/
Eiko Fried University of Leuven Network Analysis Approach to Psychopathology and Comorbidity ABCT November 14 2015 Diagnosis of Major Depression MD Reliable diagnosis is essential ID: 635022 Download Presentation
Download Presentation - The PPT/PDF document "What are 'good' depression symptoms? Com..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Presentation on theme: "What are 'good' depression symptoms? Comparing the centrality of DSM and non-DSM symptoms of depres"— Presentation transcript
Slide1
What are 'good' depression symptoms? Comparing the centrality of DSM and non-DSM symptoms of depression in a network analysis
Eiko Fried
University of Leuven
Network Analysis Approach to Psychopathology and
Comorbidity
ABCT, November 14, 2015Slide2
Diagnosis of Major Depression (MD)
Reliable diagnosis is essential to study and treat
mental disordersReliable diagnosis of MD
is difficult: biomarkers have very limited
explanatory
power,
and MD was among the least reliable diagnosis in DSM-5 field trials (kappa = 0.28)Current state: we measure depression symptoms to indicate the presence of MD. We add them to a sum-score, and suppose this adequately represents depression severity
2Slide3
Common cause model
3s1
s2
s3
s4
s5
MSlide4
Common cause model
4s1
s2
s3
s4
s5
M
Red eyesSlide5
Common cause model
5s1
s2
M
Red eyes
Fever
s3
s4
s5Slide6
Common cause model
6s1
s2
s3
s4
s5
M
Red eyes
Fever
Runny nose
Koplik's
spots
CoughSlide7
Common cause model
7s1
s2
s3
s4
s5
Red eyesFever
Runny nose
Koplik's
spots
CoughSlide8
Common cause model
8
There
is
a
specific relationship between symptoms of a disorder and the disorder itself (
common
cause
model
)
s1
s2
s3
s4
s5
M
Red eyes
Fever
Runny nose
Koplik's
spots
CoughSlide9
Common cause model
9There is a
specific relationship between
symptoms of a disorder
and
the disorder itself (common cause model)Symptoms are somewhat interchangeables1s2
s3
s5
M
Red eyes
Fever
Runny nose
Cough
s4
Koplik's
spotsSlide10
Common cause model
10There is a
specific relationship between
symptoms of a disorder
and
the disorder itself (common cause model)Symptoms are somewhat interchangeables1s3
s4
s5
M
Red eyes
Runny nose
Koplik's
spots
Cough
s2
FeverSlide11
Common cause model
11There is a
specific relationship between
symptoms of a disorder
and
the disorder itself (common cause model)Symptoms are somewhat interchangeableSymptoms are unrelated beyond their common causes1
s2
s3
s4
s5
M
Red eyes
Fever
Runny nose
Koplik's
spots
Cough
Red eyes
Fever
Runny nose
Koplik's
spotsSlide12
Common cause model
12There is a
specific relationship between
symptoms of a disorder
and
the disorder itself (common cause model)Symptoms are somewhat interchangeableSymptoms are unrelated beyond their common causeA 'good' symptom is one that indicates the latent disease well
M
s1
s2
s3
s4
s5
Red eyes
Fever
Runny nose
Koplik's
spots
CoughSlide13
Psychiatry13
s1
s2
s3
s4
s5
D
Common
cause
model
ubiquitous
in
psychiatrySlide14
Measuring Major Depression14
s1
s2
s3
s4
s5
MD
Insomnia
Fatigue
Concentration problems
Psychomotor problems
Weight loss
Common
cause
modelSlide15
Common cause modelWe
measure symptoms to indicate the disorderAdd symptoms
to total-score to indicate severity
Measuring Major Depression
15
s1
s2s3s4
s5
MD
Insomnia
Fatigue
Concentration problems
Psychomotor problems
Weight lossSlide16
Common cause modelWe
measure symptoms to indicate the disorderAdd symptoms
to total-score to indicate severity
Symptoms roughly interchangeableWe want to
treat
the disease so symptoms disappearMeasuring Major Depression16s1s2
s3
s4
s5
MD
Insomnia
Fatigue
Concentration problems
Psychomotor problems
Weight lossSlide17
Common cause model (
overly simplistic)We
measure symptoms
to indicate the
disorder
Add
symptoms to total-score to indicate severitySymptoms roughly interchangeableWe want to treat the disease so
symptoms
disappear
Measuring
Major Depression
17
s1
s2
s3
s4
s5
MD
Insomnia
Fatigue
Concentration problems
Psychomotor problems
Weight lossSlide18
Measuring Major Depression18
s1
s2
s3
s4
s5
MD
Insomnia
Fatigue
Concentration problems
Psychomotor problems
Weight loss
Problem
:
there
is
a
dramatic
lack
of
consensus
what
depression
symptoms
(
or
good
depression
symptoms
)
are
. Different
depression
instruments
measure
very
different
things
. Slide19
What are 'good' depression
symptoms?DSM-5: 9 symptomsNone of
the common rating scales
of depression measure all DSM symptoms; all of them
measure
a number of symptoms not featured in the DSMBDI: irritability, pessimism, feelings of being punished, …HRSD: anxiety, genital symptoms, hypochondriasis, insights into the depressive illness, paralysis, …CESD: frequent crying, talking less, perceiving others as unfriendly, …As a result, there is little consistency across depression studies because patients are enrolled based on very different criteria
19Slide20
20Slide21
Measurement of depression"The measurement of depression of depression is as confused as the basic construct of the state itself."
21Slide22
Network modelSymptoms co-occur
due to their common cause
22Slide23
Network modelSymptoms
co-occur because they cause
each other
23
Psychomotor problems
Insomnia
Concentration problemss1s2
s3
s4
s5
Fatigue
Weight lossSlide24
Network modelSymptoms
co-occur because they cause
each otherSymptoms are
roughly equally important indicators
24
Psychomotor problems
InsomniaConcentration problemss1
s2
s3
s4
s5
Fatigue
Weight lossSlide25
Network modelSymptoms
co-occur because they cause
each otherSymptoms
are distinct entities
with
different
characteristics25Psychomotor problemsInsomniaConcentration problems
s1
s2
s3
s4
s5
Fatigue
Weight lossSlide26
Network modelSymptoms
co-occur because they cause
each otherSymptoms
are distinct entities
with
different
characteristicsReinforcing feedback loops (attractor state)26Psychomotor problemsInsomnia
Concentration problems
s1
s2
s3
s4
s5
Fatigue
Weight lossSlide27
Network modelImportant
new questions arise: what symptoms are most
central to driving depressive processes?
27
Psychomotor problems
Insomnia
Concentration problemss1s2
s3
s4
s5
Fatigue
Weight lossSlide28
Network modelImportant
new questions arise: what symptoms are most
central to driving depressive processes?
28
Psychomotor problems
Insomnia
Concentration problemss1s2
s3
s5
Fatigue
s4
Weight lossSlide29
Network modelImportant
new questions arise: what symptoms are most
central to driving depressive processes?
29
Psychomotor problems
Concentration problems
s2s3s4
s5
Fatigue
Weight loss
Insomnia
s1Slide30
What are 'good' depression symptoms? Comparing the centrality of DSM and non-DSM symptoms of depression in a network analysis
30
Journal of Affective Disorders
Eiko I. Fried
Sacha
Epskamp
Randolph M. NesseFrancis TuerlinckxDenny BorsboomSlide31
Research questionsWhat
is the network structure of depression?
DSM symptomsA large number of symptoms
above and beyond the DSM criteria
What
symptoms are most central, i.e. most connected in the network?31Slide32
Sample3463 depressed
outpatients from the enrollment stage of the
STAR*D studyMean age 41 years (SD=13), 63%
femaleIDS-C: 28-item questionnaire that covers 15 disaggregated
DSM
symptoms
and 13 common non-DSM symptoms (e.g., anxiety, irritability)Network estimationGaussian graphical model (special case of the Pairwise Markov Random Field): edges are partial correlation coefficientsRegularization via least absolute shrinkage and selection operator (lasso); very small edges set exactly to 0, results in a conservative (sparse) network32Slide33
33
Estimation
Edges
equal
partial
correlationsSparse network InterpretationHeterogeneous networkSome clusters emergeNetwork structure of MD
DOI
|
10.1016
/j.jad.2015.09.005Slide34
34
Symptom
importance
DOI
|
10.1016
/j.jad.2015.09.005Slide35
35
Symptom
importance
DOI
|
10.1016/j.jad.2015.09.005Slide36
Permutation test to examine
differences in centrality between DSM and non-DSM symptoms:Betweenness centrality: p
= 0.12 Closeness centrality: p = 0.64 Node strength:
p = 0.03 (0.08)Controlling for outliers: Betweenness centrality: p = 0.28
Closeness centrality:
p
= 1Node strength: p = 0.13 DSM symptoms are not more central than non-DSM symptoms36Full IDS symptom networkDOI | 10.1016/j.jad.2015.09.005Slide37
Robustness analysis
37Slide38
ConclusionsCore assumptions
of the common cause model do not seem remotely
tenable for depression"Depression sum-scores don't
add up: why analyzing specific
depression
symptoms is essential" (Fried & Nesse, 2015)Centrality measures may provide new insights regarding the clinical significance of specific depression symptoms. These insights likely have major clinical implications and suggest new approaches that may better predict outcomes such as the course of illness, probability of relapse, and treatment response.38Slide39
LimitationsSTAR*D population
Cross-sectional (indegree vs outdegree centrality)
Heterogeneity of depressionTopological overlap
39Slide40
Thank youSlide41
Eiko FriedUniversity of Leuven
University of Amsterdameiko-fried.comeiko.fried@gmail.comSlide42
DiscussionRobustness: DSM
and non-DSM symptoms do not differ regarding means (W = 121, p = 0.30) or SD (W = 89, p = 0.72)
10 disaggregated symptoms not more central than the other 18 symptoms (node strength: p = 0.86; betweenness and closeness: p = 1
)42Slide43
43