Article Text

Download PDFPDF

Original research
Rheumatoid arthritis and changes on spirometry by smoking status in two prospective longitudinal cohorts
  1. Keigo Hayashi1,
  2. Gregory C McDermott1,2,
  3. Pierre-Antoine Juge1,
  4. Matthew Moll2,3,4,5,
  5. Michael H Cho2,3,4,
  6. Xiaosong Wang1,
  7. Misti L Paudel1,2,
  8. Tracy J Doyle2,3,
  9. Gregory L Kinney6,
  10. Danielle Sansone-Poe6,
  11. Kendra Young6,
  12. Paul F Dellaripa1,2,
  13. Zachary S Wallace2,7,
  14. Elizabeth A Regan8,
  15. Gary M Hunninghake2,3,
  16. Edwin K Silverman2,3,4,
  17. Samuel Y Ash9,
  18. Raul San Jose Estepar2,3,10,
  19. George R Washko2,3 and
  20. Jeffrey A Sparks1,2
  1. 1Division of Rheumatology, Inflammation, and Immunity, Brigham and Women's Hospital, Boston, Massachusetts, USA
  2. 2Harvard Medical School, Boston, Massachusetts, USA
  3. 3Division of Pulmonary and Critical Care Medicine, Brigham and Women's Hospital, Boston, Massachusetts, USA
  4. 4Channing Division of Network Medicine, Brigham and Women’s Hospital, Boston, MA, USA
  5. 5Division of Pulmonary, Critical Care, Sleep and Allergy, Veterans Affairs Boston Healthcare System, West Roxbury, MA, USA
  6. 6Department of Epidemiology, University of Colorado Anschutz Medical Campus, Aurora, Colorado, USA
  7. 7Division Rheumatology, Allergy, and Immunology, Massachusetts General Hospital, Boston, Massachusetts, USA
  8. 8National Jewish Health, Denver, Colorado, USA
  9. 9South Shore Hospital, South Weymouth, MA, USA
  10. 10Department of Radiology, Brigham and Women’s Hospital, Boston, MA, USA
  1. Correspondence to Dr Jeffrey A Sparks; jsparks{at}


Objective To compare longitudinal changes in spirometric measures between patients with rheumatoid arthritis (RA) and non-RA comparators.

Methods We analysed longitudinal data from two prospective cohorts: the UK Biobank and COPDGene. Spirometry was conducted at baseline and a second visit after 5–7 years. RA was identified based on self-report and disease-modifying antirheumatic drug use; non-RA comparators reported neither. The primary outcomes were annual changes in the per cent-predicted forced expiratory volume in 1 s (FEV1%) and per cent predicted forced vital capacity (FVC%). Statistical comparisons were performed using multivariable linear regression. The analysis was stratified based on baseline smoking status and the presence of obstructive pattern (FEV1/FVC <0.7).

Results Among participants who underwent baseline and follow-up spirometry, we identified 233 patients with RA and 37 735 non-RA comparators. Among never-smoking participants without an obstructive pattern, RA was significantly associated with more FEV1% decline (β=−0.49, p=0.04). However, in ever smokers with ≥10 pack-years, those with RA exhibited significantly less FEV1% decline than non-RA comparators (β=0.50, p=0.02). This difference was more pronounced among those with an obstructive pattern at baseline (β=1.12, p=0.01). Results were similar for FEV1/FVC decline. No difference was observed in the annual FVC% change in RA versus non-RA.

Conclusions Smokers with RA, especially those with baseline obstructive spirometric patterns, experienced lower FEV1% and FEV1/FVC decline than non-RA comparators. Conversely, never smokers with RA had more FEV1% decline than non-RA comparators. Future studies should investigate potential treatments and the pathogenesis of obstructive lung diseases in smokers with RA.

  • Rheumatoid Arthritis
  • Smoking
  • Risk Factors
  • Pulmonary Fibrosis

Data availability statement

Data are available upon reasonable request.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Restrictive and obstructive lung diseases are prevalent in rheumatoid arthritis (RA) and are associated with increased mortality.

  • There have been limited investigations that evaluated for changes in pulmonary function measures over time comparing patients with and without RA.


  • After 5–7 years of follow-up after baseline spirometry, patients with RA who never smoked had more decline in per cent-predicted forced expiratory volume in 1 s (FEV1%) than non-RA comparators.

  • Patients with RA who previously smoked ≥10 pack-years, had less decline in FEV1% and FEV1/forced vital capacity (FVC) than non-RA comparators.

  • The observed associations were most prominent in patients with RA who had baseline obstructive pattern and were not attributable to differences such as smoking levels or baseline spirometric measures.


  • These results suggest that patients who have smoked with RA and an obstructive pattern may be a unique phenotype that could have less decline than expected.

  • Further studies are required to explore the mechanisms and potential reversibility of systemic inflammation, autoimmunity and the impact of RA treatments on pulmonary function.


Rheumatoid arthritis (RA) exhibits extra-articular manifestations in nearly 40% of patients, and pulmonary involvement is one of the most common and serious manifestations.1 Although RA-associated interstitial lung disease (RA-ILD) is a recognised pulmonary manifestation,2 3 RA is also associated with an elevated risk of obstructive lung diseases such as chronic obstructive pulmonary disease (COPD), asthma and bronchiectasis.4–6 Regarding pulmonary function, patients with RA are at an increased risk for abnormalities of both restrictive and obstructive patterns on spirometric measures, which may not be explained by smoking.7 8

Recently, studies have shown that restrictive and obstructive lung diseases are not only prevalent in RA but also associated with high mortality rates. For instance, mortality rates in RA-ILD patients have been extensively described.9 10 A population-based cohort study suggests 1-year mortality rates of 13.9% in RA-ILD and 3.8% in non-ILD RA patients.11 Additionally, our research group previously showed a twofold higher risk of death in RA individuals with subclinical interstitial lung abnormalities compared with their non-RA comparators.12 Patients with RA with obstructive lung disease experience higher mortality rates compared with both patients with RA without obstructive lung disease13 14 and non-RA patients with obstructive lung disease.13

Despite these reported high mortality rates in patients with RA with pulmonary involvement, knowledge regarding the differences between pulmonary diseases in patients with and without RA is limited. This knowledge gap is crucial for understanding its pathogenesis, disease course and potential treatment approaches. Previous studies have highlighted associations between obstructive/restrictive patterns on pulmonary function tests (PFT) and the presence of rheumatoid factor and anticyclic citrullinated peptide antibodies, suggesting that lung diseases in RA patients may have a distinct pathogenesis from those in patients without RA.15 However, investigations into the progression of lung diseases in individuals with RA compared with those without RA are lacking.

In this study, we aimed to investigate whether the longitudinal changes in spirometric measurements differ between individuals with RA and non-RA comparators. We used data from two large prospective cohorts—one representing the general population and another comprising smokers at high risk for pulmonary diseases. We hypothesised that RA is associated with more decline in spirometric measurements for both restrictive and obstructive patterns, independent of smoking and other potential confounders.


Study population and design

We analysed two data sources: the UK Biobank, representing a cohort of the general population, and COPDGene, representing a cohort of individuals with or at high risk for respiratory disease due to smoking. The UK Biobank is a national prospective cohort encompassing over 500 000 participants in the United Kingdom. Details of the study regarding the cohort have been previously described.16 Briefly, adults aged 45–80 years were randomly selected from the National Health Service between 2006 and 2010.17 Baseline visits were performed at 22 sites across the UK, involving health questionnaires, spirometric assessments and laboratory and imaging tests. A subset of baseline participants was invited to attend a follow-up visit, where they underwent a second spirometry examination about 7 years later.

COPDGene is a multicentre prospective cohort primarily comprising current or former smokers, and its comprehensive description is available elsewhere.18–20 Briefly, non-Hispanic White or Black individuals aged 45–80 years who had a smoking history of at least 10 pack-years at 21 clinical centres in the USA were recruited between 2007 and 2011. The cohort included smokers both with and without baseline obstructive lung disease, with the goal of ensuring the inclusion of one-third Black participants. The baseline enrolment process involved health questionnaires, spirometric assessments and high-resolution chest CT scans. Notably, individuals with known respiratory diseases (other than asthma or COPD) and those displaying significant interstitial lung disease (ILD) or bronchiectasis on chest CT were deemed ineligible. Participants were asked to return for a follow-up evaluation 5 years after the baseline visit. COPDGene received ethical approval from each site, and all participants provided written informed consent. This analysis was approved by the Mass General Brigham Institutional Review Board.

In our study, we included participants who underwent baseline and follow-up spirometry and had results for both the forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC). Additionally, smoking data were required as a key covariate and for stratification of analyses.

RA cases and non-RA comparators

To identify participants with RA, we used a combination of self-reported physician-diagnosed RA and the use of at least one disease-modifying antirheumatic drug (DMARD) at baseline in both the UK Biobank and COPDGene cohorts. Recognising the potential limitations of relying solely on self-reported RA status, we adopted a case definition that demonstrated improved validity by incorporating DMARD use, as a prior study employing a similar case definition reported a positive predictive value (PPV) of 88%.21 We included medications approved for RA by the US Food and Drug Administration and other DMARDs previously validated for the identification of patients with RA in cohort studies.22 To establish a non-RA comparator group, we defined them as participants without a reported history of RA and who were not using any DMARDs at baseline. In addition, those reporting a history of RA but not indicating DMARD use and those on one or more DMARDs without a history of RA were excluded from the comparator group. We have previously used these definitions of RA cases and non-RA comparators in previous studies using the UK Biobank and COPDGene data.8 12 13

Spirometric measures

In this study, spirometry was performed by trained clinical coordinators or respiratory therapists using the research protocol previously described in the UK Biobank16 and COPDGene.20 FEV1 and FVC were measured at both baseline and during the follow-up visits. The per cent predicted values for FEV1 and FVC (FEV1% and FVC%, respectively) were computed, adjusting for the individual’s age (including known decline with ageing), sex and height, using race-neutral GLI global 2022 equations.23 The calculations were processed using R-package ‘rspiro’. Before spirometry, bronchodilator medication was administered only to the COPDGene cohort and postbronchodilator spirometric results were used in the analyses. Annual changes in FEV1%, FEV1/FVC ratio and FVC% were calculated by comparing values at the two visits and factoring each participant’s period in years from the baseline visit to follow-up. An obstructive pattern was defined as an FEV1/FVC ratio <0.7, whereas a preserved ratio impaired spirometry (PRISm) was characterised by the absence of an obstructive pattern and FVC% <80%, employing standard cut-off points widely used in both clinical practice and research studies.24 25


For both cohorts, we collected information on age at the baseline visit, sex and self-reported race. Additionally, baseline measurements, including height, body weight and body mass index (BMI), were recorded. Smoking-related data, such as smoking status (never, former or current) and smoking pack-years, were extracted from the baseline health questionnaire. Furthermore, the presence of chronic respiratory illnesses (including asthma, bronchiectasis, COPD, ILD and idiopathic pulmonary fibrosis) and medications for these diseases (either inhaled or systemic) were identified using self-reported information collected during the baseline visit.

Statistical analysis

We reported the baseline characteristics and spirometric measurements at both baseline and follow-up visits, using frequencies, proportions and means with SD or medians with IQRs for RA cases and non-RA comparators in the UK Biobank and COPDGene cohorts. For the UK Biobank, subgroup analyses were performed based on smoking status and pack-years. Specifically, participants were categorised into those who had never smoked, those who ever smoked less than 10 pack-years, and those who ever smoked at least 10 pack-years, the latter mirroring the COPDGene inclusion criteria.

We examined the associations between RA and non-RA status with annual changes in FEV1%, FEV1/FVC ratio and FVC% using univariable and multivariable linear regressions. The multivariable model is adjusted for age, sex, BMI, smoking status (current/past), pack-years, baseline spirometric results (FEV1%, FEV1/FVC or FVC%) and use of inhaled/systemic medication use for obstructive lung disease. Given the longitudinal study design, only participants who attended follow-up visits, including spirometry examination, were included in the analysis. To address possible differential censoring, defined by dropout or death before the follow-up visit was due, an additional model with inverse probability of censoring weighting (IPCW) was employed within the same cohort. IPCW was calculated using the following covariates: RA/non-RA status, age, sex, race, smoking status, pack-years, body weight, BMI, self-reported symptoms of limited walking, history of cancer, use of medications for obstructive respiratory diseases and spirometric measurements at baseline. Analyses were stratified based on smoking status and cohort. Stratified analyses were also performed based on the presence or absence of an obstructive spirometry pattern at baseline. Among those with at least 10 pack-years, we also performed a pooled analysis of both cohorts to enhance power in this subgroup.

In a secondary analysis, we compared annual changes in FEV1%, FEV1/FVC ratio, and FVC% in patients with RA divided into three groups according to DMARDs used: tumour necrosis factor (TNF) inhibitors (with/without other DMARDs), methotrexate (MTX; with/without other DMARDs) and other DMARDs. We chose these groups since MTX and TNF inhibitors were the most prevalent drugs used in both cohorts. For descriptive purposes, we also reported smoking changes in RA and non-RA in COPDGene, since follow-up smoking data were not available in the UK Biobank. We also stratified the main analyses by sex to investigate possible differences.

Two-sided p values <0.05 were considered statistically significant. All analyses were performed using SAS V.9.4 (Cary, North Carolina). Patients and the public were not involved in the design or implementation of this study.


Study sample

Of the 502 378 UK Biobank participants, we identified 2222 RA cases and 301 098 non-RA comparators. Of the 10 371 COPDGene participants, 85 RA cases and 9280 non-RA comparators were identified. We excluded 107 healthy never smokers in COPDGene since there was no RA case in the subgroup. Among them, 188 RA cases and 32 560 non-RA comparators in the UK Biobank and 45 RA cases and 5175 non-RA comparators in the COPDGene had follow-up spirometry results (see flow diagram in figure 1).

Figure 1

Identification of participants with RA and non-RA comparators in UK Biobank and COPDGene. BR, bronchiectasis; DMARD, disease-modifying antirheumatic drug; ILD, interstitial lung disease; PFT, pulmonary function test in spirometry; RA, rheumatoid arthritis.

Baseline characteristics

Table 1 presents the baseline characteristics of both the RA cases and non-RA comparators. RA cases were older at baseline (UK Biobank: 57.3 vs 55.4 years; COPDGene: 63.8 vs 59.3 years) and more predominantly women (UK Biobank: 64.9% vs 53.0%; COPDGene: 66.7% vs 49.3%) than non-RA comparators. Regarding smoking data, RA cases had a higher proportion of past smokers (UK Biobank, 35.1% vs 29.9%; COPDGene 64.4% vs 51.7%) than non-RA comparators. Spirometric patterns at baseline showed a higher proportion of participants with PRISm in RA cases than in non-RA comparators (UK Biobank: 8.0% vs 4.4%; COPDGene: 20.0% vs 12.3%).

Table 1

Characteristics of RA cases and non-RA comparators with longitudinal spirometric measures in the UK Biobank and COPDGene (n=37 968)

All COPDGene participants met the inclusion criterion of being current or past smokers with a smoking history of at least 10 pack-years, whereas over half of the UK Biobank participants were never smokers. Even among participants with a history of at least 10 pack-years, the median number of pack-years in COPDGene participants was approximately two times that in UK Biobank participants for both RA cases and non-RA comparators (20 vs 41 pack-years and 21 vs 38 pack-years, respectively). Regarding spirometric patterns at baseline, patients in the COPDGene cohort tended to have fewer normal patterns and more obstructive and PRISm patterns, and they received medications for obstructive respiratory diseases more frequently compared with participants in the UK Biobank cohort. The results for the UK Biobank subgroup of ever smokers with a history of less than 10 pack-years are presented in online supplemental table S1.

Pulmonary function measures at baseline and follow-up

While the mean FEV1% at baseline was 97.8% in RA cases and 100.0% in non-RA comparators within the UK Biobank cohort, it was 78.2% in RA cases and 81.1% in non-RA comparators within the COPDGene cohort. Furthermore, the mean FVC% at baseline was 102.6% in RA cases and 104.7% in non-RA comparators within the UK Biobank cohort and 86.8% in RA cases and 92.4% in non-RA comparators within the COPDGene cohort (table 2).

Table 2

Spirometric measures at baseline and follow-up (n=37 968)

Among never smokers, the annual decline in both FEV1% and FVC% was numerically higher in RA cases than in non-RA comparators (FEV1%: −0.69 vs −0.35 per year; FVC%: −0.59 vs −0.39 per year). In contrast, for ever smokers with 10 or more pack-years, the decline in FEV1% was numerically lower in RA cases, with a modest improvement in RA cases of the COPDGene cohort (UK Biobank: −0.13 vs −0.44 per year; COPDGene: 0.18 vs −0.50 per year). The results for the UK Biobank subgroup of ever smokers with less than 10 pack-years are presented in online supplemental table S2.

Change of spirometric measures in RA cases compared with non-RA comparators

Among never smokers in the UK Biobank (n=110 RA cases and n=20 701 non-RA comparators), no significant difference was observed in the annual FEV1% and FVC% change in multivariable linear regression models with IPCW (table 3). However, in never smokers without an obstructive pattern at baseline, RA was significantly associated with more decline in FEV1% (β=−0.49, p=0.04).

Table 3

Results from the linear regression of annual change of measures in pulmonary function test (PFT), comparing RA cases versus non-RA comparators among never smokers (n=20 811)

In the pooled analysis of both cohorts among ever smokers with at least 10 pack-years (n=106 RA cases and n=13 237 non-RA comparators), both FEV1% and FEV1/FVC showed significantly less decline in RA cases compared with non-RA comparators (FEV1%: β=0.50, p=0.02; FEV1/FVC: β=0.31, p<0.01; table 4) after adjusting for age, sex, BMI, smoking status (current/past), pack-years, baseline spirometry results and inhaled/systemic medication use for obstructive lung diseases. These differences were even more pronounced in participants with an obstructive pattern at baseline (RA vs non-RA: FEV1%: β=1.12, p=0.01; FEV1/FVC: β=1.32, p<0.01). Given that these smokers were from two cohorts with different follow-up statuses, no analysis could be performed using the IPCW. The results for the UK Biobank subgroup of ever smokers with less than 10 pack-years are presented in online supplemental table S3.

Table 4

Results from the linear regression of annual change of spirometric measures, comparing RA smoker cases versus non-RA smoker comparators among current/past smokers (≥10 pack-years)

Online supplemental table S4 summarises the results of the regression models based on cohort. In COPDGene, both FEV1% and FEV1/FVC showed significantly lower decline in patients with RA than non-RA comparators, especially in those with obstructive spirometry pattern at baseline (RA vs non-RA: FEV1%: β=1.16, p=0.01; FEV1/FVC: β=1.78, p<0.01) in the multivariable linear regression models with IPCW. Within the entire UK Biobank cohort, while there were no statistical differences in the annual changes in FEV1% and FVC% between the two groups, the effect size direction was similar to COPDGene.

Types of DMARDs among RA cases

Details of the specific DMARDs used in RA cases are shown in online supplemental table S5. Approximately half of the participants in both the UK Biobank and COPDGene cohorts reported the use of MTX. Among all RA cases, no significant difference was observed in the annual FEV1%, FEV1/FVC and FVC% changes among users of TNF inhibitors, MTX and other DMARDs in the multivariable logistic regression model (online supplemental table S6). However, among ever smokers with at least 10 pack-years, RA cases taking other DMARDs experienced lower decline in FEV1% compared with those taking MTX (other DMARDs vs MTX: FEV1%: β=0.88, p=0.02).

Smoking changes in COPDGene

Among current smokers at baseline in COPDGene, 6/16 (38%) of RA cases reported smoking cessation at the 5-year follow-up visit compared with 623/2497 (25%) of non-RA comparators. The median pack-years at the 5-year follow-up was 42.0 (IQR 31.6, 51.5) for RA cases and 39.9 (IQR 27.0, 54.1, online supplemental table S7).

Analyses stratified by sex

Results stratified by male or female sex are presented in online supplemental tables S8–S11.


In this longitudinal study involving two large prospective cohorts in which participants underwent spirometry at baseline and follow-up for research purposes, we investigated changes in respiratory function on spirometry in RA cases compared with non-RA comparators, both among smokers and never smokers. Among never smokers without an obstructive respiratory pattern at baseline visit, RA cases had more decline in FEV1% than non-RA comparators. This finding aligns with results of previous studies reporting a higher incidence of obstructive lung diseases, such as COPD and asthma, in patients with RA compared with non-RA patients.4–8 14 Conversely, among smokers, particularly those with a pre-existing obstructive respiratory pattern at baseline, RA cases demonstrated a lower decline in FEV1% and FEV1/FVC compared with non-RA comparators. While this result was unexpected, it raises important clinical suggestions regarding pathogenesis, treatment and potential modifiability of obstructive lung disease among smokers. It is possible that obstructive lung disease in RA cases might have a unique autoimmune or inflammatory pathogenesis distinct from that in non-RA patients. It is also possible that DMARDs used in RA may have beneficial effects on pulmonary function, particularly in those with obstructive lung disease and heavy smoking.

The association between RA and pulmonary airway diseases, such as asthma, bronchiectasis and COPD, is well established.6 Observational studies have shown that airway diseases are risk factors for the development of RA,26–28 and, conversely, RA is a risk factor for the development of airway diseases,4–8 14 with particularly strong associations seen in seropositive RA.15 26 27 These findings suggest a shared pathogenesis in mucosal immunity between these airway diseases and RA. Lymphoid aggregates near airways and interstitium are present in early patients with RA29 30 and sputum is found to detect anti-CCP antibodies and rheumatoid factors earlier than serum.31 However, Kronzer et al reported a strong association between RA and various respiratory diseases, including COPD, asthma and other chronic upper airway diseases, only in non-smokers.27 Our study also found an association between RA and annual decline of FEV1% only in non-smokers who did not have an obstructive pattern on baseline spirometry, suggesting the need to investigate its distinct pathogenesis of obstructive lung diseases in patients with RA in non-smokers and smokers separately.

In diseases characterised by an obstructive pattern such as COPD and asthma, type 2 inflammation mediated by interleukin (IL)−4 and IL-13 plays a critical role in the pathogenesis for exacerbation of pulmonary function.32–36 The effectiveness of dupilumab, a fully human monoclonal antibody that blocks the shared receptor component of IL-4 and IL-13, improved FEV1 in trials for COPD and asthma.37 38 However, in the context of RA, the cytokine profile related to type 2 inflammation differs between early RA and established RA.39 Patients in the early stages of RA exhibit elevated levels of IL-4 and IL-13 in the serum and synovial fluid,40–42 which may be a factor in the high prevalence of chronic lung diseases, including COPD and asthma before RA diagnosis,27 whereas those with established RA have lower IL-4 and IL-13 levels in their serum and synovial fluid.43–46 Less decline of obstructive respiratory pattern observed in RA cases demonstrated in our study might be attributed to the suppressive state of IL-4 and IL-13, which are associated with the progression of obstructive lung diseases. These findings are hypothesis-generating on whether obstructive lung disease in patients with RA has distinct pathogenesis from that in patients with non-RA, particularly among those with smoking history.

The favourable course of FEV1% and FEV1/FVC in patients with RA compared with non-RA comparators among smokers suggests important implications for the potential treatment of obstructive lung diseases. In our study, RA cases were limited to those with a self-reported RA diagnosis and DMARDs use to differentiate them from those with other articular diseases, such as osteoarthritis, which is more prevalent than RA. The effects of DMARDs on obstructive lung disease have not been extensively investigated, and only a few clinical trials and observational studies have been reported. In a secondary analysis in our study, spirometry outcomes were compared based on the type of DMARD among patients with RA. Despite the limited sample size, which may lack statistical power, there was a tendency for a lower decline in FEV1% with the use of DMARDs in the following order: other DMARDs, TNF inhibitors and MTX. Even MTX, which showed a greater decline in FEV1% than other DMARDs in this analysis, has been reported to result in less exacerbation of COPD in a previous large-scale observational study47 and improvement in respiratory function in patients with asthma in a randomised controlled trial.48 Other types of DMARDs, such as TNF inhibitors and hydroxychloroquine, have also been reported to improve pulmonary function in asthma in a small case series study and a clinical trial, respectively.49 50 These findings highlight the importance of investigating the effects of DMARDs on obstructive lung diseases, particularly in smokers.

Our study had several strengths. First, to our knowledge, investigations on changes in respiratory function in patients with RA using data from large-scale cohorts, with non-RA individuals as the comparison group, have not been published before. The UK Biobank includes a diverse range of individuals from the general population. Additionally, COPDGene focuses on a high-risk population for respiratory diseases due to smoking. Second, both cohorts had a follow-up period of approximately 5–7 years, allowing us to detect changes in respiratory function that might not be apparent over shorter durations. Also, spirometry was performed for research purposes, so is less biased than using clinically-indicated spirometric results. Third, both cohorts allowed us to examine respiratory factors in detail, including smoking status, smoking pack-years and medication use for obstructive lung disease. Fourth, we addressed the possible bias caused by differential censoring, a frequently encountered problem in prospective cohorts, by incorporating the IPCW calculated using respiratory and other factors obtained from these large-scale cohorts into the model of our analyses. In addition, rich details on covariates such as BMI were available for adjustment and thus did not explain our findings. Fifth, we calculated FEV1% and FVC% using the recently developed race-neutral GLI global 2022 equations,23 which could significantly affect the interpretation of pulmonary function defects, especially among Black individuals.26 While the UK Biobank participants were predominantly White, the adoption of these new equations could reduce potential bias in the COPDGene, wherein one-third of the participants were Black.

Our study also had some limitations. First, we relied on a combination of self-reported RA status and the use of DMARDs to identify cases with and comparators without RA. While there may be some misclassification of RA case status, previous studies using similar methods have demonstrated a PPV of 88% for identifying RA.21 22 Also, relying only on self-reported RA without including DMARD use would increase sample size but decrease the validity of the exposure since this is typically low likelihood (20% or less) of having RA. There were less people with RA than initially expected in the COPDGene cohort of smokers. However, people with a chronic disease such as RA may be less likely to participate in this voluntary, longitudinal study, men composed a majority (55%), and we may have missed true cases that did not report RA or were not currently on a DMARD. Second, neither the UK Biobank nor COPDGene was designed to investigate RA, resulting in limited information on RA-related covariates. Consequently, we lacked data on serostatus, systemic inflammation, disease activity, bone erosions and RA duration, which could have provided valuable insights into the mechanism of the effects observed in our study. However, this would require RA-only analyses since most of these factors are not pertinent to people without RA. While we performed analyses of DMARDs, the sample size was even smaller and many specific drugs were not able to be examined. Studies are ongoing enrolling patients with RA to investigate how these RA-specific factors may influence spirometric changes and other markers of pulmonary health. A randomised trial investigating the effect of specific DMARDs on measures of lung health is needed for definitive conclusions. Third, despite utilising data from two cohorts with a large sample size, the RA group was relatively small, resulting in insufficient statistical power to thoroughly investigate factors associated with changes in pulmonary function among patients with RA. However, the findings were robust across two large cohorts. Studies are ongoing to investigate the impact of smoking on spirometric and chest imaging measures among only patients with RA. Fourth, there may be additional factors that were unmeasured that may have affected findings. In particular, we had limited data on factors occurring between visits that may have differed by RA status and mediated findings, such as smoking changes and new-onset pulmonary diseases such as asthma or ILD. These postbaseline factors may be on the causal pathway between exposure and outcome, so may be potential mechanisms of the effects we observed but should not be adjusted for in analyses. While smoking is known to be deleterious for lung health and can induce some forms of ILD, some paradoxical improvements have also been noted.51 It is possible that RA may affect the spirometric trajectory in such a way that we investigated a plateau phase rather than during decline. We are unable to fully examine the trajectory of spirometric changes with only two measures. Fourth, we did not have serial measures of chest CT findings or other PFTs such as diffusion capacity of the lungs for carbon monoxide. In light of our findings and these limitations, it is recommended that future studies incorporate an RA-specific cohort and investigate the mechanisms of pulmonary disease in patients with RA, leveraging both detailed respiratory and RA-specific data.

In conclusion, RA cases had less decline in FEV1% and FEV1/FVC than non-RA comparators among smokers, particularly among those with an obstructive pattern on baseline spirometry. These associations were not attributable to variations in smoking, suggesting that RA with obstructive lung disease may be a distinct phenotype. However, RA cases without obstructive pattern that were never smokers had more decline of FEV1% than non-RA comparators, emphasising these patients could have deterioration of pulmonary function beyond what is expected based on ageing alone. These findings emphasise the need for further studies to explore the mechanisms and potential reversibility of systemic inflammation, autoimmunity and the impact of RA treatment on pulmonary function.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by the Mass General Brigham Institutional Review Board, Reference number: 2020P000558. Participants gave informed consent to participate in the study before taking part.


We would like to acknowledge the UK Biobank and COPDGene investigators, staff, and participants for their valuable contributions to this study.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • X @Juge_P_A, @jeffsparks

  • Contributors KH, GMD and JS had access to the study data, developed the figures and tables and vouched for the data and analyses. KH performed the statistical analyses and contributed to data quality control, data analysis, and interpretation of the data. KH, GMD, P-AJ, MM, MHC, XW, MLP, TJD, GLK, DS-P, KY, PFD, ZW, EAR, GMH, EKS, SYA, RSJE, GRW and JAS contributed to data collection, data analysis, and interpretation of the data. JS directed the work, designed the data collection methods, obtained funding, contributed to data collection, data analysis, and interpretation of the data and had final responsibility for the decision to submit for publication. All authors contributed intellectual content during the draft and revision of the work and approved the final version to be published. JAS accepts full responsibility for the work and the conduct of the study, had access to the data, and controlled the decision to publish.

  • Funding GMD is supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (T32 AR007530) and the VERITY Pilot & Feasibility Award. MHC is supported by the National Institutes of Health/National Heart, Lung, and Blood Institute (R01HL153248, R01HL149861, R01HL147148). TJD is supported by the National Institutes of Health/National Heart, Lung, and Blood Institute (R01HL155522). ZW is supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (R01 AR080659, K23 AR073334 and R03 AR0789938). SYA is supported by the National Institutes of Health/National Heart, Lung, and Blood Institute (K08HL145118). JS is supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (grant numbers R01 AR080659, R01 AR077607, P30 AR070253, and P30 AR072577), the R. Bruce and Joan M. Mickey Research Scholar Fund, and the Llura Gund Award for Rheumatoid Arthritis Research and Care. The COPDGene study (NCT00608764) is supported by grants from the NHLBI (U01HL089897 and U01HL089856), by NIH contract 75N92023D00011, and by the COPD Foundation through contributions made to an Industry Advisory Committee that has included AstraZeneca, Bayer Pharmaceuticals, Boehringer-Ingelheim, Genentech, GlaxoSmithKline, Novartis, Pfizer and Sunovion. The funders had no role in the decision to publish or preparation of this manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard University, its affiliated academic health care centers, or the National Institutes of Health.

  • Competing interests PAJ reports grant funding and other support from Novartis, Galapagos and Boehringer Ingelheim, unrelated to this work. MM reports institutional grant support from Bayer and Honoraria from Chickasaw Nation. MHC has received grant funding from Bayer, unrelated to this work. TJD received support from Bayer and has been part of a clinical trial funded by Genentech, unrelated to this study. PFD reports grant funding from Bristol Myers Squibb. ZW has received grant funding from Bristol-Myers Squibb and Principia/Sanofi and consulting fees from Viela Bio, Zenas BioPharma, Horizon Therapeutics, Sanofi, MedPace, BioCryst, Amgen, Nkarta, Inc, Adicet Bio, and Therapeutic’s and participation in data safety monitoring board or advisory board for Sanofi, Horizon, Novartis, Visterra/Otsuka and Shionogi, unrelated to this work. GMH reports consulting fees from Boehringer-Ingelheim, and the Gerson Lehrman Group, unrelated to this work. EKS has received grant support from Bayer and Northpond Laboratories, unrelated to this work. SYA reports consulting fees from Verona Pharmaceuticals and Vertex Pharmaceuticals and is cofounder and co-owner of Quantitative Imaging Solutions. RSJE reports contracts from Lung Biotechnology and Insmed, received a grant support from Boehringer Ingelheim and is cofounder and an equity holder of Quantitative Imaging Solutions. GRW reports grants from Boehringer Ingelheim, consultancy for Pulmonx, Janssen Pharmaceuticals, Novartis, and Vertex, and is founder and co-owner of Quantitative Imaging Solutions. JS has received research support from Bristol Myers Squibb and performed consultancy for AbbVie, Amgen, Boehringer Ingelheim, Bristol Myers Squibb, Gilead, Inova Diagnostics, Janssen, Optum, Pfizer, ReCor, Sobi, and UCB, unrelated to this work. Other authors report no competing interests.

  • Provenance and peer review Not commissioned; externally peer-reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.