity between studies emerged. Over two-thirds of the estimates comprised respondents who were pre-selected on the basis of medical conditions. This is problematic, because many symptoms of major depression (e.g., insomnia, fatigue) are shared with various physical conditions, or may arise as side effects of medications used to treat existing pathologies. Preselecting participants on the basis of medical conditions could therefore result in confounding by reverse causality among those who are physically unwell at baseline. Given that somatic symptoms that are not confounded by physical conditions are integral to a diagnosis of major depression, studies based on medical samples that use rating scales (instead of diagnostic interviews that query the source of these somatic symptoms) may be particularly likely to misclassify individuals who are of relatively poorer health as depressed. Furthermore, we found that over forty different instruments were used to measure depressive symptoms, which is problematic due to the considerable content heterogeneity among commonly used instruments. Even studies that used the same questionnaire frequently adopted different cutoff scores for a probable diagnosis of major depressive disorder. The interaction of three of the aforementioned points – the use of scales encompassing physical symptoms that may indicate comorbid medical conditions; the use of samples pre-selected based on medical conditions; the lack of adjustment for comorbidities when estimating the effect of major depressive disorder on mortality – points to significant weaknesses in the literature. We therefore estimated the association of depression and mortality among studies that used DSM-based structured interviews requiring the presence of core depressive symptoms (sad mood or anhedonia) prior to assessing for more general physical, somatic and cognitive symptoms, in community-based samples and based on survival analysis methodology. Only four estimates (1% of all studies) met these criteria, among which the pooled hazard ratio was 1.17 (95% CI: 0.75-1.60). Given the overall poor quality of the available evidence, we are unable to draw strong conclusions about the relationship between depression and mortality. Studies with large samples, extensive follow-up periods, adjustment for mental disorders and health behaviors, and time-to-event outcomes assessed using survival analysis methodology are especially needed. More work of a higher quality is also required to examine which variables related to depression and mortality may modify this relationship. For example, the subsequent onset of health behaviors such as smoking, drinking and physical inactivity appear to play an important role in mediating the risk of adverse cardiovascular outcomes among depressed individuals. This could account for a variety of adverse health outcomes that are not limited to cardiovascular disorders. Moreover, the risk of depression and mortality are both influenced by a subset of common variables. For example, smoking at baseline is associated with increased risk of depression onset at follow-up, and smoking is associated with many causes of death. More rigorous research is needed to better understand whether depression does, in fact, pose an increased risk of all-cause mortality. We hope that our work will encourage such efforts.