* denotes shared first authorship; SM = Supplementary Materials (often including data and syntax).
Submitted / under revision
» Epskamp, S. & Fried, E. I. (under revision). A Tutorial on Regularized Partial Correlation Networks. Psychological Methods. Preprint & SM. (Abstract)
Recent years have seen an emergence of network modeling for psychological behaviors, moods and attitudes. In this framework, psychological variables are understood to directly interact with each another rather than being caused by an unobserved latent entity. Here we introduce the reader to the most popularly used network model for estimating such psychological networks: the partial correlation network. We describe how regularization techniques can be used to efficiently estimate a parsimonious and interpretable network structure on cross-sectional psychological data. We demonstrate the method in an empirical example on post-traumatic stress disorder data, showing the effect of the hyperparameter that needs to be manually set by the researcher. In addition, we list several common problems and questions arising in the estimation of regularized partial correlation networks.
» Bos, F. M., Fried, E. I., Hollon, S. D., Bringmann, L. F., Dimidjian, S., DeRubeis, R. J., & Bockting, C. L. H. (submitted). Unraveling the working mechanisms of antidepressant medication treatment for major depressive disorder: A network approach.
» Monden, R., Wardenaar, K. J., Stegeman, A., Fried, E. I., de Jonge, P. (submitted). Decomposition of depression heterogeneity on the Person-, Symptom- and Time-level during the acute phase of antidepressant therapy: the STAR*D study. Psychological Medicine.
» Fried, E. I. & Cramer, A. O. J. (accepted). Moving forward: challenges and directions for psychopathological network theory and methodology. Perspectives on Psychological Science. Preprint & SM. (Abstract)
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? (3) And 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.
» Fried, E. I. (accepted). What are psychological constructs? On the nature and statistical modeling of emotions, intelligence, personality traits and mental disorders. Health Psychology Review. Preprint.
» Miloyan, B. & Fried, E. I. (accepted). A reassessment of the relationship between depression and all-cause mortality in 3,604,005 participants from 293 studies. World Psychiatry. Preprint & SM. (Abstract)
The objective of this study is to explain inconsistencies in the relationship between depression and all-cause mortality by performing a reassessment of the included studies of previous systematic reviews. We assessed study-level methodological variables with a focus on sample size and follow-up period, measurement and classification of depression, and model adjustment. We included the constituent studies of fifteen systematic reviews on depression and mortality, yielding 488 articles after the removal of duplicates. 333 studies were extracted, 40 of which used data that overlapped with other included studies. We included 313 estimates from 293 articles in the meta-analysis with a total sample of 3,604,005 participants and over 417,901 deaths. We identified a pronounced publication bias favoring large, positive associations in imprecise studies. Several factors moderated the relationship between depression and mortality. Most importantly, the 20 estimates adjusting for at least one comorbid mental condition (Pooled Effect: 1.08; 95% CI: 0.98-1.18), and the fraction of 8 of those estimates also adjusting for health variables (e.g., smoking, alcohol use, or physical inactivity; Pooled Effect: 1.04; 95% CI: 0.87-1.21), reported considerably smaller associations than the 204 unadjusted estimates (Pooled Effect: 1.32; 95% CI: 1.28-1.36). The sizable relationship of depression and mortality reported in previous systematic reviews is largely based on low-quality studies; controlling for important covariates attenuates the association considerably. Higher quality studies are needed based on large community samples, extensive follow-up, adjustment for health behaviors and mental disorders, and time-to-event outcomes based on survival analysis methodology.
» Haslbeck, J. M. B. & Fried, E. I. (accepted). How Predictable are Symptoms in Psychopathological Networks? A Reanalysis of 18 Published Datasets. Psychological Medicine. Preprint & SM. (Abstract)
Background. Network analyses on psychopathological data focus on the network structure and its derivatives such as node centrality. One conclusion one can draw from centrality measures is that the node with the highest centrality is likely to be the node that is determined most by its neighboring nodes. However, centrality is a relative measure: knowing that a node is highly central gives no information about the extent to which it is determined by its neighbors. Here we provide an absolute measure of determination (or controllability) of a node – its predictability. We introduce predictability, estimate the predictability of all nodes in 18 prior empirical network papers on psychopathology, and statistically relate it to centrality.
Methods. We carried out a literature review and collected 25 datasets from 18 published papers in the field (several mood and anxiety disorders, substance abuse, psychosis, autism, and transdiagnostic data). We fit state-of-the-art network models to all datasets, and computed the predictability of all nodes.
Results. Predictability was unrelated to sample size, moderately high in most symptom networks, and differed considerable both within and between datasets. Predictability was higher in community than clinical samples, highest for mood and anxiety disorders, and lowest for psychosis.
Conclusions. Predictability is an important additional characterization of symptom networks because it gives an absolute measure of the controllability of each node. It allows conclusions about how self-determined a symptom network is, and may help to inform intervention strategies. Limitations of predictability along with future directions are discussed.
» Dejonckheere, E., Brock, B., Fried, E. I., Murphy, S., & Kuppens, P. (accepted). Perceiving social pressure not to feel negative predicts depressive symptoms in daily life. Depression and Anxiety.
» Fried, E. I. (2017). The 52 symptoms of major depression: lack of content overlap among seven common depression scales. Journal of Affective Disorders. PDF, SM. (Abstract)
Background: Depression 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.
Methods: A 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.
Results: The 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 0.27–0.40, overlap among individual scales from 0.26–0.61. 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%).
Limitations: Future 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.
Conclusion: The 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.
» Epskamp, S., Borsboom, D., Fried, E. I. (2017). Estimating psychological networks and their accuracy: a tutorial paper. Behavioral Research Methods. PDF, SM, Blog. (Abstract)
The usage of psychological networks that conceptualize psychological behavior as a complex interplay of psychological and other components has gained increasing popularity in various fields of psychology. 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.
» Fried, E. I. (2017). “Moving forward: How depression heterogeneity hinders progress in treatment and research” Expert Review of Neurotherapeutics. PDF. (Abstract)
Has depression research in the last two decades led to major improvements of clinical care for patients? Pharmacological innovations have not resulted in considerable progress, and although the biology of Major Depression (MD) is among the most-studied and best-funded topics, answers remain elusive. Why is that? The current editorial argues that depression heterogeneity is at the heart of the slow progress, and suggests that a novel research framework may provide an inroad to new insights. Most researchers sum disparate depression symptoms such as sadness, insomnia, and concentration problems to one sumscore representing depression severity, and use thresholds to diagnose MD. This leads to highly heterogeneous depressed samples in which patients often have few symptoms in common. Recent work suggests that this unaddressed heterogeneity explains the lack of research progress. A novel research framework—Depression Symptomics—may offer a way forward by focusing on the study of individual depression symptoms, their causal interactions, and by paying attention to differences among patients diagnosed with MD.
» Armour, C.*, Fried, E. I.*, Deserno, M. K., Tsai, J., & Pietrzak, R. H. (2017). A Network Analysis of DSM-5 posttraumatic stress disorder symptoms and correlates in U.S. military veterans. Journal of Anxiety Disorders. PDF, SM. (Abstract)
Object: Recent developments in psychometrics enable the application of network models to analyze psychological disorders, such as PTSD. Instead of understanding symptoms as indicators of an underlying common cause, this approach suggests symptoms co-occur in syndromes due to causal interactions. The current study has two goals: (1) examine the network structure among the 20 DSM-5 PTSD symptoms, and (2) incorporate clinically relevant variables to the network to investigate whether PTSD symptoms exhibit differential relationships with suicidal ideation, depression, anxiety, physical quality of life (QoL), mental QoL, age, and sex. Method: We utilized a nationally representative U.S. military veteran’s sample; and analyzed the data from a subsample of 221 veterans who reported clinically significant DSM-5 PTSD symptoms. Networks were estimated using Gaussian graphical models and the lasso regularization. Results: The 20-item DSM-5 PTSD network revealed that symptoms were positively connected within the network. The most central symptoms were negative trauma-related emotions, flashbacks, detachment, and physiological cue reactivity. Especially strong connections emerged between nightmares and flashbacks; blame of self or others and negative trauma-related emotions, detachment and restricted affect; and hypervigilance and exaggerated startle response. Incorporation of clinically relevant covariates into the network revealed paths between self-destructive behavior and suicidal ideation; concentration difficulties and anxiety, depression, and mental QoL; and depression and restricted affect. Conclusion: These results demonstrate the utility of a network approach in modeling the structure of DSM-5 PTSD symptoms, and suggest differential associations between specific DSM-5 PTSD symptoms and clinical outcomes in trauma survivors. Implications of these results for informing the assessment and treatment of this disorder, are discussed.
» Santos, H.*, Fried, E. I.*, Asafu-Adjei, J., & Ruiz, J. (2017). Network Structure of Perinatal Depressive Symptoms in Latinas: Relationship to Stress-Related and Reproductive Biomarkers. Research in Nursing & Health. PDF, SM. (Abstract)
Emerging evidence shows that mood disorders can be plausibly conceptualized as networks of causally interacting symptoms rather than as latent variables where symptoms are passive indicators. Here we introduce an empirical application of network analysis to nursing research by estimating the network structure of 20 perinatal depressive (PND) symptoms. Further, we conduct two proof-of-principle analyses: incorporating stress-related and reproductive hormone variables into the network, and comparing the network structure of PND symptoms between healthy and depressed women. We analyzed a cross-sectional sample of 515 Latina women at the second trimester of pregnancy; networks were estimated using the Gaussian Graphical Model via lasso regularization. The main analysis yielded five strong symptom-to-symptom associations among PND symptoms (e.g., lack of happiness—lack of joy), and five symptoms of considerable clinical importance (i.e. high centrality) in the network. In exploring the relationship of PND symptoms with stress-related and reproductive biomarkers (proof-of-principle 1), biomarkers had few and weak relationships with PND symptoms. A comparison of healthy and depressed networks (proof-of-principle 2) showed that depressed participants have an overall more connected network, but that there is no difference regarding the type of relationships in the network. Our study provides preliminary evidence that PND symptoms can be conceived of as a network of interacting symptoms, and hope they will encourage future network studies in the realm of PND research, including investigations of symptom-to-biomarker mechanisms and interactions related to perinatal depression. Future directions and challenges are discussed.
» Fried, E. I.*, van Borkulo, C. D.*, Cramer, A. O. J., Lynn, B., Schoevers, R. A., Borsboom, D. (2016). Mental disorders as networks of problems: a review of recent insights. Social Psychiatry and Psychiatric Epidemiology. PDF, SM. (Abstract)
The 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. This 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. Pertaining 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 therapeutic interventions aimed at the most central symptoms may offer novel powerful clinical interventions. We conclude by sketching 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.
» Fried, E. I. (2016). Are more responsive depression scales really superior depression scales? Journal of Clinical Epidemiology. PDF. (Abstract)
Many different rating scales are used in contemporary research to measure depression. The responsiveness of these instruments—defined as a scale’s ability to detect clinical change—has become an important research topic. Responsiveness is commonly understood as a psychometric property that can inform about the quality of a scale, and scales more responsive to treatment have been advanced as superior scales that should be used in clinical trials. This conclusion is unwarranted and can severely bias research outcomes, because depression scales fail to meet three necessary requirements to interpret responsiveness as clinical change: unidimensionality, temporal measurement invariance, and content validity. The most responsive depression instruments are short scales that do not capture all clinically relevant depression symptoms, and thus cannot provide adequate assessments of the patients’ improvement. Furthermore, since scales differ considerably in symptom content, scale responsiveness depends on the particular type of treatment. The Beck Depression Inventory, for example, focuses on cognitive symptoms and will likely show higher levels of responsiveness in cognitive behavioral therapy trials than drug trials. Responsiveness is not an inherently desirable property of depression scales because it is not a valid measure of clinical change.
» Fried, E. I., van Borkulo, C. D., Epskamp, S., Schoevers, R. A., Tuerlinckx, F., & Borsboom, D. (2016). Measuring Depression over Time … or not? Lack of Unidimensionality and Longitudinal Measurement Invariance in Four Common Rating Scales of Depression. Psychological Assessment. PDF, Blog, Data1, Data2, SM. (Abstract)
In depression research, symptoms are routinely assessed via rating scales and added to construct sum-scores. These scores are used as a proxy for depression severity in cross-sectional research, and differences in sum-scores over time are taken to reflect changes in an underlying depression construct. To allow for such interpretations, rating scales must (1) measure a single construct, and (2) measure that construct in the same way across time. These requirements are referred to as unidimensionality and measurement invariance. We investigated these two requirements in two large prospective studies (combined n=3,509) in which overall depression levels decrease, examining four common depression rating scales (one self-report, three clinician-report) with different time intervals between assessments (between 6 weeks and 2 years). A consistent pattern of results emerged. For all instruments, neither unidimensionality nor measurement invariance appeared remotely tenable. At least 3 factors were required to describe each scale, and the factor structure changed over time. Typically, the structure became less multifactorial as people improved (without however reaching unidimensionality). The decrease in the sum-scores was accompanied by an increase in the variances of the sum-scores, and sharp increases in internal consistency. These findings challenge the common interpretation of sum-scores and their changes as reflecting one underlying construct. We discuss the possible causes of the violations such as response shift bias, restriction of range, and regression to the mean. The violations of common measurement requirements are sufficiently severe to suggest alternative interpretations of depression sum-scores as formative instead of reflective measures.
» Fried, E. I., Epskamp, S., Nesse, R. M, Tuerlinckx, F., & Borsboom, D. (2016). What are ‘good’ depression symptoms? Comparing the centrality of DSM and non-DSM symptoms of depression in a network analysis. Journal of Affective Disorders. PDF, Blog, SM, Data, Syntax. (Abstract)
Background: The symptoms for Major Depression (MD) defined in the DSM-5 differ markedly from symptoms assessed in common rating scales, and the empirical question about core depression symptoms is unresolved. Here we conceptualize depression as a complex dynamic system of interacting symptoms to examine what symptoms are most central to driving depressive processes.
Methods: We constructed a network of 28 depression symptoms assessed via the Inventory of Depressive Symptomatology (IDS-30) in 3,463 depressed outpatients from the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study. We estimated the centrality of all IDS-30 symptoms, and compared the centrality of DSM and non-DSM symptoms; centrality reflects the connectedness of each symptom with all other symptoms.
Results: A network with 28 intertwined symptoms emerged, and symptoms differed substantially in their centrality values. Both DSM symptoms (e.g., sad mood) and non-DSM symptoms (e.g., anxiety) were among the most central symptoms, and DSM criteria were not more central than non-DSM symptoms.
Limitations: Many subjects enrolled in STAR*D reported comorbid medical and psychiatric conditions which may have affected symptom presentation.
Conclusion: The network perspective neither supports the standard psychometric notion that depression symptoms are equivalent indicators of MD, nor the common assumption that DSM symptoms of depression are of higher clinical relevance than non-DSM depression symptoms. The findings suggest the value of research focusing on especially central symptoms to increase the accuracy of predicting outcomes such as the course of illness, probability of relapse, and treatment response.
» Fried, E. I., & Kievit, R. (2016). The volumes of subcortical regions in depressed and healthy individuals are strikingly similar: A reinterpretation of the results by Schmaal et al. Molecular Psychiatry. PDF, Blog, SM. (Abstract)
In their recent meta-analysis of magnetic resonance imaging data from 15 research samples worldwide, Schmaal et al.1 examined the structural differences of nine subcortical brain volumes between 1728 patients with major depressive disorder (MDD) and 7199 healthy participants. In the authors’ univariate analyses, none of the nine volumes was associated with depression severity, and only hippocampal volume was significantly decreased in MDD patients compared to controls, with the largest effect being observed in the recurrent MDD group (difference 1.4%, Cohen’s d 0.17). The study is the result of a huge collaborative effort, and we commend the authors on their insightful manuscript. However, as the findings present the best empirical evidence currently available, an accurate interpretation of the results is paramount, especially considering the report´s goal to “robustly discriminate MDD patients from healthy controls” (p. 1). We therefore add two observations and future research directions.
» Rhemtulla, M.*, Fried, E. I.*, Aggen, S. H., Tuerlinckx, F., Kendler, K. S., Borsboom, D. (2016). Network analysis of substance abuse and dependence symptoms. Drug and Alcohol Dependence. PDF, SM1, SM2. (Abstract)
Background: The DSM uses one set of abuse and dependence criteria to assess multiple substance use disorders (SUDs). Most SUD research aggregates across these symptoms to study the behavior of SUD as a static construct. We use an alternative approach that conceptualizes symptoms as directly interacting variables in psychopathological networks. We apply network models to symptom-level data to investigate the unique roles of individual symptoms and their interactions in SUD.
Methods: We analyzed 11 DSM III-R/IV abuse and dependence criteria in a sample of 2405 adult twins who reported use of at least one illicit substance six or more times from the Virginia Adult Twin Study of Psychiatric and Substance Use Disorders (VATSPSUD). We estimated a symptom network for each substance class as well as a global network collapsed across all substance classes. We examined similarities and differences across the 6 networks in terms of symptom-to-symptom connections and symptom centrality.
Results: The global network model revealed several interesting symptom connections, such as a strong predictive relation between tolerance and more-than-planned substance use. The most central symptom was using a drug more than planned. In addition, several interesting differences across substances emerged, both in the strength of symptom connections as well as the centrality of symptoms to each network.
Conclusions: When analyzed as networks, abuse and dependence symptoms do not function equivalently across illicit substance classes. These findings suggest the value of analyzing individual symptoms and their associations to gain new insight into the mechanisms of SUD.
» Beard, C., Millner A. J., Forgeard, M., Fried, E. I., Hsu, J. K., Treadway, M., Leonard, C. V., Kertz, S., & Björgvinsson, T. (2016). Network Analysis of Depression and Anxiety Symptom Relations in a Psychiatric Sample. Psychological Medicine. PDF. (Abstract)
Researchers have studied psychological disorders extensively from a common cause perspective, in which symptoms are treated as independent indicators of an underlying disease. In contrast, the causal systems perspective seeks to understand the importance of individual symptoms and symptom-to-symptom relations. In the current study, we used network analysis to examine the relationships between and among depression and anxiety symptoms from the causal systems perspective. We utilized data from a large psychiatric sample at admission and discharge from a partial hospital program (N = 1029, mean treatment duration = 8 days). We investigated features of the depression/anxiety network including topology, network centrality, stability of the network at admission and discharge, as well as change in the network over the course of treatment. Results revealed that individual symptoms of depression and anxiety were more related to other symptoms within each disorder than to symptoms between disorders. Sad mood and worry were among the most central symptoms in the network. The network structure was stable both at admission and between admission and discharge, although the overall strength of symptom relationships increased as symptom severity decreased over the course of treatment. Examining depression and anxiety symptoms as dynamic systems may provide novel insights into the maintenance of these mental health problems.
» Heylen, J., van Mechelen, I., Fried, E. I., Ceulemans, E. (2016). Two-mode K-Spectral Centroid analysis for studying multivariate time profiles. Chemometrics and Intelligent Laboratory Systems. PDF. (Abstract)
In many scientific areas, researchers collect multivariate time profile data on the evolution of a set of variables across time for multiple persons. For instance, clinical studies often focus on the effects of an intervention on different symptoms for multiple persons, by repeatedly measuring symptom severity for each symptom and each person. To pursue an insightful overview on how these time profiles vary as a function of both symptoms and persons, we propose two-mode K-Spectral Centroid (2M-KSC) analysis, which is a multivariate extension of K-Spectral Centroid analysis. Specifically, 2M-KSC assigns the persons to a few person clusters and the symptoms to a few symptom clusters and imposes that the time profiles that correspond to a specific combination of a person cluster and a symptom cluster have the same shape, but may vary in amplitude scaling. An algorithm for fitting 2M-KSC is proposed and evaluated in a simulation study. Finally, the new method is applied to time profiles regarding the severity of depression symptoms during a citalopram treatment.
» van Loo, H. M., Wanders, R. B. K., Wardenaar, K. J., & Fried, E. I. (2016). Problems with latent class analysis to detect data-driven subtypes of depression. Molecular Psychiatry.
» Fried, E. I., & Nesse, R. M (2015). Depression sum-scores don’t add up: Why analyzing specific depression symptoms is essential. BMC Medicine. PDF, Blog. (Abstract)
Most measures of depression severity are based on the number of reported symptoms, and threshold scores are often used to classify individuals as healthy or depressed. This method – and research results based on it – are valid if depression is a single condition, and all symptoms are equally good severity indicators. Here, we review a host of studies documenting that specific depressive symptoms like sad mood, insomnia, concentration problems, and suicidal ideation are distinct phenomena that differ from each other in important dimensions such as underlying biology, impact on impairment, and risk factors. Furthermore, specific life events predict increases in particular depression symptoms, and there is evidence for direct causal links among symptoms. We suggest that the pervasive use of sum-scores to estimate depression severity has obfuscated crucial insights and contributed to the lack of progress in key research areas such as identifying biomarkers and more efficacious antidepressants. The analysis of individual symptoms and their causal associations offers a way forward. We offer specific suggestions with practical implications for future research.
» Fried, E. I. (2015). Problematic assumptions have slowed down depression research: why symptoms, not syndromes are the way forward. Frontiers in Psychology. PDF, Blog. (Abstract)
Major depression (MD) is a highly heterogeneous diagnostic category. Diverse symptoms such as sad mood, anhedonia, and fatigue are routinely added to an unweighted sum-score, and cutoffs are used to distinguish between depressed participants and healthy controls. Researchers then investigate outcome variables like MD risk factors, biomarkers, and treatment response in such samples. These practices presuppose that (1) depression is a discrete condition, and that (2) symptoms are interchangeable indicators of this latent disorder. Here I review these two assumptions, elucidate their historical roots, show how deeply engrained they are in psychological and psychiatric research, and document that they contrast with evidence. Depression is not a consistent syndrome with clearly demarcated boundaries, and depression symptoms are not interchangeable indicators of an underlying disorder. Current research practices lump individuals with very different problems into one category, which has contributed to the remarkably slow progress in key research domains such as the development of efficacious antidepressants or the identification of biomarkers for depression. The recently proposed network framework offers an alternative to the problematic assumptions. MD is not understood as a distinct condition, but as heterogeneous symptom cluster that substantially overlaps with other syndromes such as anxiety disorders. MD is not framed as an underlying disease with a number of equivalent indicators, but as a network of symptoms that have direct causal influence on each other: insomnia can cause fatigue which then triggers concentration and psychomotor problems. This approach offers new opportunities for constructing an empirically based classification system and has broad implications for future research.
» Fried, E. I., Nesse, R. M., Guille, C., & Sen, S. (2015). The differential influence of life stress on individual symptoms of depression. Acta Psychiatrica Scandinavica. PDF, Data. (Abstract)
Objective: Life stress consistently increases the incidence of major depression. Recent evidence has shown that individual symptoms of major depressive disorder (MDD) differ in important dimensions such as their genetic and etiological background, but the impact of stress on individual MDD symptoms is not known. Here, we assess whether stress affects depression symptoms differentially.
Method: We used the chronic stress of medical internship to examine changes of the nine Diagnostic and Statistical Manual (DSM)-5 criterion symptoms for depression in 3021 interns assessed prior to and throughout internship.
Results: All nine depression symptoms increased in response to stress, on average by 173%. Symptom increases differed substantially from each other, with psychomotor problems (289%) and interest loss (217%) showing the largest increases, and suicidal ideation (146%) and sleep problems (52%) the smallest. Symptoms also differed in their severities under stress: Fatigue, appetite problems and sleep problems were most prevalent; psychomotor problems and suicidal ideation were least prevalent.
Conclusion: Stress differentially affects the DSM-5 depressive symptoms. Analyses of individual symptoms reveal important insights obfuscated by sum-scores.
» Fried, E. I., Bockting, C., Arjadi, R., Borsboom, D., Tuerlinckx, F., Cramer, A., Epskamp, S., Amshoff, M., Carr, D., & Stroebe, M. (2015). From Loss to Loneliness: The Relationship Between Bereavement and Depressive Symptoms. Journal of Abnormal Psychology. PDF, Blog, Data, SM. (Abstract)
Spousal bereavement can cause a rise in depressive symptoms. This study empirically evaluates 2 competing explanations concerning how this causal effect is brought about: (a) a traditional latent variable explanation, in which loss triggers depression which then leads to symptoms; and (b) a novel network explanation, in which bereavement directly affects particular depression symptoms which then activate other symptoms. We used data from the Changing Lives of Older Couples (CLOC) study and compared depressive symptomatology, assessed via the 11-item Center for Epidemiologic Studies Depression Scale (CES-D), among those who lost their partner (N = 241) with a still-married control group (N = 274). We modeled the effect of partner loss on depressive symptoms either as an indirect effect through a latent variable, or as a direct effect in a network constructed through a causal search algorithm. Compared to the control group, widow(er)s’ scores were significantly higher for symptoms of loneliness, sadness, depressed mood, and appetite loss, and significantly lower for happiness and enjoyed life. The effect of partner loss on these symptoms was not mediated by a latent variable. The network model indicated that bereavement mainly affected loneliness, which in turn activated other depressive symptoms. The direct effects of spousal loss on particular symptoms are inconsis- tent with the predictions of latent variable models, but can be explained from a network perspective. The findings support a growing body of literature showing that specific adverse life events differentially affect depressive symptomatology, and suggest that future studies should examine interventions that directly target such symptoms.
» Fried, E. I., & Nesse, R. M. (2015). Depression is not a consistent syndrome: an investigation of unique symptom patterns in the STAR*D study. Journal of Affective Disorders. PDF, Data, Blog. (Abstract)
Background: The DSM-5 encompasses a wide range of symptoms for Major Depressive Disorder (MDD). Symptoms are commonly added up to sum-scores, and thresholds differentiate between healthy and depressed individuals. The underlying assumption is that all patients diagnosed with MDD have a similar condition, and that sum-scores accurately reflect the severity of this condition. To test this assumption, we examined the number of DSM-5 depression symptom patterns in the “Sequenced Treatment Alternatives to Relieve Depression” (STARnD) study. Methods: We investigated the number of unique symptom profiles reported by 3703 depressed outpatients at the beginning of the first treatment stage of STARnD. Results: Overall, we identified 1030 unique symptom profiles. Of these profiles, 864 profiles (83.9%) were endorsed by five or fewer subjects, and 501 profiles (48.6%) were endorsed by only one individual. The most common symptom profile exhibited a frequency of only 1.8%. Controlling for overall depression severity did not reduce the amount of observed heterogeneity. Limitations: Symptoms were dichotomized to construct symptom profiles. Many subjects enrolled in STARnD reported medical conditions for which prescribed medications may have affected symptom presentation. Conclusions: The substantial symptom variation among individuals who all qualify for one diagnosis calls into question the status of MDD as a specific consistent syndrome and offers a potential explanation for the difficulty in documenting treatment efficacy. We suggest that the analysis of individual symptoms, their patterns, and their causal associations will provide insights that could not be discovered in studies relying on only sum-scores.
» Fried, E. I., Boschloo, L., van Borkulo, C. D., Schoevers, R. A., Romeijn, J.-W., Wichers, M. C., de Jonge, P., Nesse, R. M., Tuerlinckx, F., Borsboom, D. (2015). Symptomics as a new research paradigm in psychiatry. Frontiers in Psychiatry. PDF, SM. (Abstract)
In the past decades, almost all research in psychiatry and clinical psychology has been directed at the level of disorders, such as major depressive disorder (MDD) or schizophrenia. As has been argued bymany scholars in recent work, this organization of the psychiatric research programhas yielded limited insights,which justifies the investigation of psychopathology at amore fine-grained level: the level of symptoms (1, 2). In the present letter, we indicate two primary directions for this research program, which we propose to call symptomics.We will focus our discussion on MDD specifically and discuss possibilities in relation to the recently published work byHieronymus et al. (3).
» Costantini, G., Richetin, J., Borsboom, D., Fried., E. I., Rhemtulla, M., Perugini, M. (2015). Development of Indirect Measures of Conscientiousness: Combining a Facets Approach and Network Analysis. European Journal of Personality. PDF. (Abstract)
Because indirect measures of personality self-concepts such as the Implicit Association Test (IAT) allow tapping into automatic processes, they can offer advantages over self-report measures. However, prior investigations have led to mixed results regarding the validity of indirect measures of conscientiousness. We suggest that these results might be due to a failure to consider the different facets of conscientiousness. These facets are of crucial importance because they are associated differentially with other psychobiological constructs and they are also characterized by different mechanisms. Therefore, focusing on facets while developing indirect measures of conscientiousness may improve the validity of such measures. In Study 1, we conducted a psycholexical investigation to develop one IAT for each conscientiousness facet. In Study 2, we examined the convergent and discriminant validities of each facet IAT in relation to self-report measures, peer-report measures and self-report behavioural indicators, and we investi- gated differential associations of the conscientiousness facets with working memory capacity and self-control. We employed network analysis as a novel approach to elucidate differential relationships involving personality facets. The results corroborated the convergent and discriminant validity of the conscientiousness facet IATs with self-reports and showed that the conscientiousness facets were differentially associated with working memory capacity and with self-control.
» Fried, E. I., Tuerlinckx, F., & Borsboom, D. (2014). Mental health: more than neurobiology. Nature. PDF. (Abstract)
Starting with the next grant cycle in June 2014, the NIMH will exclusively fund research examining the neurobiological roots of mental disorders (http://www.nature.com/news/nih-rethinks-psychiatry- trials-1.14877). The idea behind this decision is that important psychiatric diagnoses, such as mood and anxiety disorders, are due to brain abnormalities.
NIMH’s reductionist approach ignores important empirical facts. Despite decades of research, consistent and causal biological evidence in support of the idea that mental disorders are brain dysfunctions is nearly completely absent, and it is well-established that etiologies of common mental disorders are massively multi-factorial, featuring biological, psychological, and environmental influences.
Moreover, the notion of underlying neural mechanisms presupposes that symptoms of specific disorders cluster because they have the same biological cause, similar to the way measles cause fever, red eyes, and Koplik’s spots. This assumption renders symptoms passive and interchangeable indicators of a common cause. However, psychopathology symptoms of disorders such as depression differ dramatically in their etiological and genetic context: they do not share a common biological background, which is in direct contradiction with the disease model underlying NIMH’s decision.
Novel models of psychopathology have demonstrated that symptoms do not cluster because they share a common cause – they cluster because they are causally connected in complex networks of direct influence: insomnia leads to fatigue which in turn causes concentration and psychomotor problems, irrespective of the particular diagnosis a patient may have.
NIMH’s decision to solely focus on the investigation of neurobiological roots is directly opposed to these new insights. The majority of patients suffering from mental health problems do not have brain disorders. Instead, they are caught in vicious circles of problems that fuel each other. Consequently, NIMH’s funding policy may stall important clinical insights pertaining to the development, prevention, and intervention of mental disorders for years to come, and we suggest that it be reconsidered.
» Fried, E. I., & Nesse, R. M. (2014). The Impact of Individual Depressive Symptoms on Impairment of Psychosocial Functioning. PLoS ONE. PDF, Data. (Abstract)
Previous studies have established that scores on Major Depressive Disorder scales are correlated with measures of impairment of psychosocial functioning. It remains unclear, however, whether individual depressive symptoms vary in their effect on impairment, and if so, what the magnitude of these differences might be. We analyzed data from 3,703 depressed outpatients in the first treatment stage of the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study. Participants reported on the severity of 14 depressive symptoms, and stated to what degree their depression impaired psychosocial functioning (in general, and in the five domains work, home management, social activities, private activities, and close relationships). We tested whether symptoms differed in their associations with impairment, estimated unique shared variances of each symptom with impairment to assess the degree of difference, and examined whether symptoms had variable impacts across impairment domains. Our results show that symptoms varied substantially in their associations with impairment, and contributed to the total explained variance in a range from 0.7% (hypersomnia) to 20.9% (sad mood). Furthermore, symptoms had significantly different impacts on the five impairment domains. Overall, sad mood and concentration problems had the highest unique associations with impairment and were among the most debilitating symptoms in all five domains. Our findings are in line with a growing chorus of voices suggesting that symptom sum-scores obfuscate relevant differences between depressed patients and that substantial rewards will come from close attention to individual depression symptoms.
» Fried, E. I., Nesse, R. M., Zivin, K., Guille, C., & Sen, S. (2014). Depression is more than the sum score of its parts: individual DSM symptoms have different risk factors. Psychological Medicine. PDF, Data. (Abstract)
Background: For diagnostic purposes, the nine symptoms that compose the DSM-5 criteria for major depressive disorder (MDD) are assumed to be interchangeable indicators of one underlying disorder, implying that they should all have similar risk factors. The present study investigates this hypothesis, using a population cohort that shifts from low to elevated depression levels.
Method: We assessed the nine DSM-5 MDD criterion symptoms (using the Patient Health Questionnaire; PHQ-9) and seven depression risk factors (personal and family MDD history, sex, childhood stress, neuroticism, work hours, and stressful life events) in a longitudinal study of medical interns prior to and throughout internship (n=1289). We tested whether risk factors varied across symptoms, and whether a latent disease model could account for heterogeneity between symptoms.
Results: All MDD symptoms increased significantly during residency training. Four risk factors predicted increases in unique subsets of PHQ-9 symptoms over time (depression history, childhood stress, sex, and stressful life events), whereas neuroticism and work hours predicted increases in all symptoms, albeit to varying magnitudes. MDD family history did not predict increases in any symptom. The strong heterogeneity of associations persisted after controlling for a latent depression factor.
Conclusions: The influence of risk factors varies substantially across DSM depression criterion symptoms. As symptoms are etiologically heterogeneous, considering individual symptoms in addition to depression diagnosis might offer important insights obfuscated by symptom sum scores.
Book chapters & dissertation
» Fried, E. I. (accepted). Psychopathological Networks. In A. E. Wenzel (Ed.), The SAGE Encyclopedia of Abnormal and Clinical Psychology. New York, NY: SAGE Publications. PDF. (Abstract)
For many decades, clinicians have observed that patients often struggle with a number of problems that reinforce each other. Such dynamic systems of mutually interacting symptoms are defined as psychopathological networks. Depression provides a good example: insomnia may trigger fatigue, which in turn can activate concentration difficulties and sad mood that in turn feed back into insomnia—a feedback loop that is hard to escape from. Around 2010, psychologists and psychiatrists started using modern statistical tools to empirically examine these causal connections among symptoms, and there is a growing interest in both the scientific and the clinical community. The majority of research so far has focused on major depression, but there is also some recent work on post-traumatic stress disorder (PTSD), attention-deficit/hyperactivity disorder (ADHD), psychosis, bipolar disorder, autism spectrum disorder, generalized anxiety disorder, bereavement and complicated grief, as well as substance abuse and dependence.
» Fried, E. I. (accepted). Zung Depression Scale. In A. E. Wenzel (Ed.), The SAGE Encyclopedia of Abnormal and Clinical Psychology. New York, NY: SAGE Publications. (Abstract)
The Zung Self-Rating Depression Scale (SDS) is a short and comprehensive clinical instrument to measure depression severity. It was developed in 1965 by William W.K. Zung (1929-1992), a psychiatrist at Duke University who studied mood and anxiety disorders. The SDS has been used widely in clinical psychology and psychiatry, and translated into over 30 languages. The original paper in which Zung introduced the SDS has accumulated over 8000 citations, stressing the popularity of the SDS as clinical tool.
» Dejonckheere, E., & Fried, E. I. (in press). Bereavement. In V. Zeigler-Hill & T. Shackelford (Eds.), Encyclopedia of Personality and Individual Differences. Preprint. (Abstract)
Losing a loved one is one of the most painful occurrences in life, and thus considered an adverse life event in psychological literature. The common reaction to such a loss is called bereavement or grief, which is often accompanied by a considerable decrease in psychological and physical well-being. Although grief is experienced as adversity, in most cases it is a normal response to a difficult life situation. In case grief persists for a prolonged time beyond what is considered an appropriate response and causes significant disturbances and impairment in daily life, this is called complex grief (CG). Synonyms are prolonged grief disorder and persistent complex bereavement disorder. Multiple factors determine whether a person is more likely to transition from a normal grief response to CG, such as personal characteristics and coping strategies, the type of relationship to the lost person, the consequences of the loss, and the reaction of the broader environment. It is not always easy to distinguish CG from Major Depression (MD) or post-traumatic stress disorder (PTSD), and various therapeutic and pharmaceutical interventions are available to help individuals suffering from complex grief disorder.
» Fried, E. I. (2014). Covert Heterogeneity of Major Depressive Disorder: Depression Is More Than the Sum-Score of its Symptoms. Dissertation. PDF. (Abstract)
Major Depressive Disorder is one of the greatest challenges of modern health. It is the leading cause of disability worldwide, highly prevalent, often recurrent, closely related to suicide, and linked to the development of life-threatening medical conditions such as diabetes and coronary heart disease. Despite decades of research, basic questions remain unresolved: genetic studies have been unable to identify loci reliably associated with depression diagnosis and treatment response, antidepressants do not work above placebo level for the majority of patients, and field trials of the recently published Diagnostic and Statistical Manual of Mental Disorders (DSM-5) show that reliability of depression diagnosis is low.
I propose that one of the main reasons for this striking lack of progress is covert heterogeneity of depression: the current diagnostic criteria lump individuals suffering from a wide range of disparate psychiatric symptoms into one undifferentiated category. Sum-scores are used instead of individual symptoms because the disease model – derived from discoveries in the field of infectious diseases at the turn of the 19th century remains unchallenged: depression is understood to exist outside classification systems as real entity, and believed to be the common cause for its symptoms. This, in turn, makes symptoms interchangeable indicators of one underlying disease, and justifies sum-scores: symptom number, not symptom nature matters.
In this dissertation I demonstrate that individual depressiv symptoms differ from each other in three important aspects. First, in a longitudinal study of 1,289 medical students undergoing the severe and chronic stressor residency, risk factors such as sex or history of depression predict increases of specific symptoms. Second, in the same sample, symptoms exhibit marked differential increases in response to severe stress. Third, in a cross-sectional study of 3,703 depressed outpatients, symptoms differ drastically in their impacts on impairment of psychosocial functioning.
Together with evidence from numerous fields of research described throughout this dissertation, these three studies illuminate that depression symptoms are more than interchangeable indicators of an underlying disease. Symptoms are distinct phenomena with particular characteristics, an the analysis of individual symptoms reveals crucial information obfuscated by sum-scores, offers important clinical utility, will substantially facilitate our understanding of depression, and lead to more efficacious prevention and intervention strategies.
» Epskamp, S., & Fried, E. I. (2016). bootnet: Bootstrap Methods for Various Network Estimation Routines. Package for the free statistical environment R (CRAN).
» Fried, E. I. (2016). Depressionsforschung im Stimmungstief – Gründe einer wissenschaftlichen Krise und mögliche Auswege. In-Mind.
» Fried, E. I., Dejonckheere, E., Tuerlinckx, F. (2016). Welke symptomen zijn het meest belangrijk bij depressie? Een netwerkanalyse van depressie. Neuron, Vol. 21 Nr. 6. (Dutch PDF) (French PDF)
» Fried, E. I. (2015). Depression — more than the sum of its symptoms. The Psychologist (British Psychological Society).
» Fried, E. I., & Chekroud, A. (2017). Treatment-resistant depression: clarifications and important steps forward. Unpublished commentary. (PDF).
» Fried, E. I., Bylsma, L., & Nesse, R. M. (2016). Is Seasonal Affective Disorder really just a “Folk Construct”? Unpublished commentary, DOI 10.13140/RG.2.1.5149.7362. (PDF).
» Fried, E. I. (2015-2016). Several blog posts on Psychosystems.org, the blog of the Psychosystems research group, University of Amsterdam. (URL).
» Fried, E. I., van der Sluis, S., & Cramer, A. O. J. (2015). The genetics of major depression remain elusive. Unpublished commentary, DOI 10.13140/RG.2.1.3480.4963. (PDF).
» Fried, E. I. (2015). SSRI treatment for Major Depression relies on problematic assumptions. Online commentary, The British Medical Journal (BMJ).
» Fried, E. I. (2015). Depression didn’t crash the plane: the co-pilot did. University of Leuven Blog.
» Fried, E. I. (2016). True facts about the giant sequoia, or: what kind of name is sequoiadendron giganteum. Science Creative Quarterly.
( Disclaimer: Manuscripts provided here are for academic purposes only and not intended for mass dissemination or copying. Please refer to applicable fair use laws, including the restrictions from publication copyright holders. )