Journal of Affective Disorders 228 (2018) 125–131

Contents lists available at ScienceDirect

Journal of Affective Disorders

journal homepage: www.elsevier.com/locate/jad

Research paper

Risk factors for recurrence in depression in the Lundby population, T

1947–1997
⁎
Linnéa Nöbbelin , Mats Bogren, Cecilia Mattisson, Louise Brådvik
Department of Clinical Sciences, Division of Psychiatry, Lund University Hospital, S-221 85 Lund, Sweden

A R T I C L E I N F O A B S T R A C T

Keywords: Background: Depression is a common disorder in both men and women, and the recurrence rate is high. The aim
Recurrence of this study was to identify risk factors for recurrence in depression in the Lundby Study.
Course Methods: The Lundby Study is a community-based longitudinal study with focus on mental health. The study
Depression started in 1947 and three follow-ups have been carried out since, the last one in 1997. The population consists of
Melancholia
3563 subjects. Data from 508 subjects afflicted by depression was gathered. Premorbid factors (gender, socio-
Risk factor
economic status, marital status, personality and heredity) and factors related to the first depressive episode (age,
Lundby Study
degree of impairment and melancholic depression) were investigated regarding their influence on the risk for
recurrence in depression. Multiple logistic regression was used in the calculations.
Results: Risk factors associated with recurrent depression were melancholic depression at first onset (OR 3.52
[95% CI 1.62–7.66, p < 0.001]), young age as compared to old age at first onset (OR 0.51 [95% CI 0.28–0.92, p
= 0.03]) and a premorbid nervous/tense personality (OR 1.77 [95% CI 1.22–2.56, p < 0.01]). Demographic
factors, including gender, had no effect on the odds of recurrence.
Limitations: The Lundby Study spans over 50 years, making the results vulnerable to changes in diagnostic
regimes and recall bias.
Conclusion: Melancholia at onset, regardless of severity of symptoms, had the greatest impact on the risk of
recurrence in depression in the Lundby Study. Information about risk factors for recurrence in depression are
useful in offering effective preventive measures in the form of psychotropic drugs and psychotherapy, and de-
ciding the length of follow-up.

1. Introduction demographic, psychosocial, cognitive, personality, co-morbidity, bio-

Depression is a common disorder in both women and men. Angst
et al. (2002) found that 22.4% of women and 13.9% of men in com-
munity-based cohorts in Belgium, France, UK, Spain and the Nether-
lands had at least one episode of depression during their life time. In the
Lundby Study, the cumulative risk of women becoming depressed was
30.7% and men 22.5% between the years 1972 and 1997 (Mattisson
et al., 2005). Recurrence rates in depression are high. In a specialized
care setting, up to 85% experienced a second episode of depression
within 15 years (Mueller et al., 1999) and in a community-based set-
ting, the Lundby Study, about 40% experienced a recurrence in de-
pression during a 30–39-year follow up (Mattisson et al., 2007). Iden-
tifying the group with a more recurrent course of depression would be a
cost-effective way to offer secondary preventive treatment (Hardeveld
et al., 2010).
Several variables have been associated with an increased risk of
first-time depression. The variables can be divided into groups of
logical and genetic risk factors. However, risk factors for first-time
depression might not overlap with risk factors for a recurrent course.
Demographic factors associated with first-time depression, e.g. gender,
socioeconomic and marital status, have been found less likely to in-
crease the risk of recurrence (Burcusa and Iacono, 2007). Although
most studies show no difference between recurrence of depression in
men and women, some studies (Kessing, 1998) have found female
gender to be a risk factor for recurrence. There is some indication that
psychosocial, cognitive and personality factors have an impact on both
first onset and recurrence of depression (Hardeveld et al., 2010).
Wainwright and Surtees (2002) investigated the relationship between
childhood stressful life events and depression and found that parental
divorce was a risk factor for both first-time depression and a recurrent
course, whereas frightening events in childhood and physical abuse
were solely associated with first-time depression. Some investigators
have suggested that a low level of social support can be a risk factor for
both first onset and recurrence in depression (Stice et al., 2004), but

⁎
Corresponding author.
E-mail address: linnea.nobbelin@med.lu.se (L. Nöbbelin).

https://doi.org/10.1016/j.jad.2017.11.038
Received 21 June 2017; Received in revised form 13 September 2017; Accepted 11 November 2017
Available online 24 November 2017
0165-0327/ © 2017 Elsevier B.V. All rights reserved.
L. Nöbbelin et al.
others have argued for a common underlying genetic vulnerability and
no causal relationship between low social support and depression
(Burcusa and Iacono, 2007). Rumination and negative cognitive styles
have been found to be risk factors for both first onset and recurrence in
depression (Iacoviello et al., 2006). Berlanga et al. (1999) found that
neuroticism in an outpatient sample was a risk factor for recurrence in
depression. Neuroticism is one of the ‘big five’ higher-order personality
traits, and is defined as a tendency to act impulsively, feel hostile, ex-
perience negative emotions, deem situations as stressful, worry, fear the
uncertain, and become easily fatigued (Ormel et al., 2004). Genetic
factors such as a family history of mental illness, especially affective
disorders, have been found to increase the risk of recurrence in de-
pression (Burcusa and Iacono, 2007).
In a systematic review of the course of depression in community-
and primary care-based surveys, Steinert et al. (2014) concluded that
the risk factors for an unfavourable course, chronicity and recurrence,
are variables inherent in the depressive disorder itself. In other words,
variables describing the characteristics of the depressive disorder, as
opposed to variables describing the afflicted individual and their life
situation, have the greatest impact on the course of depression. The
episode inherent variables that predict an unfavourable course of de-
pression are onset age (Eaton et al., 2008), baseline severity of de-
pression (Poutanen et al., 2007), dysthymia and overlying depression
(Rhebergen et al., 2009), co-morbid mental disorders (Skodol et al.,
2011), and number of episodes (Kessing et al., 2004).
The sub-type of a depressive episode could also be considered to be
an inherent factor of the depressive disorder. However, studies on the
courses of different depression subtypes have produced varied results.
In a review on the course of endogenous depression, O′Leary (1996)
concluded that most studies have shown that endogenous depression
has a shorter time to remission, lower relapse risk, but a higher risk of
late readmission than the non-endogenous depression subtypes. There
was no clear consensus on whether the endogenous subtype had a more
recurrent course or not. There are few long-term studies on depression
subtypes and recurrence and chronicity (Gili et al., 2012). One long-
term study (Angst et al., 2007) showed that melancholic depression had
a higher rate of chronicity but not recurrence.
It has been implied that the course of depression is best studied in a
general population sample, free from selection bias (Eaton et al., 2008).
The Lundby Study is one of four well known large-scale community
surveys of mental health (Hardeveld et al., 2013). The study started in
1947 and has been repeated three times since, in 1957, 1972 and 1997.
The aim of the current study is to identify risk factors associated
with recurrence in depression among subjects in the Lundby Study, by
comparing individuals who experienced a single episode of depression
with those who had at least one recurrence during the follow-up.
Factors investigated are gender, socioeconomic status, marital status,
personality, heredity, age at first depressive episode, severity of first
episode of depression, and presence of melancholic traits at first episode
of depression.
2. Methods
2.1. The Lundby cohorts
The Lundby Study is based on two overlapping cohorts comprising a
total of 3563 individuals. All residents of two adjoining parishes in the
south of Sweden in 1947 and in 1957 were included. During the period
1947–1997, the Lundby area gradually developed from a rural area to a
combined rural and suburban area. Most inhabitants of working age
commuted to other cities in 1972/1997, while agriculture was the main
source of income in 1947. About 50% of the Lundby population moved
from the Lundby area during the period 1947–1997 (Mattisson et al.,
2005).
In 1947, the first cohort of 2550 probands, 1312 men and 1238
women, aged 0–95, was gathered and examined with the main aim to
126
Journal of Affective Disorders 228 (2018) 125–131
study the distribution of various personality traits and mental disorders
in an unselected population (Essen-Möller et al., 1956). The second
cohort was gathered in 1957 and included the surviving 2297 in-
dividuals from the 1947 cohort as well as 1013 newcomers (Hagnell
et al., 1994). After 1957, no new individuals were included in the
survey. Two successive follow-ups of the participants, irrespective of
place of residence, were carried out, one in 1972 and another in 1997.
The attrition rate for the period 1947–1972 was about 1–2% and the
rate between 1972 and 1997 was 6% (men 5%, women 7%). The at-
trition rate was somewhat higher in the younger age groups among
both men and women, but did not differ in terms of socioeconomic
status (Nettelbladt et al., 2005).
2.2. Field investigations
Clinically experienced psychiatrists carried out all four field in-
vestigations. Information was gathered through semi-structured inter-
views, key informants, registers and case notes. In 1997, the interviews
were more comprehensive and a few background questions were added,
but the structure was similar to the interviews in previous surveys.
The interview started with questions about the proband's physical
and mental health between 1972 and 1997, and any contacts with the
medical services, primary care and psychiatric care during the same
period. Current mental health, including alcohol problems and drug
abuse, was discussed, and employment status was recorded. The pro-
band's prevailing satisfaction with life and their social situation be-
tween 1972 and 1997 was discussed. If there was any indication of
cognitive impairment, the Mini Mental Test (Folstein et al., 1975) was
administered. Preliminary diagnoses of mental disorder were assessed
according to the Lundby Study's diagnostic system for the period
1972–1997, and onset, termination and degree of impairment were
recorded. In 1997, corresponding diagnoses according to the DSM-IV
(APA, 1994) system, including a GAF-score, and the ICD-10
(Socialstyrelsen, 1996) system were recorded. Immediately after the
interview, the psychiatrist wrote a free-text description of the proband
(Nettelbladt et al., 2005).
In all four field investigations, additional information was gathered
from key informants, registers and case notes. In 1997, the Cause of
Death Register, the Patient Register in Sweden and a local outpatient
care register covering the Lundby district were used. The definitive
diagnoses were agreed by the entire team of psychiatrists using all
available information (Nettelbladt et al., 2005).
2.3. Diagnostic assessment
When the first Lundby survey was carried out, the DSM system was
not in place, so the system was only used in the 1997 survey. The di-
agnostic criteria for depression, used consistently throughout the
Lundby Study, was: “Lowered mood, depressive feelings, tendency to
guilt feelings, gloomy outlook, reduced activity, lack of initiative, re-
duced self-esteem, lowered enjoyment of life and a feeling of low vi-
tality, anxiety and fear. Has more difficulty than usual, and is often
unable to carry out his daily responsibilities. Sometimes retardation is
present. The subject is often worse in the morning and better towards
the evening. Often he has sleep disturbances and wakes up in the early
morning. Loss of appetite and weight” (Hagnell, 1966).
The Lundby diagnostic system also includes a diagnosis named
Depression +, which includes cases with predominant symptoms of
depression accompanied by anxiety, obsessive symptoms or any other
mental symptom. The degree of impairment for every episode was rated
as very severe, severe, medium, mild or minimal (Leighton et al., 1963).
However, very severe and severe degrees of impairment, as well as mild
and minimal degrees of impairment, have been grouped together in
later papers. In 1997 the degrees of impairment were approximated to
GAF-scores (APA, 1994): mild degree of impairment corresponds to
GAF 61–70, medium degree of impairment corresponds to GAF 51–60,
L. Nöbbelin et al.
and severe to GAF 1–50 (Mattisson et al., 2005). The Lundby diagnosis
of Depression with medium or severe degree of impairment roughly
corresponds to Major Depression in DSM-IV, whereas mild degree does
not reach the threshold for caseness (Eaton et al., 1997). According to
Mattisson et al. (Mattisson et al., 2007), 60% of the probands diagnosed
with Lundby depression of medium to severe impairment could be di-
agnosed with major depression according to the DSM-IV, whereas the
rest could be classified as other DSM-IV subtypes of depression or ad-
justment disorder with depressed mood.
Melancholia was assessed in retrospect in accordance with Taylor
and Fink (2006) concept of melancholic disorder. All first episodes with
Lundby depression of severe or very severe impairment were assessed
and labelled melancholic or non-melancholic. Taylor and Fink's concept
of melancholic disorder is defined by psychomotor disturbances, an-
hedonia, vegetative signs and psychotic features as well as a specific
course, response to treatment and dexamethasone suppression test re-
sults. Syndromes deemed melancholic according to Taylor and Fink
include MDD with melancholic and/or psychotic features and/or
catatonic features according to DSM-IV, bipolar depression, puerperal
depression, and abnormal bereavement. Taylor and Fink argue that a
division between these syndromes is of little relevance in treatment
decisions nor in prognosis.
2.4. Risk factors
2.4.1. Heredity
Heredity was assessed through the presence of a first-degree relative
with any mental illness.
2.4.2. Demographic variables
Demographic variables investigated in this study were gender,
socio-economic status and marital status. Socio-economic status was
divided into white-collar, blue-collar, self-employed and none (Sweden,
1984). Blue-collar workers encompass unskilled, semiskilled and skilled
workers. White-collar workers refer to assistant and intermediate non-
manual employees, employed and self-employed professionals, higher
civil servants and executives. Self-employed are all entrepreneurs other
than professionals. None include non-working adults as well as children
and senior citizens. Marital status was divided into unmarried, married,
divorced and widowed.
2.4.3. Personality
Personality was assessed by the interviewing psychiatrist scoring
observed behaviours, together with scores on structured questions
about the proband's subjective opinions on presence of various lifelong
dispositions/traits. Observable behaviours, such as tension and
gloominess, were rated by the interviewer as either absent, indicated,
evident, or extreme. The probands answered structured questions about
lifelong dispositions, such as “Are you nervously disposed?” and “Do
you get tired easily?” with either never, seldom, sometimes or often.
Some of the observed behaviours and reported dispositions were
grouped together into integrated items such as schizoid and abnormal.
After the interviews, the psychiatrists dichotomously assessed whether
the probands could be described by any of the integrated items.
From the observed behaviours, reported dispositions and the in-
tegrated items, dichotomous personality factors were constructed. The
personality factor nervous/tense incorporates traits such as feeling
uncertain, anxious, insecure, strained and worried. Down/semi-de-
pressed contains traits such as melancholic, heavy, gloomy, sad, low-
spirited or serious. Abnormal/antisocial refers to a severe deviance in
personality with psychotic, aggressive, fanatic or emotionally labile
traits. Blunt/deteriorated encompasses traits such as blunt, empty, in-
tellectually deteriorated and disturbed memory. Paranoid/schizotypal
personalities are unresponsive, reserved, suspicious, bizarre or para-
noid. Immature/primitive personalities are perceived as childish, rude
and undifferentiated. Ixoid personalities are slow, circumstantial and
127
Journal of Affective Disorders 228 (2018) 125–131
easily get stuck in an action or a train of thoughts. Probands could be
described by several personality factors or none. The focus in this study
on personality traits was on the interview immediately before the first
episode of depression, taking into account the possible effect of de-
pression on personality traits (Burcusa and Iacono, 2007).
2.4.4. Sample
All subjects who had an episode of Lundby depression with medium
to severe impairment, including very severe impairment, during the
period 1947–1997 were included in this survey. A total of 508 (312
women and 196 men) of the 3563 subjects in the entire Lundby cohort
met that criteria. Of the subjects, 172 had at least one recurrence, 34%
of the depressed men and 32% of the depressed women. Fifty-three
individuals had an episode of depression before inclusion, but the di-
vision into recurrent depression or single episode depression was based
on the number of episodes of depression during the period of the
Lundby survey. Twenty-nine of the subjects were, at first episode, di-
agnosed with melancholic mood disorder in accordance with Taylor
and Fink's system. Twenty-four of the melancholic cases had MDD,
thirteen with DSM-IV melancholic features, seven with DSM-IV psy-
chotic features and four with both. Five of the cases had bipolar dis-
order and two of them had melancholic and psychotic features. No
cases of puerperal depression or abnormal bereavement at first onset
were found.
2.4.5. Statistical analysis
Subjects afflicted by depression in the Lundby Study were divided
into two groups based on illness course, recurrent depression and single
episode depression. Multiple logistic regression was performed to
identify the risk factors associated with recurrence in depression. A
confidence interval of 95% was given (95% CI). Stata/SE. 9.2 for Unix
was used in the calculations. Cross tabulations of nervous/tense and
gender was performed using SPSS Statistics 24.
2.4.6. Ethical approval
Ethical approval was granted for the Lundby Study in 1997 by the
Research Ethics Committee of the Medical Faculty at the University of
Lund. Earlier field studies predated the current system of ethical ap-
proval.
3. Results
3.1. Inherent factors and heredity
Of the factors investigated, melancholia at first onset has the
greatest impact on the odds of suffering depression a second time. The
odds ratio of recurrence if the subject was melancholic at onset is 3.52
(95% CI 1.62–7.66, p < 0.001). Degree of impairment at first onset is
not a risk factor for recurrence, and recurrence in depression does not
seem to be affected by heredity. The odds of recurrence in depression
seem to decrease with age of onset, but a significant odds decrease can
only be seen when comparing age of onset between 10 and 35 with age
of onset between 66–90, OR 0.51 (95% CI 0.28–0.92, p = 0.03).
(Table 1)
3.2. Demographic variables
Demographic variables, e.g. gender, marital and socioeconomic
status, do not seem to affect the odds of recurrence in depression.
However, a tendency towards increased odds can be seen among di-
vorced subjects (OR 2.40 (95% CI 0.87 − 6.62, p = 0.09). Nine of 16
divorced subjects had recurrent depression. (Table 2)
3.3. Personality
The only personality factor significantly affecting the odds of
L. Nöbbelin et al. Journal of Affective Disorders 228 (2018) 125–131

Table 1
Multiple logistic regression of recurrence in depression among depressed subjects in the Lundby Study (N = 508) as a function of onset variables and heredity.

Variable Non-recurrent depression N (%) Recurrent depression N (%) Total N (%) OR CI (95%) P-value

Age at onset
10–35 years 82 (61.2%) 52 (38.8%) 134 (100%) 1
36–50 years 100 (61.0%) 64 (39.0%) 164 (100%) 1.01 0.63–1.61 0.97
51–65 years 86 (71.7%) 34 (28.3%) 120 (100%) 0.62 0.37–1.06 0.08
66–90 years 68 (75.6%) 22 (24.4%) 90 (100%) 0.51 0.28–0.92 0.03*
Impairment at onset
Mediuma 256 (66.1%) 131 (33.9%) 387 (100%) 1
Severeb 80 (66.1%) 41 (33.9%) 121 (100%) 1.00 0.65–1.54 0.99
Melancholia at onsetc
No 325 (67.8%) 154 (32.2%) 479 (100%) 1
Yes 11 (37.9%) 18 (62.1%) 29 (100%) 3.52 1.62–7.66 < 0.01*
Heredityd
No 117 (66.1%) 60 (33.9%) 177 (100%) 1
Yes 217 (65.7%) 114 (34.4%) 331 (100%) 0.86 0.40–1.87 0.71

The outcome variable is recurrent depression.

N = Number; OR = Odds Ratio; CI = Confidence Interval, 95%.
a
Corresponding to GAF 51–60.
b
Corresponding to GAF 51–60.
c
Melancholic mood disorder according to Taylor and Fink (2006).
d
First-degree relative with any mental illness.
* Statistically significant.

recurrence in depression is nervous/tense. The odds ratio of having a
second episode of depression if nervous/tense is 1.77 (95% CI
1.22–2.56, p < 0,01). (Table 3a)
When subjects with any other personality factor in addition to
nervous/tense are considered, the odds ratio of recurrence is 1.88 (95%
CI 1.24–2.85). Fifty-two subjects met the criteria of both the personality
factor nervous/tense and at least one other personality factor. Being in
this group gives an odds ratio of 2.02 (95% CI 1.09–3.75) to fall ill in
depression a second time. Forty subjects were exclusively described by
another personality factor than nervous/tense. Aggregating these sub-
jects gives an odds ratio of 1.65 (95% CI 0.82–3.33) of a recurrence in
depression. (Table 3b)
In the personality assessments during the field studies in 1947, 1957
and 1972, females had higher percentages of nervous/tense personality.
In 1947, 39.2% of the females and 21.1% of the males in the Lundby
population were described by the nervous/tense personality factor. In
1957, the corresponding proportions were 33.8% for females and
19.1% for males, and in 1972, 31.9% for females and 18.9% for males.
A significant association between gender and nervous/tense personality
could be seen in the Lundby population during all field investigations
(p < 0.01). Among subjects with depression in the Lundby Study,
50.8% of the females and 35.2% of the males were described as having
a nervous/tense personality. The association between gender was sig-
nificant (p < 0.01). However, among subjects with recurrent
depression, 56.9% of the females and 49.2% of the males had nervous/
tense personalities, and there was no significant association between
gender and the nervous/tense personality factor (p = 0.33). (Table 4)
4. Discussion
Risk factors associated with recurrent depression in the Lundby
study were melancholic disorder and young age at first onset of de-
pression and a premorbid nervous/tense personality. Demographic
factors, heredity and degree of impairment at first onset were not found
to be associated with recurrence in depression. In accordance with
Steinert's (2014) conclusion on risk factors for recurrent depression,
two of three risk factors found in this study are inherent in the first
episode of depression: age at onset and presence of melancholic fea-
tures. However, presence of melancholic features was not investigated
in any of the surveys included in Steinert's review.
There has been a debate about whether melancholic depression is an
entity of its own or simply on the severe end of a depression continuum
(Fink and Taylor, 2007). Taylor and Fink (2006) argue that melancholia
is a distinct disorder encompassing major depression with psychotic,
catatonic and melancholic features, bipolar disorder, puerperal de-
pression and abnormal bereavement. Their argument is based on sur-
veys showing a distinct cortisol pattern and treatment result of tricyclic
antidepressants and ECT in melancholic depression (Fink and Taylor,

Table 2
Multiple logistic regression of recurrence in depression among depressed subjects in the Lundby study (N = 508) as a function of demographic variables.

Variable Non-recurrent depression N (%) Recurrent depression N (%) Total N (%) OR CI (95%) P-value

Gender
Male 133 (67.9%) 63 (32.1%) 196 (100%) 1
Female 203 (65.1%) 109 (34.9%) 312 (100%) 1.13 0.78–1.66 0.52
Marital status
Married/Partner 217 (65.2%) 116 (34.8%) 333 (100%) 1
Unmarried 101 (70.1%) 43 (29.9%) 144 (100%) 0.80 0.52–1.22 0.29
Divorced 7 (43.8%) 9 (56.2%) 16 (100%) 2.40 0.87–6.62 0.09
Widowed 11 (73.3%) 4 (26.7%) 15 (100%) 0.68 0.21–2.18 0.52
Socioeconomic status
Blue-collar workers 201 (65.9%) 104 (34.1%) 305 (100%) 1
White-collar workers 64 (66.7%) 32 (33.3%) 96 (100%) 0.97 0.57–1.57 0.89
Self-employed 53 (67.1%) 26 (32.9%) 79 (100%) 0.95 0.56–1.60 0.95
Unemployed 18 (64.3%) 10 (35.7%) 28 (100%) 1.07 0.48–2.41 0.86

The outcome variable is recurrent depression.

N = Numbers; OR = Odds Ratio; CI = Confidence Interval, 95%.

128
L. Nöbbelin et al. Journal of Affective Disorders 228 (2018) 125–131

Table 3a
Multiple logistic regression of recurrence in depression among depressed subjects in the Lundby Study (N = 508) as function of personality factors.

Personality factor Non-recurrent depression N (%) Recurrent depression N (%) Total N (%) OR CI (95%) P-value

Nervous/Tensea
No 201 (71.8%) 79 (28.2%) 280 (100%) 1
Yes 134 (59.0%) 93 (41.0%) 227 (100%) 1.77 1.22–2.56 < 0.01*
Paranoid/Schizotypalb
No 307 (66.9%) 137 (33.1%) 446 (100%) 1
Yes 18 (54.5%) 15 (45.5%) 33 (100%) 1.68 0.83–3.42 0.18
Abnormal/Antisocialc
No 319 (66.2%) 163 (33.8%) 446 (100%) 1
Yes 16 (64.0%) 9 (36.0%) 25 (100%) 1.10 0.48–2.55 0.83
Down/Semi-depresseda
No 327 (66.1%) 168 (33.9%) 495 (100%) 1
Yes 8 (66.7%) 4 (33.3%) 12 (100%) 0.97 0.29–3.28 1.00
Blunt/Deterioratedd
No 327 (66.3%) 166 (33.7%) 493 (100%) 1
Yes 9 (60.0%) 6 (40.0%) 15 (100%) 1.31 0.46–3.75 0.59
Immature/Primitived
No 322 (66.3%) 164 (33.7%) 486 (100%) 1
Yes 14 (63.6%) 8 (36.4%) 22 (100%) 1.12 0.46–2.73 0.82
Ixoida
No 329 (66.7%) 164 (33.3%) 493 (100%) 1
Yes 6 (42.9%) 8 (57.1%) 14 (100%) 2.68 0.91–7.84 0.08

The outcome variable is recurrent depression.

N = Numbers; OR = Odds Ratio; CI = Confidence Interval, 95%; Personality factors based on factor analysis by Mattisson et al. (2009).
a
507 cases were assessed.
b
479 cases were assessed.
c
471 cases were assessed.
d
508 cases were assessed.
* Statistically significant.

Table 3b depression than subjects with the other two subtypes. This supports our
Multiple logistic regression of recurrence in depression among depressed subjects in the results of melancholic depression having a more recurrent course than
Lundby study (N = 508) as a function of personality factor groupings. non-melancholic depression. Degree of impairment at onset was not
associated with the risk of recurrence in the Lundby Study, which is not
Personality factors Non-recurrent Recurrent Total N (%) OR CI (95%)
depression N depression N in line with previous research (Burcusa and Iacono, 2007; Hardeveld
(%) (%) et al., 2013; Poutanen et al., 2007). Melancholia, regardless of the de-
gree of impairment, was associated with increased odds of recurrence in
None 176 (73.3%) 64 (26.7%) 240 (100%) 1
depression in the Lundby Study. These results give some support to the
Only Nervous/Tense 104 (59.4%) 71 (40.6%) 175 (100%) 1.88 1.24–2.85
Only non -Nervous/ 25 (62.5%) 15 (37.5%) 40 (100%) 1.65 0.82–3.33 idea of melancholia being a distinct entity with a distinct course rather
Tense than a more severe form of depression. However, the number of mel-
Nervous/Tense + 30 (57.7%) 22 (42.3%) 52 (100%) 2.02 1.09–3.75 ancholic cases were small and no conclusion on causal relationship can
any other factor be proved between melancholic depression and a recurrent course of
depression in this study.
The outcome variable is recurrent depression.
N = Numbers; OR = Odds Ratio; CI = Confidence Interval, 95%; Personality factors Young age at onset is associated with increased odds of recurrence
based on factor analysis by Mattisson et. al (2009). in depression in the Lundby Study, which is in line with previous re-
search (Eaton et al., 2008; Gilman et al., 2003). The propensity for a
recurrence in depression when young at first onset might be explained
by a greater number of years at risk of a second episode compared to
2007). Melancholic disorder, as defined by Taylor and Fink, is the factor when the first episode of depression occurs in old age.
most strongly associated with recurrence in the Lundby Study. Previous Nervous/tense personality was significantly associated with the
research on the course of melancholic depression has given no clear risk of recurrence in depression, both in individuals only described by
answer as to whether the risk of recurrence is higher in melancholic this factor and in individuals described by multiple personality fac-
than non-melancholic depression, and there have been few studies on tors, including nervous/tense. Mattisson et al. (2009) studied risk
the long-term risk of recurrence and chronicity associated with sub- factors for first-incident depression in men and women in the Lundby
types of depression. However, Angst (2007) monitored a community cohort between 1947 and 1997, and found that nervous/tense per-
sample of young adults aged 20/21 to 40/41, comparing different sonality and anxiety disorders were risk factors for both men and
variables between melancholic and atypical depression diagnosed ac- women. Nervous/tense personality is subsequently a risk factor for
cording to DSM-IV-TR. Melancholic depression had a tendency towards both first-time depression and recurrence in the Lundby Study. Ner-
a higher rate of chronicity, but there was no difference in recurrence. vous/tense personality is closely related to neuroticism and rumina-
This is not in line with our results, which might be a result of different tion, which previous studies find to be both a risk factor for first-time
diagnostic regimes and the length of follow-up. depression (Weber et al., 2013) and for recurrence in depression
To our knowledge, there have been no studies on the course of (Berlanga et al., 1999).
melancholic disorder using Taylor and Fink's concept of melancholia. Parker and Brotchie (2010) proposed a stress diathesis model of the
Gili et al. (2012) compared melancholic depression with atypical and development of depression where neuroticism or “emotional dysregu-
undifferentiated depression, and found that subjects with melancholic lation” is the higher-order diathesis factor. They argue that depression
depression had shorter first-incident episodes and more episodes of occurs when an individual with neuroticism is put under strain. The

129
L. Nöbbelin et al. Journal of Affective Disorders 228 (2018) 125–131

Table 4
Cross tabulations of nervous/tense personality factor and gender based on assessment during different field studies and on subgroups of the Lundby population.

Non-nervous/tense N (%) Nervous/tense N (%) Total N (%) P-valuec

Field study 47a

Female 737 (60.8%) 476 (39.2%) 1213 (100%)
Male 1020 (78.9%) 273 (21.1) 1293 (100%)
Total 1757 (70.1%) 749 (29.9%) 2506 (100%) < 0.01*
Field study 57a
Female 1152 (66.2%) 587 (33.8%) 1739 (100%)
Male 1474 (81.1%) 349 (19.1%) 1823 (100%)
Total 2626 (73.7%) 936 (26.3%) 3562 (100%) < 0.01*
Field study 72a
Female 1184 (68.1%) 555 (31.9%) 1739 (100%)
Male 1479 (81.1%) 344 (18.9%) 1823 (100%)
Total 2663 (74.8%) 899 (25.2%) 3562 (100%) < 0.01*
Subjects with Depressionb
Female 153 (49.2%) 158 (50.8%) 311 (100%)
Male 127 (64.8%) 69 (35.2%) 196 (100%)
Total 280 (55.2) 227 (44.8%) 507 (100%) < 0.01*
Subjects with Recurrent Depressionb
Female 47 (43.1%) 62 (56.9%) 109 (100%)
Male 33 (50.8%) 32 (49.2%) 65 (100%)
Total 80 (46.0%) 94 (54.0%) 174 (100%) 0.33

N = Numbers.
a
Assessment of the whole Lundby population at the time.
b
The results are based on the assessment of nervous/tense during the field study closest ahead of a subject's first episode of depression.
c
Pearson Chi-square test.
* Statistically significant.

results from the Lundby Study, where nervous/tense personality is a
risk factor for first-time depression in both men and women as well as a
risk factor for recurrent depression, would support the theory of neu-
roticism being a possible diathesis factor. When rare personality traits
(paranoid/schizotypal, abnormal/antisocial, down/semi-depressed,
blunt/deteriorated, ixoid and immature/primitive) were aggregated, a
tendency towards an effect on the risk of recurrence in depression was
shown.
In line with most previous research, demographic factors, including
gender, were not found to be significant risk factors associated with
recurrence in depression (Burcusa and Iacono, 2007). Female gender is
only a risk factor for first-time depression (Mattisson et al., 2005) and
not for recurrence in depression in the Lundby Study. The nervous/
tense personality factor was more common among women in the
Lundby population in the field studies in 1947, 1957 and 1972. How-
ever, no significant association between gender and nervous/tense
personality could be seen when looking at subjects with recurrent de-
pression. Our results are in line with previous research in showing an
overrepresentation of neuroticism (Jorm, 1987) and an increased risk of
depression in women (Angst et al., 2002). One possible explanation for
the higher risk of first-time but not recurrent depression among women
is that the increased risk of first-time depression reflects the over-
representation of nervous/tense personality traits among women in the
general population. However, when individuals with depression, and
especially recurrent depression, are considered, the overrepresentation
among women should be less distinct because of a singling out effect of
individuals, both male and female, with nervous/tense personalities.
4.1. Limitations and strengths
One of the advantages of the Lundby Study is the very long-term
follow-up period, but there are some inevitable limitations. Psychiatric
diagnostics changed during the follow-up period. Semi-structured in-
terviews with several questions about personality were used in the
Lundby Study instead of a validated personality inventory, so com-
parison with other studies should be made with caution. The long in-
termissions between follow-ups increase the risk of recall bias and may
result in fewer episodes of depression being reported. However, the use
of several information sources limits the recall bias. There are also
inherent strengths in longitudinal population studies. Selection and
care-seeking biases are avoided. The low attrition rate in the Lundby
Study limits the selection bias further. The long follow-up time in-
creases the likelihood of subjects having passed their risk period for
depression during the study, resulting in an increased number of cases
being found.
Among the 508 probands involved in this study, 53 had a previous
episode of depression before 1947. As a result, some individuals in the
group with a single episode of depression might have been incorrectly
classified, which may have had some effect on the results. The onset age
in the group with recurrent depression might be lower than calculated,
hiding a more significant result. However, none of the individuals as-
sessed melancholic at first onset of depression had a previous episode of
depression before study intake, limiting the effect on this variable. The
inclusion of bipolar depression in melancholic disorder might influence
the results, because bipolar disorder has been shown to have a more
recurrent course than unipolar major depression (Kessing et al., 1998).
However, among the cases of melancholic disorder, only five out of 29
subjects were diagnosed with bipolar disorder, whereas 24 had major
depressive disorder with psychotic and/or melancholic features ac-
cording to DSM- IV. This makes it less likely that the results simply
reflect the preponderance of recurrence in bipolar depression.
5. Conclusion
Factors significantly associated with the risk of recurrence in de-
pression in the Lundby Study are melancholic depression at first onset
(regardless of the degree of impairment caused by the depression),
young age at first onset, and a premorbid nervous/tense personality.
Gender was not associated with the risk of recurrence in depression.
These findings could justify a lengthier follow-up and treatment dura-
tion when melancholia is concerned. It may also be valuable to dis-
tinguish melancholia from severe impairment.
Acknowledgements
The authors would especially like to thank the Lundby population.
Furthermore, the authors would like to express their gratitude to Anna
Lindgren, statistician at the Centre of Mathematical Sciences, Lund

130
L. Nöbbelin et al.
University, for aiding with the statistics, Anders Odensten who helped
with the structuring of data and Leslie Walke for aiding with the lan-
guage revision. At last the authors would like to thank The Ellen and
Henrik Sjöbring Foundation, Skane University Hospital and the Faculty
of Medicine, Lund University, for their financial support
Role of funding source
Funding for this study was granted from The Ellen and Henrik
Sjöbring Foundation, Department of Psychiatry, Lund University. The
funding sources had no role in either the study design, the collection,
analysis and interpretation of data nor in the writing of the paper.
Furthermore, they had no role in the decision to submit the paper for
publication.
References
Angst, J., Gamma, A., Benazzi, F., Ajdacic, V., Rössler, W., 2007. Melancholia and aty-
pical depression in the Zurich study: epidemiology, clinical characteristics, course,
comorbidity and personality. Acta Psychiatr. Scand. 115, 72–84.
Angst, J., Gamma, A., Gastpar, M., Lepine, J.P., Mendlewicz, J., Tylee, A., 2002. Gender
differences in depression. epidemiological findings from the European DEPRES I and
II studies. Eur. Arch. Psychiatry Clin. Neurosci. 252, 201–209.
APA, 1994. Diagnostic and Statistical Manual of Mental Disorders, 4th ed. American
Psychiatric Association, Washington, DC (DSM-IV).
Berlanga, C., Heinze, G., Torres, M., Apiquian, R., Caballero, A., 1999. Personality and
clinical predictors of recurrence of depression. Psychiatr. Serv. 50, 376–380.
Burcusa, S.L., Iacono, W.G., 2007. Risk for recurrence in depression. Clin. Psychol. Rev.
27, 959–985.
Eaton, W.W., Anthony, J.C., Gallo, J., Cai, G., Tien, A., Romanoski, A., Lyketsos, C., Chen,
L.S., 1997. Natural history of Diagnostic Interview Schedule/DSM-IV major depres-
sion. The Baltimore Epidemiologic Catchment Area follow-up. Arch. Gen. Psychiatry
54, 993–999.
Eaton, W.W., Shao, H., Nestadt, G., Lee, H.B., Bienvenu, O.J., Zandi, P., 2008. Population-
based study of first onset and chronicity in major depressive disorder. Arch. Gen.
Psychiatry 65, 513–520.
Essen-Möller, E., Larsson, H., Uddenberg, C.E., White, G., 1956. Individual traits and
morbidity in a Swedish rural population. Acta Psychiatr. Neurol. Scand. Suppl. 100,
1–160.
Fink, M., Taylor, M.A., 2007. Resurrecting melancholia. Acta Psychiatr. Scand. Suppl.
14–20.
Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. "Mini-mental state". A practical method
for grading the cognitive state of patients for the clinician. J. Psychiatr. Res. 12,
189–198.
Gili, M., Roca, M., Armengol, S., Asensio, D., Garcia-Campayo, J., Parker, G., 2012.
Clinical patterns and treatment outcome in patients with melancholic, atypical and
non-melancholic depressions. PLoS One 7, 1–8.
Gilman, S.E., Kawachi, I., Fitzmaurice, G.M., Buka, S.L., 2003. Socio-economic status,
family disruption and residential stability in childhood: relation to onset, recurrence
and remission of major depression. Psychol. Med. 33, 1341–1355.
Hagnell, O., 1966. A Prospective Study of the Incidence of Mental Disorders. Svenska
Bokförlaget Bonniers, Lund.
Hagnell, O., Ojesjo, L., Otterbeck, L., Rorsman, B., 1994. Prevalence of mental disorders,
personality traits and mental complaints in the Lundby Study. A point prevalence
study of the 1957 Lundby cohort of 2,612 inhabitants of a geographically defined
area who were re-examined in 1972 regardless of domicile. Scand. J. Soc. Med. Suppl.
50, 1–77.
Hardeveld, F., Spijker, J., De Graaf, R., Nolen, W.A., Beekman, A.T., 2010. Prevalence and
predictors of recurrence of major depressive disorder in the adult population. Acta
Psychiatr. Scand. 122, 184–191.
Hardeveld, F., Spijker, J., De Graaf, R., Nolen, W.A., Beekman, A.T.F., 2013. Recurrence
of major depressive disorder and its predictors in the general population: results from
the Netherlands Mental Health Survey and Incidence Study (NEMESIS). Psychol.
Med. 43, 39–48.
Iacoviello, B.M., Alloy, L.B., Abramson, L.Y., Whitehouse, W.G., Hogan, M.E., 2006. The
course of depression in individuals at high and low cognitive risk for depression: a
prospective study. J. Affect. Disord. 93, 61–69.
Jorm, A.F., 1987. Sex differences in neuroticism: a quantitative synthesis of published
research. Aust. N. Z. J. Psychiatry 21, 501–506.
131
Journal of Affective Disorders 228 (2018) 125–131
Kessing, L.V., 1998. Recurrence in affective disorder. II. Effect of age and gender. Br. J.
Psychiatry 172, 29–34.
Kessing, L.V., Andersen, P.K., Mortensen, P.B., Bolwig, T.G., 1998. Recurrence in affective
disorder. Br. J. Psychiatry 172, 23.
Kessing, L.V., Hansen, M.G., Andersen, P.K., Angst, J., 2004. The predictive effect of
episodes on the risk of recurrence in depressive and bipolar disorders - a life-long
perspective. Acta Psychiatr. Scand. 109, 339–344.
Leighton, D.L., Harding, J.S., Macklin, D.B., MacMillan, A.M., Leighton, A.H., 1963. The
Charachter of Danger: The Stirling County Study of Psychiatric and Sociocultural
Environment III Basic Books, New York.
Mattisson, C., Bogren, M., Horstmann, V., Munk-Jörgensen, P., Nettelbladt, P., 2007. The
long-term course of depressive disorders in the Lundby Study. Psychol. Med. 37,
883–891.
Mattisson, C., Bogren, M., Horstmann, V., Tambs, K., Munk-Jorgensen, P., Nettelbladt, P.,
2009. Risk factors for depressive disorders in the Lundby cohort–a 50 year pro-
spective clinical follow-up. J. Affect. Disord. 113, 203–215.
Mattisson, C., Bogren, M., Nettelbladt, P., Munk-Jorgensen, P., Bhugra, D., 2005. First
incidence depression in the Lundby Study: a comparison of the two time periods
1947–1972 and 1972–1997. J. Affect. Disord. 87, 151–160.
Mueller, T.I., Leon, A.C., Keller, M.B., Solomon, D.A., Endicott, J., Coryell, W., Warshaw,
M., Maser, J.D., 1999. Recurrence after recovery from major depressive disorder
during 15 years of observational follow-up. Am. J. Psychiatry 156, 1000–1006.
Nettelbladt, P., Bogren, M., Mattisson, C., Ojesjo, L., Hagnell, O., Hofvendahl, E., Toraker,
P., Bhugra, D., 2005. Does it make sense to do repeated surveys?–the Lundby Study,
1947–1997. Acta Psychiatr. Scand. 111, 444–452.
Ormel, J., Rosmalen, J., Farmer, A., 2004. Neuroticism: a non-informative marker of
vulnerability to psychopathology. Soc. Psychiatry Psychiatr. Epidemiol. 39, 906–912.
Parker, G., Brotchie, H., 2010. Gender differences in depression. Int. Rev. Psychiatry 22,
429–436.
Poutanen, O., Mattila, A., Seppala, N.H., Groth, L., Koivisto, A.M., Salokangas, R.K., 2007.
Seven-year outcome of depression in primary and psychiatric outpatient care: results
of the TADEP (Tampere Depression) II Study. Nord. J. Psychiatry 61, 62–70.
Rhebergen, D., Beekman, A.T., Graaf, R., Nolen, W.A., Spijker, J., Hoogendijk, W.J.,
Penninx, B.W., 2009. The three-year naturalistic course of major depressive disorder,
dysthymic disorder and double depression. J. Affect. Disord. 115, 450–459.
Skodol, A.E., Grilo, C.M., Keyes, K.M., Geier, T., Grant, B.F., Hasin, D.S., 2011.
Relationship of personality disorders to the course of major depressive disorder in a
nationally representative sample. Am. J. Psychiatry 168, 257–264.
Socialstyrelsen, 1996. International Statistical Classification of Diseases and Related
Health Problems, 10th Revision (ICD-10). Almqvist & Wiksell, Uppsala.
Steinert, C., Hofmann, M., Kruse, J., Leichsenring, F., 2014. The prospective long-term
course of adult depression in general practice and the community. A systematic lit-
erature review. J. Affect. Disord. 152–154, 65–75.
Stice, E., Ragan, J., Randall, P., 2004. Prospective relations between social support and
depression: differential direction of effects for parent and peer support? J. Abnorm.
Psychol. 113, 155–159.
Sweden, S., 1984. Swedish socioeconomic classification. Rep. Stat. Coord. 1982, 4.
Taylor, M.A., Fink, M., 2006. Melancholia: the Diagnosis, Pathophysiology, and
Treatment of Depressive Illness. Cambridge University Press, New York.
Wainwright, N.W.J., Surtees, P.G., 2002. Childhood adversity, gender and depression
over the life-course. J. Affect. Disord. 72, 33–44.
Weber, K., Giannakopoulos, P., Herrmann, F.R., Bartolomei, J., DiGiorgio, S., Chicherio,
N.O., Delaloye, C., Ghisletta, P., Lecerf, T., De Ribaupierre, A., Canuto, A., 2013.
Stressful life events and neuroticism as predictors of late‐life versus early‐life de-
pression. Psychogeriatrics 13, 221–228.
Linnéa Nöbbelin obtained her master’s degree in medicine at Lund University in 2015.
She is currently a Ph.D. student under the supervision of Dr. Louise Brådvik. Her research
is centered on the epidemiology of depressive disorders.
Dr. Louise Brådvik is an associate professor in psychiatry at Lund University and the
project manager of the Lundby study. Her research is focused on suicide, addiction and
the epidemiology of psychiatric diseases. She is a senior consultant.
Dr. Mats Bogren earned his Ph.D. in psychiatry at Lund University in 2009. His thesis
was on the epidemiology of psychosis in the Lundby population. Other research interests
are melancholia and the epidemiology of affective disorders. Dr. Bogren is currently
working as a senior consultant at the psychiatric clinic at Skane University Hospital.
Dr. Cecilia Mattisson is an associate professor in psychiatry at Lund University. Her
research is focused on depressive disorders, alcohol use disorders and epidemiology. Dr.
Mattisson is currently working as a senior consultant at the psychiatric clinic at Skane
University Hospital.