1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Received: 30 January 2018 | Revised: 28 April 2018 | Accepted: 28 April 2018

DOI: 10.1002/eat.22884

ORIGINAL ARTICLE

Network analysis of specific psychopathology and psychiatric

symptoms in patients with eating disorders

Marco Solmi MD, PhD1,2,3 | Enrico Collantoni MD1 | Paolo Meneguzzo MD1 |

Daniela Degortes Psy1 | Elena Tenconi PhD1,2,3 | Angela Favaro MD, PhD1,2,3

1
Neuroscience Department, Psychiatry Unit,
University of Padua, Padua, Italy
Abstract
2
Psychiatry Unit, Padua University Hospital, Background: Network analysis of psychiatric symptoms describes reciprocal relationships of indi-
Padua, Italy
vidual symptoms, beyond categorical diagnoses. Those with eating disorders (EDs) frequently have
3
Centro Neuroscienze Cognitive, Padua,
complex patterns of comorbid symptoms and the transdiagnostic theory includes shared common
Italy
core features across diagnoses. We aim to test whether general psychiatric symptoms comprise
Correspondence components of these transdiagnostic features.
Marco Solmi, MD, PhD, University of
Padua, Neuroscience Department, Methods: Network analysis was applied on 2068 patients with EDs (955 anorexia nervosa [AN],
Psychiatry Unit, Padua, Italy. 813 bulimia nervosa [BN], and 300 binge-eating disorder [BED]). All patients underwent clinical
Email: marco.solmi83@gmail.com interviews and some self-reported questionnaires, such as the Symptom Check-List 90 (SCL-90) to
measure psychiatric symptoms, the Eating Disorder Inventory (EDI) to measure ED-specific symp-
toms, and the Tridimensional Personality Questionnaire (TPQ) for personality traits.

Results: Across EDs and within each ED, SCL-90 scores of depression, anxiety and interpersonal
sensitivity, EDI ineffectiveness, interoceptive awareness, interpersonal distrust, and drive for thin-
ness had high centrality. Notably, body mass index (BMI) and EDI bulimia played a central role
when considering the whole group, whereas they did not in individual EDs.

Discussion: The shared centrality of identified nodes in both individual and merged groups sup-
ported the transdiagnostic theory of EDs (diagnoses share core ED features), with a central role of
BMI. Moreover, the most central nodes were general psychiatric symptoms, interpersonal domain,
and self-efficacy. These findings suggest that—in addition to ED-core symptoms and BMI—depres-
sive and anxiety symptoms, interpersonal sensitivity and ineffectiveness may be important targets
to provide effective treatments across AN, BN, and BED.

KEYWORDS
anorexia nervosa, binge-eating disorder, bulimia nervosa, network analysis, transdiagnostic

1 | INTRODUCTION categorical approach in psychiatry (Maj, 2005). For example, bounda-

The categorical approach in psychiatry is based on the existence of
clearly detached conditions that are caused by disease-specific factors.
These conditions manifest through signs and symptoms which in turn
are conceptualized as manifestations of the mental condition itself
(American Psychiatric Association, 2000, 2013). Notwithstanding the
need to formulate clear diagnoses each with a range of possible treat-
ment, a growing literature summarizes the limitations of the use of a
ries across disorders are not as clear as DSM is proposing, and, in many
cases, it is possible that a diagnostic category encompasses a large
number of different neurobiological entities (Brem, Grunblatt, Drechs-
ler, Riederer, & Walitza, 2014; Dacquino, De Rossi, & Spalletta, 2015;
Goodkind et al., 2015). The dimensional approach recommended by
the Research Domain Criteria (RDoC) is one of the proposed alterna-
tive models, although other approaches are being considered by the lit-
erature in the field (Boschloo et al., 2015; Fried et al., 2017). One of

680J Eat Disord.

Int © 2018 Wiley Periodicals, Inc.
2018;1–13. wileyonlinelibrary.com/journal/eat
wileyonlinelibrary.com/journal/eat V
Int J Eat Wiley
C 2018 Disord. Inc. | 1
2018;51:680–692.
Periodicals,
 1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 |
SOLMI ET AL. SOLMI ET681
AL.

these approaches is to consider models that (Cuthbert, 2014; Insel, We applied network analysis and describe the relationship
2014; Patrick & Hajcak, 2016; Wildes & Marcus, 2015) are defined as between ED-specific symptoms, general psychiatric symptoms, person-
“psychological networks” (Epskamp, Borsboom, & Fried, 2017). As ality traits and other clinical variables in patients with AN, BN, or BED.
described by Epskamp et al. (2017), symptoms or other clinical varia- The aim of this study is to test whether transdiagnostic theory of EDs
bles are represented as nodes, connected by edges, which in turn rep- is supported by a network analysis which includes general
resent correlation measures among nodes. Such nodes are reciprocally psychiatric symptoms in addition to ED-core ones, to provide observa-
connected creating a network of interacting self-reinforcing pools of tional data to support the “rationale” underlying treatments for EDs
symptoms or clinical variables, as opposed to single manifestations of a that target psychiatric and interpersonal issues in addition to ED-core
latent underlying entity, namely the mental disorder. The importance of symptoms.
each node within a network is quantified by centrality indices (node
strength, closeness, betweenness) and the accuracy of each centrality 2 | METHODS
index can also be assessed by accuracy estimation. According to this
model, symptoms currently categorized as manifestations of specific 2.1 | Participants and measures
mental disorders according to the widely accepted categorial diagnostic
Patients were consecutively recruited among those referred to the ED
frame, have shown to map across several different mental disorders
Outpatient Unit of Padua University Hospital, Italy. To be included in
instead (Boschloo et al., 2015). Hence network analysis may better fit
the study patients had to be older than fourteen years old, have a pres-
the widely accepted evidence on frequent comorbidity among psychi-
ent diagnosis of AN, BN, or BED according to DSM-5 as defined by
atric conditions (Shevlin et al., 2017), or of diagnostic migration (Eddy
structured clinical interview at the moment of referral. Patients with
et al., 2008) compared with the ongoing categorical approach. Also,
severe comorbidity such as schizophrenia, acute manic episode, or
network analysis could be tested as a tool that identifies common fea-
alcohol addiction, were excluded from the study, as well as patients or,
tures across groups of patients, to point research towards identification
in the case of patients aged lower than 18 years, parents not accepting
of sets of clinical presentations with common etiological background,
to participate in the study.
or on the other hand to groups of variables associated with differential
We ran a network analysis built up of ED-core measuring scales,
response to treatment across or within mental disorders (i.e., response
general psychiatric symptom measure, personality variables BMI and
to second generation antipsychotics in bipolar disorder and schizophre-
duration of illness. Symptoms of EDs were measured with Eating Disor-
nia, response to cognitive-behavioral therapy in AN and BN).
ders Inventory (EDI; Garner, Olmstead, & Polivy, 1983), general psychi-
After applying network analysis within eating disorders (ED; Smith
atric symptoms with Symptom Check-list 90 (SCL-90; [Derogatis],
et al., 2018), core psychopathologic features and symptoms have dem-
personality traits with Tridimensional Personality Questionnaire (TPQ;
onstrated to map consistently across anorexia nervosa (AN), bulimia
Cloninger, Przybeck, & Svrakic, 1991). All three questionnaires were
nervosa (BN), and binge-eating disorder (BED). This supports the trans-
completed—after the diagnostic/clinical interviews—in the context of
diagnostic theory (Fairburn, Cooper, & Shafran, 2003), which cognitive-
the routine baseline assessment at the ED Unit. Questionnaires were
behavioral psychotherapy is based on(Dakanalis, Timko, Clerici, Zanetti,
used in their validated Italian version and showed to have acceptable
& Riva, 2014; DuBois, Rodgers, Franko, Eddy, & Thomas, 2017; For-
reliability in our sample (Cronbach’s alpha between 0.73 and 0.90).
bush, Siew, & Vitevitch, 2016). However, certain associations, such as
Duration of illness was defined as the duration of the current ED epi-
the one between restrictive behavior and binge-eating have been ques-
sode (Favaro, Caregaro, Tenconi, Bosello, & Santonastaso, 2009; Hin-
tioned (Dakanalis et al., 2014; Elran-Barak et al., 2015). This suggests
dler, Crisp, McGuigan, & Joughin, 1994).
that a more complex network of symptoms underlies clinical manifesta-
Among EDI subscales, body dissatisfaction, drive for thinness and
tions of EDs. Hence, a network analysis including both typical ED-core
bulimia were considered ED-core symptoms measures. All participants
features and psychiatric symptoms may provide useful information
gave informed consent for the use of data in anonymous form. The
both from theoretical and clinical perspectives. Within a network cen-
study was conducted according to the principles of the ‘‘Declaration of
tral nodes are those believed to maintain the network itself, thus they
Helsinki’’ (as amended in Tokyo, Venice and Hong Kong and Somerset
have been suggested as potential treatment targets when the studied
West) and in accordance with the Guideline for Good Clinical Practice
network is a pathological process.(Costantini et al., 2014) Conse-
(CPMP/ICH/135/95—17th July 1996).
quently, identifying central nodes in the network of aforementioned
symptoms, may provide a common frame to explain the effectiveness
2.2 | Network analysis
of psychotherapeutic interventions that target wider symptomatic
domains in addition to ED-core ones, such as family therapy (FT; Eisler All codes used are available in Supporting Information Material. Psy-
et al., 2016) and enhanced cognitive behavioral therapy (E-CBT) in ado- chological network has been estimated with RStudio (R_Core_Team,
lescents (Dalle Grave, Calugi, Doll, & Fairburn, 2013), Maudsley Model 2013) using qgraph package according to methods described by
for Treatment of Adults with AN (MANTRA; Schmidt et al., 2015) or Epskamp et al. (2017), Borsboom and Cramer (2013), Costantini et al.
interpersonal psychotherapy (IPT; Linardon, Fairburn, Fitzsimmons- (2014) (Borsboom & Cramer, 2013; Costantini et al., 2014; Epskamp
Craft, Wilfley, & Brennan, 2017) in adults affected by AN, BN, or BED. et al., 2017).

SOLMI
682 ET AL.
T AB LE 1 Characteristics of included sample
AN (n 5 955)
Age (years) 23.13 (7.58)
Age of onset (years) 18.58 (5.60)
Duration of illness (months) 35.56 (54.63)
Data are reported as mean (standard deviation).
Briefly, a network can be defined as a set of nodes reciprocally
connected through edges, without any a-priori direction or causal mod-
eling, which allows circularity of correlations, and explains the complex-
ity of psychological networks through an alternative approach to
generalized linear model, or latent component analysis. We used a pair-
wise Markov random field network model, more precisely a Gaussian
graphical model (Epskamp et al., 2016), which allows to estimate undir-
ected associations, called edges, among variables, called nodes, without
implying any directionality (i.e., they do not imply any causal inference
or direction in association). We measured the accuracy of such associa-
tions through their 95% confidence intervals, and the centrality indices
of nodes, namely how the nodes are interconnected with several other
nodes of the network, as well as the stability of the network after
removing increasing percentages of patients. Data used in the model
were observational and cross-sectional. Since data did not follow a nor-
mal distribution, they were transformed into normally distributed data
via the nonparanormal transformation (Epskamp et al., 2017). Since
psychological measures are often interconnected and since an excess
of sparse and clinically nonsignificant correlations may add confusion
without adding information to a network interpretation, we applied a
penalty to the correlations close to zero, to retain only meaningful
associations. More in detail, a “least absolute shrinkage and selection
operator” (LASSO; Friedman, Hastie, & Tibshirani, 2014) regularization
(a sort of shrinkage of small edges to zero) was applied to only retain
more solid edges (correlations). This applies a conservative/wise
approach, which makes results more interpretable. Also, the Extended
Bayesan Information Criterion (EBIC; Chen & Chen, 2008; Foygel &
Drton, 2010), a parameter that sets the degree of regularization/pen-
alty applied to sparse correlations, was set to 0.5.
Centrality of nodes was estimated with node strength (the sum of
weight of the connections for each node; Costantini et al., 2014, close-
ness inverse of the sum of the distances of the focal node from all the
other nodes in the network; Costantini et al., 2014), betweenness indi-
ces (number of shortest paths between any two nodes that pass
through the node of interest Costantini et al., 2014), whose stability
was remeasured after case-dropping subset bootstrap, namely the
recalculation of centrality indices after dropping growing percentages
of the included participants (Epskamp et al., 2017). In addition, to fur-
ther assess stability of centrality indices, the correlation stability coeffi-
cient (CS) was calculated. CS represents the maximum proportion of
population that can be dropped with recalculated indices correlating at
least 0.7 with indices of the original full sample. Networks with reliable
centrality should have a CS greater or equal to 0.25, ideally higher than
0.5 for centrality estimates). In addition, to measure edges accuracy, an
1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
|
SOLMI ET AL3.
BN (n 5 813) BED (n 5 300) Kruskal Wallis p value
26.06 (8.23) 35.31 (11.48) <.01
19.17 (6.09) 24.30 (11.29) <.01
58.03 (66.36) 100.81 (98.79) <.01
estimated 95% confidence interval of the range containing the true
correlation (edge) was calculated by means of “nonparametric” boot-
strapping has been performed (n boots 5 1000). Network analysis has
been performed in the whole group of patients with EDs, and in sepa-
rate diagnostic subgroups as well. Since sample size of different diag-
nostic groups were largely different, we could not use the Network
Comparison Test (van Borkulo et al., 2015) to compare networks
between diagnostic groups.
3 | RESULTS
3.1 | Characteristics of the included sample
Characteristics of the included sample are reported in Table 1. We
included 955 patients with AN, 813 patients with BN, and 300 with
BED according to DSM-5 criteria. The three diagnostic groups signifi-
cantly differed on age, age of onset, and duration of illness. Almost all
patients were female (96.6%).
3.2 | Characteristics of the symptom networks
Network structure of EDI scores, SCL-90 scores, TPQ and clinical varia-
bles in different diagnostic groups, as well as a comparative plot of cen-
trality indexes of the overall and three separate networks are reported
in Figures 1–5, respectively.
3.3 | Eating disorders
Figure 1 reports the network of symptoms measured with EDI, SCL-90
TPQ, and clinical variables in patients with ED, and Table 2 reports the
nodes’ centrality estimates. Central stability coefficient (maximum drop
proportions to retain correlation of 0.7 in at least 95% of the sample)
for ED network was 0.75 for betweenness, 0.75 for closeness, 0.75 for
strength (Supporting Information Figure S1). Strength, namely the sum
of the weights of the connections for each node(Costantini et al.,
2014), was the centrality index with highest central stability. The node
strength was highest for ineffectiveness (M 5 1.680), BMI (M 5 1.628),
depressive score (M 5 1.483), anxiety score (M 5 1.458), and bulimia
(M 5 1.444); it resulted to be lower for TPW persistence (M 5 0.916),
perfectionism (M 5 0.831), duration of illness (M 5 0.578), and the very
low for maturity fears (M 5 0.390; Figure 5, Table 2). The strongest cor-
relations of central nodes were with ED-core symptoms, namely EDI
bulimia, and body dissatisfaction for BMI, and with interpersonal sensi-
tivity and depressive score for EDI ineffectiveness (Supporting
 1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 |
SOLMI ET AL. SOLMI ET683
AL.

FIGURE 1 EDs network. BMI, body mass index; drt, duration of illness; EDI, Eating Disorder Inventory; edbl, EDI bulimia; edc, EDI
interoceptive awareness; edd, EDI drive for thinness; ednf, EDI ineffectiveness; edbd, EDI body dissatisfaction; edg, EDI maturity fears; edp,
EDI perfectionism; ednt, EDI interpersonal distrust; SCL90, Symptom Check-List 90; som, SCL90 somatization; oc, SCL90 obsessive compul-
sive; is, SCL-90 interpersonal sensitivity; dep, SCL90 depressive; anx, SCL90 anxiety; hos, SCL90 hostility; phob, SCL90 phobic; TPQ, Tridi-
mensional Personality Questionnaire; ns, TPQ novelty seeking; ha, TPQ harm avoidance; rd, TPQ reward dependence; prs, TPQ persistence.
Lines: blue lines, positive associations; red lines, negative associations. Thicker lines indicate stronger associations. Circles: circles indicate
variables (nodes) of the network. [Color figure can be viewed at wileyonlinelibrary.com]

Information Table S1). All relevant correlations of nodes described in
this paragraph were positive correlations.
3.4 | Anorexia nervosa
Figure 2 reports the network of symptoms measured with EDI, SCL-90
TPQ, and clinical variables in patients with AN, and Table 2 reports the
nodes’ centrality estimates. Central stability coefficient (maximum drop
proportions to retain correlation of 0.7 in at least 95% of the sample)
for AN network was 0.595 for betweenness, 0.595 for closeness,
0.751 for strength (Supporting Information Figure S2). Strength, namely
the sum of the weights of the connections for each node(Costantini
et al., 2014), was the centrality index with highest central stability. The
node strength was highest for ineffectiveness (M 5 1.511), depressive
score (M 5 1.495), anxiety score (M 5 1.379), and interoceptive aware-
ness (M 5 1.169); it resulted to be lower for body dissatisfaction
(M 5 0.619), BMI (M 5 0.438), duration of illness (M 5 0.350), and the
very low for maturity fears (M 5 0.297) (Figure 5, Table 2). Among ED-
 1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SOLMI
684 ET AL. |
SOLMI ET AL5.

FIGURE 2 AN network. BMI, body mass index; drt, duration of illness; EDI, Eating Disorder Inventory; edbl, EDI bulimia; edc, EDI
interoceptive awareness; edd, EDI drive for thinness; ednf, EDI ineffectiveness; edbd, EDI body dissatisfaction; edg, EDI maturity fears; edp,
EDI perfectionism; ednt, EDI interpersonal distrust; SCL90, Symptom Check-List 90; som, SCL90 somatization; oc, SCL90 obsessive compul-
sive; is, SCL-90 interpersonal sensitivity; dep, SCL90 depressive; anx, SCL90 anxiety; hos, SCL90 hostility; phob, SCL90 phobic; TPQ, Tridi-
mensional Personality Questionnaire; ns, TPQ novelty seeking; ha, TPQ harm avoidance; rd, TPQ reward dependence; prs, TPQ persistence.
Lines: blue lines, positive associations; red lines, negative associations. Thicker lines indicate stronger associations. Circles: circles indicate
variables (nodes) of the network. [Color figure can be viewed at wileyonlinelibrary.com]

core symptoms drive for thinness was the most central ED-core node nodes’ centrality estimates. Central stability coefficient (maximum drop
(M 5 1.154). The strongest correlations of central nodes were with proportions to retain correlation of 0.7 in at least 95% of the sample)
interpersonal sensitivity and obsessive-compulsive symptoms, in addi- for BN network was 0.361 for betweenness, 0.594 for closeness,
tion to the reciprocal strong correlation among these two nodes. Drive 0.594 for strength (Supporting Information Figure S3).
for thinness, the most central ED-core symptom, had its stronger corre- Strength and closeness, were the centrality indexes with highest
lations with EDI bulimia score, interoceptive awareness, and body dis- central stability. The node strength was highest for ineffec-
satisfaction (Supporting Information Table S2). All relevant correlations tiveness score (M 5 1.420), anxiety (M 5 1.394), depres-
described in this paragraph were positive correlations. sive (M 5 1.338), and interpersonal sensitivity
(M 5 1.229); it resulted to be lower for phobia
(M 5 0.419), TPQ persistence (M 5 0.697), maturity fears
3.5 | Bulimia nervosa
(M 5 0.371), and duration of illness (M 5 0.276) (Figure
Figure 3 reports the network of symptoms measured with EDI, SCL-90 5, Table 2). Among ED-core symptoms drive for thinness
TPQ, and clinical variables in patients with BN, and Table 2 reports the had the highest centrality (M 5 1.129). The strongest
 1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6 |
SOLMI ET AL. SOLMI ET685
AL.

FIGURE 3 BN network. BMI, body mass index; drt, duration of illness; EDI, Eating Disorder Inventory; edbl, EDI bulimia; edc, EDI
interoceptive awareness; edd, EDI drive for thinness; ednf, EDI ineffectiveness; edbd, EDI body dissatisfaction; edg, EDI maturity fears; edp,
EDI perfectionism; ednt, EDI interpersonal distrust; SCL90, Symptom Check-List 90; som, SCL90 somatization; oc, SCL90 obsessive compul-
sive; is, SCL-90 interpersonal sensitivity; dep, SCL90 depressive; anx, SCL90 anxiety; hos, SCL90 hostility; phob, SCL90 phobic; TPQ, Tridi-
mensional Personality Questionnaire; ns, TPQ novelty seeking; ha, TPQ harm avoidance; rd, TPQ reward dependence; prs, TPQ persistence.
Lines: blue lines, positive associations; red lines, negative associations. Thicker lines indicate stronger associations. Circles: circles indicate
variables (nodes) of the network. [Color figure can be viewed at wileyonlinelibrary.com]

correlations of central nodes were with depressive symptoms, interper-
sonal sensitivity, and obsessive-compulsive symptoms for both anxiety
and ineffectiveness. Drive for thinness, the most central ED-core symp-
tom, had its stronger correlations with body dissatisfaction, bulimia,
and interoceptive awareness (Supporting Information Table S3). All rel-
evant correlations described in this paragraph were positive
correlations.
3.6 | Binge-eating disorder
Figure 4 shows the network of symptoms measured with EDI, SCL-90
TPQ, and clinical variables in patients with BED, and Table 2 reports
the nodes’ centrality estimates. Central stability coefficient (maximum
drop proportions to retain correlation of 0.7 in at least 95% of the sam-
ple) for BED network was 0.205 for betweenness, 0.128 for closeness,
0.596 for strength (Supporting Information Figure S4).
Strength was the centrality index with highest central stability. The
node strength was highest for depressive score (M 5 1.390), anxiety
(M 5 1.212), ineffectiveness (M 5 1.165), and interpersonal sensitivity
(M 5 1.152); it resulted be lower for perfectionism (M 5 0.498), TPQ
persistence (M 5 0.291), maturity fears (M 5 0.265), and the lowest
duration of illness (M 5 0.260; Figure 5, Table 2). Among EDI symptoms
drive for thinness was the most central (M 5 0.943). The strongest cor-
relations of central nodes were with interpersonal sensitivity, depres-
sive, and obsessive-compulsive symptoms, in addition for reciprocal
correlation for ineffectiveness and anxiety or depressive scores. Drive
 1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SOLMI
686 ET AL. |
SOLMI ET AL7.

FIGURE 4 BED network. BMI, body mass index; drt, duration of illness; EDI, Eating Disorder Inventory; edbl, EDI bulimia; edc, EDI
interoceptive awareness; edd, EDI drive for thinness; ednf, EDI ineffectiveness; edbd, EDI body dissatisfaction; edg, EDI maturity fears; edp,
EDI perfectionism; ednt, EDI interpersonal distrust; SCL90, Symptom Check-List 90; som, SCL90 somatization; oc, SCL90 obsessive compul-
sive; is, SCL-90 interpersonal sensitivity; dep, SCL90 depressive; anx, SCL90 anxiety; hos, SCL90 hostility; phob, SCL90 phobic; TPQ, Tridi-
mensional Personality Questionnaire; ns, TPQ novelty seeking; ha, TPQ harm avoidance; rd, TPQ reward dependence; prs, TPQ persistence.
Lines: blue lines, positive associations; red lines, negative associations. Thicker lines indicate stronger associations. Circles: circles indicate
variables (nodes) of the network. [Color figure can be viewed at wileyonlinelibrary.com]

for thinness, the most central ED-core symptom, had its stronger corre-
lations with bulimia, body dissatisfaction, and interoceptive awareness
(Supporting Information Table S4). All relevant correlations described in
this paragraph were positive correlations.
4 | DISCUSSION
The aim of this work was to provide further support to transdiagnostic
theory by means of network analysis, confirming previous results
(DuBois et al., 2017), but extending the validity of transdiagnostic
theory to include general psychiatric symptoms in addition to ED-core
ones.
Four previous works applied network analysis to a sample affected
by EDs. In the first work (Forbush et al., 2016) body checking emerged
as the symptom with the shortest average distance to other symptoms
in the network, and as the most frequent symptom on the path
between any two other symptoms, as measured with Eating Pathology
Symptom Inventory (EPSI; Forbush et al., 2013). Interestingly, DuBois
et al. (2017) showed that overvaluation of weight and shape as meas-
ured with EDE-Q had the highest centrality among ED-core symptoms
measured with EPSI (Forbush et al., 2013), closely followed by body
dissatisfaction and cognitive restraint. Levinson et al. (2017) showed
that fear of weight gain and anxiety and depressive symptoms are con-
nected in BN. Olatunji, Levinson, & Calebs, 2018 measured ED
 1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 |
SOLMI ET AL. SOLMI ET687
AL.

FIGURE 5 Centrality indexes of EDs symptoms, general psychopathology, personality traits and clinical variables in patients with EDs. All,
all eating disorders; BMI, body mass index; drt, duration of illness; EDI, Eating Disorder Inventory; edbl, EDI bulimia; edc, EDI interoceptive
awareness; edd, EDI drive for thinness; ednf, EDI ineffectiveness; edbd, EDI body dissatisfaction; edg, EDI maturity fears; edp, EDI
perfectionism; ednt, EDI interpersonal distrust; SCL90, Symptom Check-List 90; som, SCL90 somatization; oc, SCL90 obsessive compulsive;
is, SCL-90 interpersonal sensitivity; dep, SCL90 depressive; anx, SCL90 anxiety; hos, SCL90 hostility; phob, SCL90 phobic; TPQ, Tridimen-
sional Personality Questionnaire; ns, TPQ novelty seeking; ha, TPQ harm avoidance; rd, TPQ reward dependence; prs, TPQ persistence.
Note: z-scores are shown on x-axis. [Color figure can be viewed at wileyonlinelibrary.com]

symptoms with EDI, and found that ineffectiveness and interoceptive
awareness were central nodes to the ED network in a large sample of
inpatients before and after treatment. Merging previous evidence with
results from this study, EDI drive for thinness (Garner et al., 1983),
EPSI cognitive restraint (Forbush et al., 2016), overvaluation of body
weight and shape (DuBois et al., 2017) bulimia and body dissatisfaction
are closely reciprocally correlated clustering together in the whole
group of patients with EDs, and within each diagnostic subgroup. Yet,
despite their close correlation, their centrality to ED network remains
debated according to results of this work in addition to the aforemen-
tioned recent work (Olatunji et al., 2018). In addition, the correlation
of ED-core symptoms with weight status across EDs remains
controversial. Such symptoms do not show a tight correlation with
BMI within ED diagnostic subgroups in our sample. This may be
due to the relatively homogeneous weight status within AN, BN, or
BED. However, the intensity of such symptoms appears to be corre-
lated with BMI in the whole group with ED, which in turn is also
among the most central nodes in the network. It may be argued
that, considering EDs as a unique diagnostic category, underweight
patients (AN) experience less body dissatisfaction or drive for thin-
ness due to their weight status. However, patients with normal or
overweight (BN or BED) tend to suffer from more severe ED-core
symptoms, possibly due to the difference between their actual
weight and the desired weight.
 1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SOLMI
688 ET AL. |
SOLMI ET AL9.

T AB LE 2 List of nodes of the network with corresponding strength estimates for EDs

Strength Strength Strength Strength

All EDs estimate AN network estimate BN network estimate BED network estimate

BMI 1.628 BMI 0.438 BMI 0.805 BMI 0.776

Duration of illness 0.578 Duration of illness 0.350 Duration of illness 0.276 Duration of illness 0.260

EDI ineffective- 1.680 EDI ineffective- 1.514 EDI ineffective- 1.420 EDI ineffective- 1.165
ness ness ness ness

EDI bulimia 1.444 EDI interoceptive 1.169 EDI interoceptive 1.196 EDI interpersonal 1.096
awareness awareness distrust

EDI drive for thin- 1.368 EDI drive for thin- 1.154 EDI interpersonal 1.175 EDI interoceptive 1.010
ness ness distrust awareness

EDI interoceptive 1.284 EDI interpersonal 1.141 EDI drive for thin- 1.129 EDI drive for thin- 0.943
awareness distrust ness ness

EDI interpersonal 1.284 EDI bulimia 1.123 EDI body dissatis- 1.100 EDI bulimia 0.749
distrust faction

EDI body dissatis- 1.092 EDI perfectionism 0.670 EDI bulimia 0.813 EDI body dissatis- 0.638
faction faction

EDI perfectionism 0.831 EDI body dissatis- 0.619 EDI perfectionism 0.749 EDI perfectionism 0.498
faction

EDI maturity fears 0.390 EDI maturity fears 0.297 EDI maturity fears 0.371 EDI maturity fears 0.265

SCL-90 depressive 1.483 SCL-90 depressive 1.495 SCL-90 anxiety 1.394 SCL-90 depressive 1.390

SCL-90 anxiety 1.458 SCL-90 anxiety 1.379 SCL-90 depressive 1.338 SCL-90 anxiety 1.212

SCL-90 interper- 1.225 SCL-90 interper- 1.144 SCL-90 interper- 1.229 SCL-90 interper- 1.152
sonal sensitivity sonal sensitivity sonal sensitivity sonal sensitivity

SCL-90 obsessive- 1.016 SCL-90 obsessive- 1.005 SCL-90 somatiza- 1.129 SCL-90 obsessive- 1.073
compulsive compulsive tion compulsive

SCL-90 somatiza- 1.016 SCL-90 somatiza- 1.005 SCL-90 obsessive- 0.980 SCL-90 hostility 0.813
tion tion compulsive

SCL-90 hostility 0.951 SCL-90 hostility 0.883 SCL-90 hostility 0.843 SCL-90 phobia 0.751

SCL-90 phobia 0.755 SCL-90 phobia 0.722 SCL-90 phobia 0.719 SCL-90 somatiza- 0.681
tion

TPQ harm avoid- 1.166 TPQ reward de- 1.026 TPQ harm avoid- 1.057 TPQ harm avoid- 0.723
ance pendance ance ance

TPQ novelty seek- 1.133 TPQ novelty seek- 1.030 TPQ novelty seek- 1.007 TPQ reward de- 0.645
ing ing ing pendence

TPQ reward de- 1.104 TPQ harm avoid- 0.909 TPQ reward de- 0.864 TPQ novelty seek- 0.499
pendence ance pendence ing

TPQ persistence 0.916 TPQ persistence 0.675 TPQ persistence 0.697 TPQ persistence 0.291

BMI, body mass index; EDI, Eating Disorder Inventory; SCL-90, Symptom CheckList-90, TPQ, Tridimensional Personality Questionnaire.

In addition to such ED-core symptoms that homogeneously map
across EDs, SCL-90 depressive, and anxious symptoms, interpersonal
functioning, ineffectiveness, and interoceptive awareness appear to
play a more central role in maintaining the symptomatic network across
EDs, suggesting that general psychiatric and interpersonal domains
should be included in an expanded transdiagnostic theory of EDs. We
interpret our findings as supporting evidence for treatments that target
psychiatric and interpersonal issues in addition to ED-core symptoms.
Anxious and depressive symptoms are frequent among patients
with either AN, BN, or BED (Brand-Gothelf, Leor, Apter, & Fennig,
€ hren et al., 2014; Godart et al., 2015; Kountza, Garyfallos,
2014; Bu
Ploumpidis, Varsou, & Gkiouzepas, 2017), and accordingly to network
analysis they appear to play a central role in mental health of such
patients. Since a network central nodes have been suggested to be
ideal treatment targets (Costantini et al., 2014), and since evidence sup-
ports depression (Franko et al., 2018) as well as anxiety (Carrot et al.,
2017) as predictors of outcome in EDs, even after controlling for dura-
tion of illness (Keski-Rahkonen et al., 2014), both should be considered
when selecting the best treatment for patients with clinically relevant
depressive and anxious symptoms.
Also, interpersonal sensitivity, another core psychopathologic
dimension measured with SCL-90 (Bech, Bille, Moller, Hellstrom, &

10 |
SOLMI ET AL.
Ostergaard, 2014; Carrozzino et al., 2016), and similarly EDI interperso-
nal distrust and ineffectiveness are shared central nodes across AN,
BN, and BED. Across all EDs the ED-core with higher centrality was
drive for thinness. Centrality of such variables may be considered in
line with evidence showing that interpersonal distrust, ineffectiveness
and interoceptive awareness are higher in patients needing inpatient
treatment versus outpatient treatment (Milos, Spindler, Buddeberg, &
Ruggiero, 2004) suggesting a more severe clinical picture related to
severity of network-central variables. Also, interpersonal distrust, inef-
fectiveness alongside with interoceptive awareness and drive for thin-
ness have even been demonstrated to have a prognostic role
determining outcome at follow-up as long as 5–10 years,(Bizeul,
Sadowsky, & Rigaud, 2001) and to be central in the networks of symp-
toms measured with EDI in inpatients treated for an ED (Olatunji et al.,
2018).
The aforementioned converging evidence supporting the role of
general psychiatric symptoms in modulating clinical outcome in EDs,
and the present evidence of a central role of such dimensions in
EDs, may be considered as theoretical frame explaining the widely
accepted effectiveness of interventions for EDs that target interper-
sonal functioning, self-esteem and emotion dysregulation in addition
to ED-core symptoms, such as FT (Eisler et al., 2016), interpersonal
therapy (Linardon et al., 2017), and enhanced cognitive behavioral
therapy (Dalle Grave et al., 2013). In addition, rates of affective
comorbidity in EDs may be inflated by the misdiagnosis of categori-
cal approach, which splits a unique clinical entity (ED-core symp-
toms, depressive symptoms, and interpersonal domain) in two or
three separate diagnoses.
Additionally, BMI does not appear to have a central role in
complex relationships among symptoms included in the analyzed
network within each individual disease, consistently with previous
findings, which show that BMI is not linked to symptoms neither
before or after weight recovery in patients with AN (Federico et al.,
2017). However, it results to be the most central node when con-
sidered the whole group of EDs. In particular, BMI has strong corre-
lations with ED-core symptoms such as EDI bulimia, drive for
thinness, and body dissatisfaction. Clinical implications of such
results should encourage not to underestimate the role of weight
recovery in AN for example, or weight loss in patients with BED,
given the crucial role of medical comorbidity in the whole health
status of patients (Solmi et al., 2016), and they suggest to carefully
consider the close relationship between weight and ED-core symp-
toms in association with diagnostic migration across EDs. Yet, no
causal inference can be drawn from the present type of analysis
between BMI and ED-core symptoms, and more importantly the
underlying affective, interpersonal and self-efficacy domains remain
also central in the whole group of EDs, despite large differences in
BMI among patients with AN, BN, and BED.
In addition, it should be noted that modeling psychological features
and psychiatric symptoms with network analysis should not be seen as
opposed to classical latent component analysis, yet the two methods
should be seen as complementary(Guyon, Falissard, & Kop, 2017), pro-
viding useful information from two different a priori assumptions
1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SOLMI ET689
AL.
whose ontological and epistemological validity remains equally justified.
Network analysis is a proper tool to look into how symptoms recipro-
cally interact within or across mental conditions and how they can be
influenced by external events. However, latent component analysis still
retains its validity when looking into expressions of a certain prede-
fined biological underlying condition, across or within diagnostic
categories.
This work should be considered in view of its limitations. First,
cross-sectional data have been used, precluding any prognostic con-
sideration, while any hypothesis regarding an eventual actual causal
inference should be addressed with properly designed prospective
cohort studies, or even better through high quality interventional
studies (namely randomized controlled studies (Smith et al., 2018).
Secondly, the three included samples had different sample size,
which precluded any quantitative comparison among networks.
Third, the three samples we included differed on clinical characteris-
tics. Fourth, any centrality comparison within or between groups is
purely descriptive. Fifth, we did not include in the model several
possibly interesting variables, such as organic/medical complications,
cognitive performance, social functioning. Sixth, we did not use
EDE-Q, which did not allow us to measure eating concerns. Finally,
network models are based on linear correlations and, for this reason,
could fail in detecting other types of associations between variables
(i.e., threshold effect or quadratic correlation among others), thus
minimizing (or neglecting) the role of BMI in influencing
psychopathology.
Several points of strength should also be considered. First, this is
the very first work assessing psychological network of general psychi-
atric symptoms, ED-specific symptoms and clinical variables in a sample
of patients affected by AN, BN, and BED. Secondly, we measured ED-
core symptoms with EDI (Garner et al., 1983) and general psychiatric
symptoms with SCL-90 (Bech et al., 2014; Carrozzino et al., 2016;
Derogatis, 2010) which has shown solid validity across international
samples. Third, we included overall 2068 participants which is the sec-
ond largest sample of patients suffering from EDs analyzed with net-
work analysis to the best of our knowledge. Fourth, we are sharing all
the used codes in Supporting Information material to allow reproduci-
bility of the study design. Fifth, the analyzed network had solid central
stability, according to stability coefficient above 0.5 in all included sam-
ples. Sixth, we used both self-rating symptom measures, duration of ill-
ness, and BMI, providing a preliminary “multimodal” approach to
network analysis (Smith et al., 2018).
In conclusion, affective symptoms, interpersonal functioning, inter-
oceptive awareness, ineffectiveness and drive for thinness are the
most central variables in individuals suffering from AN, BN, and BED.
Such results provide a theoretical background in support of “third-
wave”(Linardon et al., 2017) psychotherapies for EDs targeting a wide
range of symptoms in addition to ED-core symptoms, and confirm the
validity of transdiagnostic theory (Dakanalis et al., 2014; DuBois et al.,
2017; Elran-Barak et al., 2015; Fairburn et al., 2003; Loeb, Lock,
Grange, & Greif, 2012), which groups EDs as an homogeneous set of
conditions sharing common clinical features, that can be influence by
weight status.

SOLMI
690 ET AL.
AC KNOWLE DGE ME NT S
None.
OR CID
Marco Solmi MD, PhD http://orcid.org/0000-0003-4877-7233
Paolo Meneguzzo http://orcid.org/0000-0003-3323-6071
Angela Favaro http://orcid.org/0000-0002-6540-5194
R E FER E NCE S
American Psychiatric Association. (2000). Diagnostic and statistical manual
of mental disorders (4th ed.). Washington, DC: American Psychiatric
Association.
American Psychiatric Association (2013). Diagnostic and statistical manual
of mental disorders (5th ed.). Washington, DC. American Psychiatric
Association.
Bech, P., Bille, J., Moller, S. B., Hellstrom, L. C., & Ostergaard, S. D.
(2014). Psychometric validation of the Hopkins Symptom Checklist
(SCL-90) subscales for depression, anxiety, and interpersonal sensitiv-
ity. Journal of Affective Disorders, 160, 98–103. https://doi.org/10.
1016/j.jad.2013.12.005
Bizeul, C., Sadowsky, N., & Rigaud, D. (2001). The prognostic value of
initial EDI scores in anorexia nervosa patients: A prospective follow-
up study of 5–10 years. Eating Disorder Inventory. European Psychia-
try, 16(4), 232–238.
Borsboom, D., & Cramer, A. O. (2013). Network analysis: An integrative
approach to the structure of psychopathology. Annual Review of Clini-
cal Psychology, 9(1), 91–121. https://doi.org/10.1146/annurev-
clinpsy-050212-185608
Boschloo, L., van Borkulo, C. D., Rhemtulla, M., Keyes, K. M., Borsboom,
D., & Schoevers, R. A. (2015). The Network Structure of Symptoms
of the Diagnostic and Statistical Manual of Mental Disorders. PLoS
One, 10(9), e0137621. https://doi.org/10.1371/journal.pone.
0137621
Brand-Gothelf, A., Leor, S., Apter, A., & Fennig, S. (2014). The impact of
comorbid depressive and anxiety disorders on severity of anorexia
nervosa in adolescent girls. The Journal of Nervous and Mental Dis-
ease, 202(10), 759–762. https://doi.org/10.1097/NMD.
0000000000000194
Brem, S., Grunblatt, E., Drechsler, R., Riederer, P., & Walitza, S. (2014).
The neurobiological link between OCD and ADHD. Attention Deficit
and Hyperactivity Disorders, 6(3), 175–202. https://doi.org/10.1007/
s12402-014-0146-x
€hren, K., Schwarte, R., Fluck, F., Timmesfeld, N., Krei, M., Egberts, K.,
Bu
. . . Herpertz-Dahlmann, B. (2014). Comorbid psychiatric disorders in
female adolescents with first-onset anorexia nervosa. European Eating
Disorders Review, 22(1), 39–44. https://doi.org/10.1002/erv.2254
Carrot, B., Radon, L., Hubert, T., Vibert, S., Duclos, J., Curt, F., & Godart,
N. (2017). Are lifetime affective disorders predictive of long-term
outcome in severe adolescent anorexia nervosa? European Child &
Adolescent Psychiatry, 26(8), 969–978. https://doi.org/10.1007/
s00787-017-0963-5
Carrozzino, D., Vassend, O., Bjorndal, F., Pignolo, C., Olsen, L. R., & Bech,
P. (2016). A clinimetric analysis of the Hopkins Symptom Checklist
(SCL-90-R) in general population studies (Denmark, Norway, and
Italy). Nordic Journal of Psychiatry, 70(5), 374–379. https://doi.org/10.
3109/08039488.2016.1155235
Chen, J., & Chen, Z. (2008). Extended Bayesian information criteria for
model selection with large model spaces. Biometrika, 95(3), 759–771.
https://doi.org/10.1093/biomet/asn034
1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
|
SOLMI ET 11
AL.
Cloninger, C. R., Przybeck, T. R., & Svrakic, D. M. (1991). The Tridimen-
sional Personality Questionnaire: U.S. normative data. Psychological
Reports, 69(3), 1047–1057. https://doi.org/10.2466/pr0.1991.69.3.
1047
Costantini, G., Epskamp, S., Borsboom, D., Perugini, M., Mottus, R., Wal-
dorp, L., & Cramer, A. O. J. (2014). State of the aRt personality
research: A tutorial on network analysis of personality data in R. Jour-
nal of Research in Personality, doi: https://doi.org/10.1016/j.jrp.2014.
07.003
Cuthbert, B. N. (2014). The RDoC framework: Facilitating transition from
ICD/DSM to dimensional approaches that integrate neuroscience
and psychopathology. World Psychiatry, 13(1), 28–35. https://doi.org/
10.1002/wps.20087
Dacquino, C., De Rossi, P., & Spalletta, G. (2015). Schizophrenia and
bipolar disorder: The road from similarities and clinical heterogeneity
to neurobiological types. Clinica Chimica Acta, 449, 49–59. https://
doi.org/10.1016/j.cca.2015.02.029
Dakanalis, A., Timko, C. A., Clerici, M., Zanetti, M. A., & Riva, G. (2014).
Comprehensive examination of the trans-diagnostic cognitive behav-
ioral model of eating disorders in males. Eating Behaviors, 15(1), 63–
67. https://doi.org/10.1016/j.eatbeh.2013.10.003
Dalle Grave, R., Calugi, S., Doll, H. A., & Fairburn, C. G. (2013). Enhanced
cognitive behaviour therapy for adolescents with anorexia nervosa:
An alternative to family therapy? Behavior Research and Therapy, 51
(1), R9–R12. https://doi.org/10.1016/j.brat.2012.09.008
Derogatis, L. (2010). Symptom Checklist-90-Revised.
DuBois, R. H., Rodgers, R. F., Franko, D. L., Eddy, K. T., & Thomas, J. J.
(2017). A network analysis investigation of the cognitive-behavioral
theory of eating disorders. Behaviour Research and Therapy, 97, 213–
221. https://doi.org/10.1016/j.brat.2017.08.004
Eddy, K. T., Dorer, D. J., Franko, D. L., Tahilani, K., Thompson-Brenner,
H., & Herzog, D. B. (2008). Diagnostic Crossover in Anorexia Nervosa
and Bulimia Nervosa: Implications for DSM-V. American Journal of
Psychiatry, 165(2), 245–250. https://doi.org/10.1176/appi.ajp.2007.
07060951
Eisler, I., Simic, M., Hodsoll, J., Asen, E., Berelowitz, M., Connan, F., . . .
Landau, S. (2016). A pragmatic randomised multi-centre trial of multi-
family and single family therapy for adolescent anorexia nervosa.
BMC Psychiatry, 16(1), 422. https://doi.org/10.1186/s12888-016-
1129-6
Elran-Barak, R., Sztainer, M., Goldschmidt, A. B., Crow, S. J., Peterson, C.
B., Hill, L. L., . . . Le Grange, D. (2015). Dietary Restriction Behaviors
and Binge Eating in Anorexia Nervosa, Bulimia Nervosa and Binge
Eating Disorder: Trans-diagnostic Examination of the Restraint
Model. Eating Behaviors, 18, 192–196. https://doi.org/10.1016/j.eat-
beh.2015.05.012
Epskamp, S., Borsboom, D., & Fried, E. I. (2017). Estimating psychological
networks and their accuracy: A tutorial paper. Behavior Research
Methods, 50(1):195–212. https://doi.org/10.3758/s13428-017-0862-
1
Epskamp, S., Constantini, G., Cramer, A. O. J., Waldorp, L., Schmittmann,
V., & Borsboom, D. (2016). Qgraph: Graph plotting methods, psycho-
metric data visualization, and graphical model estimation. R package
version 1.3.5.
Fairburn, C. G., Cooper, Z., & Shafran, R. (2003). Cognitive behaviour
therapy for eating disorders: A “transdiagnostic” theory and treat-
ment. Behaviour Research and Therapy, 41(5), 509–528.
Favaro, A., Caregaro, L., Tenconi, E., Bosello, R., & Santonastaso, P.
(2009). Time trends in age at onset of anorexia nervosa and bulimia
nervosa. The Journal of Clinical Psychiatry, 70(12), 1715–1721.
https://doi.org/10.4088/JCP.09m05176blu

12 |
SOLMI ET AL.
Federico, F., Benedetta, D., Claudia, M., Emanuela, A., Valentina, L., &
Leonardo, M. (2017). A disconnection between nutritional status (in
terms of body mass index and phase angle) and psychopathology in
anorexia nervosa. Psychiatry Research, 252, 196–200. https://doi.org/
10.1016/j.psychres.2017.02.043
Forbush, K. T., Siew, C. S., & Vitevitch, M. S. (2016). Application of net-
work analysis to identify interactive systems of eating disorder psy-
chopathology. Psychological Medicine, 46(12), 2667–2677. https://doi.
org/10.1017/S003329171600012X
Forbush, K. T., Wildes, J. E., Pollack, L. O., Dunbar, D., Luo, J., Patterson,
K., . . . Watson, D. (2013). Development and validation of the Eating
Pathology Symptoms Inventory (EPSI). Psychological Assessment, 25
(3), 859–878. https://doi.org/10.1037/a0032639
Foygel, R., & Drton, M. (2010). Extended Bayesian information criteria
for Gaussian graphical models. Advances in Neural Information Process-
ing Systems, 23, 2020–2028.
Franko, D. L., Tabri, N., Keshaviah, A., Murray, H. B., Herzog, D. B.,
Thomas, J. J., . . . Eddy, K. T. (2018). Predictors of long-term recovery
in anorexia nervosa and bulimia nervosa: Data from a 22-year longi-
tudinal study. Journal of Psychiatric Research, 96, 183–188. https://
doi.org/10.1016/j.jpsychires.2017.10.008
Fried, E. I., van Borkulo, C. D., Cramer, A. O., Boschloo, L., Schoevers, R.
A., & Borsboom, D. (2017). Mental disorders as networks of
problems: A review of recent insights. Social Psychiatry and Psychiatric
Epidemiology, 52(1), 1–10. https://doi.org/10.1007/s00127-016-
1319-z
Friedman, J., Hastie, T., & Tibshirani, R. (2014). Glasso: Graphical lasso
estimation of Gaussian graphical models. R package version 1.8.
Garner, D. M., Olmstead, M. P., & Polivy, J. (1983). Development and
validation of a multidimensional eating disorder inventory for ano-
rexia nervosa and bulimia. International Journal of Eating Disorders, 2
(2), 15–34. https://doi.org/10.1002/1098-108X(198321)2:2<15::
AID-EAT2260020203>3.0.CO;2-6
Godart, N., Radon, L., Curt, F., Duclos, J., Perdereau, F., Lang, F., . . . Fla-
ment, M. F. (2015). Mood disorders in eating disorder patients: Prev-
alence and chronology of ONSET. Journal of Affective Disorders, 185,
115–122. https://doi.org/10.1016/j.jad.2015.06.039
Goodkind, M., Eickhoff, S. B., Oathes, D. J., Jiang, Y., Chang, A., Jones-
Hagata, L. B., . . . Etkin, A. (2015). Identification of a common neuro-
biological substrate for mental illness. JAMA Psychiatry, 72(4), 305–
315. https://doi.org/10.1001/jamapsychiatry.2014.2206
Guyon, H., Falissard, B., & Kop, J. L. (2017). Modeling Psychological
Attributes in Psychology - An Epistemological Discussion: Network
Analysis vs. Latent Variables. Frontiers in Psychology, 8, 798. https://
doi.org/10.3389/fpsyg.2017.00798
Hindler, C. G., Crisp, A. H., McGuigan, S., & Joughin, N. (1994). Anorexia
nervosa: Change over time in age of onset, presentation and duration
of illness. Psychological Medicine, 24(03), 719–729.
Insel, T. R. (2014). The NIMH Research Domain Criteria (RDoC)
Project: Precision medicine for psychiatry. American Journal of Psy-
chiatry, 171(4), 395–397. https://doi.org/10.1176/appi.ajp.2014.
14020138
Keski-Rahkonen, A., Raevuori, A., Bulik, C. M., Hoek, H. W., Rissanen,
A., & Kaprio, J. (2014). Factors associated with recovery from
anorexia nervosa: A population-based study. International Journal
of Eating Disorders, 47(2), 117–123. https://doi.org/10.1002/eat.
22168
Kountza, M., Garyfallos, G., Ploumpidis, D., Varsou, E., & Gkiouzepas, I.
(2017). [The psychiatric comorbidity of anorexia nervosa: A compara-
tive study in a population of French and Greek anorexic patients].
L’Encephale, https://doi.org/10.1016/j.encep.2017.07.005
1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SOLMI ET691
AL.
Levinson, C. A., Zerwas, S., Calebs, B., Forbush, K., Kordy, H., Watson,
H., . . . Bulik, C. M. (2017). The core symptoms of bulimia nervosa,
anxiety, and depression: A network analysis. Journal of Abnormal Psy-
chology, 126(3), 340–354. https://doi.org/10.1037/abn0000254
Linardon, J., Fairburn, C. G., Fitzsimmons-Craft, E. E., Wilfley, D. E., &
Brennan, L. (2017). The empirical status of the third-wave behaviour
therapies for the treatment of eating disorders: A systematic review.
Clinical Psychology Reviews, 58, 125–140. https://doi.org/10.1016/j.
cpr.2017.10.005
Loeb, K. L., Lock, J., Grange, D. L., & Greif, R. (2012). Transdiagnostic
Theory and Application of Family-Based Treatment for Youth with
Eating Disorders. Cognitive and Behavioral Practice, 19(1), 17–30.
https://doi.org/10.1016/j.cbpra.2010.04.005
Maj, M. (2005). Psychiatric comorbidity”: An artefact of current diagnos-
tic systems? British Journal of Psychiatry, 186(03), 182–184. https://
doi.org/10.1192/bjp.186.3.182
Milos, G., Spindler, A., Buddeberg, C., & Ruggiero, G. (2004). Eating disor-
ders symptomatology and inpatient treatment experience in eating-
disordered subjects. International Journal of Eating Disorders, 35(2),
161–168. https://doi.org/10.1002/eat.10243
Olatunji, B. O., Levinson, C., & Calebs, B. (2018). A network analysis of
eating disorder symptoms and characteristics in an inpatient sample.
Psychiatry Research, 262, 270–281. https://doi.org/10.1016/j.psy-
chres.2018.02.027
Patrick, C. J., & Hajcak, G. (2016). RDoC: Translating promise into pro-
gress. Psychophysiology, 53(3), 415–424. https://doi.org/10.1111/
psyp.12612
R_Core_Team (2013). R: A language and environment for statistical com-
puting. Retrieved from http://www.R-project.org/. Accessed on May
7, 2018.
Schmidt, U., Magill, N., Renwick, B., Keyes, A., Kenyon, M., Dejong, H.,
. . . Landau, S. (2015). The Maudsley Outpatient Study of Treatments
for Anorexia Nervosa and Related Conditions (MOSAIC): Comparison
of the Maudsley Model of Anorexia Nervosa Treatment for Adults
(MANTRA) with specialist supportive clinical management (SSCM) in
outpatients with broadly defined anorexia nervosa: A randomized
controlled trial. Journal of Consulting and Clinical Psychology, 83(4),
796–807. https://doi.org/10.1037/ccp0000019
Shevlin, M., Hyland, P., Vallieres, F., Bisson, J., Makhashvili, N., Javakhish-
vili, J., . . . Roberts, B. (2017). A comparison of DSM-5 and ICD-11
PTSD prevalence, comorbidity and disability: An analysis of the
Ukrainian Internally Displaced Person’s Mental Health Survey. Acta
Psychiatrica Scandinavica, 137(2):138–147. https://doi.org/10.1111/
acps.12840
Smith, K. E., Crosby, R. D., Wonderlich, S. A., Forbush, K. T., Mason, T.
B., & Moessner, M. (2018). Network analysis: An innovative frame-
work for understanding eating disorder psychopathology. Interna-
tional Journal of Eating Disorders, 51(3), 214–222. https://doi.org/10.
1002/eat.22836
Solmi, M., Veronese, N., Correll, C. U., Favaro, A., Santonastaso, P., Care-
garo, L., . . . Stubbs, B. (2016). Bone mineral density, osteoporosis,
and fractures among people with eating disorders: A systematic
review and meta-analysis. Acta Psychiatrica Scandinavica, 133(5),
341–351. https://doi.org/10.1111/acps.12556
van Borkulo, C., Boschloo, L., Borsboom, D., Penninx, B. W., Waldorp, L.
J., & Schoevers, R. A. (2015). Association of Symptom Network
Structure With the Course of [corrected] Depression. JAMA Psychia-
try, 72(12), 1219–1226. https://doi.org/10.1001/jamapsychiatry.
2015.2079
Wildes, J. E., & Marcus, M. D. (2015). Application of the Research
Domain Criteria (RDoC) framework to eating disorders: Emerging
 1098108x, 2018, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eat.22884 by Universidad De Chile, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SOLMI
692 ET AL. |
SOLMI ET 13
AL.

concepts and research. Current Psychiatry Reports, 17(5), 30. https://

doi.org/10.1007/s11920-015-0572-2 How to cite this article: Solmi M, Collantoni E, Meneguzzo P,
Degortes D, Tenconi E, Favaro A. Network analysis of specific
S UPP ORT ING IN FORMAT ION psychopathology and psychiatric symptoms in patients with eat-
ing disorders.
disorders.Int
IntJ Eat Disord.
J Eat 2018;51:680–692.
Disord. 2018;00:1–13. https://doi.org/
Additional Supporting Information may be found online in the sup-
10.1002/eat.22884
10.1002/eat.22884
porting information tab for this article.