Comorbid Anxiety and Depression, Though Underdiagnosed, Are Not Associated with High Rates of Low-Value Care in Patients with Chronic Obstructive Pulmonary Disease

Abstract

Rationale: Patients with chronic obstructive pulmonary disease (COPD) and anxiety or depression experience more symptoms and exacerbations than patients without these comorbidities. Failure to provide beneficial COPD therapies to appropriate patients (underuse) and provision of potentially harmful therapies to patients without an appropriate indication (overuse) could contribute to respiratory symptoms and exacerbations. Anxiety and depression are known to affect the provision of health services for other comorbid conditions; therefore, underuse or overuse of therapies may explain the increased risk of severe symptoms among these patients.

Objectives: To determine whether diagnosed anxiety and depression, as well as significant anxiety and depression symptoms, are associated with underuse and overuse of appropriate COPD therapies.

Methods: We analyzed data from a multicenter prospective cohort study of 2,376 participants (smokers and control subjects) enrolled between 2010 and 2015. We identified two subgroups of participants, one at risk for inhaled corticosteroid (ICS) overuse and one at risk for long-acting bronchodilator (LABD) underuse based on the 2011 Global Initiative for Chronic Obstructive Lung Disease statement. Our primary outcomes were self-reported overuse and underuse. Our primary exposures of interest were self-reported anxiety and depression and significant anxiety and depression symptoms. We adopted a propensity-score method with inverse probability of treatment weighting adjusting for differences in prevalence of confounders and performed inverse probability of treatment weighting logistic regression to evaluate all associations between the exposures and outcomes.

Results: Among the 1,783 study participants with COPD confirmed by spirometry, 667 (37.4%) did not have an indication for ICS use, whereas 985 (55.2%) had an indication for LABD use. Twenty-five percent (n = 167) of patients reported ICS use, and 72% (n = 709) denied LABD use in each subgroup, respectively. Neither self-reported anxiety and depression nor significant anxiety and depression symptoms were associated with overuse or underuse. At least 50% of patients in both subgroups with significant symptoms of anxiety or depression did not report a preexisting mental health diagnosis.

Conclusions: Underuse of LABDs and overuse of ICSs are common but are not associated with comorbid anxiety or depression diagnosis or symptoms. Approximately one-third of individuals with COPD experience anxiety or depression, and most are undiagnosed. There are significant opportunities to improve disease-specific and patient-centered treatment for individuals with COPD.

Depression and anxiety are common among patients with chronic obstructive pulmonary disease (COPD), affecting between 10% (1) and 50% (2) of such patients. Individuals with COPD and comorbid anxiety or depression have an increased risk of exacerbation, hospitalization, and death (1–12). Although there are several potential mechanisms for this association, including increased susceptibility to illness, increased systemic inflammation, and decreased self-efficacy, the simplest explanation of this observation may be that individuals with COPD and comorbid mental health conditions receive lower-quality care than those who do not (7, 13, 14).

Overall, care quality for patients with COPD is known to be poor, with many patients receiving treatments that constitute overuse of inappropriate inhaled therapies (in whom the potential harm outweighs the likely benefit) or underuse of potentially beneficial therapies (15–19). Overuse and underuse are forms of low-value care that lead to increased medication-related adverse events and waste and to increased use of health resources from unnecessary morbidity and hospitalization. Patients with COPD who suffer from comorbid anxiety or depression may be at particular risk of being prescribed potentially inappropriate medications, such as inhaled corticosteroids (ICSs) for individuals without history of recent exacerbation, to treat somatic symptoms of their mental health diagnoses, such as fatigue, anxiety, and chest tightness. Similarly, patients with COPD and comorbid anxiety or depression may be less likely to receive a complete evaluation of their symptoms or engage in motivational interviewing with their providers to ensure medication adherence (20, 21). Prior studies of the association between anxiety and depression and care quality for other acute and chronic conditions have shown that patients with mental health comorbidities are more likely to receive inappropriate treatments that constitute overuse, while also experiencing treatments that constitute underuse of potentially beneficial treatments and screenings (22–26). Understanding the role, if any, of diagnosed or undiagnosed mental health comorbidities in the treatment of COPD could help facilitate development of interventions to improve outcomes for this patient population.

We sought to evaluate whether individuals with depression or anxiety were more likely to experience low-value care through use of medications that confer more harm than benefit (e.g., ICSs among those without a history of exacerbation) or failure to receive appropriate care to mitigate COPD symptoms and prevent exacerbation (e.g., long-acting bronchodilators [LABDs] among those with significant daily symptoms) than those without depression or anxiety. Because many individuals suffer from mental health conditions that are not formally diagnosed, we evaluated the associations of patient-reported overuse and underuse with both patient-reported diagnoses of anxiety or depression and significant anxiety or depression symptoms.

Methods

Data Source and Study Design

We performed a cross-sectional secondary analysis of the baseline data collected for SPIROMICS (Subpopulations and Intermediate Outcome Measures in COPD Study), a multicenter prospective observational study that enrolled participants between 2010 and 2015 and followed them for up to 3 years (27).

SPIROMICS Enrollment and Participation

SPIROMICS enrolled 2,736 participants between 40 and 80 years of age who were either nonsmokers without airflow obstruction or current or former smokers with a smoking history of at least 20 pack-years from 2010 through 2015 (28). Participants did not require a clinical diagnosis of COPD to enroll in the study. During the enrollment encounter, all participants were asked to provide information about medications, medical history, and exacerbations of COPD via questionnaire. Participants completed multiple symptom assessments and disease screenings with the assistance of research coordinators. These assessments included the COPD Assessment Test (CAT) to quantify severity of COPD symptoms and the Hospital Anxiety and Depression Scale (HADS) to screen for anxiety and depression. All participants completed spirometric testing during the enrollment visit to enable evaluation of obstructive lung disease.

Study Populations

We identified study participants with COPD confirmed by airflow obstruction at spirometry (post-bronchodilator ratio of forced expiratory volume in 1 second [FEV₁] to forced vital capacity <0.70) and who complete the CAT and HADS assessments during their enrollment encounter to evaluate the association between comorbid anxiety and depression and appropriateness of COPD pharmacotherapy. We then created two subgroups of participants to evaluate potential overuse and potential underuse.

Potential overuse of ICSs

We identified a subgroup of participants who did not have an appropriate indication for ICS use based on the 2007 or 2011 Global Initiative for Chronic Obstructive Lung Disease (GOLD) statements, which were the statements available during the study period (29, 30). Appropriate indications included self-reported history of asthma, post-bronchodilator FEV₁ < 50% predicted or self-reported history of inpatient or outpatient exacerbation in the year before enrollment. Participants who affirmed that they had experienced an “episode of breathing problems” in the past year were considered as having had an exacerbation, regardless of treatment received, to minimize misclassification of patients appropriate for ICS use.

Potential underuse of LABDs

To evaluate underuse, we created another subgroup of participants who would be expected to benefit from LABD therapy (either long-acting muscarinic-antagonist or long-acting β-agonist therapy) because of significant self-reported dyspnea (CAT score $\ge$ 10) (31). SPIROMICS participants could belong to one subgroup, both subgroups, or neither subgroup. We excluded participants belonging to neither subgroup from the study.

Primary Outcome

The primary outcome of low-value COPD pharmacotherapy was defined as self-reported use of ICSs among participants without appropriate indication for ICS use (overuse) or failure to report use of LABDs among participants with an appropriate indication for LABD use (underuse).

Primary Exposure

The primary exposure of interest was a self-reported diagnosis of anxiety and depression or significant anxiety and depression symptoms without a self-reported diagnosis. We identified participants with a self-reported diagnosis of anxiety or depression as those with a diagnosis made by a medical provider for either condition. Participants were not asked about more specific diagnoses (e.g., major depressive disorder, generalized anxiety disorder, etc. . . .) and were not asked if they were receiving pharmacologic or behavioral therapy. We identified participants with significant anxious or depressive symptoms using the HADS assessment tool. We considered a score of 8 or higher for either the depression or anxiety subscores on the HADS as a positive screen result. We chose the cutoff score of 8 because this threshold has been shown to perform well in identifying mood disorders among patients with and without COPD (32, 33). We believed that patients who reported a known diagnosis of anxiety or depression could be significantly different from those with significant anxious or depressive symptoms who did not carry a diagnosis, so we chose to evaluate these two exposures separately in the study subgroups.

Potential Confounders

We identified income, race, marital status, education, age, and obesity as potential confounders of the relationship between mood disorders and appropriate inhaled medication use and accounted for these variables in our analysis (34–40).

Statistical Analysis

We performed separate analyses in the potential overuse and potential underuse subgroups to evaluate the association between depression and anxiety and receipt of low-value care. We first calculated descriptive statistics of the study cohort by performing univariate analyses to determine differences between participant groups with and without the primary outcomes. We considered exposures with a standardized mean difference (SMD) of greater than 0.2 between patients with and without the outcome as having at least a minor effect (41). We then examined the association between both diagnosed anxiety and depression and screened anxiety and depression and the outcome for each subgroup. To remove the effects of confounding when estimating the effects of diagnosed anxiety and depression and treatment decision, we adopted a propensity-score method with inverse probability of treatment weighting (IPTW) to adjust for differences in the prevalence of potential confounders between groups of participants who had received a diagnosis of anxiety or depression to more accurately estimate the effect of that exposure on the outcome of interest for each subgroup (average treatment effect on treated) (42). For analysis of diagnosed anxiety or depression and appropriate treatment decision, we applied propensity-score weights to adjust SMDs less than 0.1 for all potential confounders indicating balance between exposure groups. We used an IPTW logistic regression to evaluate the association of diagnosed anxiety and depression with the primary outcome for each subgroup to obtain adjusted odds ratios (ORs). We used the same logistic regression model without IPTW to evaluate the relationship between screened anxiety and depression and each outcome.

Results

The SPIROMICS study enrolled 2,877 individuals, including tobacco users and nonsmoker control subjects. The cohort included 1,783 individuals with COPD confirmed by spirometry who were eligible for inclusion in our potential overuse and potential underuse cohorts.

Potential Overuse of ICSs Subgroup

FEV₁ > 50% predicted, no recent history of exacerbation, no history of asthma, n = 657

Among the 1,783 participants enrolled in the SPIROMICS trial, with COPD confirmed by spirometry, we identified 657 (36.8%) without indication for ICS use (Figure 1A). Participants in this subgroup were predominantly white (87.8%), male (65.9%), and over age 65 (mean, 66.4; standard deviation [SD], 7.6) (Table 1). The mean FEV₁ was 75.7% predicted (SD, 16.3), and the mean CAT score was 12.5 (SD, 7.6).

Figure 1.

(A) Development of the potential overuse subgroup. We identified all patients with chronic obstructive pulmonary disease (COPD) confirmed by airflow obstruction at spirometry with no appropriate indication for an inhaled corticosteroid based on the 2011 Global Initiative for Chronic Obstructive Lung Disease statement. (B) Development of the potential underuse subgroup. We identified all patients with COPD confirmed by airflow obstruction on spirometry with an appropriate indication for a long-acting bronchodilator based on the 2011 Global Initiative for Chronic Obstructive Lung Disease statement. FEV₁ = forced expiratory volume in 1 second; SPIROMICS = Subpopulations and Intermediate Outcome Measures in COPD Study

Table 1.

Characteristics of study participants

Subgroup Characteristics	Overall
Potential overuse subgroup (n = 657)
Age, mean (SD)	66.35 (7.63)
Sex, male, n (%)	433 (65.9)
Race, white, n (%)	577 (87.8)
FEV1% predicted, mean (SD)	75.68 (16.30)
Obesity, BMI ≥ 30, n (%)	182 (27.7)
Income < 50k, n (%)	369 (56.2)
Education, less than high school, n (%)	57 (8.7)
Marital status, married, n (%)	338 (51.4)
Current smoker, n (%)	252 (38.4)
Screened mood disorder,*n (%)	179 (27.2)
Anxiety	153 (23.3)
Depression	99 (15.0)
Self-reported mood disorder, n (%)	184 (28.0)
Anxiety	107 (16.3)
Depression	152 (23.1)
CAT, mean (SD)	12.48 (7.55)
Potential underuse subgroup (n = 967)
Age, mean (SD)	64.26 (7.94)
Sex, male, n (%)	536 (55.4)
Race, white, n (%)	788 (81.5)
FEV1% predicted, mean (SD)	57.16 (22.10)
Obesity, BMI ≥ 30, n (%)	283 (29.3)
Income < 50k, n (%)	651 (67.3)
Education, less than high school, n (%)	116 (12.0)
Marital status, married, n (%)	461 (47.7)
Current smoker, n (%)	370 (38.3)
Screened mood disorder,*n (%)	390 (40.3)
Anxiety	311 (32.2)
Depression	245 (25.3)
Self-reported mood disorder, n (%)	338 (35.0)
Anxiety	215 (22.2)
Depression	274 (28.3)
CAT, mean (SD)	18.75 (6.23)

Definition of abbreviations: 50k = $50,000; BMI = body mass index; CAT = COPD Assessment Test; COPD = chronic obstructive pulmonary disease; FEV₁ = forced expiratory volume in 1 second; GOLD = Global Initiative for Chronic Obstructive Lung Disease; HADS = Hospital Anxiety and Depression Scale; ICS = inhaled corticosteroid; SD = standard deviation.

Patients in the potential overuse subgroup had no appropriate indication for an ICS use based on the 2007 and 2011 GOLD statements. Appropriate indications included a history of asthma, a COPD exacerbation in the past year, and/or a post-bronchodilator FEV₁ < 50% predicted. Patients in the potential underuse subgroup had an appropriate indication for a long-acting bronchodilator use based on the 2011 GOLD statement. All patients reported a CAT score of 10.

^*HADS score ≥ 8 for total score, anxiety-specific score, or depression-specific score.

One-quarter of participants in this subgroup screened positive for either anxiety or depression using the HADS (n = 179, 27.2%), and a similar number reported a preexisting diagnosis of either condition (n = 184, 28.0%). Approximately 30% of participants either had a positive screen result or a self-reported a diagnosis of anxiety or depression (n = 203, 30.9% for anxiety; n = 201, 30.6% for depression; Table 2). Among these participants, 23.3% (n = 153) screened positive for anxiety and 16.3% (n = 107) reported an existing diagnosis, whereas 15.1% (n = 99) screened positive for depression and 23.1% (n = 152) reported an existing depression diagnosis. The majority of participants who screened positive for anxiety (62.7%) did not report an existing diagnosis of anxiety, indicating that one in seven (14.6%) participants in the potential overuse subgroup had a potentially undiagnosed anxiety disorder. Half (49.5%) of participants who screened positive for depression did not report an existing diagnosis of depression, indicating that 1 in 13 (7.5%) participants in the potential overuse subgroup had a potentially undiagnosed depressive disorder.

Table 2.

Agreement between self-reported diagnosis and positive screen result for anxiety or depression

	Positive Screen Result	Negative Screen Result
Patients without indication for an ICS according to the 2007 and 2011 GOLD statements (n = 657)
Anxiety
Self-reported anxiety diagnosis	57	50
No self-reported anxiety diagnosis	96	454
Depression
Self-reported depression diagnosis	50	102
No self-reported depression diagnosis	49	456
Patients with severe daily dyspnea (CAT score = 10) (n = 967)
Anxiety
Self-reported anxiety diagnosis	116	99
No self-reported anxiety diagnosis	195	557
Depression
Self-reported depression diagnosis	122	152
No self-reported depression diagnosis	123	570

Definition of abbreviations: CAT = COPD Assessment Test; COPD = chronic obstructive pulmonary disease; GOLD = Global Initiative for Chronic Obstructive Lung Disease; ICS = inhaled corticosteroid.

One out of four participants in this subgroup (n = 165, 25.1%) reported regular use of an ICS, despite reporting no appropriate indication (Table 3). Patients with overuse of ICSs had lower lung function (FEV₁% predicted: 64.8% vs. 78.3%; SMD, 0.68), more severe dyspnea (CAT score: 15.6 vs. 11.4; SMD, 0.59), and were more likely to identify as white (88.5 vs. 87.6; SMD, 0.23) than participants without overuse. Neither self-reported, screened, or potentially undiagnosed anxiety or depression were associated with overuse of ICSs (Table 4). IPTW reduced SMDs to 0.1 or less between exposure groups among those diagnosed with anxiety and depression for calculation of the adjusted ORs (see Figures E1A and E2A in the online supplement). One out of three individuals with overuse of ICSs (31.5%, n = 52) reported ICS monotherapy (Table E1A).

Table 3.

Participant characteristics by overuse of ICSs and underuse of LABDs

	Overall	Overuse or Underuse	No Overuse or Underuse	SMD
Overuse of ICSs in participants without an indication for an ICS according to the 2007 and 2011 GOLD statements
n (%)	657	165 (25.1)	492 (74.9)
Age, mean (SD)	66.35 (7.63)	65.61 (7.63)	66.60 (7.62)	0.130
Sex, male, n (%)	433 (65.9)	107 (64.8)	326 (66.3)	0.030
Race, white, n (%)	577 (87.8)	146 (88.5)	431 (87.6)	0.229
FEV1% predicted, mean (SD)	75.68 (16.30)	67.96 (14.18)	78.27 (16.15)	0.678
Obesity, BMI ≥ 30, n (%)	182 (27.7)	54 (32.7)	128 (26.0)	0.148
Income < 50k, n (%)	369 (56.2)	97 (58.8)	272 (55.3)	0.071
Education, less than high school, n (%)	57 (8.7)	12 (7.3)	45 (9.1)	0.068
Marital status, married, n (%)	338 (51.4)	87 (52.7)	251 (51.0)	0.034
Current smoker, n (%)	252 (38.4)	54 (32.7)	198 (40.2)	0.157
Screened mood disorder,*n (%)	179 (27.2)	42 (25.5)	137 (27.8)	0.054
Anxiety	153 (23.3)	36 (21.8)	117 (23.8)	0.047
Depression	99 (15.0)	29 (17.6)	70 (14.2)	0.092
Self-reported mood disorder, n (%)	184 (28.0)	49 (29.7)	135 (27.4)	0.050
Anxiety	107 (16.3)	29 (17.6)	78 (15.9)	0.046
Depression	152 (23.1)	40 (24.2)	112 (22.8)	0.035
CAT, mean (SD)	12.48 (7.55)	15.62 (7.07)	11.41 (7.42)	0.589
Underuse of LABDs in participants with an indication for an LABD according to the 2011 GOLD statement
n (%)	967	269 (27.8)	698 (72.2)
Age, mean (SD)	64.26 (7.94)	64.35 (7.93)	64.23 (7.95)	0.016
Sex, male, n (%)	536 (55.4)	149 (55.4)	387 (55.4)	0.001
Race, white, n (%)	788 (81.5)	211 (78.4)	577 (82.7)	0.173
FEV1% predicted, mean (SD)	57.16 (22.10)	49.18 (19.89)	60.24 (22.15)	0.525
Obesity, BMI ≥ 30, n (%)	283 (29.3)	75 (27.9)	208 (29.8)	0.042
Income < 50k, n (%)	651 (67.3)	180 (66.9)	471 (67.5)	0.012
Education, less than high school, n (%)	116 (12.0)	31 (11.5)	85 (12.2)	0.020
Marital status, married, n (%)	461 (47.7)	131 (48.7)	330 (47.3)	0.028
Current smoker, n (%)	370 (38.3)	73 (27.1)	297 (42.6)	0.328
Screened mood disorder,*n (%)	390 (40.3)	103 (38.3)	287 (41.1)	0.058
Anxiety	311 (32.2)	77 (28.6)	234 (33.5)	0.106
Depression	245 (25.3)	71 (26.4)	174 (24.9)	0.034
Self-reported mood disorder, n (%)	338 (35.0)	93 (34.6)	245 (35.1)	0.011
Anxiety	215 (22.2)	67 (24.9)	148 (21.2)	0.088
Depression	274 (28.3)	73 (27.1)	201 (28.8)	0.037
CAT, mean (SD)	18.75 (6.23)	19.74 (6.67)	18.36 (6.02)	0.217
Hospitalized for exacerbation in year before enrollment, n (%)	110 (11.4)	42 (15.6)	68 (9.7)	0.177

Definition of abbreviations: 50k = $50,000; BMI = body mass index; CAT = COPD Assessment Test; COPD = chronic obstructive pulmonary disease; FEV₁ = forced expiratory volume in 1 second; GOLD = Global Initiative for Chronic Obstructive Lung Disease; HADS = Hospital Anxiety and Depression Scale; ICS = inhaled corticosteroid; LABD = long-acting bronchodilator; SD = standard deviation; SMD = standardized mean difference.

Participants experiencing overuse of ICSs had lower FEV₁ values and more COPD-specific symptoms. Participants experiencing underuse of LABDs had high post-bronchodilator FEV₁ values but similar COPD-specific symptom scores.

^*HADS score ≥ 8 for total score, anxiety-specific score, or depression-specific score.

Table 4.

Neither self-reported nor screened anxiety and depression are associated with overuse of ICSs or underuse of LABDs

	OR (95% CI)	P Value	AOR (95% CI)	P Value
Overuse of ICSs in participants without indication for ICS use
Self-reported mood disorder	1.12 (0.76–1.65)	0.58	1 (0.63–1.58)	0.99
Self-reported anxiety	1.13 (0.71–1.81)	0.60	1.03 (0.62–1.71)	0.90
Self-reported depression	1.09 (0.72–1.64)	0.70	0.97 (0.61–1.55)	0.90
Screened mood disorder	0.89 (0.59–1.37)	0.61	0.92 (0.57–1.49)	0.75
Screened anxiety	0.91 (0.6–1.39)	0.66	0.91 (0.56–1.5)	0.72
Screened depression	1.29 (0.8–2.07)	0.30	1.38 (0.81–2.36)	0.24
Screened positive without self-reported diagnosis	0.89 (0.52–1.52)	0.67	1.21 (0.61–2.39)	0.59
Underuse in participants with indication for LABD use
Self-reported mood disorder	0.98 (0.73–1.31)	0.8	1.1 (0.78–1.54)	0.60
Self-reported anxiety	1.23 (0.89–1.72)	0.22	1.39 (0.97–2)	0.07
Self-reported depression	0.92 (0.67–1.26)	0.61	1.0 (0.71–1.41)	0.99
Screened mood disorder	0.89 (0.67–1.19)	0.42	0.91 (0.65–1.28)	0.58
Screened anxiety	0.8 (0.58–1.08)	0.14	0.79 (0.55–1.13)	0.20
Screened depression	1.08 (0.78–1.49)	0.64	1.09 (0.76–1.57)	0.64
Screened positive without self-reported diagnosis	0.81 (0.56–1.17)	0.25	0.7 (0.43–1.13)	0.14

Definition of abbreviations: AOR = adjusted OR; CI = confidence interval; GOLD = Global Initiative for Chronic Obstructive Lung Disease; ICS = inhaled corticosteroid; LABD = long-acting bronchodilator; OR = odds ratio. Indications for use are based on the 2011 GOLD statement.

Potential Underuse Subgroup

CAT score ≥ 10, n = 967

Among the 1,783 participants enrolled in the SPIROMICS trial, with COPD confirmed by spirometry, we identified 967 (54.2%) participants with significant dyspnea (CAT score ≥ 10, Figure 1B). Participants in this subgroup were predominantly white (81.5%), male (55.4%), and over the age of 64 (mean, 64.3; SD, 7.9) (Table 1). The mean FEV₁ was 57.2% predicted (SD, 22.1), and the mean CAT score was 18.8 (SD, 6.2). In this subgroup, 11.4% of participants had been hospitalized for an exacerbation in the year before enrollment.

Forty percent of participants (n = 400, 40.6%; Table 2) screened positive for anxiety or depression, whereas only 35.0% (n = 338) of participants reported a preexisting diagnosis. Among these participants 32.2% (n = 311) screened positive for anxiety and 25.3% (n = 245) screened positive for depression, whereas 22.2% (n = 215) reported a diagnosis of anxiety and 28.3% (n = 274) reported a diagnosis of depression. Approximately 40% of participants either screened positive or had a preexisting diagnosis of anxiety (n = 410, 42.4%; Table 2) or depression (n = 397, 41.1%). The majority of participants who screened positive for anxiety (63%) did not have a preexisting diagnosis of anxiety, indicating that one in five (20.2%) participants in this subgroup had potentially undiagnosed anxiety. Half (50.2%) of participants who screened positive for depression did not report a preexisting diagnosis of depression, indicating that one in eight (12.7%) participants in this subgroup had potentially undiagnosed depression.

Nearly 3 out of 10 participants with CAT scores ≥10 (27.8%, n = 279) denied using an LABD regularly (Table 3). Participants with underuse had worse lung function (FEV₁% predicted: 48.9% vs. 60.2%; SMD, 0.54), more severe dyspnea (CAT score: 19.9 vs. 18.4; SMD, 0.24), and were less likely to be current smokers (27.1% vs. 42.6%; SMD, 0.33) than participants on an LABD. Self-reported, screened, and potentially undiagnosed anxiety and depression were not associated with underuse of LABDs (Table 4). IPTW reduced SMDs to 0.1 or less between exposure groups among those diagnosed with anxiety and depression for calculation of the adjusted ORs (Figures E1B and E2B). Three out of four individuals not using an LABD (75.3%, n = 210) were using ICS monotherapy, including 20% of individuals without recent exacerbation (Table E1B).

Discussion

Among outpatients with spirometrically confirmed COPD, we found that overuse and underuse of appropriate medications is common, suggesting that the overall quality of care is low. However, our study did not find that the substantial burden of diagnosed and undiagnosed anxiety and depression explains this gap in care quality. In addition, the prevalence of both diagnosed and undiagnosed anxiety and depression was high, even among outpatients with mild manifestations of COPD. Our study demonstrates multiple opportunities to improve the quality of care and quality of life of outpatients with COPD.

Ours is the first study that we are aware of to investigate the link between anxiety and depression and the quality of care experienced by patients with COPD in order to determine whether low-quality care could explain the previously observed links between these conditions and disease-specific COPD outcomes (e.g., dyspnea and exacerbation). Taken together, the self-reported underuse of LABDs among patients with COPD and the underdiagnosis of significant anxiety and depression symptoms suggest that providers and patients are not effectively assessing or discussing patient symptoms. More than half of the patients with COPD enrolled in this study reported significant daily dyspnea (CAT score ≥ 10), and yet 30% of these patients were not using an LABD. Among individuals not on an LABD, 75% received ICS monotherapy, which has not been recommended for management of COPD, suggesting that the patients with more severe COPD symptoms are receiving the lowest quality of care. Similarly, a majority of patients with significant mood-disorder symptoms were undiagnosed or unaware of their diagnosis. Although we found that there is no direct link between the quality of the care received and comorbid anxiety and depression, we found that a high proportion of patients with significant anxious or depressive symptoms are underdiagnosed and that a high proportion of patients with severe dyspnea symptoms are currently undertreated.

Our findings suggest that both anxiety and depression occur in higher rates among patients with mild COPD than in the general population and that a large proportion of patients are potentially unaware of and untreated for these diagnoses. We found that patients with mild COPD and no recent exacerbation are significantly more likely to report a diagnosis of depression than members of the general population in the United States and nearly twice as likely to either report a diagnosis or screen positive (38). Half of those with COPD and significant depressive symptoms have not received a diagnosis or are unaware of the diagnosis. The prevalence of diagnosed anxiety was lower in this group than in the overall population in the United States, yet the overall prevalence of serious anxiety symptoms was high, with nearly half of patients who screened positive receiving no diagnosis or treatment (38). Adding to prior studies that evaluated patients with severe COPD, we found that even patients with mild manifestations of COPD have increased odds of anxiety and depression (5–7). Our findings also suggest that providers may be better at identifying depression in patients with mild COPD than they are at identifying anxiety in such patients. A possible explanation is that, when providers and patients do engage in a discussion of symptoms, patients may report symptoms of anxiety that appear to overlap with those of dyspnea (e.g., tachypnea, tachycardia, light-headedness) and do not respond to appropriate disease-specific COPD therapy (e.g., a short-acting bronchodilator), occasionally leading providers to escalate disease-specific COPD therapy without investigating whether the symptoms may not be somatic (43). Some providers may add an unnecessary second LABD, whereas others add an ICS, a heterogeneity that could explain the lack of association in our analysis of overuse. However, as we are unable identify overuse of LABD therapy, we cannot be certain how often this misattribution of symptoms occurs. It is not clear whether primary care physicians or pulmonologists are better at recognizing undiagnosed or untreated mental health symptoms in their patients with COPD, suggesting that solutions targeted at improving mental health care for patients with COPD must not rely on one group of providers more than the other to improve care quality but must foster a collaborative approach to manage this complex patient population.

Limitations

Despite our study having many strengths, there were potential limitations of this observational study. The availability of a single year of look-back may have potentially misclassified overuse for patients enrolled after 2011. To minimize the risk of misclassification, we excluded individuals from the potential overuse cohort with an FEV₁ < 50%, which was required for appropriate ICS initiation in the 2007 GOLD statement and remained an appropriate indication for ICS use under the 2011 statement (30). The 2011 statement also recommended ICSs for patients with one or more inpatient or two or more outpatient exacerbations in the prior year, which led us to exclude individuals with any exacerbation in the prior year (44). Neither guideline indicated whether patients should be continued on this medication after completing 365 days of therapy without an exacerbation. The 2019 GOLD statement was the first to explicitly recommend deescalation of ICS use (45). However, there was considerable evidence available in 2010 that long-acting muscarinic-antagonist therapy was as effective as ICS therapy at preventing exacerbations and that ICS therapy came with an increased risk of pneumonia and other adverse events (46–48). In addition, in 2014, multiple trials demonstrated that ICS therapy can be safely discontinued among individuals who no longer experience severe exacerbations (49–51). In light of the evidence available to providers during the study and our current understanding of the risks and benefits of ICS therapy, we believe that our definition of overuse captures the provision of low-value care. In addition, our study is potentially limited by patient recall and stigma surrounding mental health diagnoses. Patients may not recall receiving a diagnosis of depression or anxiety or may be ashamed to admit this diagnosis to study personnel. To overcome this limitation, we evaluated both self-reported diagnosis and screened symptoms to capture the majority of patients with anxiety and depression. There may have been residual misclassification among patients who did not remember or were ashamed to admit their diagnosis but whose symptoms were being effectively managed with treatment. However, although prior studies have demonstrated that patients often underreport depressive symptoms, they do not suggest that patients would underreport established diagnoses (52). Medication recall and health literacy, which was not assessed in this study, potentially limit our ability to accurately assess our primary outcomes, as we relied on self-reported data rather than pharmacy dispensation data to determine COPD medication use. However, our observed rates of overuse and underuse were similar to those of prior studies that examined inhaled therapies among patients with COPD using pharmacy dispensation data (15, 53). When evaluating underuse of LABD therapy, we could not ascertain whether a patient self-discontinued an LABD or was never prescribed one. As all patients in this study underwent spirometry confirming the presence of airflow obstruction, they had already received a higher quality of care than most individuals with diagnosed COPD and are more likely, therefore, to receive appropriate medications (54, 55). Finally, it is possible that more medically complex patients with a higher number of comorbidities (e.g., hypertension, coronary artery disease, diabetes, etc.) would receive lower-quality outpatient care for COPD. However, the literature is mixed on this subject (56). We therefore did not include other comorbidities in our analysis, which is a potential limitation.

Conclusions

In summary, we found that undertreated COPD symptoms and undiagnosed anxiety and depression are common among patients with COPD. Both conditions require effective patient and provider communication for appropriate assessments of disease severity to be made. Ideally, both primary care and pulmonary providers would improve their adherence to the depression and anxiety screening recommendations put forward by the U.S. Preventative Services Task Force, National Institutes of Health, and Society for Hospitalist Medicine, which suggest routine screening of all adults and recommend the use of therapies that effectively address both mood disorders and COPD (e.g., pulmonary rehabilitation) instead of reliance on pharmacotherapy alone for those adults with COPD (57–61). However, primary care providers and pulmonologists face substantial time constraints and have priorities that shift care away from evaluating mental health symptoms and quantifying dyspnea severity. These competing priorities lead to prioritization of other conditions, undertreatment of symptoms, and use of potentially inappropriate therapies (15, 62–68). Therefore, for patients with COPD, we advocate decentralizing both the burden of evaluation and referral for mental health symptoms and the burden of management of symptoms of chronic severe dyspnea. There are multiple potential strategies that would accomplish this goal, including proactive models of mental health and pulmonary consultation for patients with COPD that are focused on identifying and referring populations of patients who would benefit from specialist care rather than relying on referrals of individual patients, as well as complex dyspnea clinics staffed by mental health and internal medicine specialty providers (pulmonary and/or palliative care) that specialize in managing patients with symptomatic lung disease. We are currently evaluating the effectiveness, feasibility, and acceptability of these programs in multiple healthcare systems (69, 70). Given the projected shortage of pulmonologists and mental health providers, we prefer that these interventions be delivered through virtual consultation, which has been effective in identifying patients and recommending treatment for other conditions (71–74). Proactively consulting appropriate pulmonary and mental health providers and creating complex dyspnea clinics to assist in the management of patients with comorbid dyspnea and mental health symptoms has the potential to maximize the potential for symptom relief among patients with chronic disease (75).