Abstract
Keywords
Introduction
Chronic obstructive pulmonary disease (COPD) is an important health problem that affects 80 million people worldwide [World Health Organization, 2010]. COPD is the fourth leading cause of death in the United States and Canada [Statistics Canada, 2003; American Thoracic Society, 1995], and the economic and human burden of COPD remains substantial. For example, in 2006, the direct costs of healthcare resource utilization (e.g. hospitalization, home care) associated with moderate to severe COPD exacerbations were estimated to be CAD$646–736 million per year in Canada [Mittmann et al. 2008]. Exacerbations are common in COPD [American Thoracic Society and European Respiratory Society, 2004] and contribute significantly to healthcare costs (e.g. frequent hospitalizations) [Mittmann et al. 2008], but also to increased morbidity, including reduced health-related quality of life [Miravitlles et al. 2004; Seemungal et al. 1998] and mortality [Soler-Cataluna et al. 2005]. Both prevention and treatment of exacerbations is therefore necessary in order to reduce the overall burden of this disease [O’Donnell et al. 2007]. However, our knowledge about the determinants of exacerbations remains incomplete.
COPD is associated with elevated levels of psychological distress, with 6–80% of patients suffering from anxiety and depressive symptoms, and up to 55% suffering from actual psychiatric disorders [Yohannes et al. 2009]. Given the high prevalence of these psychological factors, some studies have evaluated their association to COPD-related outcomes. According to reports, elevated levels of psychological distress have been associated with increases in the frequency and duration of hospitalizations for exacerbations [Ng et al. 2007; Dahlen and Janson, 2002] among COPD inpatients and outpatients. Despite these associations, relatively few studies have examined the extent to which psychological distress (i.e. anxiety and depression) are independent risk factors for COPD exacerbation. Furthermore, we are aware of only two studies that have evaluated clinical levels anxiety or depression (i.e. anxiety and depressive disorders) [Laurin et al. 2009; Yohannes et al. 2000a]. Therefore, to date, the existing literature on the association between psychological distress and exacerbation risk remains largely misunderstood.
In order to guide future research, the purpose of this report is threefold: (1) to critically review the existing literature on the cross-sectional relationship between psychological distress and COPD exacerbations, and the impact of psychological distress on future exacerbations in patients suffering from COPD; (2) to highlight the limitations of the current literature; and (3) to provide recommendations for future research. To the best of authors’ knowledge, this is the first literature review on this topic.
Methods
Study eligibility and literature search
This review included English-language peer-reviewed research and was not restricted to a specific period of time because of the small number of available studies. The last search was conducted on 10 April 2010. All cross-sectional and retrospective studies on the relationship between psychological distress (defined as anxiety and depressive symptoms and anxiety and depressive disorders) and COPD exacerbations and all prospective studies on the impact of psychological distress on further exacerbations in patients suffering from COPD were included. Study samples must have comprised COPD patients of any age or severity, either exclusively or in combination with other disease populations (e.g. asthma). However, in studies including combined samples, data must have been available on COPD patients separately from other subgroups. This review used pertinent MeSH search terms in the Ovid portal for Medline, EMBASE and PsycINFO. A final sample of 12 studies was included.
Definitions and measures of psychological distress
In the COPD literature, a wide range of screening and diagnostic tools have been used to assess psychological distress. They fall into two categories: those that measured symptom severity (i.e. using self-report questionnaires) and those that evaluated psychiatric disorders (i.e. using structured clinical interviews) [Yohannes et al. 2009]. Experiencing symptoms of psychological distress can be acute and transient (e.g. feeling anxious before a first pulmonary rehabilitation session), but may become chronic and persist (e.g. feeling so fearful of having a panic attack and being unable to leave the house as a result). Clinical levels of psychological distress (i.e. psychiatric disorders) can only be diagnosed using structured interviews that take DSM-IV criteria and levels of functional impairment into account [American Psychiatric Association, 2000]. In this review, studies have to include either a clinical interview or a validated self-reported questionnaire. The term ‘psychological distress’ will refer to either symptoms of anxiety or depression, or a diagnosis of an anxiety or depressive disorder according to DSM-IV criteria; both levels of psychological distress and how they were assessed will be evaluated as part of this review.
Definitions and measures of exacerbations
Several definitions of what constitutes a COPD exacerbation have been proposed, and there is still no consensus definition. For the purposes of this review, careful attention was paid to the way in which exacerbations were defined, evaluated, analyzed, and reported, as per the recommendations of Suissa [Suissa, 2006] and Aaron and colleagues [Aaron et al. 2008].
An exacerbation generally refers to a discrete period of symptom deterioration that usually requires some form of intervention [Burge and Wedzicha, 2003]. An initial consensus of European and American respiratory health specialists, defined an exacerbation as ‘a sustained worsening of the patient’s condition, from the stable and beyond normal day-to-day variations, that is acute in onset and necessitates a change in regular medication in a patient with underlying COPD’ [Rodriguez-Roisin, 2000]. In 2003, this definition was amended to no longer require a change in regular treatment [Burge and Wedzicha, 2003].
In general, exacerbation definitions fall into two categories [Pauwels et al. 2004; Rodriguez-Roisin, 2000]. One set are the
Results
Study characteristics
Studies that focused on the retrospective relationship between clinical levels and symptoms of psychological distress and exacerbation of COPD.
Abbreviations: AECOPD, acute exacerbation of COPD; BAI, Beck Anxiety Inventory; BASDEC, Brief Assessment Schedule Depression Cards; CES-D, Center for Epidemiologic Studies Depression Scale; COPD, chronic obstructive pulmonary disease; ED, emergency department; GDS, Geriatric Depression Scale; GMS, Geriatric Mental State Schedule; HADS, Hospital Anxiety and Depression Scale; hospit., hospitalizations; MADRS, Montgomery Asberg Depression Rating Scale; signif., significantly.
Studies that focused on the prospective relationship between clinical levels and symptoms of psychological distress and exacerbation of COPD.
Abbreviations: ADIS-IV, Anxiety Disorders Interview Schedule-IV; AECOPD, acute exacerbation of COPD; BDI-II, Beck Depression Inventory; COPD, chronic obstructive pulmonary disease; ED, emergency department; HADS, Hospital Anxiety and Depression Scale; hospit., hospitalizations; NS, not significant; pts, patients; signif., significantly; Zung SDS, Zung Self-Rating Depression Scale.
Six studies were conducted among stable outpatients [Laurin et al. 2009; Quint et al. 2008; Xu et al. 2008; Fan et al. 2007; Kim et al. 2000; Yohannes et al. 2000a], five during a hospitalization [Ng et al. 2007; Almagro et al. 2006; Cao et al. 2006; Chen and Narsavage, 2006; Gudmundsson et al. 2005] and one in the ED [Dahlen and Janson, 2002]. The majority of studies were prospective (n = 7), with follow ups ranging from 90 days to 2.2 years [Laurin et al. 2009; Xu et al. 2008; Fan et al. 2007; Ng et al. 2007; Almagro et al. 2006; Chen and Narsavage, 2006; Gudmundsson et al. 2005]. Four studies evaluated exacerbation rates during the previous year [Quint et al. 2008; Cao et al. 2006; Kim et al. 2000; Yohannes et al. 2000a]. The average sample size was 2845 (from 43 to 610), and the average proportion of women included was 35%. For those studies that reported it, the mean age of the participants ranged from 65 to 73 years and the mean forced expiratory volume in one second (FEV1), from 27% to 49% of predicted.
In addition to exacerbations, mortality was evaluated in two studies [Fan et al. 2007; Ng et al. 2007]. Eleven of the 12 studies (92%) employed an event-based definition of an exacerbation, alone or in combination with a symptom-based definition. Only one study defined an exacerbation using a symptom-based definition alone [Quint et al. 2008]. When symptom definitions were used, they varied, with one study requiring the worsening of only one key symptom [Xu et al. 2008] and the other requiring the worsening of two or more key symptoms [Quint et al. 2008]. Of the 11 studies defining exacerbations according to events, eight (73%) limited their definition to hospital admissions (or readmissions) and/or ED visits with or without a change in medication [Fan et al. 2007; Ng et al. 2007; Almagro et al. 2006; Cao et al. 2006; Chen and Narsavage, 2006; Gudmundsson et al. 2005; Dahlen and Janson, 2002; Yohannes et al. 2000a]. Further, only one study using an event-based definition distinguished between outpatient-treated events (i.e. those involving patient initiation of an action plan in patient’s own environment after consultation with a case nurse) and inpatient-treated events (i.e. those involving an ED or hospital visit) [Laurin et al. 2009].
Exacerbations were assessed by patient interview or self reports in six studies (50%) [Laurin et al. 2009; Quint et al. 2008; Xu et al. 2008; Ng et al. 2007; Chen and Narsavage, 2006; Gudmundsson et al. 2005]. With the exception of two studies [Quint et al. 2008; Ng et al. 2007], all self-reported events were confirmed by hospital chart review. Two studies (18%) assessed exacerbations by accessing national administrative databases [Almagro et al. 2006] or Medicare claims data [Fan et al. 2007]. Only three studies (25%) [Laurin et al. 2009; Quint et al. 2008; Xu et al. 2008] specified how they distinguished a new exacerbation from a relapse of a previous exacerbation. Finally, only two studies [Laurin et al. 2009; Xu et al. 2008] reported the weighted number of events/patient-year [Suissa, 2006] in order to take into account each patient’s duration in the study.
With the exception of two studies [Laurin et al. 2009; Yohannes et al. 2000a], the studies assessed psychological distress using self-report questionnaires rather than clinical interviews. Five studies used the Hospital Anxiety and Depression Scale (HADS) [Xu et al. 2008; Ng et al. 2007; Cao et al. 2006; Gudmundsson et al. 2005; Dahlen and Janson, 2002] to assess both depressive and anxiety symptoms. Other depressive symptom assessment tools included the Zung Self-Rating Depression Scale (Zung SDS) [Chen and Narsavage, 2006], the Centre for Epidemiologic Studies Depression Scale (CES-D) [Quint et al. 2008], the Yesavage Scale [Almagro et al. 2006], the Brief Assessment Schedule Depression Cards (BASDEC) [Yohannes et al. 2000a], the Geriatric Depression Scale (GDS) [Kim et al. 2000], the Beck Depression Inventory-II (BDI-II) [Fan et al. 2007], and the Montgomery Asberg Depression Rating Scale (MADRS) [Yohannes et al. 2000a]. Anxiety symptom tools used included the Beck Anxiety Inventory (BAI) [Kim et al. 2000] and the State–Trait Anxiety Inventory [Fan et al. 2007]. The two studies that assessed anxiety and/or depressive disorders used the Anxiety Disorder Interview Schedule for DMS-IV (ADIS-IV) [Laurin et al. 2009] and the Geriatric Mental State Schedule (GMS) [Yohannes et al. 2000a]. Psychological distress was measured in all studies at baseline (n = 10 or 83%), except for two studies, who evaluated it only 1 month after the baseline evaluation [Cao et al. 2006; Dahlen and Janson, 2002]. Some studies, in addition to the baseline evaluation, conducted follow-up assessments at 6 and 12 months [Ng et al. 2007], or at an exacerbation visit [Quint et al. 2008]. With respect to the wide variety of tools used with specific cutpoints, the prevalence of significantly elevated depressive and anxiety symptoms was found to range from 9% to 46% and 5% to 41%, respectively, and the prevalence of depression and anxiety disorders, from 17% to 42% and 18% to 46%, respectively.
A total of seven studies (58%) evaluated the prospective association between psychological distress and exacerbation risk, three among stable outpatients [Laurin et al. 2009; Xu et al. 2008; Fan et al. 2007] and four among treated inpatients (emergency care or hospitalized) [Ng et al. 2007; Almagro et al. 2006; Chen and Narsavage, 2006; Gudmundsson et al. 2005]. The remainder examined the retrospective associations between exacerbations and psychological distress [Quint et al. 2008; Cao et al. 2006; Dahlen and Janson, 2002; Kim et al. 2000; Yohannes et al. 2000a].
The following sections will review each of the 12 studies in turn, beginning with those that focused on the retrospective links between clinical levels and symptoms of psychological distress, and exacerbation risk, followed by those that focused on the prospective associations between clinical levels and symptoms of psychological distress, and exacerbation risk.
Association between clinical levels of psychological distress (anxiety and depressive disorders) and exacerbations, retrospective studies
Yohannes and colleagues assessed the prevalence of clinical depression and anxiety via one of the most widely used mental health instrument (the GMS) and their association with COPD-related hospitalizations in the last year [Yohannes et al. 2000a]. Univariate analysis in this study of 137 stable COPD outpatients found a higher proportion of patients with depression who had hospital admissions episodes (60%) compared with not depressed patients (35%) (X2 = 7.20, p = 0.0007). They also had significantly higher hospitalization length compared with non-depressed patients (mean ± SEM = 9.8 ± 1.7 versus 2.3 ± 0.6 days, respectively, p < 0.0001). However, multiple regression analysis revealed that anxiety (and not depression) was independently associated with hospitalizations (t = 2.35, p = 0.02). However, it should be noted that the authors excluded patients with a prior history of non-depressive psychiatric disorders, which may have underestimated the impact of psychiatric morbidity on exacerbations.
Association between symptoms of psychological distress (anxiety and depressive symptomatology) and exacerbations, retrospective studies
Four studies retrospectively assessed the association between anxiety and depressive symptoms, and exacerbations. The size of the samples were small (n = 43) to modest (n = 186), comprised COPD and asthmatics patients, and two of them included no women [Kim et al. 2000] or very few (<20% of the sample) [Cao et al. 2006]. Two of the four studies observed non-significant associations between higher levels of psychological distress and exacerbations.
The first of these studies examined the characteristics of 61 frequent (n = ≥ 3) versus 106 infrequent (n < 3) ‘exacerbators’ in the preceding year using a symptom-based definition [Quint et al. 2008]. The results showed that significantly more patients with high depressive symptoms (CES-D ≥ 16) were likely to be frequent exacerbators compared with those with low depressive symptoms (54% versus 35%, p = 0.01). In this study, depressive symptoms were also evaluated prospectively at an exacerbation visit. The change in depression score from baseline to an exacerbation was not related to the length of the symptom-based exacerbation, the time to the next exacerbation or the exacerbation’s symptoms (no statistics are shown). Although the significant relationships were reproduced using multiple linear regression analyses, details about what covariates were included in the models were not specifically described.
The second retrospective study was conducted by Cao and colleagues with 186 stable COPD inpatients from Singapore and found that those patients who were prescribed psychotropic drugs had a near 14-fold increased risk (95% CI, 1.5–122.9) of having had frequent hospital COPD readmissions (≥2) in the past year, independent of covariates [Cao et al. 2006]. However, having high levels of depressive (OR = 1.73; 95% CI, 0.89–3.37) or anxiety (OR = 1.53; 95% CI, 0.52–4.47) symptoms (as indicated by a score of ≥8 on the HADS) was not significantly associated with frequent hospital readmissions. It is important to note that the authors excluded seven patients with psychiatric disorders (no details available), potentially underestimating the associations.
The third study by Dahlen and Janson was able to demonstrate an independent association between symptoms of psychological distress and increased rates of relapses (any hospital contacts that necessitated an additional treatment) among a sample of 43 patients with COPD and asthma presenting to the ED [Dahlen and Janson, 2002]. In this sample, in the 4 weeks after their ED visit, patients with high anxiety and/or depressive symptoms (HADS ≥ 8) had significantly more relapses (53%), compared with patients with low anxiety and/or depressive symptoms (19%) (p < 0.05). When patients with and without evidence of relapse following their ED visit were compared, those with a relapse had a significantly higher HADS total score than those without (mean ± SD; 8.6 ± 5.1 versus 12.4 ± 5.9 points, respectively, p < 0.05). However, when asthma and COPD patients were analyzed separately, this association only remained among asthmatic patients (11.1 ± 7.4 versus 7.8 ± 4.6 points respectively, p = 0.40). Finally, asthmatics and COPD patients with anxiety and/or depressive symptoms were over seven times more likely to have a relapse, controlling for age, sex, pack-years, emergency treatment, and atopic status (adjusted OR = 7.40; 95% CI, 1.10–50.0).
The fourth retrospective study assessed the relationship between psychological distress and use of healthcare resources in the past year, among 43 men with COPD [Kim et al. 2000]. Results revealed that anxiety and depressive symptoms, measured with the GDS (>10) and the BAI (>15), respectively, did not contribute significantly to the overall variance in outpatient (total outpatient visit days, primary care visit days, mental health visit days) or inpatient (total inpatient hospitalization discharges, hospitalization days) use of healthcare resources. It is important to note that this study evaluated a small sample size of male veterans. Therefore, the external validity can be criticized, as well as the reliability and accuracy of the data, inherent to a retrospective design.
Association between clinical levels of psychological distress (anxiety and depressive disorders) and exacerbations, prospective studies
Only one study assessed psychiatric disorders and prospective risk for exacerbations. Using a structured psychiatric interview (ADIS-IV) among 110 stable COPD outpatients, Laurin and colleagues found that patients with anxiety and depressive disorders had higher annual rates of event-based exacerbations (weighted mean = 3.81 in psychiatric versus weighted mean = 2.73 in controls; F = 6.90, p = 0.009) [Laurin et al. 2009]. The presence of an anxiety or depressive disorder was also associated with a 1.56-fold increase risk of having an exacerbation (95% CI, 1.02–2.37) over an average of 2.2 years. Finally, having an anxiety or depressive disorder accounted for 22% of the population attributable risk of having an exacerbation. It is noteworthy that this association seemed to be driven by exacerbations treated on an outpatient basis in consultation with a case nurse, and not those treated in hospital. These results were adjusted for several a priori determined covariates including age, sex, recruitment site, time between the patient’s last hospital exacerbation and the baseline interview, smoking (pack-years), medical comorbidities, and COPD duration and severity.
Association between symptoms of psychological distress (anxiety and depressive symptomatology) and exacerbations, prospective studies
Six studies prespectively assessed the association between symptoms of depression and anxiety, and exacerbations. In these studies, sample sizes were modest (n = 129) to large (n = 610). Among these studies, five were either conducted among inpatients and/or focused on hospitalizations (or hospital contact) only [Fan et al. 2007; Ng et al. 2007; Almagro et al. 2006; Chen and Narsavage, 2006; Gudmundsson et al. 2005]. Five of these six studies highlighted an overall trend for an absence of an association between symptoms of psychological distress and hospitalizations [Fan et al. 2007; Ng et al. 2007; Almagro et al. 2006; Chen and Narsavage, 2006; Gudmundsson et al. 2005].
Xu and colleagues assessed the impact of psychological distress, as measured by the HADS, on event-based as well as symptom-based exacerbations among a sample of 491 stable, moderate to severe, COPD outpatients from China [Xu et al. 2008]. Compared with patients unlikely to have depression (HADS ≤ 7), patients with ‘possible depression’ (HADSD ≥ 8) were found to have significantly more symptom-based exacerbations (≥2 exacerbations: 46% versus 38%, p = 0.04), more event-based exacerbations (≤2 exacerbations: 25% versus 20%, p = 0.003), more COPD hospitalizations (≤2 hospitalizations: 14% versus 8%, p = 0.03), and longer hospital stays (mean days ± SD = 38 ± 37 versus 26 ± 19, p = 0.05) during a 1-year follow up. Similarly, patients with ‘possible anxiety’ (HADSA ≥ 8) were found to have significantly more symptom-based exacerbations (≥2 exacerbations: 43% versus 40%, p = 0.003) and longer hospital stays (mean days ± SD = 49 ± 48 versus 27 ± 21, p = 0.03) compared with patients unlikely to have anxiety (HADSA ≤ 7). Moreover, patients with versus without ‘probable depression’ (HADSD ≥ 11) had at least a 50% higher risk of having a symptom (adjusted incidence rate ratio [IRR], 1.51, 95% CI, 1.01–2.24) or an event-based exacerbation (adjusted IRR, 1.56, 1.02–2.40, respectively). They also had a higher risk of having a subsequent COPD hospitalization (adjusted IRR, 1.72; 95% CI, 1.04–2.85). However, there was no difference between patients with and without ‘probable anxiety’ (HADSA ≥ 11).
The National Emphysema Treatment Trial (NETT) study assessed the prospective association between depressive symptoms (measured using the BDI), anxiety symptoms (measured with the State–Trait Anxiety Inventory), and COPD hospitalizations or ED visits among 610 patients with emphysema and severe airflow limitation [Fan et al. 2007]. Depressive symptoms (measured continuously) were not associated with a higher risk for hospitalizations at 1 year after adjusting for covariates. When BDI scores were analyzed by quintile, the results revealed a higher risk of COPD-related hospitalizations and ED visits in patients with depressive scores ≥15 (OR, 2.49; 95% CI, 1.40–4.45), compared with patients with low depressive scores (<5). However, this association became non-significant after adjusting for covariates (i.e. sex, antidepressant use, disease severity, partial pressure of oxygen, prior COPD hospitalization, medical comorbidity, and dyspnea severity). In another BDI quintile analysis, patients with BDI scores of 11–14 and ≥15 had an increased risk of 3-year COPD mortality (adjusted OR, 1.97; 95% CI, 1.03–3.77 and adjusted OR, 2.74; 95% CI, 1.42–5.29, respectively) compared with patients with a BDI score in the lowest quintile (BDI < 5). The authors mentioned that anxiety symptoms were not associated with COPD-related hospitalizations, or ED visits, or 1- or 3-year mortality.
Ng and colleagues found no significant association between depressive symptoms (measured using the HADS) and readmissions for COPD exacerbations (adjusted HR, 0.93; 95% CI, 0.68–1.28) among 376 COPD patients (56 women) hospitalized for an acute exacerbation [Ng et al. 2007]. However, compared with patients with low levels of depressive symptoms, those with high depressive symptoms (HADS ≥ 8) at baseline had significantly higher total number of days spent in the hospital in the following year (14.2 and 11.2 days per patient per year, respectively, p = 0.04), even after controlling for covariates (including comorbid anxiety and COPD severity). Patients with high depressive symptoms also had a 1.93 higher risk of all-cause mortality (adjusted HR, 1.93; 95% CI, 1.04–3.58), even after controlling for COPD duration and severity, body mass index, dyspnea, blood gas, smoking, length of hospitalization during the baseline evaluation, and socioeconomic variables. On the other hand, high anxiety symptoms (HADS ≥ 8) were not independently associated with all-cause mortality (HR, 1.97; 95% CI, 0.96–4.02). No COPD specific mortality data were presented.
Among 145 patients hospitalized for chronic bronchitis, emphysema, or COPD as the primary or secondary diagnosis in Taiwan, Chen and Narsavage were not able to find an association between high depressive symptoms (Zung Self-Rating Depression Scale >50) and readmissions at 14 and 90 days post-discharge, controlling for age, sex, and education (OR, 1.05; 95% CI, 0.99–1.12 at 90 days) [Chen and Narsavage, 2006]. However, it should be noted that they excluded psychiatric illnesses (no details available).
Almagro and colleagues assessed risk factors for hospital readmissions among a sample of 129 COPD patients (9 women) hospitalized for COPD [Almagro et al. 2006]. Patients who were readmitted during the 1-year follow up demonstrated higher baseline depressive symptoms (Yesavage Depression Scale ≥ 5: mean ± SD, 5.0 ± 3.4 versus 3.7 ± 3.1, p < 0.05). However, multivariate analysis revealed no association between depressive symptoms at baseline and readmissions 1 year later.
Finally, Gudmundsson and colleagues evaluated 406 consecutive patients hospitalized for an acute COPD exacerbation, and analyzed determinants of readmissions over 1 year [Gudmundsson et al. 2005]. The authors reported finding no evidence of a significant association between 1-year readmissions and depressive or anxiety symptoms (measured using the HADS ≥ 8). However, among patients with poorer health status (as measured by the St George’s Respiratory Questionnaire), higher anxiety symptoms (but not higher depressive symptoms) were independently associated with an increased risk of readmission at 1 year (adjusted HR, 1.36; 95% CI, 1.12–1.65). This suggests that anxiety symptoms may be more common among patients in poorer health, and may be a specific risk factor for more rapid deterioration among the sickest patients.
Discussion
Summary of the evidence
Among the 12 studies included in this review, five (42%) revealed positive associations between various levels of psychological distress and increased risk for exacerbations, although seven (58%) failed to support such an association. Two studies reported a positive and independent association between depressive symptoms and mortality. So although there is some evidence to suggest that psychological distress confers greater risk for COPD exacerbations, particularly those defined using symptom-based and those treated on an outpatient basis, the low number of relevant studies and overall heterogeneity limit the conclusions that can be drawn. The following sections will highlight some of the limitations of this literature, and provide some recommendations for future research.
Characterization of exacerbation
Despite the fact that many COPD exacerbations are now often being treated on an outpatient basis [Seemungal et al. 1998], e.g. after consultation with a case manager or nurse, and represent a significant healthcare burden [Mittmann et al. 2008], the vast majority of studies limited their definition of an exacerbation to inpatient or hospital-treated events, which does not take into account the full spectrum of exacerbations. This may have grossly underestimated the rate of exacerbations occurring across studies, and diminished the probability of observing more associations between outpatient-treated exacerbations and psychological distress. This hypothesis is supported by the findings of Laurin and colleagues, who observed significant associations between clinical levels of psychological distress and outpatient but not inpatient-treated exacerbations [Laurin et al. 2009]. This suggests that psychological distress may have a greater impact on exacerbations managed on an outpatient basis, where the patient has a greater role to play in assessing the severity of their symptoms, and in initiating self-management strategies (e.g. action plan). Thus, the assessment of both types of exacerbation management strategies (outpatient and inpatient) appears important to understand to true nature of the association between anxiety and depression and COPD exacerbation risk.
While the majority of studies used only event-based exacerbations, the way in which exacerbations were defined tended to vary across studies. For example, it was not always specified whether hospital-treated exacerbations included ED visits that did not result in hospital admissions. Few studies specified the type of treatment needed (e.g. reliever medication and/or systemic corticosteroids and/or antibiotics). Moreover, when looking at the symptom-based exacerbations, the minimal number and the type of symptoms required (i.e. major versus minor) was inconsistent across studies.
Another important limitation of the current literature is that all but three studies [Laurin et al. 2009; Quint et al. 2008; Xu et al. 2008] failed to discuss how they dealt with possible overlapping events (i.e. relapses of previous exacerbations or slow-to-resolve exacerbations). As a result, it is impossible to ensure the independence of events [Aaron et al. 2008], especially in patients with recurrent symptom episodes over short periods of time. Consequently, it is crucial for future studies to determine the independence of individual exacerbations, and report how independence was determined.
Statistical approach
The robustness of the statistical approaches used also varied significantly across studies. For example, all but two studies [Laurin et al. 2009; Xu et al. 2008] failed to report the weighted number of events/patient-year which takes varying follow up for each patient into account (e.g. patients who drop out of the study before the end of the follow up), which may lead to bias estimates of mean exacerbations rates [Suissa, 2006]. What constitutes the end of the follow up (e.g. death, attrition) should be clearly specified and controlled for using unbiased time-weighted methods [Aaron et al. 2008]. Furthermore, while most results adjusted for potential confounding (e.g. disease severity), information about the nature of the statistical adjustments was not always provided, making it difficult to assess the independence of the associations observed. Not only should the information about covariates be clearly provided, but it should be determined a priori (as recommended by statistical guidelines) and not chosen after conducting univariate analyses [Freedland et al. 2005].
Psychological assessment
With the exception of two studies [Laurin et al. 2009; Yohannes et al. 2000a], the studies in this area assessed psychological distress levels using self-report questionnaires rather than structured psychiatric interviews. The use of self-reported questionnaires to measure anxiety and depressive symptomatology among COPD patients merits some psychometric comments. First, no study reported any justification for their choice of instrument based on evidence of its measurement equivalence (i.e. invariance) across distinct (i.e. COPD) samples of depressive or anxious participants. This is problematic given that these instruments are specifically designed to identify potential clinical depression and/or anxiety in community samples, not specific disease populations. Testing for measurement invariance allows for the verification that high scores on the instrument, which tend to be observed among depressed and anxious individuals, are really due to higher levels of anxiety and depression, and not to a different construct in the sample of interest [Meredith and Teresi, 2006]. Second, the depression inventories (e.g., BDI-II, HADS), being measures of ‘depression’, should be able to differentiate between convergent measures of anxiety and depression [Beck and Steer, 1993]. However, it has been shown previously [Abramson et al. 2002; Roberts and Monroe, 1999] that both the subscale and full-scale scores of several depression inventories (e.g. BDI-II, CES-D, HADSD) are actually highly correlated with anxiety measures (e.g. BAI, HADSA). This indicates that the discriminant validity of these instruments is far from perfect, and is likely to confound levels of depressive and anxiety symptoms in COPD patients. Third, few studies investigated the appropriateness of the proposed cut-off scores used in their samples to designate ‘high/significant depression’ or ‘high/significant anxiety’. It has been shown that both the cultural origin and the sex of the participants could substantially affect the classification accuracy of the cut-off score recommended in the initial validation studies of the questionnaires. For example, the use of a similar cut-off point (i.e. 8) for the HADS is doubtful in Chinese [Xu et al. 2008] and Scandinavian [Gudmundsson et al. 2005; Dahlen and Janson, 2002] samples. To illustrate this problem, previous studies cross-culturally investigated the screening properties of the CES-D and their results are also divergent. In two Spanish studies, these cut-offs ranged from 16 [Soler et al. 1997] to 26 [Vázquez et al. 2007]; in Portuguese and Greek samples, the cut-off score ranged from 20 [Gonçalves and Fagulha, 2004] to 24 [Fountoulakis et al. 2001]. Scholars and clinicians should be aware of the limitations of these self-report measures when interpreting the results of studies in disease-specific populations. Therefore, we encourage the use of standardized clinical interviews (e.g. ADIS-IV, SCID) to diagnose anxiety and depression in COPD patients, because these interviews are much more sensitive and specific for the determination of significant levels of psychological distress. If design, personnel, and/or time constraints limit the ability to use psychiatric interviews, future studies should clearly justify their choice of self-report questionnaire, which should include references to its measurement equivalence and demonstration of sound psychometric properties. If a questionnaire is scored continuously, future studies should conduct analyses using scores in their continuous form, as cut-off scores established in one population (e.g. a community sample) may not be relevant to another (e.g. COPD sample). Finally, given the high correlation between anxiety and depression measures, it is recommended that future studies examine the impact of anxiety and depression on exacerbation rates independently of each other (i.e. entering each as a covariate in their respective models), in order to determine the unique contribution of anxiety and depression to exacerbation risk.
Participant characteristics
Several participant characteristics may have influenced the results of this review. First, the studies that failed to show a positive association between psychological distress and exacerbations tended to examine inpatients rather than outpatients [Ng et al. 2007; Almagro et al. 2006; Cao et al. 2006; Chen and Narsavage, 2006; Gudmundsson et al. 2005]. This suggests that associations between psychological distress and outpatient-treated exacerbations may have been obscured. Second, two studies specifically excluded patients with psychiatric illnesses, which likely underestimated the strength of the associations observed [Cao et al. 2006; Chen and Narsavage, 2006]. Finally, half of the studies in this review included samples containing fewer than 35% women [Xu et al. 2008; Ng et al. 2007; Almagro et al. 2006; Cao et al. 2006; Chen and Narsavage, 2006; Kim et al. 2000]. Given that rates of COPD are rapidly increasing among women and approaching rates observed in men [Gershon et al. 2010], and that women are at least 1.5 times more likely to have anxiety and depression than men [Laurin et al. 2007], this suggests that the current literature has likely underestimated the overall association between anxiety and depression and exacerbation risk in COPD patients. Future studies should therefore include comparable rates of men and women in order to increase external validity and clinical relevance.
Mechanisms linking psychological distress to increased exacerbation risk
While a detailed discussion on the mechanisms underlying the possible relationship between psychological distress and exacerbations goes beyond the scope of this review, we can still speculate on some of the potential pathways.
First, anxiety and depression are known to be associated with feelings of hopelessness, helplessness, withdrawal, and fear [Burgess et al. 2005; Yohannes et al. 2000b; Seligman, 1975], which may lower self confidence and feelings of self efficacy [Yohannes et al. 2009]. This may in turn, lead to worse disease-related coping [Hadjistavropoulos et al. 1998], including poor adherence to medical regimens [Di Matteo et al. 2000] and poor health behaviors (e.g. smoking, physical inactivity) [Burgess et al. 2005; Yohannes et al. 2000b; Pomerleau, 1997].
Second, anxiety and depression are often associated with impaired (distorted) cognitions [Yohannes et al. 2009], which could lead patients to catastrophically misinterpret and over-report symptoms such as dyspnea [Dales et al. 1989], leading to an increase in healthcare resource and medication use.
Finally, anxiety and depression could also lead to exacerbations through autonomic nervous system and immune system dysregulation, including decreased immune responses to infection and/or increased progression of viral and bacterial infections [Padgett and Glaser, 2003; Liu et al. 2002; Cohen et al. 1993; O’Leary, 1992]. For example, it has been shown that participants exhibiting increased chronic psychological stress (e.g. stressful life events) are at higher risk of developing the common cold; an association that also seems to be dose-dependent [Cohen et al. 1991].
Limitations of the present review
The main limitation of this review is that the high heterogeneity of patient populations, sample sizes, methods (e.g. setting, design, timing of the evaluations, length of follow up, measurement instruments, and statistical approaches), and outcome (exacerbation) definitions, makes it difficult to compare results across studies and draw any firm conclusions on the strength of the association between psychological distress and COPD exacerbations.
This review can also be criticized for the small number of studies included, especially those that focused on clinical levels of psychological distress. Other limitations include the restricted number of women sampled across studies (981 on a total of 2845 patients or less than 35%), which limits the generalizability of the results as well as our understanding of sex-related differences. As a result, women seem to be underrepresented in these studies since around 50% of the Americans with COPD are women [Gershon et al. 2010; Varela et al. 2010]. A final limitation to the present review is that we did not assess the presence of publication bias, although the presence of negative studies makes the existence of this bias less likely.
Recommendations
Considering the important heterogeneity found between studies on the topic, it is the authors’ opinion that there is a need for more studies looking at the prospective association between psychological distress and exacerbations among stable cohorts of COPD patients.
One of the most important recommendations emanating from this review is the necessity to standardize the assessment of psychological distress and exacerbations. In particular, for research purposes, the use of standardized clinical interviews (e.g. ADIS-IV, SCID) to diagnose anxiety and depression is important considering their higher sensitivity and specificity for the determination of significant levels of psychological distress and should be considered to be the ‘gold standard’ [Yohannes et al. 2009]. However, when it is not possible to use a clinical interview due to time and resource limitations inherent to research, the use of questionnaires that are well justified and specifically validated in the population evaluated is imperative.
Problems associated with the absence of a standard consensus definition of an exacerbation and relapse, as well as the ongoing debate about the most appropriate definition of an exacerbation (event versus symptom-based), is beyond the scope of this review. However, we emphasize the importance of including quality control measures to standardize methods across studies in order to facilitate cross-study comparisons. More specifically, we recommend ensuring the independence of sequential events and using time-weighted methods to account for each patient’s duration in the study. We also recommend to assess exacerbations treated on an inpatient and outpatient basis [Seemungal et al. 1998]. Exacerbations should also be assessed by a blind adjudication committee, as per Aaron and colleagues [Aaron et al. 2008], who will review study data to ensure that exacerbations meet prestated study definitions of a COPD exacerbation.
Finally, there is a critical need to include an equal number of men and women in studies, and not to exclude patients with psychiatric illnesses such as major depressive disorder, in order to get a clearer and more representative picture of the anxiety/depression-COPD exacerbation risk association.
Conclusions
To the best of the authors’ knowledge, this is the first review of literature on the link between psychological distress (e.g. depressive and anxiety symptoms and clinical diagnosis) and COPD exacerbations. It appears that psychological distress confers greater risk for symptom-based rather than event-based defined exacerbations. However, methodological weaknesses and a high heterogeneity across studies make it difficult to draw any firm conclusions on the nature and strength of these associations. Further prospective studies are needed to more comprehensively assess the question, particularly in light of the high levels of both anxiety and depression in COPD patients.
Footnotes
Funding
This work was supported by a postdoctoral scholarship from the Fonds de la Recherche en Santé du Québec (FRSQ) (CL), a Gender and Sex Determinants of Cardiovascular Disease: From Bench to Beyond (GENESIS) Canadian Institutes of Health Research (CIHR) Scholarship (CL), and an Asthma in the Workplace Scholarship (CIHR) (GM), FRSQ Chercheur boursier awards (SLB and KLL), and a Canadian Institutes of Health Research New Investigator Award (SLB).
Conflict of interest statement
None declared.