Factors associated with new onset of depression and anxiety symptoms differ by sex in the COPDGene study

Introduction

Depressive symptoms and anxiety symptoms are elevated in smokers and patients with chronic obstructive pulmonary disease (COPD) [1,2] but often go unrecognized. [3–6] It is important to understand factors that increase risk for depression and anxiety among smokers and people at risk for or with COPD. For example, smokers who are at increased risk of COPD have been found to have higher rates of depression and anxiety. [7–13] Among subjects with COPD during the COVID-19 pandemic, a previous study found associations with new onset of depression and the following: loneliness, functional limitations, family conflict, limited access to healthcare resources, and higher income. [14] In the Evaluation of COPD Longitudinally to Identify Predictive Surrogate Endpoints (ECLIPSE) study, risk factors for new onset of depression over a 3 year follow-up period included: worse health status, fatigue, and moderate to severe dyspnea. [1] Among COPD patients entering a pulmonary rehabilitation (PR) program, anxiety was associated with younger age, impaired quality of life, and more prevalent in females. [15] Symptoms of anxiety and depression also differ by sex. [16] Among 202 COPD patients, dyspnea was more strongly correlated with depression in women than in men. [17] However, caution is required in interpreting this study findings had a limited sample size and did not account for potential confounders.

To address this gap in this literature, the Genetic Epidemiology of COPD (COPDGene) study provides an opportunity to examine clinical and demographic variables associated with the progression of anxiety and depressive symptoms among adults with 10 or more pack-years of smoking history and at risk for or expressing one of the stages of COPD (Global Initiative for Obstructive Lung Disease (GOLD) grades 1–4). [18] While sex has been associated with anxiety and depression symptoms in the COPDGene study [19], potential sex differences for risk factors have not been tested using moderation analyses and the onset of anxiety and depressive symptoms has not been evaluated. Here, we examined clinical and demographic variables associated with new onset symptoms of depression and anxiety and sought to assess moderation by sex in COPDGene. Given the prevalence and impact of depression and anxiety on quality of life and daily functioning and emerging evidence that COPD may manifest differently among men and women, understanding which clinical factors may drive depression and anxiety between the sexes is important in order to personalize the focus of clinical intervention.

Methods

With written informed consent for the Institutional Review Board-approved protocols, participants were enrolled in COPDGene with baseline measurements between 2007–2012 and with comprehensive assessment occurring over three phases approximately five years apart. In addition to clinical variables for COPD, including COPD Assessment test (CAT) and modified Medical Research Council (mMRC) dyspnea scale, the Hospital Anxiety and Depression Scales for anxiety (HADS-A) and depression (HADS-D) were collected at phases 2 and 3 of the COPDGene study. [20] Table 1 lists clinical and demographic characteristics of the cohort. Note that the HADS-A and HADS-D were self-reported. We restricted the analyses to participants with available phenotypic information at both phases 2, without clinically elevated symptoms of depression (HADS-D<8) and anxiety (HADS-A<8) at phase 2, and excluding non-smoking controls due to limited sample size. We defined new onset depression symptoms as not having elevated HADS-D at phase 2 (HADS-D<8) but having elevated HADS-D approximately 5 years later at phase 3 (HADS-D≥8) versus no new onset as not having elevated HADS-D at phases 2 and 3 (HADS-D<8). New onset anxiety symptoms were defined identically using HADS-A. There were 2,653 participants without clinically elevated depression symptoms at phase 2, of which 194 (7.3%) had new onset depression symptoms at phase 3. There were 2,413 participants without clinically elevated anxiety symptoms at phase 2, of which, 188 (7.8%) had new onset anxiety symptoms at phase 3. To examine demographic and clinical variables associated with new onset depression and anxiety symptoms, we used logistic regression models among all participants and stratified by sex fitting the full model with all 20 variables, which were collected at phase 2, listed in Table 2 with a Bonferroni-corrected p-value threshold (0.05/20=2.5×10⁻³ for the overall analysis, 0.05/19=2.6×10⁻³ for the sex stratified analyses). Note that the Bonferroni correction is conservative, especially given the correlation between variables (i.e, CAT and mMRC). Crude/ unadjusted models are included in the supplement. In addition, for variables associated with the outcome for only one sex, we considered sex interactions adjusting for the same set of covariates as in the starfield models. In secondary analyses, all models were adjusted for medication use collected at phase 2 and whether the phase 3 HADS measurements were collected before or after the pandemic. Since very similar results were observed for these models as compared to Table 2, the results are presented in the supplement. All analyses were run in R (version 4.3.2).

Results

As seen in Table 2, increasing CAT score (β=0.05, p=3.0×10⁻⁵) and current smoking status (β=0.82, p=1.8×10⁻⁵) were significantly associated with new onset depression symptoms and the following variables were marginally associated (2.5×10⁻³≤p<0.05): increasing mMRC dyspnea score (β=0.15, p=0.04), low income (β=0.54, p=5.5×10⁻³), and being married or member of an unmarried couple was negatively associated with depression symptoms (β=−0.51, p=6.5×10⁻³). In the sex stratified analyses, CAT score (β=0.08, p=1.9×10⁻⁵) was significantly associated with new onset depression symptoms among males, but not females and the following variables were marginally associated among females, but not males: mMRC dyspnea score (β=0.23, p=0.03), race (β=−0.69, p=0.01), low income (β=0.71, p=8.2×10⁻³), and internet access (β=−0.66, p=0.01). When testing for a sex interaction on new onset depression symptoms with the 5 variables that were associated in the sex stratified analyses, there was a CAT score by sex interaction on new onset depression symptoms (β=−0.05, p=0.04).

Female sex (β=0.56, p=8.4×10⁻⁴) was significantly associated with new onset anxiety symptoms and the following variables were marginally associated (2.5×10⁻³≤p<0.05): younger age (β=−0.03, p=0.02), severe exacerbations (β=0.58, p=0.02), CAT score (β=0.04, p=4.9×10⁻³), low income (β=0.54, p=8.0×10⁻³), and internet access (β=−0.52, p=9.3×10⁻³). In the sex stratified analyses, new onset anxiety symptoms were significantly associated with mMRC among females (β=0.33, p=1.4×10⁻³), CAT score among males (β=0.06, p=1.4×10⁻³), and marginally associated with low income among females (β=0.70, p=7.2×10⁻³). When testing for a sex interaction on new onset anxiety symptoms with the 3 variables that were associated in the sex stratified analyses, there was a mMRC dyspnea score by sex (β=0.42, p=5.1×10⁻³) and CAT score by sex (β=−0.06, p=0.02) interaction on new onset anxiety symptoms.

Discussion

Supplementary Material