Respiratory Medicine 117 (2016) 154e161

Contents lists available at ScienceDirect

Respiratory Medicine
journal homepage: [Link]/locate/rmed

Review article

Prevalence of depression in COPD: A systematic review and meta-

analysis of controlled studies
Darlan L. Matte a, b, Marcia M.M. Pizzichini a, c, *, Andrea T.C. Hoepers a, Alexandre P. Diaz a,
Manuela Karloh a, Mirella Dias a, Emilio Pizzichini a, c
a
Postgraduate Program in Medical Sciences, Department of Internal Medicine, Health Science Center, Federal University of Santa Catarina (UFSC),
polis, Brazil
Floriano
b
Department of Physical Therapy, Health Science and Sports Center, State of Santa Catarina University (UDESC), Floriano polis, Brazil
c polis, Brazil
Department of Internal Medicine, Health Science Center, Federal University of Santa Catarina (UFSC), Floriano

a r t i c l e i n f o a b s t r a c t

Article history: Background: Depression is frequently reported in association with COPD. However, the prevalence of
Received 20 October 2015 depression in these patients ranges largely. This study aimed to systematically review the prevalence of
Received in revised form depression in COPD and controls and to explore remaining causes of inter-study variability in the re-
3 May 2016
ported prevalence.
Accepted 6 June 2016
Available online 7 June 2016
Methods: A systemic review of the literature and a meta-analysis was performed to evaluate the source
of variability in the reported rates of depression in stable COPD. Main eligibility criteria were: controlled
studies with a sample size >100, outpatients with COPD diagnosed by spirometry and, use of a validated
Keywords:
COPD
depression screening instrument.
Depression Results: From 1613 studies identified, eight controlled studies were included in the review. The number
Review of participants in the pooled studies was of 5.552 COPD subjects and 5.211 controls. Using stricter criteria
Systematic review for study selection reduced the variability of the depression prevalence in COPD and controls, which was
Meta-analysis 27.1% [25.9e28.3] in COPD subjects and 10.0% [9.2e10.8] in the control group. The pooled odds ratio and
95% CI was 3.74 [2.4e5.9]. However, the heterogeneity across studies was high. Possible explanatory
factor included sample sizes, COPD/controls ratio, smoker’s/nonsmokers ratio and qualitative differences
(source of subjects, instruments to screen depression, COPD severity, smoking status, and comorbidities).
Conclusion: The study highlights the variability in estimates of depression prevalence in COPD. It could
be explained by methodological differences across the included studies. This suggests that a standard-
ization is critical to improve precision of the estimates.
© 2016 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
2.1. Search strategy and selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
2.2. Study selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
2.3. Data extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
2.4. Quality assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
2.5. Meta-analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
3.1. Subject country, source, sampling, and matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
3.2. Instrument used to screen depression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
3.3. Clinical diversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158

* Corresponding author. Division of Internal Medicine, Universidade Federal de Santa CatarinaeUFSC, NUPAIVA, Hospital Universit rio, Trindade,
ario Campus Universita
88040-970 Florianopolis, SC, Brazil.
E-mail address: mpizzich@[Link] (M.M.M. Pizzichini).

[Link]
0954-6111/© 2016 Elsevier Ltd. All rights reserved.
 D.L. Matte et al. / Respiratory Medicine 117 (2016) 154e161 155

4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
5. Take home message . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Conflict of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Author contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161

1. Introduction share common features such as fatigue, low physical activity, sleep
disturbance, and loss of interest in life.
Chronic obstructive pulmonary disease (COPD) and depression Therefore, besides the inherent heterogeneity of COPD and
are common heterogeneous conditions that are frequently associ- depression, it could be argued that the lack of methodological
ated [1]. Both diseases generate an elevated global economic and standardization across studies is a major contributor to the vari-
social burden [2]. While the association between COPD and ability of the reported prevalence of depression in COPD patients.
depression should be anticipated because of social isolation of more We thought that an objective and critical appraisal of well-designed
severe patients, the prevalence of depression in COPD appears to be studies examining the prevalence of depression in COPD patients is
more frequent than in other chronic disabling diseases [3,4]. timely. Thus, this study aimed to perform a systematic review and
However, the relationship between depression and COPD is not meta-analysis of the literature on depression prevalence in COPD
linear [5]. using stricter inclusion criteria for article selection than those
Smoking, the leading risk factor for COPD, is often associated described in previous metanalysis [15,16]. We also sought to
with depression and both frequently potentiate each other [6]. explore remaining causes of inter-study variability in the reported
Therefore, smoking, COPD and depression are deeply interrelated, prevalence.
with depression playing a role in the initiation and maintenance of
smoking, smoking frequently leading to the development of COPD 2. Methods
and COPD, in turn, contributing to the establishing or worsening of
depression [5,7,8]. In addition, the association between depression 2.1. Search strategy and selection
and COPD appears to be bidirectional, as shown by Atlantis et al. [9]
in a recent meta-analysis of longitudinal studies. This meta-analysis The literature searches of articles published in English was
demonstrated that not only COPD increases the risk of developing conducted in MEDLINE, PsycINFO, Embase, the Cochrane Library
depression (relative risk 1.69; 95% CI 1.45e1.96) but also, depres- and Science Direct from their respective dates of inception up to
sion increases 1.43 (95% CI 1.20e1.71) the risk of COPD adverse January 1rst, 2015. The search terms used included ‘COPD’, ‘Chronic
outcomes and mortality. Obstructive Pulmonary Disease”, and “depression” and derivative
However, it is well known that the reported prevalence of terms. It was also conducted hand searches in the included studies
depression in COPD varies largely, from 10 to 57% [10], with the and other sources.
highest rates being reported in patients with more severe COPD
[7,11], particularly in long-term oxygen users [12] or in patients
recovering from COPD exacerbation [11]. This may in part be due to 2.2. Study selection
the heterogeneity and complexity of both diseases [13,14]. Also, the
expression of depression in COPD patients can be very heteroge- This systematic review is in accordance with the Preferred
neous depending on several factors, including genetic predisposi- Reporting Items for Systematic Reviews and Meta-Analyses [19].
tion, exposure to losses and stressors, as well as direct damage to Eligibility and exclusion criteria for articles selection are described
the brain mediated by the pathophysiological effects of the chronic in Table 1 [12,20e24]. In addition, studies were excluded if they
respiratory illness [5]. were not the primary data source and/or if they were duplicate of
For almost two decades, authors have emphasized that this high an article that had been previously identified. Two raters screened
variability between estimates of depression in COPD is also asso- the titles and abstracts independently to find potentially eligible
ciated with methodological dissimilarities, including sampling, studies. If necessary, a consensus meeting was done in case of
screening instruments, and diagnosis of depression and/or COPD disagreement.
[15,16]. Furthermore, the heterogeneity of both diseases across
studies is an important factor, often overlooked [7,17]. In a previous 2.3. Data extraction
meta-analysis and meta-regression Zhang et al. [16] aimed to
identify factors associated with the heterogeneity of the prevalence Data extraction was performed by two independent raters using
of depressive symptoms in COPD. These authors noted a six-fold a standardized form. The following information were extracted:
variation in point prevalence of depression in COPD subjects and authors, year of publication, design, sample size, sampling, settings,
an 18-fold variation in controls. They concluded that demographic country, age and gender of subjects, COPD diagnosis, COPD severity
and clinical factors (age, gender and smoking history) were not based on GOLD guidelines [25], smoking status, comorbidities, in-
determinants of the heterogeneity across studies. However, the struments and diagnostic criteria for depression, depression prev-
study had a number of limitations including differences in COPD alence, stratification and, adjustment in the analysis. Quality
definition among the selected studies. More recently, Hegerl and control checks comparing the summarized data with the original
Mergl [18] explored other aspects that may affect the prevalence of articles were performed by an independent quality-control
depression in COPD patients including the fact that both diseases specialist.
156 D.L. Matte et al. / Respiratory Medicine 117 (2016) 154e161

Table 1
Study selection: inclusion and exclusion criteria.

Inclusion criteria
Controlled studies with a sample size >100 subjects.
Subjects selected from outpatient clinics from primary, secondary or tertiary care, from community or random population-based samples.
COPD diagnosis confirmed by spirometry: post-bronchodilator FEV1/FVC
to 0.7 [25].
Depression determined by a structured interview to establish a diagnosis based on standardized and widely used criteria (e.g., SCID- DSM) [22] or by a validated
depression questionnaire (e.g., CES-D, GDS, HADS-D,SDS (Zung), HRSD) [23,24].
Exclusion criteria
Studies focused on subpopulations, such as patients on continuous oxygen therapy [12].
Studies that included COPD patients recovering from an exacerbation and/or hospital admission at the time of evaluation [20].
Studies comprising convenience samples.
Studies focusing on pre-existing conditions that potentially could increase the risk for depression (such as living alone, having cancer) [21].

SCID-DSM ¼ Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders; CES-D ¼ Center for Epidemiological Studies e Depression Inventory; SDS
(Zung) ¼ Zung Self-Rating Depression Scale; GDS ¼ Geriatric Depression Scale; HADS-D (Hospital Anxiety and Depression Scale: depression scores); HRS-D ¼ Hamilton Rating
Scale for Depression.

2.4. Quality assessment prevalence studies of dementia and Parkinson’s disease, adapta-
tions were necessary to rate the diagnosis of COPD and its severity
The methodological quality of the studies was assessed by two and the diagnosis of depression. We also rated matching of COPD
independent reviewers using the criteria adapted from Aarsland and control groups. The quality of studies was appraised as
et al. [26] and, when required, a third reviewer decided the described previously described [26] on a four-point scale (0, none;
disagreement. Because these quality criteria [26] were designed for 1, poor; 2, fair; and 3, good). The following four characteristics were

Fig. 1. PRISMA Flowchart of the studies selection process.

 D.L. Matte et al. / Respiratory Medicine 117 (2016) 154e161 157

rated: subject source, diagnostic criteria for COPD, diagnostic variables were: (1) lack of matching between COPD and controls,
criteria for depression, and matching. This quality score ranges (2) control groups including exclusively current and former
from zero to [Link] subject source (rated as previously described smokers, (3) differences in instruments to screen depression, (4)
[26]): population based sample (3); multiple sources of patients quality score. The analysis was performed using the Review Man-
identified from general practitioners [GPs]; tertiary care clinics, ager Software 5.2 Copenhagen (The Nordic Cochrane Center, The
nursing homes, private specialists, hospital records, drug pre- Cochrane Collaboration, 2012). All tests were two-tailed and a
scriptions, multi-centric studies, etc.; (2); single source of patients statistical significance was accepted for p values < 0.05.
(a GP registration, hospital files, or convenience sample of out-
patients) (1); no information (0). Diagnostic criteria for COPD: 3. Results
GOLD criteria for COPD diagnosis and functional severity classifi-
cation [25] (3); GOLD criteria for COPD diagnosis [25] without The search strategy resulted in 1613 citations of potentially
functional severity classification (2); limited description of clinical relevant studies. Of these, 19 eligible studies were submitted to
exclusion and inclusion criteria (1); and no or very limited infor- quality assessment criteria. Eleven studies were excluded after
mation (0). Diagnostic criteria for depression: use of a structured quality assessment criteria and eight studies [29e36] were
interview by a psychiatrist to establish a diagnosis based on stan- included in this systematic review (Fig. 1 and Table 2). The quality
dardized and widely used criteria (e.g. Structured Clinical Interview assessment rates of the included studies ranged from 5 to 10
(SCID) for Diagnostic and Statistical Manual of Mental Disorders (5 ¼ one study [29], 7 ¼ three studies [30,31,34], 9 ¼ one study [33]
(DSM) [22] (3); standardized and widely used criteria (DSM) and, 10 ¼ three studies [32,35,36]. The pooled sample size was
without a (semi) structured clinical interview [22] (2); cut-off score 5.502 COPD subjects and 5.211 controls. The prevalence of
on a validated depression rating scale [22] (1); no or inadequate depression ranged from 15.2 [32] to 35.7% [31] in COPD and from
criteria (e.g., subjective clinical impression, self-reported diag- 3.6 [32] to 17.5% [21] in the controls (Table 1). The mean prevalence
nosis). Matching of controls: random sample or matching by three of depression and 95% CI was 27.1% [25.9e28.3] in COPD and 10.0%
or more criteria (e.g., sex, age, race, schooling, social status, [9.2e10.8] in the control group. The pooled odds ratio and 95% CI
comorbidities) (3); matching by two criteria (2); matching by one for depression in COPD of 3.74 [2.36e5.93] (Fig. 2). The heteroge-
criteria (1); no matching (0). Studies that scored <3 in subject neity among studies was high (Fig. 2). However, sensitivity analyses
source and/or, <3.0 in diagnostic criteria for COPD and/or scored <1 showed that the exclusion of studies with a distinctive character-
in diagnostic criteria for depression were excluded. istic from the remaining (similar proportions of men and woman
[34], OR with 90% of COPD stages I and II [31], OR with 100% COPD
2.5. Meta-analysis stages II-IV [33], OR using the SCID-DSM-IV [36]) the heterogeneity
fell from 89%, p < 0.0001e25%, p ¼ 0.26. (Table E1).
Because of potential challenges in performing a meta-analysis of
observational studies related to biases and diversity in the original 3.1. Subject country, source, sampling, and matching
studies, the technique of accepted recommendations [27] was fol-
lowed in this meta-analysis. The odds ratio (OR) and 95% CI of There was a large variability among studies regarding subjects’
depression in COPD patients compared with controls was calcu- country, setting, sampling and matching (Table 2). Subjects were
lated using a random-effects model. Heterogeneity was examined selected from cohort studies (n ¼ 3) [31e33], from primary (n ¼ 2)
using the chi-square test (with p < 0.1 considered to indicate sig- [29,35] and tertiary cares (n ¼ 2) [30,36], and from general popu-
nificant heterogeneity), the I2 parameter [28]. A sensitivity analysis lation (n ¼ 1) [34]. The ratio between COPD and controls ranged
was performed to explain heterogeneity. Potential explanatory from 0.08 [31] to 3.9 [32]. In four studies [29,31,35,36] the ratio was

Table 2
Studies characteristics.

Reference Subjects source and design Subjects (n) Mean Gender COPD COPD severity Psychological Depression
age, males, % severity I-II, III-IV, n (%) measure, cutoff prevalence
years n (%) point

Van Manen et al., 2002 Primary Care. Random sample. COPD ¼ 162 65.6 73.3 102 (63.0) 60 (37.0) CES-D  16 21.6%
[29], Netherlands Cross-sectional study Control ¼ 359 66.8 40.4 17.5%
Di Marco et al., 2006 Tertiary Care. Matching by age and sex. COPD ¼ 202 68.0 76.7 114 (56.4) 88 (43.6) SDS (Zung)  50 18.8%
[30] Italy Cross-sectional study Control ¼ 114 65.1 70.1 12.9%
Ng et al., 2009 [31] Population based cohort. Cross-sectional COPD ¼ 189 NA 35.4 171 (90.5) 18 (9.5) GDS  5 22.8%
Singapore study Control ¼ 2.213 NA 36.8 12.4%
Omachi et al., 2009 Cohort of COPD and controls matched by age, COPD ¼ 1.202 58.2 43.0 832 (69.2) 370 (30.8) GDS  6 27.0%
[32] United States sex, and race. Cross-sectional study Control ¼ 302 58.5 39.0 6.0%
Hanania et al., 2011 Cohort of COPD and controls matched by age COPD ¼ 2.118 54.6 65.0 Stages II-IV, 2118 (100) CES-D16 26.0%
[33] Multinationala and sex. Cross-sectional study Control ¼ 578 63.0 51.9 9.1%
Persson et al., 2012 General population. Cross-sectional study COPD ¼ 433 58.6 60.0 199 (46.0) 234 (54.0) CES-D  16 21.0%
[34] Norway Control ¼ 325 63.5 54.0 6.0%
Lou et al., 2012 [35] Primary Care. Cross-sectional study COPD ¼ 1.100 63.2 75.2 623 (56.6) 477 (43.4) HADS-D  8 35.7%
China Control ¼ 1.100 62.1 75.2 7.2%
Wong et al., 2014 [36] Tertiary Care. Matching by age and sex. COPD ¼ 146 73.4 75.5 31 (21.2) 114 (78.1) SCID-DSM-IV 15.2%
China Cross-sectional study Control ¼ 220 74.8 83.6 3.6%
OVERAL NA COPD ¼ 5.552 NA NA NA NA NA 27.1% [25.9
Control ¼ 5.211 e28.3]b
10.0% [9.2
e10.8]b

COPD ¼ Chronic Obstructive Pulmonary Disease; GOLD ¼ Global Initiative for Chronic Obstructive Lung Disease) NA ¼ not applicable; other abbreviations as in Table 1.
a
Bulgaria, Canada, Czech Republic, Denmark, Netherlands, New Zealand, Norway, Slovenia, Spain, UK, Ukraine and United States.
b
Mean and 95% CI.
158 D.L. Matte et al. / Respiratory Medicine 117 (2016) 154e161

Fig. 2. Forest plot of Odds Ratio and 95% CI for prevalence of depression in COPD subjects and controls of all included studies.

1.0 and in another four [30,32e34] was >1.0. In five studies of depression in COPD subjects according to disease severity was
[30,32,33,35,36] controls were matched by age and gender with described in all included studies [29e36]. After adjustment for
COPD subjects. confounders disease severity was associated with higher preva-
lence of depression in five studies [29,32,33,35,36]. The analysis
3.2. Instrument used to screen depression was adjusted for gender [29e33], smoking status [29,31e34] and
comorbidities in five studies [29,31e34]. Only one study investi-
All but one study [36] used self-completed depression ques- gated past history of depression [33]. The diversity among studies is
tionnaires. However, the screening questionnaires differed in their illustrated in Fig. 4.
construct, contents and rating scales (Table 3). Three studies
[29,33,34] used the Center for Epidemiological Studies and 4. Discussion
Depression Inventory (CES-D), two [31,32] used the Geriatric
Depression Score (GDS); one [30] used the Zung Self-Rating To our knowledge, this is the first study to systematically
Depression Scale [SDS (Zung)] and, another [35] used the Hospi- explore clinical and methodological diversities that may explain the
tal Anxiety and Depression Scale (HADS-D) (Table 1). In only one observed variability across studies in the estimates of depression
study [36] the Structured Clinical Interview for DSM-I (SCID-DSM- prevalence in subjects with COPD. This systematic review showed
IV) was used by a psychiatrist to determine the presence of that using stricter criteria for study selection resulted in reduction
depression. of the variability of depression prevalence of depression in COPD
subjects and controls. In fact, there was a 2.1-fold variation in point
3.3. Clinical diversity prevalence of depression in subjects with COPD and a 4.9-fold
variation in controls, as opposed to a previous meta-analysis [16]
Apart from methodological variability, we identified diversity in that reported a six-fold variation in the prevalence of depression
COPD severity, smoking history, and presence and/or type of in COPD subjects and an 18-fold variation in controls. The direction
comorbidities (Fig. 3). Subgroup analysis comparing the prevalence of the odds ratio was consistent across all studies and the relatively

Table 3
Summary of main characteristics of depression instruments used in the studies selected.

Questionnaires Purpose Rating Recall Number Responses Content

used in the period of items
different studies

CES-D Designed to cover most of Self-rated Past 20 4 points scale (0 ¼ rarely or Four factor structure representing depressed affect,
depression symptoms with week none of the time or <1 day anhedonia, somatic activity or inactivity, and
emphasis on affective through 4 ¼ most or all of the interpersonal challenges
components time or 5e7 days)
GDS (short Designed for screening Self-rated Past 15 YES/NO Affective (e.g., sadness, apathy, crying) and
version) depression in older week cognitive domains)
populations
HADS - Designed for screening Self-rated Past 14 4 points Likert scale (0e3) 7 depression items measuring cognitive and
Depression depression in general week emotional aspects of depression
medical population
SDS- (Zung) Developed to quantify Self-rated Past 20 4 points scale (1 ¼ a little of the Covers psychological, affective, cognitive,
depression in patients with several time, through 4 ¼ most of the behavioral, and somatic aspects of depression
a previous diagnosis of days time)
depression
Structured Semi-structured interview Administered by a Past 15 Clinical judgment Covers, among others, depressive mood,
Clinical for making the major DSM- clinician or trained month anhedonia, change in body weight or appetite,
Interview for IV Axis I diagnoses. mental health hypersomnia or insomnia, fatigue or loss of energy,
DSM-I (SCID- professional feelings of guilt or worthlessness, loss of
DSM-IV) concentration.

Based on references [22e24,37,38].

 D.L. Matte et al. / Respiratory Medicine 117 (2016) 154e161 159

Fig. 3. Forest plot of Odds Ratio and 95% CI for prevalence of depression in COPD subjects and controls after exclusion of three studies, showing a reduction in heterogeneity across
studies (see discussion).

narrow confidence interval suggests precision of the pooled odds ratios between COPD subjects and controls) and qualitative dis-
ratio [3.74 (2.36e5.93)] between COPD and controls. However, the parities (source of subjects, COPD severity, instruments used do
elevated heterogeneity among studies indicated great variability screen depression, smoking status, and comorbidities). The most
and therefore, and the size of the odds ratio should be regarded relevant sources of clinical heterogeneity were the variability in
with caution as it could double when the heterogeneity is reduced. disease severity, the association with other comorbidities (number
The prevalence of COPD according to disease severity was also and type) and, discrepancies of smoking history between COPD
examined. Two thirds of the selected studies reported that, after subjects and controls. The number, but not the type or severity, of
adjustment for confounders, the worse the severity of COPD the comorbidities was reported in only half of the included studies
higher the prevalence of depression [29,32,33,35,36]. The increase [29,31e33]. The exclusion of four studies [29,31,33,36] from the
in the prevalence of depression according to COPD severity is not meta-analysis reduced the heterogeneity measured by the I2 sta-
surprising because the associated physical limitation and social tistic from 89% to 25%. These exclusions happened for different
isolation. However, considering the high heterogeneity among reasons. One study [31] contrasted from the remaining by having a
these studies and the small number of selected studies, subgroup predominance of woman in both COPD and control groups, the
analysis was not performed. highest proportion (90.5%) of mild to moderate COPD and, the
In the present review, the heterogeneity among studies was lowest proportion of smokers and former smokers in both COPD
likely to be due to both quantitative (different sample sizes and and control groups. The second excluded study [29] differed by

Fig. 4. Qualitative and quantitative diversity across studies. Lines represent each study. Columns represent diversity among study. Similar colors indicate similarity among studies
for that respective characteristic.*Ratio between number of COPD and controls, **Average number of family members and marital status, xSelf-reported depression questionnaire.
160 D.L. Matte et al. / Respiratory Medicine 117 (2016) 154e161

having the greatest gender imbalance between COPD and control number and severity), knowledge of the disease, stressors factors,
groups. The third excluded study [33] may have added heteroge- cognitive impairment]. In fact, in the present review only half of the
neity because the absence of mild COPD (100% of disease severity studies [30,32,33,36] matched COPD subjects and controls by age
ranged between GOLD stages II and IV) and, by having discrep- and gender but none by smoking status or comorbidities, although
ancies in smoking status between COPD (100% with a smoking differences in comorbidities were adjusted in the analysis of some
history of at least 10 pack-years) and controls (42% of non- of the studies [29,31e33]. Only one study investigated past history
smokers). Finally, the exclusion of the fourth study [36] was done of depression [33] and one considered cognitive impairment be-
because this was the only one that used a structured interview tween COPD and controls [31]. Together, these diversities encom-
administered by a psychiatrist. pass an alternative explanation for the variability in the estimates
Another possible source of heterogeneity across the studies was of depression prevalence in COPD. Another strength of this sys-
the use of different instruments to assess depression. Apart from the tematic review is that after identifying and minimizing the het-
one study [32] that used a psychiatrist structured clinical interview, erogeneity of the published studies it was presented a pooled
all the other studies used a variety of self-reported questionnaires, estimation of depression prevalence that could be considered a
including the CES-D [29,33,34], GDS [31,32], SDS-Zung [30] and, more robust data in relation to the available literature.
HADS [35]. Despite of the popularity of self-reported questionnaires Possible limitations of the present review are related to the
as screening tools for depression (or significant symptoms of studies selection. For instance, the decision to include only studies
depression) in large studies, it is important to stress that there is with a sample size greater than a hundred subjects may have led to
substantial variability among these questionnaires. Moreover, a selection bias. By using this selection criteria, small studies per-
although validated, these questionnaires differ in their purpose, formed in nursing homes or home centers were not included.
symptom areas that are covered and in the number of questions in Nonetheless, because all small studies also failed in one or more of
each domain [22e24,37,38]. In addition, many researchers have selection criteria (quality of assessment, diagnostic criteria for
questioned the validity and psychometric properties of the items of COPD and/or depression criteria) it could be argued that, possibly,
various self-rated depression scales and their performance in the restriction of the sample size has not lead to a bias towards
different populations. For example, there are concerns regarding the selection.
specificity of CES-D scale particularly in elder patients [24]; the
inability of the HADS to consistently distinguish between the con- 5. Take home message
structs of anxiety and depression [37]; the performance of GDS scale
in different severity degrees of cognitively impaired patients [38]; Identifying the causes of variability across studies may increase
and the appropriateness of the SDS-Zung as a screening research the accuracy of the estimates of depression in COPD.
tool [23]. Finally, there is a lack of studies validating the use of these
questionnaires in subjects with COPD. These observations highlight 6. Conclusion
the importance of the proper assessment of the screening tools
when measuring depression in COPD patients. The current systematic review of the literature, based on
Apart from methodological differences, there are several well- controlled well conducted studies, provides evidence suggesting
established factors inherent to COPD and depression that are that the use of stricter criteria reduces the variability of the esti-
frequently overlooked and may be a source of variability across mates of depression prevalence in COPD subjects and controls.
studies developed to estimate the prevalence of depression in However, the observed lack in matching clinical variables that
COPD. First, both COPD and depression are common heterogeneous might increase the prevalence of depression and the absence of
diseases that may be associated, and it is not easy to distinguish uniformity in choosing the instruments used to screen depression
whether depression preceded or not COPD [5]. In addition, both in COPD subjects indicate that standardization is critical to improve
diseases are related with smoking [5]. Second, COPD is a disease of the accuracy of the estimates and to allow comparisons among
elderly patients. Ageing is frequently associated with other mor- studies.
bidities that, in turn, are also associated with depression [39]. Third,
as the severity of COPD increases, the prevalence of depression is Conflict of interest
also expected to increase [7]. However, the increase in COPD
severity is likely to be accompanied by symptoms such as fatigue, The authors of this report are responsible for its content. All
decreased physical activity that may be misinterpreted when using authors declare that they have no competing interests. This sys-
self-reported questionnaires [18]. Therefore, the precision of the tematic review involved only the extraction of data from other
estimates of depression prevalence in COPD is greatly related to the published studies, thus, there is no need of patients’ consent and
recognition of these factors when selecting subjects and controls ethics committee approval was not necessary. The study did not
and considering them in the analysis. require funding.
One of the strengths of this systematic review was the use of
several criteria to select the studies. For example, we have selected Author contribution
only controlled studies including patients with COPD diagnosed by
spirometry aiming to avoid dissimilarities in COPD diagnosis. Also, DLM, ATCH, and ADP conceived the study design. DLM, ATCH,
the exclusion of manuscripts focusing on patients with exacerbated and MK searched for and selected the trials and extracted, analyzed
COPD or receiving long-term oxygen therapy, is likely to have and interpreted the data. DLM, ATCH, ADP, MD and MK drafted the
minimized spectrum bias. We also tried to decrease the influence of manuscript. MMMP and EP helped with the study design, analysis,
the environment by excluding hospital admitted or institutional- writing and critically reviewed the manuscript. All authors read
ized patients. Although these criteria resulted in decreased range of and approved the final version of the manuscript.
the estimates of depression prevalence in COPD subjects and in
controls, they did not prevent the heterogeneity that possibly Acknowledgements
resulted from the variability in COPD severity, or differences in the
report of comorbidities or how the authors dealt with differences This work was supported by NUPAIVA Research Center, Federal
between cases and controls [(smoking history, comorbidities (type, University of Santa Catarina and State of Santa Catarina University
 D.L. Matte et al. / Respiratory Medicine 117 (2016) 154e161 161

(UDESC). The views expressed in this article are those of the au- chronic obstructive pulmonary disease, Eur. J. Epidemiol. 19 (9) (2004)
851e859.
thors and do not necessarily reflect the position or policy of the
[21] R.E. Schane, L.C. Walter, A. Dinno, et al., Prevalence and risk factors for
Universities. We thank the translator, Eliana Beatriz Sartoretto depressive symptoms in persons with chronic obstructive pulmonary disease,
Matte, for her help in proofreading this manuscript. J. general Intern. Med. 23 (11) (2008) 1757e1762.
[22] M.B. First, M. Gibbon, The structured clinical interview for DSM- IV Axis I
disorders (SCID-I) and the structured clinical interview for DSM-IV Axis II
Appendix A. Supplementary data disorders (SCID-II), in: M.J. Hilsenroth, D.L. Segal, M. Hersen (Eds.), Compre-
hensive Handbook of Psychological Assessment Hoboken, John Wiley and
Supplementary data related to this article can be found at http:// Sons, Inc;, New Jersey, 2003.
[23] I. McDowell, Measuring Health: a Guide to Rating Scales and Questionnaires,
[Link]/10.1016/[Link].2016.06.006. New York Oxford University Press, Inc, 2006.
[24] R.L. Schein, H.G. Koenig, The Center for Epidemiological Studies-Depression
References (CES-D) Scale: assessment of depression in the medically ill elderly, Int. J.
geriatric psychiatry 12 (4) (1997) 436e446.
[25] J. Vestbo, S.S. Hurd, A.G. Agusti, et al., Global strategy for the diagnosis,
[1] J. Maurer, V. Rebbapragada, S. Borson, et al., Anxiety and depression in COPD:
management, and prevention of chronic obstructive pulmonary disease:
current understanding, unanswered questions, and research needs, Chest 134
GOLD executive summary, Am. J. Respir. Crit. Care Med. 187 (4) (2013)
(4 Suppl) (2008) 43se56s.
347e365.
[2] C.J. Murray, A.D. Lopez, Measuring the global burden of disease, N. Engl. J.
[26] D. Aarsland, J. Zaccai, C. Brayne, A systematic review of prevalence studies of
Med. 369 (5) (2013) 448e457.
dementia in Parkinson’s disease. Movement disorders, official J. Mov. Disord.
[3] L. van den Bemt, T. Schermer, H. Bor, et al., The risk for depression comorbidity
Soc. 20 (10) (2005) 1255e1263.
in patients with COPD, Chest 135 (1) (2009) 108e114.
[27] D.F. Stroup, J.A. Berlin, S.C. Morton, et al., Meta-analysis of observational
[4] P.J. Barnes, Chronic obstructive pulmonary disease: effects beyond the lungs,
studies in epidemiology: a proposal for reporting. Meta-analysis of Observa-
PLoS Med. 7 (3) (2010) e1000220.
tional Studies in Epidemiology (MOOSE) group, Jama 283 (15) (2000)
[5] R.J. Norwood, A review of etiologies of depression in COPD, Int. J. Chron.
2008e2012.
Obstruct Pulmon Dis. 2 (4) (2007) 485e491.
[28] J.P. Higgins, S.G. Thompson, Quantifying heterogeneity in a meta-analysis,
[6] R.D. Goodwin, K.L. Lavoie, A.R. Lemeshow, et al., Depression, anxiety, and
Statistics Med. 21 (11) (2002) 1539e1558.
COPD: the unexamined role of nicotine dependence, Nicotine Tob. Res. 14 (2)
[29] J.G. van Manen, P.J. Bindels, F.W. Dekker, et al., Risk of depression in patients
(2012) 176e183.
with chronic obstructive pulmonary disease and its determinants, Thorax 57
[7] C. Schneider, S.S. Jick, U. Bothner, et al., COPD and the risk of depression, Chest
(5) (2002) 412e416.
137 (2) (2010) 341e347.
[30] F. Di Marco, M. Verga, M. Reggente, et al., Anxiety and depression in COPD
[8] R. Norwood, Prevalence and impact of depression in chronic obstructive
patients: the roles of gender and disease severity, Respir. Med. 100 (10)
pulmonary disease patients, Curr. Opin. Pulm. Med. 12 (2) (2006) 113e117.
(2006) 1767e1774.
[9] E. Atlantis, P. Fahey, B. Cochrane, et al., Bidirectional associations between
[31] T.P. Ng, M. Niti, C. Fones, et al., Co-morbid association of depression and COPD:
clinically relevant depression or anxiety and COPD: a systematic review and
a population-based study, Respir. Med. 103 (6) (2009) 895e901.
meta-analysis, Chest 144 (3) (2013) 766e777.
[32] T.A. Omachi, P.P. Katz, E.H. Yelin, et al., Depression and health-related quality
[10] M.I. Pumar, C.R. Gray, J.R. Walsh, et al., Anxiety and depression-Important
of life in chronic obstructive pulmonary disease, Am. J. Med. 122 (8) (2009)
psychological comorbidities of COPD, J. Thorac. Dis. 6 (11) (2014) 1615e1631.
778.
[11] G. Gudmundsson, T. Gislason, C. Janson, et al., Depression, anxiety and health
[33] N.A. Hanania, H. Mullerova, N.W. Locantore, et al., Determinants of depression
status after hospitalisation for COPD: a multicentre study in the Nordic
in the ECLIPSE chronic obstructive pulmonary disease cohort, Am. J. Respir.
countries, Respir. Med. 100 (1) (2006) 87e93.
Crit. Care Med. 183 (5) (2011) 604e611.
[12] Y. Lacasse, L. Rousseau, F. Maltais, Prevalence of depressive symptoms and
[34] L.J. Persson, M. Aanerud, P.S. Hiemstra, et al., Chronic obstructive pulmonary
depression in patients with severe oxygen-dependent chronic obstructive
disease is associated with low levels of vitamin D, PloS one 7 (6) (2012)
pulmonary disease, J. Cardiopulm. Rehabil. 21 (2) (2001) 80e86.
e38934.
[13] A. Agusti, P.M. Calverley, B. Celli, et al., Characterisation of COPD heterogeneity
[35] P. Lou, Y. Zhu, P. Chen, et al., Prevalence and correlations with depression,
in the ECLIPSE cohort, Respir. Res. 11 (2010) 122.
anxiety, and other features in outpatients with chronic obstructive pulmonary
[14] T.A. Widiger, L.A. Clark, Toward DSM-V and the classification of psychopa-
disease in China: a cross-sectional case control study, Bmc Pulm. Med. 12
thology, Psychol. Bull. 126 (6) (2000) 946e963.
(2012) 53.
[15] L. van Ede, C.J. Yzermans, H.J. Brouwer, Prevalence of depression in patients
[36] T.S. Wong, Y.T. Xiang, J. Tsoh, et al., Depressive disorders in older patients with
with chronic obstructive pulmonary disease: a systematic review, Thorax 54
chronic obstructive pulmonary disease (COPD) in Hong Kong: a controlled
(8) (1999) 688e692.
study, Aging & Ment. health 18 (5) (2014) 588e592.
[16] M.W. Zhang, R.C. Ho, M.W. Cheung, et al., Prevalence of depressive symptoms
[37] K.L. Smarr, A.L. Keefer, Measures of depression and depressive symptoms:
in patients with chronic obstructive pulmonary disease: a systematic review,
beck depression inventory-II (BDI-II), center for epidemiologic studies
meta-analysis and meta-regression, General Hosp. psychiatry 33 (3) (2011)
depression scale (CES-D), geriatric depression scale (GDS), hospital anxiety
217e223.
and depression scale (HADS), and patient health Questionnaire-9 (PHQ-9),
[17] A. Jain, S. Lolak, Psychiatric aspects of chronic lung disease, Curr. psychiatry
Arthritis care & Res. 63 (Suppl 11) (2011) S454eS466.
Rep. 11 (3) (2009) 219e225.
[38] M. Conradsson, E. Rosendahl, H. Littbrand, et al., Usefulness of the Geriatric
[18] U. Hegerl, R. Mergl, Depression and suicidality in COPD: understandable re-
Depression Scale 15-item version among very old people with and without
action or independent disorders? Eur. Respir. J. 44 (3) (2014) 734e743.
cognitive impairment, Aging & Ment. health 17 (5) (2013) 638e645.
[19] D. Moher, A. Liberati, J. Tetzlaff, et al., Preferred reporting items for systematic
[39] K. Barnett, S.W. Mercer, M. Norbury, et al., Epidemiology of multimorbidity
reviews and meta-analyses: the PRISMA statement, BMJ Clin. Res. ed) 339
and implications for health care, research, and medical education: a cross-
(2009) b2535.
sectional study, Lancet (London, Engl. 380 (9836) (2012) 37e43.
[20] R. Andenaes, M.H. Kalfoss, Psychological distress in hospitalized patients with