Introduction
Depression and anxiety are common mental disorders associated with increased comorbidities, premature death, and healthcare costs (Charlson et al., Reference Charlson, van Ommeren, Flaxman, Cornett, Whiteford and Saxena2019; Patel et al., Reference Patel, Saxena, Lund, Thornicroft, Baingana, Bolton and UnÜtzer2018). According to the Global Burden of Disease 2019 and World Health Organization report, the prevalences of depression and anxiety are 3.4 and 3.8%, which account for 7.5 and 3.4% of global disability, respectively (GBD, 2019 Mental Disorders Collaborators, 2022; World Health Organization, 2017). Depression and anxiety have been reported to co-occur with chronic obstructive pulmonary disease (COPD), which is the most common chronic lung disease with progressive lung function decline (Cavaillès et al., Reference Cavaillès, Brinchault-Rabin, Dixmier, Goupil, Gut-Gobert, Marchand-Adam and Diot2013; Yohannes & Alexopoulos, Reference Yohannes and Alexopoulos2014). Previous studies have also indicated that COPD was an independent risk factor for depression and anxiety, and the prevalences of depression and anxiety in COPD patients were reported to be as high as 50% (Matte et al., Reference Matte, Pizzichini, Hoepers, Diaz, Karloh, Dias and Pizzichini2016; Pumar et al., Reference Pumar, Gray, Walsh, Yang, Rolls and Ward2014; Riblet, Gottlieb, Hoyt, Watts, & Shiner, Reference Riblet, Gottlieb, Hoyt, Watts and Shiner2020). Given the severity of depression and anxiety in obstructive lung function decline, the potential causal associations of non-obstructive lung function abnormalities with depression and anxiety also merit exploration in prospective studies.
In population-based studies, 7.1 to 20.3% of individuals undergoing spirometry had proportional declines in forced expired volume in the first second (FEV1) and forced vital capacity (FVC), normal FEV1 to FVC ratio, and decreased FEV1% predicted (Wan, Reference Wan2022). This pattern of lung function impairment is defined as preserved ratio impaired spirometry (PRISm), which is often neglected in previous studies (Wan et al., Reference Wan, Castaldi, Cho, Hokanson, Regan, Make and Silverman2014). Recently, PRISm is increasingly recognized to have a clinically important role in the health of the general population (Godfrey & Jankowich, Reference Godfrey and Jankowich2016). Several population-based studies have shown that PRISm was related to increased respiratory symptoms, higher incidence of comorbidities and greater mortality, compared to normal spirometry (Higbee, Granell, Davey Smith, & Dodd, Reference Higbee, Granell, Davey Smith and Dodd2022; Li et al., Reference Li, Jankowich, Wu, Lu, Shao, Lu and Ke2023b; Wan et al., Reference Wan, Balte, Schwartz, Bhatt, Cassano, Couper and Oelsner2021; Washio et al., Reference Washio, Sakata, Fukuyama, Honda, Kan, Shibata and Ninomiya2022; Zheng et al., Reference Zheng, Zhou, Zhang, Su, Chen, Li and Wu2023).
The relationships of COPD with depression and anxiety have been reported previously (Yohannes & Alexopoulos, Reference Yohannes and Alexopoulos2014). Although not fully elucidated, some factors have been reported to contribute to the development of depression and anxiety in COPD patients. COPD patients have higher smoking exposure, more respiratory symptoms, reduced exercise capacity and more comorbidities, contributing to depression and anxiety (Huang et al., Reference Huang, Huang, Xu, Yang, Zhao, Zhang and Wang2021). Chronic systemic inflammation may also lead to psychological distress in patients with COPD (Lu et al., Reference Lu, Feng, Feng, Nyunt, Yap and Ng2013). However, the risks of depression and anxiety in individuals with PRISm remain to be elucidated.
To address these knowledge gaps, we aimed to use data from the UK Biobank to investigate the associations of baseline PRISm with adverse outcomes of depression and anxiety during follow-up, as well as their subtypes. We also assessed the associations of lung function trajectories with depression and anxiety in a subpopulation with repeated spirometry.
Methods
Study population
The UK Biobank study is a nationwide prospective study of approximately 500 000 middle-aged and older adults in the UK (Sudlow et al., Reference Sudlow, Gallacher, Allen, Beral, Burton, Danesh and Collins2015). The participants were recruited between 19 Dec 2006 and 10 Oct 2010, and baseline data on sociodemographic characteristics, lifestyle, physical measurements and health-related outcomes were collected through brief interviews, questionnaires and linkage to national health records. Participants with missing information on spirometry, depression, anxiety, or covariates were excluded from the study.
Three main populations were identified in this study (online Supplementary Fig. S1). In population 1, we included participants without depression or anxiety at baseline. This population was used to explore the risks of baseline lung function categories on incident depression or anxiety during follow-up. In population 2, participants with depression or anxiety at baseline were included to investigate the association of baseline lung function categories with mortality during follow-up. The relationships of lung function trajectories with depression and anxiety were evaluated in the subpopulation with repeated spirometry.
The research protocol for this study has been approved by the review committee of UK Biobank. The study was approved by the North-West Research Ethics Committee and all participants provided written informed consent before enrollment.
Assessments of spirometry
The participants were requested to complete 2–3 pre-bronchodilator spirometry at recruitment if they consented and had no contraindications. Participants were invited for repeated spirometry between 30 Apr 2014, and 13 Mar 2020 if they resided close to an assessment center. The highest FEV1 and FVC values from acceptable blows were used in this study.
PRISm was defined as FEV1/FVC of 0.70 or higher and FEV1 of less than 80% predicted. Airflow obstruction (AO) was defined as FEV1/FVC of less than 0.70 according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) (Agustí et al., Reference Agustí, Celli, Criner, Halpin, Anzueto, Barnes and Vogelmeier2023). A FEV1/FVC of 0.70 or higher and a FEV1 of 80% predicted or higher were considered normal spirometry. Predicted FEV1 was calculated according to the Global Lung Initiative (GLI) established prediction equations considering race, gender, age and height (Quanjer et al., Reference Quanjer, Stanojevic, Cole, Baur, Hall, Culver and Stocks2012). FEV1% value was used to evaluate the severity of lung function impairment. We categorized lung function at baseline into three groups: normal spirometry, PRISm and AO, and lung function trajectories into nine groups: normal to normal, normal to PRISm, normal to AO, PRISm to PRISm, PRISm to AO, PRISm to normal, AO to AO, AO to PRISm, and AO to normal.
Outcome measurement and covariates
The outcome of population 1 was time to incident depression, anxiety or the subtypes, and the outcome of population 2 was the time to mortality. Depression and anxiety were diagnosed by medical doctors according to the guidelines of the National Institute for Health and Care Excellence. The assessments were conducted using records derived from death register, primary care, hospital admission, and self-report information. Depression was defined by the first record of five subtypes: depressive episode (F32), recurrent depressive disorder (F33), persistent mood disorder (F34), other mood disorder (F38) or unspecified mood disorder (F39), coded by International Classification of Diseases version-10 (ICD-10). Anxiety included phobic anxiety disorders (F40) and other anxiety disorders (F41).
Based on previous research on risk factors for depression and anxiety in UK Biobank participants, several potential confounders were considered as covariates in this study (Gao et al., Reference Gao, Geng, Jiang, Huang, Zheng, Belsky and Huang2023; Yang, Wang, Huang, Kelly, & Li, Reference Yang, Wang, Huang, Kelly and Li2023). Age, sex (female and male), ethnicity (White, Black, and other), smoking status (never, previous, and current), drinking status (never, previous, and current), and income level (<£ 31 000 and ⩾£ 31 000) were collected using baseline questionnaires. Height and weight were obtained through physical examinations, and body mass index (BMI) was calculated as weight (kg)/height (m)2. Physical activity was evaluated by the number of days/week that the participant walked for 10+ minutes, and categorized into 0 days/week, 1–3 days/week and 3–7 days/week. Hypertension was defined by ICD-10 codes I10–I13 and I15, diabetes by E10–E14, and cardiovascular disease (CVD) by I5–I9, I11, I13, I20–I28 and I30–I52. The variable assignments for these covariates were provided in online Supplementary Table S1.
Statistical analysis
Baseline characteristics were presented as the number of observations and percentage for categorical variables and mean and standard deviation (s.d.) for continuous variables. The demographic differences between participants with PRISm and those with normal spirometry or AO were compared using Pearson's χ2 test or t test as appropriate.
Cumulative survival curves were calculated using the Kaplan-Meier method grouped by lung function categories and compared using the log-rank test. The Cox proportional hazard model was used to estimate the hazard ratio (HR) with a 95% confidence interval (CI) for incident depression, anxiety or subtypes in population 1 and mortality in population 2, in relationship to lung function categories at baseline and lung function trajectories. The non-linear associations between FEV1% predicted and outcomes were also investigated with penalized cubic splines fitted in Cox proportional hazard models (online Supplementary Method) (Govindarajulu, Malloy, Ganguli, Spiegelman, & Eisen, Reference Govindarajulu, Malloy, Ganguli, Spiegelman and Eisen2009). The coefficient curves of FEV1% predicted were zeroed at 80% (HR = 1) in the non-linear models. Likelihood ratio tests were used to test the statistical significance of overall and non-linear part of FEV1% predicted.
Statistical analyses were performed using SAS 9.4 (SAS Institute Inc, Cary, North Carolina) and R 4.2.1 (R Foundation for Statistical Computing, Vienna). Two-sided values of p < 0.05 were considered statistically significant.
Results
Baseline characteristics
Of the participants in the UK Biobank, 369 597 were included in this study, in which the prevalences of PRISm and AO at baseline were 10.5 and 14.7%, respectively. The participant characteristics by baseline lung function categories were presented in Table 1. Participants with PRISm were older, were less likely to be White, had higher BMI, were more likely to be current smokers, were less likely to be current drinkers, had less physical activity, had lower income level, and had higher prevalences of hypertension, diabetes and CVD than those with normal spirometry.
Demographic information of the total population in this study
a p < 0.05 for the comparison of PRISm and normal spirometry.
b p < 0.05 for the comparison of PRISm and AO.
Note: Age, BMI and lung function were presented as mean ± standard deviation, and other characteristics as number (percentage).
The prevalences of depression (9.3%), depressive episode (8.9%), recurrent depressive disorder (0.6%), persistent mood disorder (0.2%), anxiety (3.9%) and other anxiety disorder (3.7%) were significantly higher in participants with PRISm than in those with normal spirometry and AO (online Supplementary Table S2). In population 1, the incidences of depression, depressive episode, anxiety, phobic anxiety disorder and other anxiety disorder in participants with PRISm were 5.3 5.2, 5.5, 0.7 and 5.1% after a median follow-up period of 13.9 years, which were also significantly higher than those in the other two groups. In population 2, the mortality rates of participants with depression (12.5%), depressive episode (12.3%), recurrent depressive disorder (13.5%), anxiety (12.2%) and other anxiety disorder (12.0%) were significantly higher than those with normal spirometry, and lower than those with AO.
Incident depression and anxiety by lung function categories
As presented in Fig. 1, participants with PRISm had the highest unadjusted incidences of depression (Figs 1a to c) and anxiety (Figs 1d to 1f). Table 2 showed the associations of baseline lung function categories with incident depression and anxiety in Cox proportional hazard models. In the multivariate-adjusted models, the risk of depression was 12% higher in participants with PRISm compared to those with normal spirometry (HR: 1.12; 95% CI 1.07–1.18), and the increased risk was 11% for anxiety (HR: 1.11; 95% CI 1.06–1.15). The higher risks of PRISm were significant in the subtypes of depression episode (HR: 1.14; 95% CI 1.08–1.20), phobic anxiety disorder (HR: 1.35; 95% CI 1.18–1.54) and other anxiety disorder (HR: 1.10; 95% CI 1.04–1.15).
Kaplan-Meier curves of incident depression, anxiety and mortality by baseline lung function categories in different populations. (a) Depression in population 1. (b) Depressive episode in population 1. (c) Recurrent depressive disorder in population 1. (e) Anxiety in population 1. (e) Phobic anxiety disorder in population 1. (f) Other anxiety disorder in population 1. (g) Mortality of participants with depression in population 2. (h) Mortality of participants with depressive episode in population 2. (i) Mortality of participants with recurrent depressive disorder in population 2. (j) Mortality of participants with anxiety in population 2. (k) Mortality of participants with phobic anxiety disorder in population 2. (l) Mortality of participants with other anxiety disorder in population 2.
Hazard ratios (95% CIs) of incident depression and anxiety with different baseline lung function categories in individuals without depression and anxiety
a Adjusted for age, sex, ethnicity, BMI, smoking status, drinking status, income level, physical activity, hypertension, diabetes and CVD.
The non-linear associations of FEV1% predicted with incident depression and anxiety in participants without AO were shown in Fig. 2. The FEV1% predicted of more than 99% of these participants ranged between 54% and 140%. The non-linear association between FEV1% predicted and incident depression was marginally significant (p Nonlinear = 0.07), which was mainly contributed by depressive episode (Figs 2a to c). Furthermore, we found a non-linear increase of anxiety risk when FEV1% predicted decreased in participants with PRISm (p Nonlinear < 0.01), which was mainly contributed by other anxiety disorder (Figs 2d to f).
Distribution of FEV1% predicted and penalized cubic spline analyses for the association of FEV1% predicted with incident depression and anxiety. (a) Depression in population 1. (b) Depressive episode in population 1. (c) Recurrent depressive disorder in population 1. (d) Anxiety in population 1. (e) Phobic anxiety disorder in population 1. (f) Other anxiety disorder in population 1. (g) Mortality of participants with depression in population 2. (h) Mortality of participants with depressive episode in population 2. (i) Mortality of participants with recurrent depressive disorder in population 2. (j) Mortality of participants with anxiety in population 2. (k) Mortality of participants with phobic anxiety disorder in population 2. (l) Mortality of participants with other anxiety disorder in population 2.
Mortality of participants with depression and anxiety by lung function categories
The associations of baseline lung function categories with the mortality of participants with depression and anxiety were presented in Fig. 1 and Table 3. The unadjusted mortality rates for PRISm were intermediate between normal spirometry and AO among participants with depression (Figs 1g to i) and anxiety (Figs 1j to l). In Cox proportional hazard models adjusted for potential confounders, PRISm increased 46 and 70% of mortality rates in participants with depression (HR: 1.46; 95% CI 1.31–1.62) and anxiety (HR: 1.70; 95% CI 1.44–2.02) (Table 3). The higher risks of mortality in participants with PRISm were observed in the subtypes of depression episode (HR: 1.43; 95% CI 1.28–1.60) and other anxiety disorder (HR: 1.68; 95% CI 1.41–2.00). The non-linear association between FEV1% predicted and mortality was not statistically significant in participants with depression (p Nonlinear = 0.18) (Figs 2g to i), but it was statistically significant in those with anxiety and other anxiety disorder (p Nonlinear < 0.01) (Figs 2j to 2l).
Hazard ratios (95% CIs) of all-cause mortality with different baseline lung function categories in individuals with depression and anxiety
a Adjusted for age, sex, ethnicity, BMI, smoking status, drinking status, income level, physical activity, hypertension, diabetes and CVD.
Characteristics of different lung function trajectories
Of the participants included in this study, 35 319 were invited to undergo repeated spirometry after an average of 9.5 years. Among 2654 participants with PRISm at baseline, 1172 (44.2%) reverted to normal spirometry, 1137 (42.8%) had persistent PRISm, and 345 patients (13.0%) developed AO (online Supplementary Fig. S2). Participants with PRISm had a higher risk of developing AO than those with normal spirometry (13.0% v. 9.4%; p < 0.01).
Participants who transitioned from normal spirometry to PRISm were more likely to be women, had higher BMI, were more likely to be current smokers, were less likely to be White, were less likely to be current drinkers, had less physical activity, had lower income level, and had higher prevalences of hypertension, diabetes and CVD than participants with persistent normal spirometry (p < 0.05) (online Supplementary Table S3). Participants with persistent PRISm were less likely to be White and current drinkers, had lower income level, and had higher prevalences of hypertension and diabetes than those who reverted to normal spirometry (p < 0.05). Compared with participants who transitioned from PRISm to AO, those with persistent PRISm were less likely to be White, and had higher BMI (p < 0.05). Among the participants with normal spirometry at baseline, those with lower FEV1/FVC and FEV1% predicted tended to transition to PRISm (p < 0.05). The participants with PRISm at baseline who had a higher level of FEV1% predicted were more likely to transition to normal spirometry, and those who had a lower level of FEV1/FVC tended to develop AO (p < 0.05).
The prevalences of depression and anxiety at baseline and mortality rates of participants with depression and anxiety were also statistically different among nine lung function trajectories (p < 0.05) (online Supplementary Table S4).
Depression and anxiety by lung function trajectories
After a median follow-up period of 4.4 years, the incidences of depression and anxiety were not statistically different among nine lung function trajectories (online Supplementary Table S5). Although not significant, the HRs were over 1 in most of the trajectories, especially those with PRISm at baseline. Compared with consistent normal spirometry, the transition from normal spirometry to PRISm had a higher risk of mortality in participants with depression (HR: 4.27; 95% CI 2.10–8.69) and anxiety (HR: 3.08; 95% CI 1.19–7.94).
Discussion
In the general population from the UK Biobank study, we found significant associations of PRISm with increased risks of depression, anxiety, and mortality among participants with depression and anxiety. These risks increased non-linearly with the severity of PRISm. The magnitudes of these associations were similar in the phenotypes of lung function impairment and the subtypes of mental disorders. In the lung function trajectory analysis, the transition from normal spirometry to PRISm showed a higher risk of mortality in participants with depression and anxiety.
PRISm is a new lung function impairment phenotype that increases the risks of multiple adverse outcomes. Population-based studies conducted in Europe, the United States and Japan revealed that PRISm was associated with increased all-cause and cause-specific mortality (Higbee et al., Reference Higbee, Granell, Davey Smith and Dodd2022; Wan et al., Reference Wan, Fortis, Regan, Hokanson, Han, Casaburi and Investigators2018, Reference Wan, Balte, Schwartz, Bhatt, Cassano, Couper and Oelsner2021; Washio et al., Reference Washio, Sakata, Fukuyama, Honda, Kan, Shibata and Ninomiya2022; Wijnant et al., Reference Wijnant, De Roos, Kavousi, Stricker, Terzikhan, Lahousse and Brusselle2020). Recent findings also reported the increased risks of developing diabetes, CVD and frailty in individuals with PRISm (He et al., Reference He, Yan, Zhou, Ge, Zhang, Xu and Zhu2023; Krishnan et al., Reference Krishnan, Tan, Farias, Aaron, Benedetti, Chapman and Bourbeau2023; Li et al., Reference Li, Jankowich, Lu, Wu, Shao and Ke2023a; Zheng et al., Reference Zheng, Zhou, Zhang, Su, Chen, Li and Wu2023). An interesting phenomenon noted in most of these studies was the similar risks of PRISm and COPD on these adverse outcomes, which was consistent with our findings regarding depression and anxiety. Therefore, PRISm and COPD may have similar comorbidities despite different lung function impairment phenotypes, emphasizing the importance of identifying PRISm.
Previous studies have examined the associations of COPD with depression and anxiety, most of which were cross-sectional designs (Matte et al., Reference Matte, Pizzichini, Hoepers, Diaz, Karloh, Dias and Pizzichini2016). A meta-analysis of six longitudinal studies showed that COPD increased the risk of depression by 69% (Atlantis, Fahey, Cochrane, & Smith, Reference Atlantis, Fahey, Cochrane and Smith2013). Another longitudinal study in China observed that chronic lung disease was associated with a 17% increased risk of developing depressive symptoms (Zheng et al., Reference Zheng, Li, Pei, Zhu, Cheong, Li and Wang2022). In addition to confirming and extending prior knowledge about COPD, we comprehensively evaluated the associations of PRISm with depression, anxiety and mortality among participants with depression and anxiety. Our study supported the hypothesis that non-obstructive lung function abnormality could also lead to adverse outcomes of depression and anxiety. More importantly, the potentially reverse J shaped associations suggested the accelerated progression of depression and anxiety with the decrease of FEV1% predicted in PRISm, highlighting the necessity of preventing PRISm progression. To the best of our knowledge, this is the first study to validate the associations of baseline category and trajectory of PRISm with adverse outcomes of depression and anxiety.
The underlying mechanisms linking PRISm with depression and anxiety have not been understood completely, but several mechanisms may explain the associations. Lung function impairment can cause chronic systemic inflammation and chronic hypoxemia, which may restrict the function of brain cells and induce symptoms of depression and anxiety (Barnes, Reference Barnes2016; Giltay, Nissinen, Giampaoli, Zitman, & Kromhout, Reference Giltay, Nissinen, Giampaoli, Zitman and Kromhout2010). Lung function decline, together with extra-pulmonary comorbidities of PRISm reported in previous studies, can change lifestyle and impair quality of life (e.g. more sedentary, less mobility, dyspnea, pain), resulting in depression and anxiety disorders (Ali et al., Reference Ali, Stone, Skinner, Robertson, Davies and Khunti2010; Li, Ge, Greene, & Dunbar-Jacob, Reference Li, Ge, Greene and Dunbar-Jacob2019; Vreijling et al., Reference Vreijling, van Haeringen, Milaneschi, Huider, Bot, Amin and Jansen2023). In addition, similar risk factors may also explain the associations of PRISm with depression and anxiety. Further studies are needed to clarify the underlying mechanisms.
Several limitations should be acknowledged in the current study. First, the diagnosis of PRISm and AO with pre-bronchodilator spirometry may overestimate the prevalences of diseases in individuals with reversible airflow obstruction. Second, depression and anxiety in some participants were defined based on self-reported diagnoses, which may introduce recall bias. Third, the limited sample size of some subtypes prevented further exploration on the heterogeneity of associations between lung function impairment phenotype and subtypes of mental disorders. Fourth, the follow-up period was not long enough for the participants with repeated spirometry, possibly leading to an underestimate of the differences between lung function trajectories. Finally, although we adjusted for several important confounders, residual or unmeasured confounding may remain.
In conclusion, this prospective study reveals that PRISm is associated with an increased risk of depression and anxiety, as well as a higher mortality of participants with depression and anxiety. Individuals with PRISm show a similar risk of depression and anxiety as those with AO. Therefore, these findings emphasize the importance of recognizing PRISm and lung function rehabilitation in preventing the adverse outcomes of depression and anxiety.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291724002162
Acknowledgments
This research has been conducted using the UK Biobank Resource under Application Number 97223. The authors thank the participants and those involved in building the UK Biobank study.
Funding statement
This work was supported by the National Key R&D Program of China (grant numbers: 2022YFF0710800, 2022YFF0710802) and the National Natural Science Foundation of China (grant numbers: 82103941, 82100023, 82170042).
Competing interests
None.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Open access
