Consistency of the NEO-FFI during an episode of major ... 6.pdf · The FFM can be reliably meas...
Transcript of Consistency of the NEO-FFI during an episode of major ... 6.pdf · The FFM can be reliably meas...
-
VU Research Portal
Combination treatment for depressed outpatients: efficacy and prediction of outcome
Blom, M.B.J.
2007
document versionPublisher's PDF, also known as Version of record
Link to publication in VU Research Portal
citation for published version (APA)Blom, M. B. J. (2007). Combination treatment for depressed outpatients: efficacy and prediction of outcome.
General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ?
Take down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.
E-mail address:[email protected]
Download date: 06. Apr. 2021
https://research.vu.nl/en/publications/9c92c672-b2d5-4912-8e13-a846dff765c4
-
105
6
Consistency of the NEO-FFI during
an episode of major depression
Blom M.B.J., Hoffman, T., Spinhoven, Ph. Jonker, K., Hoencamp, E., Haffmans P.M. J.,
and van Dyck, R. Manuscript submitted for publication.
-
106
ABSTRACT Five ways of assessing continuity of personality factors were examined in 138
outpatients treated for major depressive disorder. Structural, absolute, individual-level,
ipsative and differential continuity, using the NEO-FFI, were all shown to be stable over
time. Changes in Neuroticism and Agreeableness were significant over time, but much
smaller compared to changes in depression severity. Changes on the HAMD and BDI
showed important differences in overlap with changes on personality dimensions, of
which Neuroticism was foremost. In this study it is shown that personality factors can be
reliably assessed in the course of treatment effective in reducing depression severity, but
that correlations between changes in depression scores and changes in personality
dimensions are more pronounced on a self-rated compared to an expert-rated instrument.
-
107
Reliable assessment of personality factors during an episode of major depressive disorder
(MDD) is thought as questionable. The discussion has been much influenced by an early
study by Hirschfeld et al. (1983b). In this study personality factors were found to be
strongly influenced by state factors. Hence it was concluded that personality factors
cannot be reliably assessed during an episode of MDD. This view is supported by a large
number of studies showing that scores on personality scales change significantly after
treatment (Coyne & Gotlib, 1983; Bagby, Joffe et al., 1995; Griens et al., 2002).
Especially for the personality dimension Neuroticism (Ne), described as a “general
tendency to experience negative affects…” (Costa & McCrae, 1992, page 14), this has
been found. If Ne and symptoms of depression show a significant overlap, Ne cannot be
evaluated separately from current symptoms. Clinicians then should be cautioned to draw
conclusions from assessments made during an episode of MDD (Enns & Cox, 2005;
Griens et al., 2002).
Since then several authors have argued that although personality factors are influenced
by symptom level, this does not imply that personality factors cannot be reliably assessed
during a depressive episode. For instance Santor, Bagby et al. (1997) found that in a
sample of 71 depressed outpatients measures of Ne and Extraversion (E) did change in
relationship with depression severity, but at the same time changes in these scores were
only moderately accounted for by changes in depression scores. Costa, Bagby et al.
(2005a) compared test-retest correlations of the NEO-PI-R in a sample of 250 outpatients
with MDD. Their study showed that scores on the Ne scale of the NEO-PI-R decreased
as did scores on the Openness (O) and Conscientiousness (C) scales in patients who
responded to treatment. Scores on the E-scale increased. The authors however concluded
that since changes were small and the results of factor analysis of the NEO-PI-R was
similar before and after treatment, this did not mean that measurements were unreliable
but that an accentuation of personality factors occurred during an episode of MDD.
Also, as has been shown before, Ne scores remained high in remitted patients (Bagby et
al., 1995) suggesting either a scar effect or a true personality characteristic.
Much of the confusion about whether or not personality factors can or cannot be reliably
measured during an episode of MDD is due to the nature of personality change. Both
-
108
Roberts and DelVecchio (2000) and De Fruyt, Van Leeuwen et al. (2006) distinguish
several types of personality change. In most studies in this area only one of these types of
consistency is studied.
The five types of personality change described by De Fruyt et al. (2006) are differential
continuity, referring to relative differences among individuals over time, absolute
continuity, referring to the change over time of a group of patients, individual-level
continuity, referring to the magnitude of change of each individual on any trait, ipsative
continuity, referring to stability of the ordering and organization of traits within the
individual and structural continuity referring to the stability of the co-variation among
traits over time. Each of these five types of continuity describes a unique facet of stability
of personality dimensions over time. Studying all these five types of change and
continuity has several important advantages. It provides insight into the nature of change
on a group level as well as on an individual level. Ideally patients should be examined
before the onset of depression to get a true estimate of personality factors. In everyday
clinical practice this is almost never possible. However if it is shown that personality
factors can be measured reliably during an episode of MDD, the study of the role of
personality factors in treatment outcome, maintenance of symptoms and recurrence could
be more easily elucidated.
The currently most often used model to describe normal personality is the Five Factor
Model (FFM). The FFM can be reliably measured using the NEO-PI-R or the NEO-FFI
(Costa & McCrae, 1992). In the study by de Fruyt et al (2002) a French instrument, the
Système de Description en Cinq Dimensions (DSD) (Rolland & Mogenet, 2001) was
used for the measurement of the big Five. They found that the DSD showed consistency
on all five types of possible personality change and concluded that “personality states
remain stable … during the course of different antidepressant therapies”. They however
recommend that a replication be carried out using a more commonly used instrument to
assess the Five Factor Model. They also pointed to an inconsistency between several
studies (Du, Bakish et al., 2002), and their own (De Fruyt et al., 2006), in the correlation
between depression change scores during treatment and change scores of personality
traits. They hypothesized that this discrepancy could partly be due to a different method
-
109
of scoring: i.e. self-rating versus expert-rated instruments. To solve this important
clinical question we compared in this study a self-rated instrument for the measurement
of depression severity (Beck Depression Inventory (BDI) (Beck et al., 1961) with an
expert rated instrument (Hamilton Rating Scale for Depression (HAMD) (Hamilton,
1960).
We hypothesized that we would confirm the findings by de Fruyt et al. (2006) in showing
continuity of personality dimensions during treatment of MDD. We further hypothesized
that these findings would be independent from the method of administering an instrument
(self-rating versus expert-rating). To our knowledge this is the first study using the NEO-
FFI before and after treatment of depression which investigates all five types of
dimension consistency for this instrument while assessing depression severity with both a
self-rated and expert-rated instrument.
Measurement of dimension consistency Intra-individual continuity can be analyzed using the residualized change index (RCI)
(Jacobson & Truax, 1991) or growth curves estimates of change (Tate & Hokanson,
1993). In this article the method described by De Fruyt et al (2006) using RCI will be
performed.
Ipsative continuity is as the name implies, exclusively person-oriented and refers to the
stability of the ordering of dimensions within the individual. For this, participants are
assessed at least two times on measures of personality and are further examined on
consistency of the configuration of the personality dimensions.
Ipsative personality is examined using dissimilarity (D) indices discussed by Cronbach
and Gleser (1953): D2, D’2 and D’’2. Dissimilarity indices describe the similarity of two
personality dimensions of the same person by quantifying this relationship. A large value
indicates great dissimilarity and a small value indicates less dissimilarity. These indices
are used to describe different parts of profile variation. These variations can be divided in
elevation (the mean level of dissimilarity for a given person), scatter (the individual
deviation about his own mean level of dissimilarity), and shape (the residual information
-
110
indicating a persons’ individual patterning of scores). D2 is sensitive to all three parts of
variation (elevation, scatter and shape), D’2 is only sensitive to scatter and shape by
removing the effect of elevation, and D’’2 is only sensitive to shape.
Mean-level consistency is the most well known type of personality change and refers to
the extent in which personality scores of the cohort change over time.
Rank-order consistency or differential consistency refers to the degree in which the
relative difference among individuals remain stable over time (De Fruyt et.al, 2006).
Method
Design and procedure All subjects participated in a randomized clinical efficacy study. The full design and
outcome of the study have been described elsewhere (Blom, et al., 2007). Participants
were required to be 18 years or older, with non-psychotic, non-bipolar major depressive
disorder (MDD) and a score ≥ 14 on the Hamilton Depression Rating Scale (HAMD, 17-
item) (Hamilton, 1960). Participants fulfilled the criteria of MDD as specified in the
DSM-IV (APA, 1994) and were assessed with the Structured Clinical Interview for Axis
I DSM-IV Disorders, Dutch translation (SCID) (Groenestijn, Akkerhuis et al., 1998).
Excluded from the study were participants with substance abuse, serious medical
condition, organic psychiatric disorder, severe suicidality, history of psychotic disorder
or schizophrenia, bipolar disorder, current use of psychotropic medication and ongoing
psychotherapy. The study was approved by an independent Ethical Commission. All
participants gave written informed consent before entering the study.
Participants Participants who were referred to treatment either by their primary care physician or by
another mental health professional were assessed for the study. No participants were
contacted through advertisements.
Measures Depression severity was measured using the Hamilton Depression Rating Scale
(Hamilton, 1960), 17-item version (HAMD). The HAMD is an expert rated measure of
-
111
depression severity. Before entering treatment, the SCID and the HAMD were
administered by experienced clinical raters not informed of the treatment condition.
Monthly interrater trainings were held to ensure good interrater reliability. Internal
consistency has been found to be between .46 and .97 (Bagby et al., 2004;Santor et al.,
1997). In the present study Crohbach’s alpha was .533, .716 and .845 for the three points
of measurement.
The Beck Depression Inventory (BDI) (Beck et al., 1961), a 21-item self report measure
of depression severity, was also completed at baseline, after 6 weeks and upon
completion of the study. The BDI has a high internal consistency, with alpha coefficients
of .86 for psychiatric populations (Beck, Steer, & Garbin, 1988). In this study
Cronbach’s alpha was .825, .879 and .925 respectively.
Personality dimensions were studied using the NEO-Five Factor Inventory (NEO-FFI)
(Costa & McCrae, 1992). It is the short form of the NEO-PI-R and has excellent
correlation (.88-.94) with the longer NEO PI-R (Costa & McCrae, 1992; Hoekstra, Ormel
et al., 1996). Correlations with the individual scales of the NEO-PI-R were found to be
for respectively Ne: .93; E: .90; O: .94; A: .88 and C: .89 (Costa & McCrae, 1992). The
NEO-FFI was administered at baseline and at endpoint.
Analytic Method Differential continuity was examined by using Pearson correlation coefficients between
the two assessments periods and structural continuity was examined by using structural
equations modelling (SEM) as described by (Robins, Fraley et al., 2001). After a visual
inspection of the intercorrelations, this was followed by a formal test of equivalence of
the correlation matrix. This was done by comparing the fit of a model in which the
intercorrelations among the NEO-FFI were freely estimated (baseline model) and a
model in which the intercorrelations were constrained to be equivalent across the two
assessment periods. If the fit of the constrained model compared to the fit of the baseline
model (unconstrained model) resulted in decline of the overall fit, it would indicate
significant changes in the intercorrelations among the NEO-FFI. The baseline model was
-
112
specified as a single-indicator latent variable model with one latent variable associated
with each of the 10 scale scores. The model was identified by fixing the variances of the
latent variables to 1 and the variances of the residuals to 0. The paths among all latent
variables and between each latent variable and its scale score were also estimated. As a
result, the matrix of covariances among the latent variables is equivalent to the matrix of
correlations. The baseline model is fully saturated. The constrained model was estimated
by placing 10 pairwise equality constraints between the scale scores across the
assessment periods.
Absolute continuity or mean-level change was examined by descriptive statistics,
Pearson correlations between the personality dimensions within and across time periods.
The correlations were corrected for unreliability (Hoekstra et al., 1996; Schmitt, 1996).
Next, a repeated measures analysis on the two assessment periods was used to test the
mean-level change. Effect sizes were calculated using an adjusted Cohen’s d for repeated
measures designs (Dunlop, Cortina et al., 1996). A regression analyses with partial
change scores of the personality measures as dependent variables and partial change
scores for depression as the independent variable was used to predict the personality
change by accounting for changes in depression. Partial change scores were calculated to
account for the time-dependent interrelation of same personality dimensions (Cohen,
Cohen et al., 2006). Rather than using simple change scores, partial change scores were
used. If we had used simple change scores, this would have suggested that regression of
the post-scores on the pre-scores had a slope of 1.0 instead of the regression coefficient
beta. This would risk an overcorrection of the postscores by the prescores (Cohen et al.,
2006).
Partial change scores of depression were regressed on partial change scores of all
personality dimensions using hierarchical multiple regression. The order of inclusion of
independent variables was as follows: Ne, E, A, C, and finally O. This order of
independent variables was based on Mulder (2002) indicating that Ne is the best
predictor followed by E.
-
113
Individual-level continuity was assessed by using the Reliable Change Index (RCI)
(Jacobson & Truax, 1991). This index was used to describe participants showing
decreased, unchanged or increased personality dimension scores. Participants were
classified as having a reliable increase or decrease in NEO-FFI scores when the
probability of the RCI was less than 5%. These categories were further tested as in
Robins et al. (2001) from the expected frequencies of 2.5% (reliable decrease), 95% (no
change), and 2.5% (reliable increase) for each of the personality dimension scores by
using chi-square analysis. Using the RCI requires reliability estimates, Cronbach’s alpha
and descriptive statistics from a normative sample. These data were obtained from the
Dutch NEO-FFI manual (Hoekstra, et al., 1996).
Ipsative continuity was examined by using D2, D’2 and D’’2 indices (Cronbach & Gleser,
1953). Participants were classified as having changed as the probability of the D2, D’2 and
D’’2 indices were less than 5%. Parallel to De Fruyt et al. (2006), these probabilities were
estimated by simulating dimension scores on a sample of 100,000 persons in which there
was no change in means, variances, covariances and coefficient alpha reliabilities as
estimated from the real data.
Results The sample consisted of 193 participants. Of these, 138 (71.5%) completed the study.
For a description of clinical and demographical variables see table 1. Some participants
were not able or refused to complete the NEO FFI either at baseline or at endpoint. A
total of 126 (91.3%) completed the NEO-FFI both at baseline and upon completion of the
study..
Participants who dropped out (DO) differed in some aspects significantly from the
completer sample: they were more often single (chi-square (1) = 4.473; p = 0.034) and
were more often from a non-European (NE) background (chi-square (1) = 6.840; p =
0.009). Illness-related factors, such as duration and severity of the index episode and
number of previous episodes, did not differ between DO and completers (F (1, 178) =
.003; p = .956; F (1, 190) = 1.069; p = .302, respectively chi-square (1) = .769; p = .380).
-
114
DO scored significantly higher on the Ne-dimension of the NEO-FFI (F=7.821; p =
0.006), but not on any of the other four NEO-FFI dimensions.
Table 1: Baseline demographic and clinical characteristics of all
participants who completed the trial.
Mean (SD) / %
(frequency) N
Age 40.82 (11.18) 137Sex (% female) 64.2 (88) 137Age first diagnosis 35.8 (12.58) 130Length index episode 11.69 (13.73) 132First episode (% first) 72.6 (98) 135Marital status (% single) 39.6 (53) 134Cultural background (% non-European)
14.5 (20) 138
Melancholic features (% melancholic)
56.6 (73) 129
Baseline Neuroticism 36.63 (5.48) 126Baseline Extraversion 35.77 (4.72) 125Baseline Openness 34.10 (3.85) 124Baseline Agreeableness 36.92 (4.45) 126Baseline Conscientiousness 38.65 (4.21) 125Baseline HAMD 21.14 (4.73) 138Baseline BDI 31.20 (9.33) 138
Differential continuity and structural stability Table 2 shows the intercorrelations among the Big Five Dimensions assessed at 0 weeks
and 12 weeks. On visual inspection one sees that the intercorrelations reported above the
diagonal (week 12) strongly parallel those below the diagonal (week 0). To formally test
the structural stability of the Big Five, SEM was used to compare the correlation matrices
on equivalence. A chi-squared difference test indicated that the constrained model did not
lead to a significant reduction in fit (chi-square (10) = 14.0255; p = .172 and CFI = .98;
-
115
SRMS = .026; RMSEA = .063). The intercorrelations are invariant across the two
assessment periods.
-
116
Tabl
e 2:
Str
uctu
ral C
ontin
uity
of P
erso
nalit
y: In
terc
orre
latio
ns A
mon
g B
ig F
ive
Dim
ensi
ons
Ass
esse
d at
Bas
elin
e an
d En
dpoi
nt.
N
E
O
A
C
Neu
rotic
ism
-
-.02
.07
.41*
**
-.06
Ext
rave
rsio
n .0
5 -
.37*
**
-.07
.53*
**
Ope
nnes
s .1
2 .2
7***
-
.05
.36*
* A
gree
able
ness
.26*
* .1
5 .1
6**
- .2
0*
Con
scie
ntio
usne
ss
.20*
.4
1***
.2
8**
.23*
-
Not
e In
terc
orre
latio
ns a
t bas
elin
e ar
e re
port
ed b
elow
the
diag
onal
, and
in
terc
orre
latio
ns a
t end
poin
t are
repo
rted
abo
ve th
e di
agon
al.
*
p >
05
** p
> .0
1 **
* p
> .0
01
Abso
lute
Con
tinui
ty o
r Mea
n-Le
vel C
hang
e fo
r HAM
D
The
desc
riptiv
e st
atis
tics a
nd c
orre
latio
ns a
mon
g th
e va
riabl
es a
re d
escr
ibed
in T
able
3.
-
117
Tabl
e 3:
Des
crip
tive
Stat
istic
s an
d C
orre
latio
ns A
mon
g Va
riabl
es
1 2
3 4
5 6
7 8
9 10
11
12
13
14
15
M
SD
1 : B
DI_
0T
1
31
,09
9,47
2:
BD
I_6T
.6
5 1
25
,89
10,3
3 3:
BD
I_12
T
.51
.76
1
21
,65
11,5
4 4
: HA
MD
_0T
.5
3 .4
5 .3
6 1
20
,95
4,40
5:
HA
MD
_6T
.4
0 .7
2 .5
4 .5
5 1
17,3
4 5,
86
6: H
AM
D_1
2T
.36
.56
.73
.41
.64
1
14,8
0 8,
34
7: N
_0T
.3
4 .2
4 .0
7 .2
0 .0
8 -.0
5 1
36,4
1 4,
95
8: E
_0T
-.1
5 -.1
5 -.1
6 -.1
9 -.1
5 -.0
9 .0
5 1
35
,50
4,81
9:
O_0
T
-.03
.02
.08
-.06
-.03
-.05
.12
.27
1
33
,89
3,76
10
: A_0
T
.21
.10
.05
-.02
.08
-.04
.26
.15
.16
1
36,8
4 4,
42
11: C
_0T
-.1
0 -.0
7 -.1
7 -.1
3 -.0
3 -.1
4 .2
0 .4
1 .2
8 .2
3 1
38,2
0 4,
53
12: N
_12T
.2
2 .3
6 .3
8 .1
2 .2
5 .1
8 .4
7 .1
1 .0
5 .1
4 -.0
7 1
35
,08
4,35
13
: E_1
2T
-.20
-.30
-.48
-.13
-.33
-.45
.11
.55
.25
.08
.39
-.02
1
35
,42
4,31
14
: O_1
2T
-.24
-.19
-.25
-.16
-.06
-.27
.10
.22
.53
.13
.21
.07
.37
1
34,4
7 3,
44
15: A
_12T
.2
5 .3
1 .3
1 .1
2 .1
9 .1
1 .1
9 -.0
4 .1
0 .4
7 .0
8 .4
1 -.0
7 .0
5 1
35,4
9 3,
88
16: C
_12T
-.1
6 -.0
5 -.2
1 -.0
9 .0
4 -.1
8 .0
4 .2
9 .2
1 .1
5 .6
1 -.0
6 .5
3 .3
6 .2
0 37
,89
3,64
BD
I: B
eck
Dep
ress
ion
Inve
ntor
y; H
AM
D: H
amilt
on D
epre
ssio
n R
atin
g Sc
ale,
Ne:
Neu
rotic
ism
; E: E
xtra
vers
ion;
O:
Ope
ness
; A: A
gree
able
ness
; C: C
onsc
ient
ious
ness
; 0t:
base
line;
6T:
ratin
g at
6 w
eeks
; 12T
: end
poin
t rat
ing.
-
118
Ta
ble
4: A
bsol
ute
chan
ge in
per
sona
lity
and
depr
essi
on
Wee
k 0
Wee
k 12
Scor
es
M
SD
M
SD
N
F (1
, 94
) p
Part
ial-e
ta
squa
red
Adj
uste
d C
ohen
’s d
N
euro
ticis
m
36.4
1 4.
9535
.08
4.35
95
7.25
90.
008
0.07
2 0.
285
Ext
rave
rsio
n 35
.50
4.81
35.4
24.
31
950.
035
0.85
1 0.
000
0.01
8 O
penn
ess
33.8
9 3.
7634
.47
3.44
95
2.46
60.
120
0.02
6 0.
159
Agr
eeab
lene
ss
36.8
4 4.
4235
.49
3.88
95
10.2
00.
002
0.09
8 0.
324
Con
scie
ntio
usne
ss
38.2
0 4.
5337
.89
3.64
95
0.69
70.
406
0.00
7 0.
075
Dep
ress
ion
seve
rity
(HA
MD
) 20
.95
4.40
14.8
08.
34
9554
.33
0.00
0 0.
366
0.82
0
Dep
ress
ion
seve
rity
(BD
I)
31.0
9 9.
4721
.65
11.5
495
81.3
40.
000
0.46
4 0.
884
-
119
Personality factors before and after treatment are highly correlated, with correlations of
.469, .552, .527, .472 and .609 for Ne, E, O, A and C respectively. After correction for
unreliability (Hoekstra et al., 1996) these correlations achieved high levels of .558, .756,
.798, .694 and .883.
Absolute continuity was assessed by comparing mean-level scores at week 0 and week
12 using repeated measures ANOVA. Absolute changes in personality and depression are
reported in Table 4.
Depression scores decreased considerably after treatment, but participants described
themselves on average also as slightly less neurotic, more open to experience, and less
agreeable after treatment. The changes in personality represent small changes because
none of the effect sizes indicated large values (small = .02, medium = .15, large = .35).
Transformed to Cohen’s d for repeated measures designs (Dunlop et al., 1996) the effect
sizes lead to the same conclusion (see table 4). Personality dimensions of the Big Five
remain stable across the two assessments. The reduction in depression score represents a
large and significant change (adj. Cohen’s d = 0.845).
Regressing personality partial change scores on depression partial change scores shows
that personality change is only to a limited extent accounted for changing levels of
depression. A and C are unrelated to changes in depression scores (on the HAMD) (p =
.484 respectively p = .807). Ne (p = .016), E (p < .0001), and O (p = .022) are related to
the Depression Scores and explain 4.6% of variance, 12.2% of variances, and 4.2% of the
variance of the HAMD score respectively. Regressing HAMD change scores on
personality traits change scores resulted in explaining 18.0% of variance with only E
showing significance (p < .001). Change scores of Ne, O, A, and C showed no significant
prediction on depression change scores (p = .071, p = .060, p = .827, and p = .104).
Absolute Continuity or Mean-Level Change for the BDI The same procedure as described above for the HAMD was applied for the BDI. The
descriptive statistics and correlations among the variables are described in table 3.
After correction for unreliability, the FFM before and after treatment is highly correlated,
with correlations of .56, .76, .80, .70, and .88 for Ne, E, O, A and C respectively.
-
120
Regressing personality partial change scores on BDI partial change scores shows that
personality change is only to an extent accounted for changing levels of the BDI. C is
unrelated to changes in BDI scores (p = .872). Ne (p
-
121
expected frequencies of 2.5% (reliable decrease), 95% (no change), and 2.5% (reliable
increase) for each of the dimensions scores (Ne: p = .099; E: p = .641; A: p = .187; C: p
= .385). O was not tested, for this dimension was a constant.
Ipsative Continuity D2 correlated .872 with D’2, and .393 with D’’2. D’2 correlated .522 with D’’2. The
distribution of D2 ranged from 5.0 to 398.0, with a mean of 87.014 and a standard
deviation of 77.961. The 25th , 50th, and 75th quartile values were 35.25, 60.00 and
113.75. Only 1.9% had D2 values greater than expected on the basis of chance. The
distribution of D’2 ranged from 3.2 to 297.10, with a mean of 61.399 and a standard
deviation of 63.222. The 25th , 50th, and 75th quartile values were 25.20, 39.40 and 70.70.
None of the participants exceeded levels of significance. The distribution of D’’2 ranged
from -.11 to 3.04, with a mean of .765 and a standard deviation of .767. The 25th , 50th,
and 75th quartile values were .174, .479 and 1.171. Of the participants 12.5% had D’’2
values greater than would be expected by measurement error alone. Consequently, none
of the profile changes reflected changes due to changes in elevation and scatter. Most of
the profile changes reflected changes due to changes in shape.
Discussion Our study is principally a replication of the findings of De Fruyt et al (2006), although
the instrument to measure the big Five used by De Fruyt et al (2006) was different from
the NEO-FFI used in this study. We also extended the findings by comparing results on
the HAMD as well as on the BDI.
As in the aforementioned study, we found that the structure of the Big Five personality
factors remained stable during treatment for depression. Treatment did not have a
significant influence on the intercorrelations of the Big Five.
Participants did, however change over time: Ne and A scores were lower compared to
baseline. Although changes were significant, the absolute change was very small. For
instance, Ne scores went from the 98th percentile down to the 97th percentile, A moved
from the 86th to the 77th percentile and the other three factors did not move at all. As
-
122
stated by Costa et al (2005) some personality traits are more enhanced during a period of
MDD, but remain essentially the same. Of note is that participants differed substantially
from normative data obtained in the general population (Costa & McCrae, 1992). Since
we did not perform a prospective study we do not know if this reflects a vulnerability for
depression prior to the onset of the episode or a ‘scar’ effect (Akiskal, Hirschfeld et al.,
1983). In one of the few prospective studies (Ormel et al., 2004) Ne was high before the
onset of depression, even higher during an episode of MDD and returned to the original
level upon recovery.
We found that changes in depression scores on the HAMD were relatively independent
from changes in personality traits across the duration of the study. This is completely in
line with the finding by De Fruyt et al (2006) and differs somewhat from other studies in
this area (Du et al., 2002; Santor et al., 1997). The reason for this difference could be due
to the method of administering the measurement (self-rating vs. independent rating). In
our study we both examined a self-rating scale (the BDI) and an expert-rated scale
(HAMD). We did indeed find important differences in correlations between both scales.
Overlap between the HAMD and the NEO-FFI is not strong, with HAMD scores
predicting 18.8% of variance in NEO-FFI scores of which only E, and not Ne, has a
significant relationship with depression change score. Using the BDI, on the other hand,
Ne does have a large and significant relationship with depression change scores (on the
BDI) explaining 30.0% of the variance. Vice versa Ne explains 17.1% of variance in
depression change scores. Compared to the HAMD, the use of the BDI may give serious
distortions which can to a large extent be contributed to differences in personality
characteristics (Enns et al., 2000). It has also been argued that the BDI includes not only
state variables but also trait variables (Groth-Marnat, 1990). If this is true, an overlap
with personality factors is to be expected.
Differential or rank-order stability was also studied. Stability coefficients were high. In
an earlier study by Vaidya, Gray et al. (2002) comparable coefficients were found.
Differences between personality traits were very small. This is noteworthy since Ne is
often found to be less stable than the four other traits (Robins et al., 2001). This finding is
different from that in the De Fruyt et al. (2006) study. The difference could be accounted
-
123
for by the shorter duration of our study and the somewhat smaller effectsize for reduction
of depression in our study. If Ne is influenced by state effects, one would presume that a
smaller effectsize would mean a smaller change in (especially) Ne scores. Future studies
should take this into account. A second reason for the difference between our study
results and those of De Fruyt et al. (2006) could lie in the relative high mean score on Ne.
As has been found before (Brown, 2007), high initial scores of negative affect tend to
predict less change over time as compared to lower baseline scores. This could point to a
higher trait versus state relationship in these participants.
Individual-level continuity was also found in our study. By far the majority (86.1%) of
participants stayed stable on all five personality factors. Six point five percent showed a
reliable change on one personality factor and only 1.4% of participants showed a reliable
change on two factors. Of the five factors, Ne was the least stable, and O the most stable
factor.
In the same line, ipsative continuity was also found to be high. Only variability in shape
was in 12.5% of participants greater than expected. Variability in scatter and elevation
were either not significant or in a small minority of participants (1.9%) greater than
expected. Again this is in accordance with the earlier study by de Fruyt et al (2006).
In conclusion our study demonstrates that personality factors remain stable during
treatment of depression. Only a small minority of participants exhibit significant changes
in scores on any personality factor. We did however find large and significant differences
between the two instruments most widely used for measuring depression severity, the
HAMD and the BDI, with respect to the association with Ne scores in particular. When
using the BDI as (only) outcome measure this could mean that scores are strongly
influenced by personality characteristics of participants studied. The question remains if
this difference is due to difference in the method of administration (self-rated or expert-
rated) or difference in items of the two instruments. This problem could be solved using
an instrument having both a self-rated as an expert rated version (e.g. Inventory of
Depressive Symptoms, (Rush et al.,1996).
-
124
Since personality factors do not predict short term outcome in the treatment of depression
(Blom et al., 2007), but more likely influence the long term prognosis of MDD (Du et al.,
2002; Jang, Clay et al., 2004), measurement of personality factors can be relevant for
clinicians in predicting long term vulnerability to the occurrence of relapse in MDD. We
have shown in our study that personality factors can be reliably assessed by the NEO-FFI
during an episode of MDD.
-
125
-
126