16000447, 2010, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1600-0447.2009.01519.x by Cbua - Consorcio De Bibliotecas, Wiley Online Library on [20/01/2023].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Acta Psychiatr Scand 2010: 122: 184–191  2009 John Wiley & Sons A/S
All rights reserved ACTA PSYCHIATRICA
DOI: 10.1111/j.1600-0447.2009.01519.x SCANDINAVICA

Review
Prevalence and predictors of recurrence of
major depressive disorder in the adult
population
Hardeveld F, Spijker J, De Graaf R, Nolen WA, Beekman ATF. F. Hardeveld1,2, J. Spijker1,2,
Prevalence and predictors of recurrence of major depressive disorder in R. De Graaf2, W. A. Nolen3,
the adult population. A. T. F. Beekman4
1
De Gelderse Roos, Institute for Mental Health Care,
Objective: Knowledge of the risk of recurrence after recovery of a Ede, 2Netherlands Institute of Mental Health and
major depressive disorder (MDD) is of clinical and scientific Addition, Utrecht, 3Department of Psychiatry, University
importance. The purpose of this paper was to provide a systematic Medical Center Groningen, University of Groningen,
review of the prevalence and predictors of recurrence of MDD. Groningen and 4Department of Psychiatry and Institute
Method: Studies were searched in Medline en PsychINFO using the for Research in Extramural Medicine, VU University
search terms Ôrecur*Õ, Ôrelaps*Õ, Ôdepress*Õ, Ôpredict*Õ and course. Medical Centre ⁄ GGZ Buitenamstel, Amsterdam, The
Results: Recurrence of MDD in specialised mental healthcare settings Netherlands
is high (60% after 5 years, 67% after 10 years and 85% after 15 years)
Key words: major depressive disorder; recurrence;
and seems lower in the general population (35% after 15 years). predictor
Number of previous episodes and subclinical residual symptoms
F. Hardeveld, De Gelderse Roos, PO Box 70, 6710 RR
appear to be the most important predictors. Gender, civil status and
Ede, The Netherlands.
socioeconomic status seem not related to the recurrence of MDD. E-mail: f.hardeveld@degelderseroos.nl
Conclusion: Clinical factors seem the most important predictors of
recurrence. Data from studies performed in the general population and
primary care on the recurrent course of MDD are scarce. Accepted for publication November 11, 2009

Summations
• The percentage of recurrence of major depressive disorder (MDD) in specialised mental healthcare
settings is high (85% after 15 years) and seems lower in the general population (35% after 15 years).
• The two most important predictors of recurrence of MDD appear to be the number of previous
episodes and subclinical residual symptoms after the depression.
• Demographic factors, which are risk factors for the onset of MDD, seem not to be risk factors for
recurrence.

Considerations
• Studies are difficult to compare due to differences in methodology.
• The number of studies performed in the general population and primary care setting on the recurrent
course of MDD is small.

leading cause of burden of disease and will rise to

Introduction
the first place in 2030 (2).
Major depressive disorder (MDD) is one of the Long-term studies performed in the US (3)
most prevalent mental disorders with a high showed that MDD has a less favourable course
morbidity. The lifetime prevalence in the United than initially thought. Knowledge about the course
States of America (US) is estimated at 16.2% (1). of MDD and its determinants is of clinical and
Currently, MDD is worldwide and the third scientific importance. It would seem reasonable to

184
 16000447, 2010, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1600-0447.2009.01519.x by Cbua - Consorcio De Bibliotecas, Wiley Online Library on [20/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Recurrence of major depressive disorder

treat patients with a high risk of recurrences as a

Aims of the study
chronic disorder (4), while treatment of those with
a low risk of recurrence may be limited to the index To perform a systematic review of prevalence and
episode. Accurate differentiation of these groups risk factors predicting recurrence of major depres-
may help to improve both the efficacy and the sive disorder, comparing patients in specialised
efficiency of mental health care. The treatment of mental health with those seen in primary care and
the high risk group should not only include community settings.
maintenance treatment with antidepressants, but
also aim for the reduction of risk factors for
Material and methods
recurrence. This approach has been successful in
other disciplines in medicine. The prevention of Relevant articles were searched in PsychINFO and
myocardial infarction for example targets its risk Medline with the search terms Ôrecur*Õ, Ôrelaps*Õ,
factors (e.g. smoking, high blood pressure and Ôdepress*Õ, Ôpredict*Õ and course. The flow chart is
hypercholesterolemia) and this approach has low- shown in Fig. 1. Only articles written in English
ered the incidence of myocardial infarction (5). which studied the adult population and which were
Moreover, prevention programmes for diabetes published between January 1980 and August 2008
have been successful in the prevention of the long- were included (n = 736). These articles were
term complications, such as reducing diabetic screened on relevance, methodology, and study
retinopathy and nephropathy (6). Identification population by the first author (FH), using the title
of predictors of recurrence can help to select which and summary. Ninety-nine articles were retrieved
patients should be offered longer term preventive for further evaluation. Another three articles were
care. Furthermore, it is likely that patients treated included after examining the references. Of these
in specialised mental health care have a higher risk 102 articles, two authors (FH and JS) studied the
for recurrence than patients treated in primary full text. The following selection criteria were
care. If proven true, this would lead to different applied: i) naturalistic cohort study, ii) including
recommendations for patients treated for MDD in subjects with MDD, iii) MDD diagnosed using a
specialised mental health care vs. primary care. diagnostic interview or checklist based on Research
Over the past 20–30 years, several studies have Diagnostic Criteria ⁄ Bedford College Criteria
been able to assess putative risk factors for (RDC ⁄ BCC) criteria, DSM-III ⁄ III-R ⁄ DSM-IV
recurrence of MDD. The factors that have been criteria or ICD-9 ⁄ 10 criteria, iv) the course had
considered include demographic data (such as age, to be measured with a standardised instrument or
sex and level of education), psychosocial charac- checklist, v) with a minimum follow-up of
teristics (such as personality and social support) 6 months, vi) and including at least 50 subjects.
and clinical characteristics of the depressive disor- Moreover, vii) the criteria of remission, recovery,
der (such as the severity and duration of the index relapse and recurrence according to Frank et al. (7)
episode). had to be applied; relapse is a return of symptoms

Potential relevant studies: 736

Medline: 675
PsychINFO: 208

Potential relevant studies screened on

title and abstract (n = 736)

Excluded studies (n = 637)

Methodology (n = 36)
Patient population (n = 122)
Diagnosis MDD (n = 94)
No predictor measured (n = 215)
Retrieved for further evaluation (n = 99) No recurrence measured (n = 170)

Excluded studies (n = 74)

Time to recurrence was not
measured properly (n = 31)
No diagnostic instrument used
(n = 21)
3 crossreferences Patient population (n = 12)
Fig. 1. Flow chart of selection of stud- No predictor measured (n = 4)
ies that researched the prevalence and Follow-up <6 months (n = 2)
Included in review (n = 27)
predictors of recurrence of major No naturalistic cohort study (n = 5)
depressive disorder.

185
 16000447, 2010, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1600-0447.2009.01519.x by Cbua - Consorcio De Bibliotecas, Wiley Online Library on [20/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Hardeveld et al.

satisfying the full syndrome criteria for an episode (CDS). In this study 431 patients with a MDD
that occurs during the period of remission but were followed 15 years, the first 5 years half yearly
before recovery, recurrence is the appearance of a and then yearly. During the first year after recov-
new episode of MDD and can occur only during a ery, the risk of recurrence was 25%, increasing to
recovery, remission is a period during which an 85% over 15 years (10).
improvement of sufficient magnitude is observed
that the individual no longer meets syndromal Primary care and general population. Only one
criteria for the disorder, recovery is a remission primary care study was included. In this study,
that lasts for 10 days or longer. Frank et al. (7) did which included 386 patients, a percentage of
not formulate the time to recovery but made recurrence of 31% after a follow-up of 1 year
suggestions depending on the course instrument was found (11). This percentage is comparable with
used. All included studies had to fulfil all the results from the abovementioned studies in specia-
applied selection criteria. Studies that included lised mental health care. We also included only one
patients with bipolar disorder, depression with a prospective population-based study. This is the
seasonal pattern, postpartum depression or specific Baltimore Epidemiologic Catchment Area follow-
age groups were excluded because this may unduly up (ECA) (12, 13) which included 78 patients at
influence the risk of recurrence. We did not risk for a recurrence of MDD. After a follow-up of
subdivide the studies according to MDD sub- 15 years, the percentage of recurrence after recov-
groups (e.g. melancholic, with psychotic features, ery was 35%.
catatonic, atypical) because too few studies differ-
entiated according to subgroups and in those who
Predictors
did the subgroups were usually too small to
compare. The articles were screened independently Demographic factors. Many studies reported on
on the selection criteria by FH and JS. When there the association between demographic factors and
were inconsistencies (n = 3) both reviewers the risk of recurrence. These studies were mainly
reached consensus by analysing the articles on the performed in specialised mental health care.
applied selection criteria together. Demographic factors such as gender, age, socio-
economic status (SES) and marital status were not
found to be related to the recurrence of MDD
Results
(Table 3). However, there are exceptions. Mueller
Twenty-seven studies were included for further et al. (10) and Winokur et al. (14), both part of the
evaluation. The articles were subdivided in studies earlier mentioned CDS study, found a greater risk
that included patients from the general population, of recurrence among women. Mueller et al. (10)
primary care and specialised mental health care. observed a higher risk of recurrence among those
The characteristics of the included studies are who have never been married. And lastly, Holma
summarised in Table 1. et al. (15) and Eaton et al. (12) reported that a
younger age was related to recurrence.
Prevalence
Life events, social support, impairment of psychosocial
The prevalence depends on the setting in which is functioning and personality factors. The included
measured. Table 2 gives an overview of the per- studies (15–17) reported that a severe life event and
centage of recurrence by setting. lack of social support were not related to recur-
rence. Patients with impaired functioning in areas
Specialised mental health care. The majority of the such as work, relationships and leisure after
available studies were carried out in specialised (complete) remission appear to have a higher risk
mental health care. Considering all studies, the for recurrence of a MDD (11, 18). A personality
overall risk of recurrence after recovery during the disorder can be a predictor for recurrence (19). In a
first year ranges between 21% and 37%. The high study with 50 patients who were followed
percentage of 37% was observed in a study that 33–84 months, both cluster B and C personality
selected patients with a high number of previous characteristics were related to a higher risk for
episodes (8). The lowest percentage of 21% was recurrence. There are, however, also studies which
found in a Japanese study, that included patients observed no such association (15, 16, 20). Patients
treated in specialised mental health care as well as with a high neuroticism score could also have a
primary care (9). An important study is the high risk for recurrence (11). However, Holma
National Institute of Mental Health Collaborative et al. (15) observed no such association. Further-
Program on the Psychobiology of Depression more, it is possible that coping skills are associated

186
 16000447, 2010, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1600-0447.2009.01519.x by Cbua - Consorcio De Bibliotecas, Wiley Online Library on [20/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Recurrence of major depressive disorder

Table 1. Characteristics of studies that researched the association between predictors and the risk of recurrence

Maximum Number of Duration of

follow-up Diagnostic patients remission Course Predictor
Author (years) instrument Diagnosis Ôat riskÕ (months) instrument measured

Specialised mental health care

Holma et al. 2008 (15) 5 SCAN DSM-IV 140 2 LCI Demographic, clinical, personality, social
Ilardi et al. 1997 (19) 7 DIS DSM-III-R 50 0.5 LIFE Personality
Judd et al. 1998 (22) 10 SADS RDC 238 2 LIFE Clinical
Judd et al. 2000 (28) 12 SADS RDC 238 2 LIFE Clinical
Kanai et al. 2003 (9) 6 COALA DSM-IV 82 2 COALA Demographic, clinical
Kennedy et al. 2003 (26) 11 CID RDC 65 2 LIFE Demographic, clinical
Kennedy et al. 2004 (31) 10 CID RDC 59 2 LIFE Clinical
Keller et al. 1983 (32) 2 SADS RDC 234 2 LIFE Demographic, clinical
Lavori et al. 1994 (23) 5 SADS RDC 359 2 LIFE Demographic, clinical
Maj et al. 1992 (8) 9 SADS RDC 72 2 LIFE Demographic, clinical
CPRS
SADS
Melartin et al. 2004 (16) 1,5 SCAN DSM-IV 198 2 LCI Demographic, clinical, personality, social
Mueller et al. 1999 (10) 15 SADS RDC 380 2 LIFE Demographic, clinical
Naz et al. 2007 (27) 4 SCID DSM-III-R 60 2 SCID Demographic, clinical
Paykel et al. 1995 (29) 1.25 CID RDC 60 2 BDI Clinical
Paykel et al. 1996 (17) 1.25 CID RDC 60 2 BDI Demographic, social
Pintor et al. 2003 (30) 2 SCID DSM-III-R 183 2 HAMD Clinical
Pintor et al. 2004 (24) 4 SCID DSM-III-R 138 2 HAMD Clinical
Ramana et al. 1995 (20) 1.25 CID RDC 60 2 BDI Clinical
Simpson et al. 1997 (42) 15 SADS RDC 176 2 LIFE Demographic
Solomon et al. 2000 (25) 10 SADS RDC 318 2 LIFE Clinical
Solomon et al. 2004 (18) 15 SADS RDC 290 2 LIFE Social
Winokur et al. 1993 (14) 5 SADS DSM-III-R 172 2 LIFE Demographic, clinical
Primary care
Conradi et al. 2008 (21) 3 CIDI DSM-IV 110 2 CIDI Demographic, personality, social, clinical
BDI
Gopinath et al. 2007 (11) 1 SCID DSM-III-R 386 3 LIFE, SCID Demographic, clinical, social
General population
Eaton et al. 1997 (12) 15 DIS DSM-IV 80 12 DIS, LCI Demographic, clinical
Eaton et al. 2008 (13) 23 DIS DSM-IV 78 12 LCI, DIS Demographic, clinical, social

BDI, Beck Depression Index; CID, Clinical interview for Depression; CIDI, Composite International Diagnostic Interview; COALA, Comprehensive Assessment List for Affective
disorders Raskin Scale; CPRS, Comprehensive Psychopathological Rating Scale; DIS, Diagnostic Interview Schedule; DSM, Diagnostic and Statistical Manual of Mental
Disorders; LCI, Life Chart Interview; LIFE, Longitudinal Interval Follow-up Evaluation; HAMD, Hamilton Rating Scale of depression; RDC, Research Diagnostic Criteria; SADS,
Schedule for Affective Disorders and Schizophrenia; SCID, Structured Clinical Interview for DSM-IV; SCAN, Schedules for Clinical Assessment in Neuropsychiatry.

Table 2. Percentage of patients with a recurrence

of MDD during follow-up Follow-up (years) Setting Author 1 5 10 15

Percentage (%) Specialised mental health care Holma et al. 2008 (15) – 70.7 – –
Kanai et al. 2003 (9) 21 42 – –
Maj et al. 1992 (8) 37* 75 – –
Mueller et al. 1999 (10) 25 60 – 85
Melartin et al. 2004 (16) 38  – – –
Ramana et al. 1985 (20) 40à – – –
Solomon et al. 2000 (25) – – 67 –
Primary care Gopinath et al. 2007 (11) 31 – – –
General population Eaton et al. 1997 (12) – – – 35

MDD, major depressive disorder.

*Percentage of recurrence after 6 months is 24%.
 Percentage of recurrence after 1.5 years.
àPercentage of recurrence after 15 months.

with risk for a recurrence of MDD. In a study patients with lower self-efficacy had a higher risk
among 110 patients in a primary care population, for recurrence (11).
patients with moderate coping skills had a shorter
time to recurrence in comparison to those with Clinical factors. Numerous clinical variables have
better skills (21). In agreement with this result, been studied in relation to the recurrence of MDD

187
 16000447, 2010, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1600-0447.2009.01519.x by Cbua - Consorcio De Bibliotecas, Wiley Online Library on [20/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Hardeveld et al.

Table 3. Predictors associated with recurrence of MDD

Significant association between Significant association between

recurrence and MDD present recurrence and MDD absent

Demographic factors
Age 12, 15 8–11, 16, 20, 23, 26, 27
Female 10, 14 8, 9, 12, 13, 15, 16, 20, 21, 23, 26, 27, 42
Marital status 10 8, 9, 12, 13, 15–17, 20, 21, 26, 27
Low socioeconomic status ⁄ educational level – 8, 9, 11–13, 15, 16, 20, 21, 26, 27
Life events, social support, psychosocial impairment and personality factors
Severe life event before depressive episode – 15–17
Lack of social support – 13, 15–17
Decreased psychosocial functioning 11, 18 –
Comorbid personality disorder 19 15, 16, 20
Neuroticism 11 15, 16
Coping skills 11, 21 –
Clinical factors
Number of previous episodes 8, 10, 11, 14, 21–25 9, 15, 16, 19, 20, 26, 27
Subclinical residual symptoms 9, 22, 24, 28–30 16, 19, 31
Age of onset 13, 14, 24 8, 15, 16, 20, 23, 31
Severity of the last episode 11, 15, 16, 19, 20, 24, 26 8, 10, 23, 27
Duration of last episode 10, 23 8, 9, 15, 16, 20
Comorbid psychiatric disorder 8, 15, 16, 32 9, 19, 23
Family history of MDD 8 9, 13, 14, 20, 26, 27

MDD, major depressive disorder.

Each row represents the studies that performed research on the specific predictor in relation to recurrence of MDD. The figures refer to the literature in the reference list.

(Table 3). The main variables are age of onset of relapsed three times faster than an asymptomatic
the first depressive episode, duration and severity group (median 68 vs. 23 weeks) and found an odds
of the last episode, the number of previous ratio of 3.65 for residual symptoms. Three included
episodes, the presence of comorbid axis I disorders, studies did not find an association. However, the
subclinical residual symptoms and a family history study by Ilardi et al. (19) included a relatively small
of MDD. One of the variables that seems associ- number of patients (n = 50) and Melartin et al.
ated with the recurrence of MDD is the number of (16) did find a bivariate association with an high
previous episodes. A larger number of included odds ratio of 2.12 but in the multivariate models it
studies (8, 10, 11, 14, 21–25), mainly performed in was not significant anymore (16).
specialised mental health care, found an associa- Nonetheless, although there is strong evidence
tion between the number of previous episodes and that subclinical residual symptoms are a predictor
the risk of recurrence. Within these studies are for a shorter time to recurrence, this does not imply
CDS studies (10, 14, 22, 23, 25) which included a that in the long run the overall risk of recurrence
high number of patients with a long follow-up necessarily differs. Kennedy et al. (31) reported in a
using the life chart method. There are, however, study population of 60 patients with recurrent
also some studies that did not find an association depression and residual symptoms a percentage of
(9, 15, 16, 19, 20, 26, 27), but most of these studies recurrence of 42% after 1 year and 56% after
included smaller numbers of patients at risk (9, 19, 2 years in comparison with respectively 20% and
20, 26, 27) or did not use the life chart method (20, 42% in the group without residual symptoms.
27). The association between number of previous After a follow-up of 8–10 years, the percentage of
episodes and recurrence of MDD seems strong, the recurrence between the groups was no longer
odds ratio is estimated at 1.64 (22). Another strong significantly different (74% in the group with
predictor for recurrence of MDD seems to be the subclinical residual symptoms and 70% in the
presence of subclinical residual symptoms. Also, a group without residual symptoms). One of the
larger number of included studies found an asso- major limitations in this study was the use of a
ciation between this predictor and the risk of single longitudinal interview to cover the long
recurrence (9, 22, 24, 28–30) and the included follow-up period of 8–10 years. Studies that
studies had a long follow-up using the life chart assessed the association between age of onset of
method (22, 28). The included studies suggest that the first depressive episode and the risk of recur-
this is the strongest predictor for a recurrence. rence are inconsistent. For example, Eaton et al.
Judd et al. (22) described that patients with resid- (13) observed that a younger age of onset was a
ual subclinical symptoms after recovery of MDD significant predictor of time to recurrence, with

188
 16000447, 2010, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1600-0447.2009.01519.x by Cbua - Consorcio De Bibliotecas, Wiley Online Library on [20/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Recurrence of major depressive disorder

each additional year of age at onset lowering the Our review has several limitations. First, despite
risk for recurrence by 0.96 (95% confidence inter- the use of strict selection criteria, the included
val 0.93–0.99), but this result was not corroborated studies remain difficult to compare. This is also the
in other studies (8, 23). There is also inconclusive reason why we did not perform a formal meta-
data that severity or duration of the last episode is analysis, pooling the results. The studies used
a risk factor for recurrence. A high number of different diagnostic and course instruments and
included studies found an association between the duration of follow-up varied between the
severity and risk of recurrence of MDD (11, 15, studies. Our conclusions are therefore not based
16, 19, 20, 24, 26) but the studies that did not find on a strong evidence-based method but merely on
an association also included CDS studies (10, 23). number of studies with a significant association
The opposite seems true for duration of the last and design of the study (number of patient
episode; a high number of these studies did not find included, course instrument). However, other not
an association (8, 9, 15, 16, 20), but the CDS included but well performed studies show similar
studies did (10, 23). Comorbid axis I disorders results (35, 36). These studies were not included
seems to increase the risk for recurrence (16). This because the data did relate to readmissions because
is shown for dysthymia (8, 32) and for social of MDD and not to recurrence. Furthermore, in a
phobia (15). A family history of MDD did have an number of studies the difference between a relapse
association with recurrence in one study (8), but and a recurrence of MDD could not be fully
this could not be corroborated in a number of specified because these studies also included
other studies (9, 13, 14, 20, 26, 27). patients with partial remission. Subclinical residual
symptoms can also be seen as a limited prolonga-
tion of an active disease process and is therefore a
Discussion
difficult predictor to interpret in relation to recur-
To our knowledge, this is the first systematic rence. Moreover, selection of specific patient sam-
review which outlines the prevalence and predic- ples probably influenced the outcome. For
tors of recurrence after remission of an episode of instance, in the CDS study patients were all
MDD. Several conclusions can be drawn from the referrals to a tertiary care centre (10, 25), while in
selected studies. i) The percentage of recurrence in other studies only patients with already recurrent
the specialised mental health care is very high (up MDD were included (8, 11). It is likely that this
to 85% after 15 years) and in this population it is would lead to a higher risk of recurrence. Further-
better to ask when instead of whether the patient more, this review only included naturalistic studies,
will have a recurrence. ii) The included study thus treatment is a factor that was not controlled
suggest that the percentage of recurrence in the for. Patients receiving more vigorous or more
primary care population is similar. iii) The per- extended treatment would be less likely to recur. As
centage of recurrence in the general population is more vigorous or extended treatment is more likely
lower (up to 35% after 15 years). iv) Two main in tertiary care centres, this would offset the
predictors of recurrence can be identified: the previous effect of selection. Finally, the number
number of previous episodes and subclinical resid- of well carried out studies in primary care and in
ual symptoms after recovery for the last episode. the general population is limited. Therefore, no
Demographic factors such as gender, SES, and clear conclusions can be drawn on the prevalence
civil status, which are risk factors for the onset of and predictors of recurrence in these settings; the
the first depressive episode (33, 34), do not appear conclusion that prevalence of recurrence of MDD
to predict recurrence after recovery. The age of the in the general population is 35% is only based on
index episode also does not seem to predict one study (12). Most studies in primary care and in
recurrence. The evidence is (still) inconclusive for the general population did not use clear defined
the following predictors: neuroticism, lack of social criteria for remission and recurrence and did not
support, severe life event, the presence of a use a diagnostic instrument in commencement of
comorbid axis I and II disorder, the severity and the study. The results suggest that the percentage
duration of the previous episode, a younger age of of recurrence of MDD in primary care is compa-
onset, a family history of MDD and psychosocial rable with the specialised mental health care.
impairment after a depressive episode. The severity However, the study of Gopinath et al. (11) only
of the previous episode and psychosocial impair- included patients that experienced at least three
ment after a depressive episode seem most likely depressive episodes, so this is not a representative
associated, whereas a family history of MDD, lack sample. In other (not included) studies, in primary
of social support and duration of the previous care a lower percentage of recurrence was found.
episode seem less likely predictors of recurrence. For example, a highly cited study is a Dutch study

189
 16000447, 2010, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1600-0447.2009.01519.x by Cbua - Consorcio De Bibliotecas, Wiley Online Library on [20/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Hardeveld et al.

among 222 patients with a follow-up of 10 years factors could predict different outcome variables
(37). In this study, a percentage of recurrence of (21). If we know which predictors play a role in
40% was found after 10 years. This study was not recurrences and which interventions could help to
included because the diagnosis was not made with prevent recurrences there is enough knowledge
a standardised instrument and the course was available to develop well-founded guidelines for
measured retrospectively with a registration treatment and recurrence prevention for the
system. MDD.
In conclusion, it becomes increasingly clear that
MDD (especially in the specialised mental health
Declaration of interest
care) is often a chronic and ⁄ or recurrent disorder
with consequences over the entire lifespan. Demo- Dr Nolen has received speaking fees from Astra Zeneca, Eli
graphic determinants play a role in the onset of the Lilly, Pfizer, Servier, Wyeth; unrestricted grants from the
Netherlands Organisation for Health Research and Develop-
first depressive episode but seem not important ment, European Union, Stanley Medical Research Institute,
during the course. The risk factors for recurrency Astra Zeneca, Eli Lilly, GlaxoSmithKline, Wyeth; and served
and chronicity of MDD seem to have similarities. on advisory boards for Astra Zeneca, Cyberonics, Pfizer,
For both, clinical factors seem most important. Servier. Dr Spijker has received speaking fees from Astra
The knowledge that clinical variables are signifi- Zeneca and Wyeth.
cant risk factors could be taken into account
during the initial phase of the treatment. More- References
over, it is possible that chronicity and recurrency 1. Kessler RC, Berglund P, Demler O et al. The epidemiology
are both expressions of the same dimension: of major depressive disorder: results from the National
everything what predicts chronicity (subclinical Comorbidity Survey Replication (NCS-R). JAMA
residual symptoms, number of previous episodes) 2003;289:3095–3105.
predicts recurrences. With the increasing know- 2. World Health Organisation. The global burden of disease:
2004 update. Geneva: World Health Organisation, 2008.
ledge on genetic and neurobiological determinants 3. Keller MB, Shapiro RW, Lavori PW, Wolfe N. Relapse in
of MDD, it may be possible to identify better major depressive disorder: analysis with the life table. Arch
predictors of its course (38). Future research Gen Psychiatry 1982;39:911–915.
should pay attention to this area. 4. Andrews G. Should depression be managed as a chronic
What are the consequences for clinical practice? disease? BMJ 2001;322:419–421.
5. Pais PS. Early intervention and prevention of myocardial
Knowledge of predictors of recurrence is essential infarction. J Hypertens Suppl 2006;24:S25–S30.
for making treatment decisions. Most patients 6. Nathan DM. Some answers, more controversy, from
with MDD in specialised mental health care will UKPDS. United Kingdom Prospective Diabetes Study.
have a recurrence whereas clinical variables prob- Lancet 1998;352:832–833.
ably determine the time to recurrence. It is 7. Frank E, Prien RF, Jarrett RB et al. Conceptualization and
rationale for consensus definitions of terms in major
important to closely monitor and treat patients depressive disorder. Remission, recovery, relapse, and
with subclinical residual symptoms because these recurrence. Arch Gen Psychiatry 1991;48:851–855.
patients have a high risk of recurrence. Patients 8. Maj M, Veltro F, Pirozzi R, Lobrace S, Magliano L. Pattern
with multiple recurrences should receive mainte- of recurrence of illness after recovery from an episode of
nance treatment with antidepressants, with special major depression: a prospective study. Am J Psychiatry
1992;149:795–800.
attention to treatment adherence. The risk of 9. Kanai T, Takeuchi H, Furukawa TA et al. Time to recur-
relapse or recurrence by maintenance treatment rence after recovery from major depressive episodes and its
with antidepressants can be reduced by 25% when predictors. Psychol Med 2003;33:839–845.
compared with placebo (39, 40). However, we do 10. Mueller TI, Leon AC, Keller MB et al. Recurrence after
not know the efficacy of antidepressants on the recovery from major depressive disorder during 15 years of
observational follow-up. Am J Psychiatry 1999;156:1000–
recurrence rate over a long period. In addition, 1006.
preventive interventions should be considered in 11. Gopinath S, Katon WJ, Russo JE, Ludman EJ. Clinical fac-
patients with multiple recurrences, focusing on tors associated with relapse in primary care patients with
residual symptomatology and specific coping styles chronic or recurrent depression. J Affect Disord
(41). Nonetheless, knowledge about predictors of 2007;101:57–63.
12. Eaton WW, Anthony JC, Gallo J et al. Natural history of
recurrence is still incomplete and prospective Diagnostic Interview Schedule ⁄ DSM-IV major depression.
research, particularly in the general population, The Baltimore Epidemiologic Catchment Area follow-up.
is necessary to give better-founded conclusions Arch Gen Psychiatry 1997;54:993–999.
which predictors play a role in recurrences. It 13. Eaton WW, Shao H, Nestadt G, Lee HB, Bienvenu OJ, Zandi
could also be useful to take along different P. Population-based study of first onset and chronicity in
major depressive disorder. Arch Gen Psychiatry
outcome variables (severity, time between recur- 2008;65:513–520.
rences, number of recurrences) because different

190
 16000447, 2010, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1600-0447.2009.01519.x by Cbua - Consorcio De Bibliotecas, Wiley Online Library on [20/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Recurrence of major depressive disorder

14. Winokur G, Coryell W, Keller M, Endicott J, Akiskal H. a chronic course of illness? Am J Psychiatry
A prospective follow-up of patients with bipolar and pri- 2000;157:1501–1504.
mary unipolar affective disorder. Arch Gen Psychiatry 29. Paykel ES, Ramana R, Cooper Z, Hayhurst H, Kerr J,
1993;50:457–465. Barocka A. Residual symptoms after partial remission: an
15. Holma KM, Holma IA, Melartin TK, Rytsala HJ, Isometsa important outcome in depression. Psychol Med
ET. Long-term outcome of major depressive disorder in 1995;25:1171–1180.
psychiatric patients is variable. J Clin Psychiatry 30. Pintor L, Gastó C, Navarro V, Torres X, Fañanas L.
2008;69:196–205. Relapse of major depression after complete and partial
16. Melartin TK, Rytsala HJ, Leskela US, Lestela-Mielonen remission during a 2-year follow-up. J Affect Disord
PS, Sokero TP, Isometsa ET. Severity and comorbidity 2003;73:237–244.
predict episode duration and recurrence of DSM-IV major 31. Kennedy N, Paykel ES. Residual symptoms at remission
depressive disorder. J Clin Psychiatry 2004;65:810–819. from depression: impact on long-term outcome. J Affect
17. Paykel ES, Cooper Z, Ramana R, Hayhurst H. Life events, Disord 2004;80:135–144.
social support and marital relationships in the outcome of 32. Keller MB, Lavori PW, Endicott J, Coryell W, Klerman
severe depression. Psychol Med 1996;26:121–133. GL. ‘‘Double depression’’: two-year follow-up. Am J Psy-
18. Solomon DA, Leon AC, Endicott J et al. Psychosocial chiatry 1983;140:689–694.
impairment and recurrence of major depression. Compr 33. Anthony JC, Petronis KR. Suspected risk factors for
Psychiatry 2004;45:423–430. depression among adults 18-44 years old. Epidemiology
19. Ilardi SS, Craighead WE, Evans DD. Modeling relapse in 1991;2:123–132.
unipolar depression: the effects of dysfunctional cognitions 34. Graaf Rde, Bijl RV, Ravelli A, Smit F, Vollebergh WAM.
and personality disorders. J Consult Clin Psychol Predictors of first incidence of DSM-III-R psychiatric
1997;65:381–391. disorders in the general population: findings from the
20. Ramana R, Paykel ES, Cooper Z, Hayhurst H, Saxty M, Netherlands Mental Health Survey and Incidence Study.
Surtees PG. Remission and relapse in major depression: a Acta Psychiatr Scand 2002;106:303–313.
two-year prospective follow-up study. Psychol Med 35. Kessing LV, Andersen EW, Andersen PK. Predictors of
1995;25:1161–1170. recurrence in affective disorder-analyses accounting for
21. Conradi HJ, de Jonge P, Ormel J. Prediction of the three- individual heterogeneity. J Affect Disord 2000;57:139–
year course of recurrent depression in primary care 145.
patients: different risk factors for different outcomes. 36. Kessing LV, Hansen MG, Andersen PK, Angst J. The pre-
J Affect Disord 2008;105:267–271. dictive effect of episodes on the risk of recurrence in
22. Judd LL, Akiskal HS, Maser JD et al. Major depressive depressive and bipolar disorders – a life-long perspective.
disorder: a prospective study of residual subthreshold Acta Psychiatr Scand 2004;109:339–344.
depressive symptoms as predictor of rapid relapse. J Affect 37. Van Weel-Baumgarten E, van den Bosch W, van den Hoogen
Disord 1998;50:97–108. H, Zitman FG. Ten year follow-up of depression after
23. Lavori PW, Keller MB, Mueller TI, Scheftner W, Fawcett diagnosis in general practice. Br J Gen Pract 1998;48:1643–
J. Recurrence after recovery in unipolar MDD: an obser- 1646.
vational follow-up study of clinical predictors and somatic 38. Robinson OJ, Sahakian BJ. Recurrence in major depressive
treatment as a mediating factor. Int J Methods Psychiatr disorder: a neurocognitive perspective. Psychol Med
Res 1994;4:211–229. 2008;38:315–318.
24. Pintor L, Torres X, Navarro V, Matrai S, Gastó C. Is the 39. Geddes JR, Carney SM, Davies C et al. Relapse prevention
type of remission after a major depressive episode an with antidepressant drug treatment in depressive disorders:
important risk factor to relapses in a 4-year follow up? a systematic review. Lancet 2003;361:653–661.
J Affect Disord 2004;82:291–296. 40. Kaymaz N, van Os J, Loonen AJ, Nolen WA. Evidence that
25. Solomon DA, Keller MB, Leon AC et al. Multiple recur- patients with single versus recurrent depressive episodes
rences of major depressive disorder. Am J Psychiatry are differentially sensitive to treatment discontinuation: a
2000;157:229–233. meta-analysis of placebo-controlled randomized trials.
26. Kennedy N, Abbott R, Paykel ES. Remission and recurrence J Clin Psychiatry 2008;69:1423–1436.
of depression in the maintenance era: long-term outcome 41. Bockting CL, Spinhoven P, Koeter MW, Wouters LF, Schene
in a Cambridge cohort. Psychol Med 2003;33:827–838. AH. Prediction of recurrence in recurrent depression and
27. Naz B, Craig TJ, Bromet EJ, Finch SJ, Fochtmann LJ, the influence of consecutive episodes on vulnerability for
Carlson GA. Remission and relapse after the first hospital depression: a 2-year prospective study. J Clin Psychiatry
admission in psychotic depression: a 4-year naturalistic 2006;67:747–755.
follow-up. Psychol Med 2007;37:1173–1181. 42. Simpson HB, Nee JC, Endicott J. First-episode major
28. Judd LL, Paulus MJ, Schettler PJ et al. Does incomplete depression. Few sex differences in course. Arch Gen Psy-
recovery from first lifetime major depressive episode herald chiatry 1997;54:633–639.

191