Content uploaded by Lars Kessing
Author content
All content in this area was uploaded by Lars Kessing on Apr 06, 2022
Content may be subject to copyright.
Research report
The effect of prolonged duration of untreated depression on
antidepressant treatment outcome
Jens Drachmann Bukh
n
, Camilla Bock, Maj Vinberg, Lars Vedel Kessing
Psychiatric Center Copenhagen, Rigshospitalet, University Hospital of Copenhagen, Department 6233, Blegdamsvej 9, DK-2100 Copenhagen, Denmark
article info
Article history:
Received 13 April 2012
Received in revised form
6 July 2012
Accepted 6 July 2012
Available online 30 July 2012
Keywords:
Depression
Duration of untreated illness
Remission
abstract
Background: The duration of untreated illness has been considered a likely predictor of the course of
psychotic disorders. However, there is only sparse data concerning the influence of treatment delay on
the outcome of mood disorders. The present study aimed to assess the effect of prolonged untreated
depression on the outcome of antidepressant treatment.
Method: Patients aged 18–70 years with recent onset of the first lifetime depressive episode were
systematically recruited by the Danish Psychiatric Central Research Register during a 2-year period. A
total number of 399 individuals out of 1006 potential participants in the Register were interviewed, and
270 fulfilled the inclusion criteria. The validity of the diagnosis, duration of untreated illness, remission
on first-line antidepressant treatment and a number of covariates, including psychiatric co-morbidity,
personality disorders and traits, stressful life events prior to onset, and family history of psychiatric
illness, were assessed by structured interviews.
Results: The remission rate was significantly decreased among patients with six months or more of
untreated depression as compared to patients who were treated with antidepressant medication earlier
after onset (21.1% versus 33.7%, OR ¼0.5, 95% CI 0.3 to 0.9, p¼0.03). The negative influence of a
prolonged DUI on the outcome did not seem confounded by any of a wide range of demographic and
clinical variables.
Limitations: The outcome was evaluated retrospectively. The findings cannot be generalized to patients
outside hospital settings.
Conclusion: Initiation of antidepressant treatment more than six months after onset of first episode
depression reduces the chance of obtaining remission. The results emphasize the importance of early
recognition and treatment of patients suffering from depression.
&2012 Elsevier B.V. All rights reserved.
1. Introduction
The duration of untreated illness (DUI), defined as the time
interval between first onset of the illness and initiation of
(pharmacological) treatment, has increasingly been considered
to influence the clinical course of different psychiatric disorders
(Dell’osso and Altamura, 2010), especially in schizophrenia and
other psychotic disorders (Nordentoft et al., 2009). Similarly,
knowledge of the potential consequences of postponing medical
treatment of depression is of crucial importance, when the choice
stands between antidepressants, other therapies or watchful
waiting. Unfortunately, there is little evidence for the effect of
DUI on the outcome of affective disorders. Scott et al. (1992)
reported, that the length of the index episode prior to instituation
of antidepressant treatment was the most important predictor of
time to recovery in a cohort of 55 patients with major depression.
Gormley et al. (1999) replicated the findings of an association
between the ‘‘no-treatment interval’’ of a depressive episode and
the time to remission following antidepressant treatment
(N¼83). Later, Kisely et al. (2006) found a significant association
between the duration of untreated symptoms and the overall
psychiatric morbidity in a 12-month follow-up among 351
patients in primary care presenting with a range of psychiatric
disorders (201 with major depression). However, the duration of
untreated episode or untreated symptoms during recurrent
depression is essentially different from the concept of DUI,
generally defined as untreated first onset illness. Furthermore, a
number of patients suffering from recurrent depression will be
taking maintenance antidepressant medication; hence, the dura-
tion of untreated episode cannot be established in a meaningful
way for those individuals. Only two small studies (de Diego-
Adelino et al., 2010;Okuda et al., 2010)–including 83 and 133
patients, respectively–have recently provided preliminary evidence
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/jad
Journal of Affective Disorders
0165-0327/$ - see front matter &2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jad.2012.07.008
n
Correspondence to: Psychiatric Center Copenhagen, Rigshospitalet, Research
Unit for Affective Disorders, Department 6233, Blegdamsvej 9, DK-2100 Køben-
havn Ø, Denmark
E-mail addresses: jens.bukh@regionh.dk, jensbukh@privat.dk (J.D. Bukh).
Journal of Affective Disorders 145 (2013) 42–48
Downloaded for Anonymous User (n/a) at Capital Region of Denmark from ClinicalKey.com by Elsevier on April 06,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.
for a negative influence of prolonged DUI in the more strict sense
(that is, the duration of untreated first episode depression) on the
outcome of antidepressant treatment. In addition, Altamura et al.
(2007,2008) has reported an unfavourable effect of a longer DUI on
the subsequent course of the illness (a higher number of recur-
rences, a longer total duration of illness and more frequent
co-morbid axis I disorder).
The present study aimed to assess the influence of DUI on the
remission rate following antidepressant treatment in a larger
sample of patients suffering exclusively from first episode depres-
sion, investigating the hypothesis that longer DUI is associated
with a poorer treatment outcome. Further, a broad range of
demographic and clinical covariates were examined in order to
detect potential associations with the DUI.
2. Method
2.1. Design
The present investigation is part of a study of gene–environment
interactions on depression and antidepressant outcome. The sample
and methodology of the study have been described in more details
elsewhere (Bukh et al., 2010). In short, 401 in- and out-patients
discharged from a psychiatric hospital in eastern Denmark (Zealand
(Sjælland)) with the main diagnosis of a single depressive episode
(ICD-10, code DF32-32.9) were sampled consecutively from the
Danish Psychiatric Central Research Register every second month in
a 2-year period from 2005 through 2007. Patients with organic,
schizoaffective or non-affective psychotic disorders, or a history of
recurrent depression, manic, hypomanic, or mixed state episodes
were not included (see flowchart in Fig. 1 for further details). The
Danish Psychiatric Central Research Register is a nation-wide
registration of all psychiatric hospitalisations and outpatient con-
tacts in Denmark (Munk-Jorgensen and Mortensen, 1997) compris-
ing information on treatment settings, duration of contact to
psychiatric care, and psychiatric diagnoses. Further inclusion cri-
teria were treatment with antidepressant medication, age at
discharge was 18 to 70 years, and Danish ethnicity. Since the
present study was part of a study of gene–environment interac-
tions, rather strict criteria were used for Danish ethnicity, which
was established, when the patient as well as both parents were
born in Denmark, and none of the grandparents were born outside
Europe. Exclusion criteria were significant physical illness, demen-
tia or mental retardation.
The participants were invited to the present study consecutively
2–3 months after discharge from inpatient treatment or completion
of outpatient treatment. They completed the Eysenck Personality
Questionnaire (EPQ) (Eysenck and Eysenck, 1975) and went through
a number of standardized semi-structured interviews conducted
after discharge by two experienced medical doctors (CB and JDB).
Firstly, the clinical diagnoses reported to the Register were validated
by the Schedules for Clinical Assessment in Neuropsychiatry (SCAN)
(Wing et al., 1990) based on the interview with the patient and
data from case reports. ICD-10 diagnoses were established for the
episode leading to psychiatric hospital contact and for the lifetime
Individuals in register
N = 1486
Excluded
N = 480
owing to data protection (N=78)
non-pharmacological treatment
(N=78), ethnicity (N=291), death
(N=9), disability (N=14), or
migration (N=10)
Refused to participate
or did not answer
N = 369
Short telephone interview or
questionnaire
N = 238
Affective disorder
N = 360
Potentiel participants
N = 1006
Bipolar affective disorder
N = 13
Depressive disorder
N = 345
Dysthymia
N = 2
Participants
N = 399
No affective disorder
N = 39
Recurrent depression
N = 44
Single depressive episode
N = 301
Included in the analyses
N = 270
Never received an adequate
antidepressant trial
N = 31
Fig. 1. Participants in the study.
J.D. Bukh et al. / Journal of Affective Disorders 145 (2013) 42–48 43
Downloaded for Anonymous User (n/a) at Capital Region of Denmark from ClinicalKey.com by Elsevier on April 06,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.
before. Secondly, diagnoses of personality disorders were assessed
according to DSM-IV criteria using the Structured Clinical Interview
for DSM-IV Axis II Personality Disorders (SCID-II) (First et al.,1997).
Thirdly, the presence of life events in a period of six months prior to
onset of the depressive symptoms was examined by means of the
Interview for Recent Life Events (IRLE) (Paykel, 1997). Stressful life
events were defined as independent events with moderate to severe
negative impact. Fourthly, the family history of psychiatric illness
among first degree relatives was evaluated using a modification of
the Family History Method (Andreasen et al., 1977). Fifthly, a
complete medical treatment history was assessed by the Treatment
Response to Antidepressants Questionnaire (TRAQ) (Posternak et al.,
2004), which includes a recording of the duration, dosage, compliance
and outcome (rated from no effect to very good effect on a five-point
scale) of each antidepressant treatment trial. Sixthly, the severity of
depressive symptoms at the time of the interview was measured by
the 17-item Hamilton Depression Rating Scale (Ham-D 17)
(Hamilton, 1960). An adequate antidepressant trial was defined in
accordance with the Antidepressant Treatment History Form (ATHF)
by Sackheim (Sackeim, 2001) as treatment with an antidepressant
drug for Z4 weeks in a sufficient dosage (corresponding to a score of
3oraboveinATHF)andwithZ85% compliance (corresponding to a
score of 4 in the TRAQ). Remission following the first adequate
antidepressant trial was defined as a response score of 4 or 5 in TRAQ
as well as a Ham-D 17 scorer7 at the time of assessment.
Non-remission was also assumed, when the initial antidepressant
treatment had been discontinuated because of insufficient effect.
Discontinuation was defined as either a shift to another antidepres-
sant or add-on treatment with lithium, another antidepressant drug
(except for mianserin in doses not higher than 20 mg and amitriptylin
or nortriptylin in doses not higher than 25 mg day
1
), or an
antipsychotic drug in a dosage equivalent to 100 mg chlorpromazine
per day or above.
The onset of the first depressive episode (the presence of
significant symptoms meeting the criteria according to ICD-10)
was established by the SCAN interview, the starting date of
antidepressant treatment was determined by the TRAQ, and DUI
was defined as the time interval in between.
The two investigators (CB and JDB) conducted co-ratings of 10
interviews in a pilot study and of 16 interviews during the 2-year
inclusion period. The Cohen’s kappa coefficients of reliability were
acceptable (the diagnosis of a single depressive episode: 1.0, the
diagnosis of a co-morbid personality disorder: 0.76, the presence
of one or more SLE: 0.82, family history of depression within
first degree relatives: 0.91, remission at the time of assessment
(Ham-Dr7): 0.95).
Register information on age, gender, treatment settings (inpatient/
outpatient), discharge diagnoses, and the duration of psychiatric
hospital contact was available for all participants as well as non-
participants in the study. Additional information was collected for
non-participants on ethnicity, adverse life events prior to onset of
depression, treatment with antidepressants, and overall subjective
improvement during the treatment period by a structured telephone
interview or in the alternative by means of a questionnaire including
the same questions, which was posted to all non-participants.
The investigation was conducted in accordance with the latest
version of the Declaration of Helsinki. After complete description
of the study to the subjects, written informed consent was
obtained. The Danish Ethic Committee (KF 01.209/04) and the
Data Inspection (2004-41-4562) approved the study.
2.2. Statistical analyses
In the two previous studies of the effect of DUI on treatment
outcome, DUI was categorized into two and five groups, respec-
tively (de Diego-Adelino et al., 2010;Okuda et al., 2010). Both
studies applied a cut-off of 8 weeks, and the study with five
groups sat the highest cut-off in 6 months. To make the present
results comparable with these studies, we categorized DUI into
the following three groups: short DUI (r2 months), intermediate
DUI (2 to 6 months), and long DUI (46 months). The association
between DUI (categorized into the 3-point ordinal scale) and
remission was estimated by
g
-coefficients, where a
g
-coefficient
between numerically 0.2 and 0.3 indicates a moderate correlation
and a
g
-coefficient above 0.3 indicates a strong correlation
(Goodman and Kruskall, 1954). Odds ratios for remission among
patients with intermediate and long DUI, respectively, as com-
pared to patients with short DUI were assessed in a multinomial
logistic regression model with inclusion of age-of-onset and
gender as covariates. Further, the associations between DUI and
a number of other variables were assessed in univariate analyses
in order to include those covariates, which showed a significant
association with DUI, in the regression model. Specifically, chi-
square test was used in the analyses of associations between DUI
and categorical variables. The association with age-of-onset,
which was not normally distributed, was assessed by Mann–
Whitney test, and the associations with EPQ neuroticism scale
and duration of treatment by one-way ANOVA. P-values (two-
sided)o0.05 were regarded to indicate statistical significance in
all analyses. The tests were performed with SPSS 19 for windows.
3. Results
3.1. The sample
As can be seen from Fig. 1, a total of 399 participants took part
in the full face-to-face interview, corresponding to a participation
rate of 40%. At the time of assessment, the mean duration of the
total treatment period was 10.3 months (95% CI 9.4–11.3
months). The participants did not differ from the non-participants
with respect to age at discharge, severity of depression (mild,
moderate or severe according to Register diagnoses), setting (in-
or out-patients), duration of hospital contact according to Register
information, nor with respect to the prevalence of SLE, Danish
ethnicity, or overall subjective improvement during treatment
according to the non-participants interview (all p40.1). A minor,
however statistically significant, excess of women were included
in the study (64.9% versus 58.5%, p¼0.04). Among the 399
participants, only 301 patients fulfilled the ICD-10 criteria for a
single depressive episode according to the SCAN research inter-
view, even though they were all discharged with that clinical
diagnosis (see flowchart in Fig. 1). The remaining patients
obtained diagnoses of recurrent depression (11.0%), bipolar dis-
order (3.3%), dysthymia (0.5%), schizophrenia (1.0%) and various
other diagnoses (8.8%) and were consequently excluded from
further analyses. Among the non-remitters, 31 individuals never
received at least one adequate antidepressant trial according to
the Sackheim criteria, leaving a total of 270 patients for the
analyses of remission on first-line antidepressant treatment
(Fig. 1). The sample consisted of 179 female (66.3%) and 91
(33.7%) male patients with a median age-of-onset of 37 years
(quartiles: 27 to 50 years). The severity of depression according to
the ICD-10 diagnoses was distributed as follows: mild: 67 (24.8%),
moderate: 146 (54.1%), and severe: 57 (21.1%).
All participants were treated in naturalistic settings. Selective
serotonin reuptake inhibitors (SSRIs) constituted the most frequent
first-line drugs (69.6%), and among these citalopram and escitalo-
pram were the most commonly used medications (citalopram:
n¼113, escitalopram: n¼39, fluoxetine: n¼15, sertraline: n¼16,
paroxetine: n¼3, unknown SSRI: n¼2). Mirtazapine (a noradrener-
gic and specific serotonergic antidepressant) was used as first-line
J.D. Bukh et al. / Journal of Affective Disorders 145 (2013) 42–4844
Downloaded for Anonymous User (n/a) at Capital Region of Denmark from ClinicalKey.com by Elsevier on April 06,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.
drug for 41 participants (15.2%), and 13.7% were given a serotonin
and norepinephrine reuptake inhibitor as first-line treatment (ven-
lafaxine: n¼20, duloxetine: n¼17). Tricyclic antidepressants were
rarely used as first-line treatment (n¼4). As can be seen from
Table 2, there were no significant associations between DUI and the
class of antidepressants administrated as first-line treatment and
also no differences between mean dosages of individual antide-
pressants and DUI (data not shown).
3.2. Effect of DUI on remission
The median DUI was 3.4 months (quartiles: 1.0–9.6 months).
A total of 91 patients (33.7%) received antidepressant medication
within 2 months after onset of the depressive episode, 84 (31.1%)
were treated between 2 and 6 months after onset, and 95 patients
(35.2%) experienced 6 months or more of untreated illness. As can
be seen from Table 1, a longer DUI resulted in decreased remission
rate. The correlation between DUI and remission rate was of
moderate strength (
g
¼0.23) and reached significance (p¼0.03).
In order to examine, whether the effect of DUI was a result of
outliers with a very long DUI, we repeated the analysis with
exclusion of patients with a DUI436 months (15 individuals) and
DUI424 months (27 individuals), respectively, and found a largely
unchanged
g
-coefficients of 0.24 (p¼0.03) in both analyses.
In a multinomial logistic regression model with gender and
age-of-onset as covariates and the short DUI (r2 months) as
reference, there was a significant effect of long DUI compared to
short DUI on remission (OR¼0.5, 95% CI 0.3–1.0, p¼0.04), but no
effect of intermediate DUI as compared to short DUI (OR¼0.9,
95% CI 0.5–0.7, p¼0.8). In this model, there were no significant
effects on remission of either gender (OR¼0.8, 95% CI 0.4–1.4,
p¼0.4) or age-of-onset (OR¼1.0, 95% CI 1.0–1.0, p¼0.5). These
results indicate, that DUI could be expressed as a dichotomous
variable with a cut-off of 6 months. Hence, we also analysed the
effect of DUIr6 months versus DUI46 months in a regression
model with age-of-onset and gender as covariates and found an
OR for remission of the same size as when comparing the two
extreme groups (OR¼0.5, 95% CI 0.3–0.9, p¼0.03), confirming
that the DUI in our sample was meaningfully described by only
one cut-off point of 6 months.
In order to identify potentially confounding variables, we
assessed the associations between DUI and a number of demo-
graphic and clinical characteristics that might either have influ-
enced the DUI or have been a consequence of prolonged DUI. To
increase the possibility of finding significant associations, we
analysed differences between the patients with short (r2months)
and long (46 months) DUI as well as between individuals with
DUIr6 months and DUI46 months. As can be seen from Table 2,
there were no significant differences between the groups with
respect to gender, age-of-onset, working and marital status, severity
of depression, prevalence of psychotic and melancholic features, co-
morbidity including personality disorders of cluster A, B, and C, the
level of neuroticism, family history of psychiatric disorders or the
experience of stressful life events during a 6 months period
preceding onset of depression. The prevalence of depressive per-
sonality disorder (which is not included in any of the clusters) was
lower among patients with a DUI46 months as compared to those
with shorter DUI (4.2% versus 11.4%), and this difference was
borderline significant (p¼0.05). However, including co-morbidity
of depressive personality disorder as a covariate in the regression
model did not substantially change the results.
AscanbeseenfromTable 2, the mean duration of the treatment
period was longer among patients with DUI46 months as com-
pared to the patients with shorter DUI, though this difference did
not reach significance. Correcting for the effect of the total duration
of treatment in the regression models did not substantially change
the results, which remained significant.
4. Discussion
4.1. Principal findings
This is the largest study to date of the effect of DUI on
antidepressant outcome and the first study that has assessed
differences in the clinical presentation of depression according to
DUI. We found a substantial effect of DUI with an OR of 0.5 (95%
CI 0.3–0.9) for remission on first-line antidepressant treatment
among patients with a DUIZ6 months as compared to patients
with a shorter DUI. The results did not indicate any associations
between DUI and either gender, age-of-onset, working and
marital status, severity of depression, psychotic and melancholic
features, psychiatric co-morbidity, neuroticism, family history of
psychiatric disorders or the prevalence of stressful life events.
4.2. Comparison with previous findings
As discussed in the introduction, only two studies have
assessed the association between outcome of antidepressant
treatment and DUI, defined as untreated first episode of the
illness. In an observational study, Okuda et al. (2010) examined
133 patients treated in hospital settings for first episode depres-
sion in a period from 1997 to 2008. The participants were aged 20
to 75 years and there were no exclusion criteria except a history
of schizophrenia, schizoaffective disorders, mania or mixed epi-
sodes, thus the sample seems to conform well to ours. The
authors classified DUI in five blocks (cut-offs at 1, 2, 3, and
6 months) and found a significant impact of DUI on response and
remission rates after 8 weeks of treatment with fluvoxamine. In
another study by de Diego-Adelino et al. (2010), 141 participants
(83 with first episode depression) were recruited from two
randomised 6-week clinical trials of fluoxetine and citalopram.
Patients with co-morbid psychiatric disorders, psychotic symp-
toms and drug and alcohol abuse and patients with a placebo-
response in the run-in period were excluded. Thus, the partici-
pants were–as in the majority of clinical trials (Zimmerman et al.,
2002)–selected on a number of variables that might be associated
with DUI, implying a risk of biased results. Nevertheless, the
authors found, that patients with a short DUI (r8 weeks) were
significantly more likely to attain response on 6-week antide-
pressant treatment (OR¼3.8, 95% CI 1.5–9.7), and there was also a
trend towards higher remission rates in that group. The non-
significance of the latter result might be due to the small sample
size (N¼83), i.e. a type II error. In accordance with previous
studies (Scott et al., 1992;Gormley et al., 1999), they also found
an association between the no-treatment interval and outcome of
a depressive episode during recurrent depression (N¼58). Even
though the samples were smaller, both studies assessed baseline
severity be means of Ham-D score, and they have evaluated the
outcome prospectively, hence these studies were able to measure
Table 1
Remission on first-line antidepressant treatment of 270 patients suffering from
first episode depression according to duration of untreated illness (DUI).
DUI Remission, N(%)
Yes No
r2 months 32 (35.2) 59 (64.8)
2–6 months 27 (32.1) 57 (67.9)
46 months 20 (21.1) 75 (78.9)
J.D. Bukh et al. / Journal of Affective Disorders 145 (2013) 42–48 45
Downloaded for Anonymous User (n/a) at Capital Region of Denmark from ClinicalKey.com by Elsevier on April 06,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.
the outcome more in-depth, i.e. both response and remission
rates, all of which are notorious advantages compared to the
present study. On the other hand, none of the two prior studies of
DUI were able to control for potential effects of covariates other
than age-of-onset and gender. The study by Gormley et al. (1999)
assessed 60 patients with first episode depression and 23 patients
with recurrent depression. They included a number of potential
explanatory variables (age, severity on admission, history of
depression, social class, and neuroticism) and found only pre-
morbid neuroticism and the no-treatment interval of the index
episode to be significant predictors of outcome in a Cox regression
analysis. The authors did not present any separate analyses for
the group with onset of first episode.
4.3. Strengths
The use of a nationwide case Register enabled us to recruit study
participants systematically and to assess differences between parti-
cipants and non-participants. Only ethnically Danish patients were
included in the study, and that might also reduce a potentially
confounding effect from ethnicity, since differences in the cultural
background might influence the view of depressive symptoms and
thereby the DUI. We used comprehensive semi-structured inter-
views to ensure a high validity of the data and assessed a broad
range of clinical variables in order to identify potential confounders.
To our surprise, the patients with a DUI of 6 months or more did not
differ from patients with a shorter DUI on any of these variables
except a lower prevalence of depressive personality disorder. The
latter might be a false positive result owing to multiple testing.
However, more important in this context is the fact that long DUI
was not associated with–and thus, the unfavourable treatment
outcome not explained by–an increased prevalence of depressive
personality traits among the participants with a longer DUI.
In addition, the naturalistic settings and few exclusion criteria
increased the generalisability of the results.
4.4. Limitations
The evaluation of DUI is necessarily a retrospective measure.
In a prospective observational follow up study of healthy subjects,
Table 2
Gender, age-of-onset, working and marital status, severity of depression, melancholic and psychotic features, co-morbidity, level of neuroticism, family history, and
stressful life events preceding onset according to duration of untreated illness (DUI) among 270 patients suffering from first episode depression.
Characteristics DUI P
a
P
b
o2 months N¼91 o6 months N¼175 46 months N¼95
Gender (female), N(%) 59 (64.8) 120 (68.6) 59 (62,1) 0.7 0.3
Age-of-onset (years), median (quartiles) 38 (27–52) 37 (29–51) 39 (24–50) 0.4 0.7
Working status, N(%)
Unemployed, sick leave or retired 26 (31.0) 44 (27.5) 19 (20.4) 0.1 0.2
Employed or student 58 (69.0) 116 (72.5) 74 (79.6)
Marital status, N(%)
Married/living together 48 (57.8) 92 (57.9) 40 (46.5) 0.1 0.1
Living alone 35 (42.2) 67 (42.1) 46 (53.5)
Severity of depression, N(%)
Mild 26 (28.6) 48 (27.4) 19 (20.0) 0.4 0.3
Moderate 48 (52.7) 89 (50.9) 57 (60.0)
Severe 17 (18.7) 38 (21.7) 19 (20.0)
Melancholic depression, N(%) 58 (63.7) 115 (65.7) 62 (65.3) 0.8 0.9
Psychotic depression, N(%) 3 (3.3) 5 (2.9) 3 (3.2) 1.0 0.9
Psychiatric co-morbidity, N(%)
Anxiety/OCD 48 (52.7) 85 (48.6) 45 (47.4) 0.5 0.9
Alcohol abuse 13 (14.3) 23 (13.1) 18 (18.9) 0.4 0.2
Drug abuse 6 (6.6) 10 (5.7) 9 (9.5) 0.5 0.2
Somatoform and eating disorders 4 (4.4) 7 (4.0) 6 (6.3) 0.6 0.4
Co-morbid personality disorder, N(%)
Cluster A 4 (4.4) 4 (2.3) 3 (3.2) 0.7 0.7
Cluster B 10 (11.0) 21 (12.0) 9 (9.5) 0.7 0.5
Cluster C 16 (17.6) 29 (16.6) 16 (16.8) 0.9 1.0
Depressive 12 (13.2) 20 (11.4) 4 (4.2) 0.03 0.05
Any personality disorder 30 (33.0) 55 (31.4) 28 (29.5) 0.6 0.7
Neuroticism, mean (S.D.) 12.0 (6.5) 11.4 (6.2) 12.0 (6.3) 1.0 0.5
Family history of psychiatric illness, N(%)
Depression 30 (33.0) 53 (30.3) 27 (28.4) 0.5 0.7
Any psychiatric illness 61 (67.0) 111 (63.4) 66 (69.5) 0.7 0.3
Stressful life event prior to onset, N(%) 58 (63.7) 109 (62.3) 62 (65.3) 0.8 0.6
First-line treatment, N(%)
Selective serotonin reuptake inhibitors 70 (67.3) 134 (70.2) 70 (63.6) 0.9 0.5
Serotonin and norepinephrine reuptake inhibitors 14 (13.5) 26 (13.6) 15 (13.6)
Noradrenergic and specific serotonergic antidepressants 16 (15.4) 26 (13.6) 21 (19.1)
Tricyclic antidepressants 2 (1.9) 2 (1.0) 3 (2.7)
Other or unknown 2 (1.9) 3 (1.6) 1 (0.9)
Total duration of medical treatment (months), mean (S.D.) 9.9 (5.8) 9.9 (6.7) 11.2 (10.3) 0.3 0.2
a
P-values (2-sided) in bivariate analyses comparing patients with DUI o2 months with patients with DUI 46 months (
w
2
-test for categorical data and one-way
ANOVA/Mann–Whitney test for continuous data)
b
P-values (2-sided) in bivariate analyses comparing patients with DUIo6 months with patients with DUI 46 months (
w
2
-test for categorical data and one-way
ANOVA/Mann–Whitney test for continuous data)
J.D. Bukh et al. / Journal of Affective Disorders 145 (2013) 42–4846
Downloaded for Anonymous User (n/a) at Capital Region of Denmark from ClinicalKey.com by Elsevier on April 06,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.
the assessment itself would influence the DUI by drawing attention to
depressive symptoms and, further, observing upcoming depressions
without intervention would be ethically unacceptable. A randomised
study, in which the treatment is initiated with varying delay, can for
ethical reasons only include patients with milder depressions, and the
participants, who are waiting for later commencement of treatment,
will necessarily be influenced by the awareness of their diagnosis
during the waiting time. Consequently, a prospective determination
of DUI seems methodologically unattainable. The naturalistic, retro-
spective nature of the study also implies, that no homogenous
treatment has been used, and no objective baseline severity of
depressive symptoms was available. However, the assessment of
treatment outcome by means of TRAQ has been found valid as
compared to a prospective evaluation (Posternak and Zimmerman,
2003). We cannot exclude the possibility that patients with a poorer
treatment outcome are inclined to report an earlier onset of depres-
sive symptoms, and consequently a longer DUI. On the other hand,
patients with residual symptoms might fail to recall more remote
symptoms due to cognitive impairment. Hence, we have no clear
indication of the direction of a potential recall bias. A random
misclassification because of uncertain measurement of DUI would
tend to weaken a potential association with outcome. Further, the
natural course of the depression or the patient’s expectancy of benefit
from treatment could be associated with DUI and thereby confound
the results. However, we assume that the prevalence of spontaneous
remission as well as the expectancy effect is smaller in our sample of
patients referred to hospital settings with more severe depression as
compared to community samples or patients treated in primary care.
The evaluation of outcome was hampered by the fact, that not all of
the patients had actually received an adequate antidepressant trial,
reducing the number of patients included in the analyses. However,
we find it less likely, that this has caused selection bias, since there is
no reason to believe, that the chance of obtaining an adequate
treatment was associated with DUI. Only 40% of the potential
participants were included in the study. Though the sample seemed
representative for Danish patients referred to in- or out-patient
psychiatric hospital care and diagnosed with a single depressive
episode, we cannot exclude unpredictable effects due to selection
bias. Finally, our results cannot be generalized to patients treated for
depression outside hospital settings.
4.5. Clinical implications
The results of the present study point to prolonged DUI as a
statistically significant predictor of poorer outcome of antidepres-
sant treatment. In fact, initiation of antidepressant treatment
more than six months after onset of first episode depression
reduces the odds of obtaining remission to the half, which is a
substantial and clinically important detrimental effect. This effect
was not mediated by increased co-morbidity, increased severity
of depression, demographic variables, personality factors or a
different distribution of genetic and environmental risk factors
(i.e. family history of affective disorder and the relation to
stressful life events). Rather, the impact on the outcome seems
to be a result of DUI itself, and the poorer outcome of antide-
pressant treatment among patients with a long DUI might there-
fore be influenceable by earlier initiation of treatment. Hence, the
present findings underline the importance of a precise diagnostic
procedure and initiation of treatment within a time interval
around half a year after onset of the first depressive episode.
4.6. Future research
The effect of DUI may be due to progressive neurobiological
changes during the course of the depressive episode that some-
how reduces the susceptibility to treatment. Some studies have
correlated the total duration of depressive episodes to loss of
hippocampal volume (Sheline et al., 1999), possibly caused by
dysregulation of the hypothalamic–pituitary–adrenal axis and
hypersecretion of glucocorticoids (Sapolsky, 2000) or a decrease
in trophic factors such as Brain Derived Neurotrophic Factor
(BDNF) (Duman and Monteggia, 2006), and this effect might be
associated with a reduced responsiveness to treatment
(MacQueen, 2009). However, these results are not fully consis-
tent, and direct associations between DUI and neurobiological
parameters have never been investigated. Therefore, the biologi-
cal mechanisms underlying alterations of the responsiveness to
antidepressant treatment over time (e.g. hypercortisolism,
increased level of corticotropin-releasing hormone or decreased
expression of BDNF) remain to be clarified.
There is also a need for replications of the results among
patients with first episode depression treated in other settings,
e.g. in primary care, and for studies that assess the long-term
consequences of a prolonged DUI, e.g. on the risk of chronicity and
recurrence of depression and on the outcome of subsequent
depressive episodes. Moreover, factors that may affect the DUI
require further investigation, including society-related variables
(e.g. social stigma and differences in local mental health services
influencing the detection of psychiatric disorders and access to
treatment), patient-related variables (e.g. knowledge of psychia-
tric illness, premorbid function, and coping style), and physician-
related variables (e.g. diagnostic skills and attitudes to the
treatment of depression), in order to decrease the duration of
untreated depression. Thereby, the treatment outcome of first
episode depression could likely be improved, and hopefully also
the long-term prognosis.
Role of funding source
The studywas a part of Centre for Pharmacogenomics, University of Copenhagen,
which has received funding from the Danish Research Councils (2052-03-0025).
The Danish Research Councils had no further role in study design, in the
collection, analysis and interpretation of data, in the writing of the report and in
the decision to submit the paper for publication.
Conflict of interest
MV has received payment for board membership, consultancy or lectures from
Eli Lilly, Servier, Lundbeck, Janssen–Cilag, AstraZenica, and Bristol–Meyers Squibb;
LVK has received payment for board membership, consultancy or lectures from Eli
Lilly, Servier, Lundbeck, Janssen–Cilag, and Bristol–Meyers Squib, AstraZenica,
Pfizer, and Wyeth; JDB and CB have no financial relationships with any organisa-
tions that might have an interest in the submitted work.
Acknowledgements
The study was supported by the Danish Research Councils (2052-03-0025)
and the Eli Lilly’s Psychiatric Research Foundation. MV has received payment for
board membership, consultancy or lectures from Eli Lilly, Servier, Lundbeck,
Janssen–Cilag, AstraZenica, and Bristol–Meyers Squibb. LVK has received payment
for board membership, consultancy or lectures from Eli Lilly, Servier, Lundbeck,
Janssen– Cilag, and Bristol–Meyers Squib, AstraZenica, Pfizer, and Wyeth. JDB has
received grants from the Lunbeck Foundation and the Novo Nordisk Foundation.
References
Altamura, A.C., Dell’osso, B., Mundo, E., Dell’Osso, L., 2007. Duration of untreated
illness in major depressive disorder: a naturalistic study. International Journal
of Clinical Practice 61, 1697–1700.
Altamura, A.C., Dell’osso, B., Vismara, S., Mundo, E., 2008. May duration of
untreated illness influence the long-term course of major depressive disorder?
European Psychiatry 23, 92–96.
Andreasen, N.C., Endicott, J., Spitzer, R.L., Winokur, G., 1977. The family history
method using diagnostic criteria. Reliability and validity. Archives of General
Psychiatry 34, 1229–1235.
Bukh, J.D., Bock, C., Vinberg, M., Werge, T., Gether, U., Kessing, L.V., 2010. No
interactions between genetic polymorphisms and stressful life events on
outcome of antidepressant treatment. European Neuropsychopharmacology
20, 327–335.
J.D. Bukh et al. / Journal of Affective Disorders 145 (2013) 42–48 47
Downloaded for Anonymous User (n/a) at Capital Region of Denmark from ClinicalKey.com by Elsevier on April 06,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.
de Diego-Adelino, J., Portella, M.J., Puigdemont, D., Perez-Egea, R., Alvarez, E., Perez, V.,
2010. A short duration of untreated illness (DUI) improves response outcomes in
first-depressive episodes. Journal of Affective Disorders 120, 221–225.
Dell’osso, B., Altamura, A.C., 2010. Duration of untreated psychosis and duration of
untreated illness: new vistas. CNS Spectrums 15, 238–246.
Duman, R.S., Monteggia, L.M., 2006. A neurotrophic model for stress-related mood
disorders. Biological Psychiatry 59, 1116–1127.
Eysenck, H.J., Eysenck, S.G.B., 1975. The Manual of the Eysenck Personality
Questionnaire. Hodder and Stoughton, London.
First, M.B., Gibbon, M., Spitzer, R.L., Williams, J.B.W., Benjamin, L.S, 1997.
The Structured Clinical Interview for DSM-IV Axis II Personality Disorders
(SCID-II). American Psychiatric Press, Washington.
Goodman, L.A., Kruskall, W.H., 1954. Measures of association for cross-classifica-
tion. Journal of the American Statistical Association 63, 732–764.
Gormley, N., O’Leary, D., Costello, F., 1999. First admissions for depression: is the
‘no-treatment interval’ a critical predictor of time to remission? Journal of
Affective Disorders 54, 49–54.
Hamilton, M., 1960. A rating scale for depression. Journal of Neurology, Neuro-
surgery, and Psychiatry 23, 56–62.
Kisely, S., Scott, A., Denney, J., Simon, G., 2006. Duration of untreated symptoms in
common mental disorders: association with outcomes: International study.
\British Journal of Psychiatry 189, 79–80.
MacQueen, G.M., 2009. Magnetic resonance imaging and prediction of outcome in
patients with major depressive disorder. Journal of Psychiatry and Neu-
roscience 34, 343–349.
Munk-Jorgensen, P., Mortensen, P.B., 1997. The Danish Psychiatric Central Regis-
ter. Danish Medical Bulletin 44, 82–84.
Nordentoft, M., Jeppesen, P., Petersen, L., Bertelsen, M., Thorup, A., 2009. The
rationale for early intervention in schizophrenia and related disorders. Early
Intervention in Psychiatry 3 (Suppl. 1), S3–S7.
Okuda, A., Suzuki, T., Kishi, T., Yamanouchi, Y., Umeda, K., Haitoh, H., Hashimoto,
S., Ozaki, N., Iwata, N., 2010. Duration of untreated illness and antidepressant
fluvoxamine response in major depressive disorder. Psychiatry and Clinical
Neurosciences 64, 268–273.
Paykel, E.S., 1997. The interview for recent life events. Psychological Medicine 27,
301–310.
Posternak, M.A., Young, D., Sheeran, T., Chelminski, I., Franklin, C.L., Zimmerman,
M., 2004. Assessing past treatment history: test–retest reliability of the
treatment response to antidepressant questionnaire. Journal of Nervous and
Mental Disease 192, 95–102.
Posternak, M.A., Zimmerman, M., 2003. How accurate are patients in reporting
their antidepressant treatment history? Journal of Affective Disorders 75,
115–124.
Sackeim, H.A., 2001. The definition and meaning of treatment-resistant depres-
sion. Journal of Clinical Psychiatry 62 (Suppl. 16), 10–17.
Sapolsky, R.M., 2000. The possibility of neurotoxicity in the hippocampus in
major depression: a primer on neuron death. Biological Psychiatry 48,
755–765.
Scott, J., Eccleston, D., Boys, R., 1992. Can we predict the persistence of depression?
The British Journal of Psychiatry 161, 633–637.
Sheline, Y.I., Sanghavi, M., Mintun, M.A., Gado, M.H., 1999. Depression duration but
not age predicts hippocampal volume loss in medically healthy women with
recurrent major depression. Journal of Neuroscience 19, 5034–5043.
Wing, J.K., Babor, T., Brugha, T., Burke, J., Cooper, J.E., Giel, R., Jablenski, A., Regier,
D., Sartorius, N., 1990. SCAN. Schedules for clinical assessment in neuropsy-
chiatry. Archives of General Psychiatry 47, 589–593.
Zimmerman, M., Mattia, J.I., Posternak, M.A., 2002. Are subjects in pharmacological
treatment trials of depression representative of patients in routine clinical
practice? American Journal of Psychiatry 159, 469–473.
J.D. Bukh et al. / Journal of Affective Disorders 145 (2013) 42–4848
Downloaded for Anonymous User (n/a) at Capital Region of Denmark from ClinicalKey.com by Elsevier on April 06,
2022. For personal use only. No other uses without permission. Copyright ©2022. Elsevier Inc. All rights reserved.