Understanding the Association Between Chronic Obstructive Pulmonary Disease and Current Anxiety: A Population-Based Study

Abstract

This study's objectives were to investigate the independent relationship between COPD and past-year Generalized Anxiety Disorder (GAD) in a population-based sample of adult Canadians and to identify significant correlates of GAD among COPD patients. A series of logistic regression analyses were conducted with a sample of 11,163 respondents aged 50+ from the 2012 Canadian Community Health Survey–Mental Health to determine the degree to which the direct association between COPD and GAD was attenuated by socio-demographic factors, social support, health behaviors, sleep problems, pain, functional limitations, and early childhood adversities. Additional analyses were completed using the sub-sample of those diagnosed with COPD (n = 746) to determine predictors of GAD. One in 17 (5.8%) of older individuals with COPD had past-year GAD, in comparison to 1.7% of those without (p < .001). The age-sex-race adjusted odds of GAD were four times higher for those with COPD compared to those without COPD (OR = 3.90, 95%CI: 2.64, 5.77). After full adjustment for 18 characteristics, these odds declined to 1.72 (95%CI: 1.10, 2.71). Factors associated with GAD among those with COPD include not having a confidant (OR = 7.85, 95%CI: 3.47, 17.75), exposure to parental domestic violence (OR = 5.63, 95% CI: 2.07, 15.34) and lifetime depressive disorders (OR = 3.59, 95% CI:1.61,7.98). Those with COPD have substantially higher odds of GAD even after most known risk factors for GAD are accounted for. These findings have implications for targeted outreach and screening, particularly for patients with pain and functional limitations. The importance of a multidisciplinary healthcare team is underscored by the multiple issues that may impact GAD among COPD patients.

KEYWORDS:

• Adult
• COPD
• generalized anxiety disorder
• population-based
• pulmonary emphysema

Introduction

Chronic obstructive pulmonary disease (COPD) refers to a group of diseases that cause blockage of airflow and subsequent breathing-related problems, including emphysema and chronic bronchitis. In 2011, 15,000,000 Americans reported a diagnosis of COPD (1) with an overall prevalence ranging from 4% to 9% (2). COPD is the third-leading cause of death in the United States, with an estimated 135,000 Americans dying of the disease annually (2). COPD can be severely debilitating (3), impairing people's ability to work and participate fully in their family and social life (2). In the United States, COPD-related direct costs to the healthcare system and the indirect costs related to lost productivity total approximately $50 billion dollars annually (2).

In addition to death and disability, COPD is associated with mental health problems, including anxiety. Anxiety disorders are the most commonly diagnosed mental illnesses in the United States (4), affecting approximately 1 in 10 over the past year (5, 6) and 17% to 29% of the population over their lifetime (5, 7). A strong association between COPD and anxiety disorders has been shown in both clinical and population-based studies (8–10). A common symptom of COPD—dyspnea (or difficulty breathing) — can be anxiety-provoking (8). Additionally, people with COPD may experience anxiety due to deteriorating functioning and decreased ability to perform activities of daily living (8).

Anxiety disorders are associated with poor health outcomes and increased mortality among patients with COPD (10). The relationship between anxiety and COPD outcomes is thought to be cyclical – elevated anxiety may increase negative coping strategies, such as smoking, which, in turn, exacerbate COPD symptoms (8). Elevated COPD symptoms may then further exacerbate anxiety. Although there is strong evidence that COPD is related to anxiety disorders, less is known about correlates of anxiety disorders among those with COPD. Improved understanding of these factors will enhance targeting and outreach to those patients with COPD who are most vulnerable to anxiety.

Risk factors for COPD and anxiety

Several population-based studies have documented factors that may influence both COPD and anxiety, shown in Table 1. These factors may be key to understanding the relationship between COPD and anxiety, including demographic characteristics, health behaviors, adult socioeconomic status, exposure to childhood stressors, physical and mental health co-morbidities, and emotional and spiritual wellbeing (11–39).

Table 1. Shared risk factors for COPD and anxiety.

Factor	Association
Socioeconomic status	some studies show that lower socioeconomic status (e.g. income, education, occupation) predicts COPD, while other studies show that it is unrelated to COPD diagnosis low socioeconomic status is associated with higher odds of anxiety disorder diagnosis
Social support	people with COPD are less likely to be engaged in social activities older men and women who are divorced or separated have a significantly higher risk of mortality from COPD increased anxiety is related to being unmarried loneliness increases the risk of hospital admission for an anxiety disorder
Religiosity	religiosity/spirituality is associated with increased anxiety among COPD patients, due to the use of negative religious coping (e.g. believing that COPD is a punishment from God) or as a result of the increased likelihood of becoming religious/ spiritual when health is deteriorating religiosity is found to be associated with less anxiety among those who believe in the afterlife
Obesity	obesity is associated with COPD obesity is associated with anxiety disorders, especially among women
Smoking	the odds of COPD increase with the amount smoked the odds of having a lifetime anxiety disorder are increased for people who smoke
Drug and/or alcohol abuse	COPD is associated with alcohol dependence and illicit drug use, particularly cannabis cannabis and alcohol dependence are strongly associated with anxiety disorders
Sleep problems, pain, and functional limitations	COPD has been linked to chronic pain, insomnia and functional limitations and limitations in Activities of Daily Living (ADLs) anxiety disorders are associated with chronic pain, insomnia, and lower levels of physical functioning
Adverse childhood experiences	sexual abuse and witnessing parental domestic violence are associated with increased odds of developing COPD in adulthood anxiety is associated with different types and/or number of childhood adversities, including physical abuse, sexual abuse, and family violence
Depression	there are strong links between COPD and depression depression and anxiety disorders are often co-morbid

^a References [11–39].

Study objective

The objectives of this study are twofold: 1) To investigate the independent relationship between COPD and past-year Generalized Anxiety Disorder (GAD) in a nationally representative sample of older adult Canadians; 2), To identify the significant independent correlates of past-year GAD in a representative national sample of those with COPD.

Methods

For the current study, we analyzed data from the nationally representative 2012 Canadian Community Health Survey-Mental Health (CCHS-MH). The CCHS-MH is a cross-sectional survey that collects data on mental health status, access to and perceived need for formal and informal services supports, functioning and disability, and covariates of community dwelling Canadians (40, 41). The CCHS-MH covers the population of individuals age 15 and older living in the 10 provinces. Less than 3% of the target population are excluded from the survey's sampling frame. Excluded Canadians include residents of the three territories, those living on reserves or other Aboriginal settlements, full-time members of the armed forces, and people who are institutionalized (41).

Sample

The sample for the CCHS-MH was selected using a three-stage design determining first, geographical areas, second, households within each geographical area, and third, a randomly-selected eligible respondent from each household (41). Out of all selected units, 36,443 were in-scope for the survey and, of these, 29,088 households agreed to participate, yielding an overall household-level response rate of 79.8%. An overall person-level response rate of 86.3% was observed, as 25,113 out of 29,088 persons selected to participate from responding households provided a valid questionnaire. Thus, the national level overall response rate was 68.9% (41).

For the current study, we analyzed two sub-samples of the full CCHS-MH. We were only interested in those respondents age 50 and older because of the long latency period for COPD and subsequent older age onset of the disease. The first sub-sample is made up of adults aged 50 and over with complete data on COPD and current anxiety, as well as each of the independent measures included in the final analysis (n = 11,163), which excluded 10.9% of participants. Of these, 746 reported that they had been diagnosed with COPD and these comprised our second sub-sample. Approximately 12.1% of those aged 50-plus with COPD were excluded due to missing data.

Measures

Respondents were asked if they had any “conditions diagnosed by a health professional” which are “expected to last or have already lasted 6 months or more.” Included in the list of potential health problems was COPD, which was assessed as follows: “Do you have chronic bronchitis, emphysema or chronic obstructive pulmonary disease or COPD?” Individuals were coded as having a current anxiety disorder if they met the World Health Organization version of the Composite International Diagnostic Interview (WHO-CIDI) criteria for Generalized Anxiety Disorder. The WHO-CIDI is a rigorous structured diagnostic interview that generates diagnosis according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) and the International Classification of Disease (ICD-10). GAD is assessed using a series of 59 questions (40).

Demographic characteristics included gender, self-identified race/ethnicity (Non-Aboriginal White versus Visible Minority and/or Aboriginal) and age (in decades). Socioeconomic status (SES) was measured by the highest level of achieved education (less than university degree versus post-secondary degree or higher) and household income. The latter was determined based on Statistics Canada's measure of household income as a ratio related to the national low-income cut-off, divided into deciles. This variable takes into account the number of people in the household and the size of the community.

Social support measures included marital status, and having an advisor/confidant available for important decisions. Marital status was measured according to the following categories: Married/common-law versus divorced/widowed/never married. Having a confidant available for important decisions was ascertained using those who agreed or strongly agreed to the statement “There is someone I could talk to about important decisions in my life.”

Religious/spiritual coping was assessed using the question “To what extent do your religious or spiritual beliefs give you strength to face everyday difficulties,” categorized as a little/somewhat/a lot versus not at all.

Health behaviors included obesity, smoking status, and drug and alcohol abuse. Individuals were determined to be obese if they had a BMI greater than or equal to 30. Those who reported ever smoking were considered ever-smokers. Substance abuse/dependence was based on the WHO-CIDI using a combination of the derived variables created by Statistics Canada “Drug Abuse or Dependence (including Cannabis)” and “Alcohol Abuse or Dependence” and coded as yes/no. Both were derived using the lifetime algorithm. For more information, please see Statistics Canada (40).

We included sleep problems, pain, and functional limitations. Individuals were ascertained as having sleep problems if they answered most or all of the time vs none/a little/some of the time in response to “How often do you have trouble going to sleep or staying asleep?” Individuals were considered to be in debilitating pain if they answered that pain prevented a few/some/most activities versus no activities prevented. Individuals were defined as having functional limitations based on the question “In the last 30 days, how much difficulty did you have in taking care of your household responsibilities?” Responses were coded as a binary variable (none versus mild/moderate/severe difficulties).

Adverse childhood experiences (ACES) included witnessing parental domestic violence (PDV), childhood sexual abuse (CSA), and childhood physical abuse (CPA). Individuals were defined as having been exposed to chronic intimate partner violence if they reported they had seen or heard 11 or more times one of their “parents, step-parents or guardians hit each other or another adult” aged 18 or over in their home. CSA was measured by the question “How many times did an adult force you or attempt to force you into any unwanted sexual activity, by threatening you, holding you down or hurting you in some way?” This variable was coded never versus ever. CPA was determined by asking participants if an adult had slapped them on the face, head, or ears, or hit or spanked them with something hard to hurt them at least three times and/or pushed, grabbed, shoved or threw something at them to hurt them at least three times and/or an adult had at least once kicked, bit, punched, choked, burned, or physically attacked them.

Individuals were coded as having a lifetime major depressive disorder if they met the WHO-CIDI lifetime criteria for Major Depressive Episode. For more details please see Statistics Canada (40).

Statistical analyses

Prior to analysis, a weighted multicollinearity matrix was run. All correlations were less than .30. To better understand the potential bias introduced by missing data, we investigated the percentage of anxiety disorders among those with COPD, comparing those with and without complete data. The difference was not significant (p = .48) and therefore we do not believe non-completion biased the responses with respect to those who had started the survey. Two sets of logistic regression analyses were conducted, the first based on those who completed the CCHS-MH who had no missing data on all included measures (n = 11,163) and the second with a sub-sample of this group who had COPD (n = 746).

In the first series, the outcome was anxiety and the key exposure variable was COPD. The first model and each subsequent model controlled for sex, age and race/ethnicity. The second model adjusted for socioeconomic status. The third and fourth models took into account social support and positive coping strategies (e.g., religious values), respectively. The fifth model controlled for health behaviors (smoking, obesity, alcohol/drug abuse or dependence) while the sixth model controlled for sleep problems, pain, and functional limitations. The seventh model included early adversities (witnessing parental domestic violence, being physically abused, and being sexually abused as a child). The eighth model controlled for depression. Finally, the ninth and final model controlled for all the above-name factors simultaneously.

The subsequent logistic regression analysis examined the predictors of anxiety among those with COPD (n = 746). Because the sample of those with COPD was relatively small, we undertook bivariate screening to determine what factors should be included in a final model determined by those variables which reached a significance level at p < 0.20 (analyses not shown). We included the following factors in the model: socioeconomic status, social support (marital status, having a confidant), drug/alcohol abuse, depression, insomnia, pain, and functional limitations and adverse childhood experiences. Data from all the analyses were weighted to adjust for the probability of selection and non-response. Sample sizes are reported in their unweighted form.

Results

One in 17 older adults with COPD (5.8%) reported generalized anxiety disorder (GAD), in comparison to 1.7% of those without COPD (p < .001). The descriptive characteristics of those with COPD versus those without COPD are shown in Table 2. In this sample, those with COPD were more likely to be female, older, white, and poorer compared to those without COPD (p < .01). Those with COPD were also more likely to be obese, depressed, have ever smoked, to experience sleep problems, pain, and/or functional limitations, to have witnessed parental domestic violence and to have experienced sexual abuse, compared to those without COPD (p < .001). Those with COPD were less likely to have a post-secondary degree, be married, or to have a confidant (p < .001).

Table 2. Description of those aged 50+ with COPD and those without COPD (n = 11,163)^a.

	No. COPD	COPD
	n = 10,437 (93.5%)	n = 726 (6.5%)	p-value^b
Generalized Anxiety Disorder Past Year
Yes	1.7%	5.8%	<.001
Demographics
Sex
Male	49.8%	43.8%	<.01
Age
50–59	45.0%	26.9%	<.001
60–69	30.5%	34.5%
70–79	16.8%	25.3%
80+	7.7%	13.3%
Race
White	85.8%	92.5%	<.001
Socioeconomic Status
Education
Post-secondary graduate	58.6%	46.5%	<.001
Household income level
Lowest (bottom 10%)	7.7%	13.8%	<.001
Low-middle (11–49%)	30.7%	41.9%
High-middle (50–79%)	28.7%	27.6%
Highest (80–100%)	32.9%	16.7%
Social Support
Marital status
Married/common-law	72.9%	58.9%	<.001
Advisor/confidant available for important decisions
Yes	97.1%	92.0%	<.001
Positive Coping Strategies
Religion/spirituality gives strength
Strong	61.8%	69.5%	<.001
Health Behaviors
Obesity
Yes (BMI ≥ 30)	22.3%	28.5%	=.001
Smoking status
Ever smoker	68.5%	86.5%	<.001
Drug/alcohol abuse
Yes	20.2%	29.0%	<.001
Sleep, Pain, and Functional Limitations
Sleep problems
Most or all of the time	18.7%	40.5%	<.001
Debilitating pain
Yes	15.9%	38.4%	<.001
Difficulty managing household tasks
Yes	13.7%	27.8%	<.001
Adverse Childhood Experiences
Parental domestic violence
Witnessed 11 or more times	3.7%	8.4%	<.001
Physical abuse
Abused	26.6%	29.2%	=.177
Sexual abuse
Abused	6.4%	13.6%	<.001
Mental Health
Lifetime major depressive disorder
Yes	9.9%	14.4%	=.001

^a Sample sizes are presented in their unweighted form. Percentages are weighted to adjust for the probability of selection and non-response according to Statistics Canada data release guidelines.

^b p-value is derived from a chi-square test.

Source: Canadian Community Health Survey 2012.

Table 3^{Table 3} contains a series of models predicting anxiety, each which include age, race and gender plus a different group of potential explanatory factors. In the first model controlling for the demographics only, those with COPD are shown to have almost four time higher odds of past-year GAD than those without this disorder (OR = 3.90, 95% CI: 2.64, 5.77). Adding in different clusters of factors lowers the odds of GAD among those with COPD modestly.

Table 3. Logistic regression of anxiety among those with copd models 1–9, aged 50+ (n = 11,163).

	Model 1 Demog^a	Model 2 Demog+ SES^b	Model 3 Demog + Social Support	Model 4 Demog + Positive Coping Strategies	Model 5 Demog + Health Behaviors	Model 6 Demog + Sleep, Pain, and Functional Limitations	Model 7 Demog + ACEs^c	Model 8 Demog + Mental Health	Model 9 Fully Adjusted
COPD
No (Ref)	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Yes	3.90 (2.64, 5.77)	3.16 (2.13, 4.70)	3.15 (2.11, 4.71)	3.86 (2.61, 5.70)	3.58 (2.40, 5.34)	1.98 (1.32, 3.00)	3.26 (2.19, 4.87)	3.37 (2.24, 5.07)	1.72 (1.10, 2.71)
Demographics
Sex
Male (Ref)	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Female	1.59 (1.20, 2.09)	1.47 (1.12, 1.95)	1.55 (1.17, 2.05)	1.52 (1.15, 2.02)	2.06 (1.52, 2.79)	1.06 (0.79, 1.42)	1.61 (1.21, 2.15)	1.21 (0.91, 1.61)	1.05 (0.74, 1.48)
Age
Decade (continuous)	0.66 (0.56, 0.78)	0.58 (0.49, 0.68)	0.61 (0.51, 0.72)	0.65 (0.55, 0.77)	0.72 (0.60, 0.85)	0.64 (0.55, 0.76)	0.71 (0.60, 0.84)	0.76 (0.64, 0.91)	0.69 (0.57, 0.83)
Race
White	1.52 (0.97, 2.39)	2.07 (1.31, 3.29)	1.67 (1.06, 2.64)	1.58 (1.00, 2.49)	1.46 (0.92, 2.32)	1.30 (0.82, 2.07)	1.47 (0.93, 2.31)	1.30 (0.82, 2.06)	1.77 (1.08, 2.90)
Visible minority (Ref)	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Socioeconomic Status
Education
No post-secondary degree (Ref)	–	1.00	–	–	–	–	–	–	1.00
Post-secondary graduate	–	0.75 (0.56, 1.01)	–	–	–	–	–	–	0.91 (0.67, 1.25)
Household income
Decile (continuous)	–	0.82 (0.78, 0.86)	–	–	–	–	–	–	0.93 (0.87, 0.98)
Social Support
Marital status
Single/divorced/ widowed	–	–	2.13 (1.61, 2.83)	–	–	–	–	–	1.51 (1.10, 2.07)
Married/common-law (Ref)	–	–	1.00	–	–	–	–	–	1.00
Advisor/confidant available for important decisions
No	–	–	6.07 (4.11, 8.96)	–	–	–	–	–	3.09 (1.96, 4.87)
Yes (Ref)	–	–	1.00	–	–	–	–	–	1.00
Positive Coping Strategies
Religion/spirituality gives strength
Not at all	–	–	–	0.80 (0.60, 1.08)	–	–	–	–	0.81 (0.58, 1.12)
A little/somewhat/a lot (Ref)	–	–	–	1.00	–	–	–	–	1.00
Health Behaviors
(Continued)

Table 3.  (Continued)

Obesity
No (Ref)	–	–	–	–	1.00	–	–	–	1.00
Yes (BMI ≥ 30)	–	–	–	–	1.37 (1.02, 1.85)	–	–	–	0.96 (0.70, 1.33)
Smoking status
Never (Ref)	–	–	–	–	1.00	–	–	–	1.00
Ever smoker	–	–	–	–	0.58 (0.43, 0.79)	–	–	–	0.54 (0.39, 0.75)
Drug/Alcohol Abuse
No (Ref)	–	–	–	–	1.00	–	–	–	1.00
Yes	–	–	–	–	3.03 (2.19, 4.19)	–	–	–	1.74 (1.22, 2.47)
Sleep, Pain, and Functional Limitations
Sleep problems
Most or all of the time (Ref)	–	–	–	–	–	1.00	–	–	1.00
None or some of the time	–	–	–	–	–	2.78 (2.08, 3.71)	–	–	1.80 (1.31, 2.46)
Pain prevents activities
No (Ref)	–	–	–	–	–	1.00	–	–	1.00
Yes	–	–	–	–	–	2.12 (1.54, 2.91)	–	–	1.71 (1.22, 2.38)
Difficulty managing household tasks
No (Ref)	–	–	–	–	–	1.00	–	–	1.00
Yes	–	–	–	–	–	4.27 (3.09, 5.88)	–	–	3.52 (2.52, 4.93)
Adverse Childhood Experiences
Parental Domestic Violence
Never witnessed or witnessed ≤ 10 times (Ref)	–	–	–	–	–	–	1.00	–	1.00
Witnessed 11 or more times	–	–	–	–	–	–	1.47 (0.91, 2.37)	–	1.07 (0.62, 1.85)
Physical Abuse
None (Ref)	–	–	–	–	–	–	1.00	–	1.00
Abused	–	–	–	–	–	–	1.87 (1.38, 2.53)	–	1.40 (1.00, 1.95)
Sexual Abuse
None (Ref)	–	–	–	–	–	–	1.00	–	1.00
Abused	–	–	–	–	–	–	2.02 (1.37, 2.96)	–	0.91 (0.58, 1.41)
Mental Health
Lifetime major depressive disorder
No (Ref)	–	–	–	–	–	–	–	1.00	1.00
Yes	–	–	–	–	–	–	–	9.15 (6.90, 12.13)	5.11 (3.76, 6.95)
Nagelkerke R Square Value	.037	.066	.086	.038	.063	.167	.062	.144	.270
−2 Log Likelihood	2070	2013	1973	2068	2019	1811	2020	1859	1603

^a Demog = Demographics.

^b SES = Socioeconomic Status.

^c ACEs = Adverse Childhood Experiences.

The largest reduction is seen when sleep problems, pain, and functional limitations are included in the model (model 6: OR = 1.98, 95% CI: 1.32, 3.00). This is followed by social support (model 3: OR = 3.15, 95% CI: 2.11, 4.71) and SES (model 2: OR = 3.16, 95% CI: 2.13, 4.70). After these factors, the next largest reduction in odds is observed when ACEs are accounted for (model 7: OR = 3.26, 95% CI: 2.19, 4.81), followed by depression (model 8: OR = 3.37, 95% CI: 2.24, 5.07) and health behaviors (model 5: OR = 3.58, 95% CI: 2.40, 5.34). The COPD-anxiety relationship is attenuated least when accounting for religious coping (model 4: OR = 3.86, 95% CI: 2.61, 5.70). In the last model that controls for all the variables simultaneously, the odds of past-year GAD among those with COPD remain highly significant, but are reduced to 1.72 (95% CI: 1.10, 2.71).

The second part of the analysis focuses on identifying the significant independent correlates of past-year GAD among those with COPD. Table 4 presents the factors in the bivariate analyses (analyses not shown) that were associated with past-year GAD at p < .20. As can be seen, the strongest independent correlate of GAD, based on magnitude, is not having a confidant (OR = 7.85, 95% CI: 3.47, 17.75), followed by having had witnessed parental domestic violence (OR = 5.63, 95% CI: 2.07, 15.34), current depression (OR = 3.59, 95% CI: 1.61, 7.98) and being unmarried (OR = 2.28, 95% CI: 1.06, 4.89).

Table 4. Logistic regression of current anxiety among those with COPD, aged 50+ (n = 746).

	Model 1 Fully Adjusted
Socioeconomic Status
Household income
Decile (continuous)	0.91 (0.78, 1.06)
Social Support
Marital Status
Single/divorced/widowed	2.28 (1.06, 4.89)
Married/common-law (Ref)	1.00
Advisor/confidant available for important decisions
No	7.85 (3.47, 17.75)
Yes (Ref)	1.00
Health Behaviors
Drug/alcohol abuse
No (Ref)	1.00
Yes	1.52 (0.73, 3.17)
Sleep, Pain, and Functional Limitations
Sleep problems
None or some of the time	1.11 (0.52, 2.39)
Most or all of the time (Ref)	1.00
Debilitating pain
No (Ref)	1.00
Yes	1.82 (0.85, 3.87)
Difficulty managing household tasks
No (Ref)	1.00
Yes	1.43 (0.67, 3.04)
Adverse Childhood Experiences
Parental Domestic Violence
Never witnessed or witnessed ≤ 10 times (Ref)	1.00
Witnessed 11 or more times	5.63 (2.07, 15.34)
Physical Abuse
None (Ref)	1.00
Abused	0.55 (0.22, 1.41)
Sexual Abuse
None (Ref)	1.00
Abused	0.58 (0.19, 1.74)
Depression
Lifetime major depressive disorder
No (Ref)	1.00
Yes	3.59 (1.61, 7.98)
Nagelkerke R Square Value	.283
−2 Log Likelihood	246.81

Discussion

The current study shows that COPD is associated with 70% higher odds of past-year generalized anxiety disorder (GAD) in a population-based sample of Canadians, even when controlling for eighteen potentially explanatory factors. When controlling for demographics only, those with COPD are shown to have almost four times higher odds of GAD than those without COPD. The rate from our Canadian study is comparable to that of other population-based studies conducted in the Netherlands, for example Wagena, Ludovic, van Amelsvoort, Kant, and Wouters (42), who found the odds of anxiety was 5 times higher among those with COPD compared to those without. To our knowledge, no other population-based Canadian study has assessed the association between COPD and anxiety or associated predictors.

We found sleep problems, pain, and functional limitations explained the largest share of the relationship between COPD and past-year GAD. With regards to pain, these results are consistent with previous reports that 45% of COPD patients experience pain, many of whom rate their pain as severe or very severe (14). Research has identified shared pathophysiological mechanisms for the processing of chronic disease pain and anxiety (43). Individuals may respond to chronic pain by misinterpreting this as a sign of imminent death, thereby exacerbating anxiety (43).

Ohayon (15) found a higher prevalence of co-morbid anxiety and insomnia among adults with COPD compared to those without. Furthermore, this co-morbidity increased the likelihood of hospitalizations (44), indicating the importance of assessing for both anxiety and insomnia among COPD patients to increase health-related quality of life and decrease avoidable healthcare system costs. In addition, COPD disease progression is associated with decreased physical functioning and subsequent disability (16, 17). Although respiratory impairment is most often discussed, COPD patients may also experience non-respiratory impairment and functional limitations that result in disability (17). As impairment increases, anxiety is also likely to increase (20), which may further prevent peoples' abilities to engage in activities of daily living. However, the fact that the odds of anxiety decrease significantly when pain, insomnia and disability are accounted for within our sample underlines the importance for healthcare providers to assess and treat these issues in addition to the more well-known COPD symptoms.

Social support accounted for the next largest reduction in odds of past-year GAD observed. COPD can result in social isolation, particularly for older adults with functional limitations (16). Social isolation, in turn, exacerbates anxiety (45). It is plausible that addressing functional limitations may help patients to engage in social activities and access more extensive social support.

The third-largest reduction occurred when socioeconomic status was added to the model. Rates of exposure to second hand smoke and to air pollution are significantly higher for people with lower income than the rest of the population (46), contributing to increased risk of COPD. Concomitantly, people with low socioeconomic status experience higher levels of anxiety (36, 37), possibly due to the strain caused by inadequate financial resources, a problem which may be exacerbated by health related costs associated with COPD.

When health behaviours were added to the model, the direct association between COPD and anxiety was only modestly attenuated (from OR = 3.90 to OR = 3.58). These results are surprising given the well-known association between smoking cigarettes and the onset of anxiety disorders (42) and between smoking and reduced lung functioning (8), which may further exacerbate anxiety symptoms.

Despite controlling for 18 factors, we showed that the odds of anxiety were still 1.72 times higher among those with COPD compared to those without. Both biological and psychological explanations for a direct association have been posited (9). From a biological perspective, the anxiety-provoking symptom of dyspnea may play a role in the development of anxiety disorders, particularly panic disorder (9). Studies suggest a common brain area for the processing of both the affective dimension of dyspnea and fear, anxiety, and pain (47, 48). From a psychological perspective, Clark's seminal model proposes that panic and anxiety result from a misinterpretation of certain bodily sensations such as laboured breathing (49). As patients experience progressively worse symptoms, they may experience increasingly catastrophic thoughts, resulting in increased risk of anxiety (50–53). Future studies should assess disease severity in order to enhance understanding of biological-psychological complexities.

The final analysis focused on identifying the significant correlates of past-year GAD among those with COPD. We found that not having a confidant is associated with almost 8 times the odds of having anxiety. Consistent with other studies (54), these findings suggest that screening for anxiety may be particularly important among patients who lack a strong social network. The actual social network size of people with COPD may be hindered by the aforementioned functional limitations, which are also linked to decreased engagement in social activities (16). Our findings also suggest that social network size is not the only risk factor for anxiety among COPD patients, but that the quality of the relationships matters greatly as measured by whether or not the patient perceives that their social support contains at least one confidant. This finding is consistent with DiNicola and colleague's study (55) assessing the types of social support associated with anxiety among 452 COPD patients. They found that the perceived use of unsympathetic and insensitive responses by social network members and the perception that others might let the patient down were both independently associated with anxiety among COPD patients.

Childhood physical or sexual abuse did not have a significant effect on the odds of GAD among respondents with COPD. This lack of association may be due to the fact we included depressive disorders in the analysis. Previous research in the general population found that the link between childhood physical abuse and adult anxiety disorders declined to the point that it was no longer significant once depression was included in the analysis (25). This suggests that childhood maltreatment led to depression, and it was the depression that led to anxiety disorders, rather than a direct relationship between the original physical abuse and anxiety. In contrast, witnessing parental domestic violence more than 10 times during one's childhood was associated with increased odds of having anxiety among those with COPD. The chronic chaotic and violent home environment may have predisposed individuals to anxiety disorders in adulthood (56). Further research is needed to understand the pathways through which witnessing chronic parental domestic violence during the respondent's childhood may impact the development of anxiety disorders among adults with COPD.

Depression was associated with 3.6 times the odds of anxiety among those with COPD. These findings are not surprising given the well-established co-morbidity of anxiety and depression (27, 28). Clearly, considering co-morbid depression is essential in all studies examining anxiety among those with COPD. Unfortunately, the large number of questions used to derive lifetime depression in this study impedes its adoption in short surveys. Future research should consider promising shorter standardized tools with COPD patients (e.g., the Primary Care Evaluation of Mental Disorders (PRIME-MD) (57)). A few studies have assessed the accuracy of standard depression screening tools with COPD patients (e.g., Beck Depression Inventory (BDI), Short-Form Questionnaire (58); 15-item Geriatric Depression Scale (GDS) (59)) finding them to be accurate. These results are promising and suggest the viability of including short but valid depression and anxiety tools in a wider range of studies on COPD.

Limitations

Findings of this study must be interpreted with caution due to several limitations. The CCHS-MH utilized a self-reported measure of COPD, which probably resulted in an under-reporting of COPD. A Canadian study assessing the difference in prevalence between reported diagnosis of COPD versus assessed COPD measuring obstructed airflow found the latter had two to six times higher prevalence than self-reported diagnosis of COPD, supporting that COPD symptoms are under-recognized and under-diagnosed (60). Under-reporting of COPD would bias our report towards the null. There is some suggestion that those with the most severe anxiety are least likely to respond to surveys such as the CCHS- MH (61). However, we have no reason to believe that this bias would be more prevalent among those with COPD than those without.

Given that this is a cross-sectional study, it is impossible to determine a temporal order between the onset of COPD and the onset of anxiety. However, by using a measure of past-year anxiety rather than lifetime anxiety, we increase the chance that COPD (which had to be diagnosed for a period of 6 months or more) preceded the current anxiety symptoms.

There are some potentially important explanatory factors that were not measured using the CCHS-MH dataset, including breathlessness, illness severity, symptom severity, the presence/absence of highly stress inducing co-morbidities such as lung cancer and other stressful events such as job loss, bereavement and family crises. The lack of measurement of breathlessness in the CCHS-MH is a primary limitation. Breathlessness may increase anxiety, as it is often a marker of disease progression for those with COPD (50). Additionally, panic attacks can increase breathlessness which may further exacerbate COPD symptoms (51–53). Although most anxiety was accounted for by clusters of other factors in both analyses, it is possible that some of the remaining variance could be attributed to breathlessness. Standardized measures have been developed to assess anxiety associated with breathlessness for COPD patients [e.g., The Dyspnea Management Questionnaire (51)]. Future studies should use such standardized measures to take into account dyspnea/breathlessness-related anxiety and/or assess breathlessness among patients with COPD in order to carefully elucidate this association.

Illness and symptom severity indicate disease progression, increased trouble breathing, and potentially concomitant increased anxiety. Future studies should include these measures to understand how and/or if different factors are associated with anxiety at different states of disease progression. Additionally, lung cancer, in particular, is more common among COPD patients due to the common exposure of smoking and increased cancer susceptibility after damage to the lungs caused by the chronic inflammation associated with COPD (62). Population-based studies have shown that anxiety is associated with lung cancer (63) therefore, including this factor is important for future studies. Despite these limitations, this study has several strengths: the CCHS is a population-based national probability sample, which utilizes a strong and valid measure of current anxiety disorders and the study includes a large number of known risk factors for anxiety.

Conclusions

One in every 17 older adults with COPD have a current anxiety disorder. Those with COPD have substantially higher odds of GAD even after accounting for most of the known risk factors for GAD. However, these results also suggest that a variety of controllable factors may partially attenuate the relationship between COPD and anxiety, particularly pain and disability, suggesting that healthcare providers may play a significant role in reducing anxiety for this population by screening for and addressing pain and functional limitations. The importance of a multidisciplinary healthcare team is underscored by the multiple issues (e.g., pain, functional limitations, co-morbid mental health, social isolation) that may impact GAD among COPD patients.

Acknowledgments

The authors would like to thank Rachel Max for her help with the review of the COPD literature and Sarah Brennenstuhl for her help with manuscript preparation.

Funding

Esme Fuller-Thomson would like to gratefully acknowledge support received from the Sandra Rotman Endowed Chair at the University of Toronto. This research was undertaken, in part, thanks to this support.

Declaration of interest statement

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.