Authors: Henrik Albæk Jacobsen, Anastasia Karachalia Sandri, Ulla Møller Weinreich, Tine Jess, Lone Larsen
Categories: Inflammatory Bowel Disease, Crohn's disease, asthma, bronchiectasis, chronic obstructive pulmonary disease (COPD), epidemiology, pulmonary disease, ulcerative colitis, Original Article
Source: United European Gastroenterology Journal
Doi: 10.1002/ueg2.12527
Inflammatory bowel disease (IBD) is associated with disease manifestations in organs other than the gastrointestinal tract. In this study, we aimed to estimate the odds of obstructive lung disease (OLD) before IBD onset and the risk of OLD after IBD onset.
In a nationwide population‐based Danish cohort study from 1999 to 2018, individuals with IBD and OLD were identified using the Danish registries. Between 2003 and 2013, 24,238 individuals with IBD were identified and matched 10 with non‐IBD individuals. Logistic regression was used to estimate the prevalence odds ratio for OLD before IBD onset. Time‐to‐event analysis was performed to explore the risk of OLD after IBD onset. In a sensitivity analysis, the time‐to‐event analysis was repeated using the composite outcome OLD and the separate outcomes, chronic obstructive pulmonary disease (COPD), asthma, and bronchiectasis.
Individuals with IBD were 60% more likely to have OLD before onset (adjusted odds 1.60, 95% confidence interval [CI]: 1.53–1.67). Furthermore, their risk of OLD was more than 40% higher after IBD diagnosis (adjusted hazard ratio [aHR]: 1.43, 95% CI: 1.37–1.49). The sensitivity analysis increased the risk to 60% (aHR: 1.63, 95% CI: 1.53–1.73). Similar results were found for COPD and asthma separately, whereas the risk of bronchiectasis increased more than 2‐fold (aHR: 2.44, 95% CI: 1.91–3.11).
The odds of OLD before‐ and the risk following an IBD diagnosis were increased. We encourage physicians to be vigilant of pulmonary symptoms in persons with IBD and gastrointestinal symptoms in individuals with OLD.
Keywords: asthma, bronchiectasis, chronic obstructive pulmonary disease (COPD), Crohn's disease, epidemiology, pulmonary disease, ulcerative colitis
Manifestations of inflammatory bowel disease (IBD) are, despite what the name suggests, not restricted to the gastrointestinal tract. ^1^ , ^2^ Extraintestinal manifestations involving the liver, joints, skin, and eyes have been known in IBD for decades. Also, an interrelation between IBD and obstructive lung diseases (OLD), including bronchitis, chronic obstructive lung disease (COPD), asthma, and bronchiectasis, has long been reported. ^3^ Historically, airway diseases have been thought of as extraintestinal manifestations occurring after IBD onset. ^4^ However, growing evidence suggests that the risk of IBD is also increased in patients already diagnosed with OLD. ^5^ , ^6^ Hence, the increased risk of developing OLD might be bidirectional in relation to IBD onset. Previous epidemiological studies that have addressed this topic have primarily been prevalence studies. These studies have reported an increased co‐existence of chronic lung diseases and IBD without providing insight into the temporal sequence of disease occurrence. ^7^ , ^8^ , ^9^ , ^10^ , ^11^
The respiratory and gastrointestinal tracts share structural similarities and microbial colonization present in both organs is believed to play important roles in the OLD and IBD pathogenesis. Based on current evidence, a gut‐lung axis has been described where dysbiosis and inflammation in the gut may influence the respiratory tract and vice versa. ^12^ , ^13^
In the present study of more than 24,000 individuals with IBD, we used a nationwide population‐based cohort to investigate the odds of OLD before IBD onset. We also examined the incidence and risk of OLD after IBD onset, including asthma, COPD and bronchiectasis, compared to a matched population without IBD.
The present study applied prospectively recorded data from the Danish Civil Registration System, the Danish National Patient Registry (DNPR), and the Danish National Prescription Registry (see Supporting Information S1). ^14^ , ^15^ , ^16^ The study period was from 1 January 1999, to 31 December 2018.
We identified newly diagnosed IBD cases in Denmark from 1 January 2003 to 31 December 2013 who resided in Denmark for at least 4 years before diagnosis (or since birth for those ≤4 years old). IBD diagnosis was defined as two IBD registrations based on the International Classification of Diseases, 10^th^ revision (ICD‐10) codes (Crohn's disease [CD]: K50; ulcerative colitis [UC]: K51) up to two years apart, either through outpatient or inpatient visits. ^17^ Those with prior IBD registrations before 2003 were excluded.
The IBD subtype was determined by the ICD‐10 codes from the two contacts. If both CD and UC were recorded, the subtype was determined by the latest code. The first IBD registration date was the diagnosis date, and the second IBD registration was used for matching. Each eligible IBD case was matched 10 with individuals from the general population without IBD, based on age at diagnosis, birth sex, calendar period, and municipality of residence at the time of diagnosis. Each matched individual was assigned an index date according to the date of IBD diagnosis of the corresponding IBD person.
The primary (composite) outcome was OLD defined based on either at least one relevant diagnosis, that is, one ICD‐10 diagnosis code (J40–J44 chronic bronchitis not specified as acute or chronic, simple and mucopurulent chronic bronchitis, unspecified chronic bronchitis, emphysema, and COPD; J45 asthma; J47 bronchiectasis), or relevant prescriptions, that is, two Anatomical Therapeutic Chemical (ATC) codes (Supporting Information S1, Table S1) on two different dates within 365 days, whichever came first. We also considered diagnoses based only on ICD‐10 codes of OLD (secondary composite outcome) as well as COPD (J40–J44), asthma (J45) and bronchiectasis (J47) separately in sensitivity analyses.
Logistic regression was used to estimate the prevalence Odds ratio (OR) for OLD (primary outcome) among persons with IBD compared to matched controls up to 4 years before IBD diagnosis/index date (overall for IBD, and separately for UC and CD). Time‐to‐event analysis was performed to estimate the risk of OLD (primary outcome) after IBD diagnosis/index date among persons with IBD compared to matched controls. The Kaplan‐Meier approach was used to assess the cumulative OLD incidence, and Cox proportional regression estimated OLD risk after IBD diagnosis/index date in IBD individuals compared to those without IBD. Individuals diagnosed with OLD 4 years prior to IBD diagnosis/index date were excluded from the time‐to‐event analysis. Eligible individuals were followed up until OLD (study outcome), emigration, death, or end of follow‐up (31 December 2018).
In the sensitivity analyses, time‐to‐event analysis was repeated using the secondary composite outcome (OLD based on ICD‐10 codes only) and separate outcomes (COPD, asthma, bronchiectasis). Incidence rates (IRs) per 1000 person‐years were calculated for COPD, asthma, and bronchiectasis in individuals with and without IBD.
All analyses were adjusted for sex, age group at IBD diagnosis/index date, and year of entry in the cohort based on IBD diagnosis/index date. Cox regression analysis was also adjusted for IBD subtype. Descriptive statistics included frequency (percentage) for categorical variables and mean (standard deviation) or median (interquartile range) for numeric variables. OR and hazard ratio estimates from logistic and Cox regression analyses, respectively, were presented with corresponding 95% confidence intervals (CIs). Log‐rank and chi‐square tests compared Kaplan‐Meier curves and regression estimates, respectively, with a significance threshold set at 0.05. SAS 9.4 (SAS Institute) was used for all analyses.
The study population comprised 266,618 persons, including 24,238 persons with IBD, and 242,380 matched persons without IBD (Table 1). Of the 24,238 persons identified with IBD, 16,869 (69.6%) and 7369 (30.4%) had a diagnosis of UC and CD, respectively. A total of 21,103 (7.9%) persons were identified with a diagnosis of OLD prior to the IBD diagnosis/index date, out of whom 2785 (11.5%) were in the IBD group and 18,318 (7.6%) were in the control group. In the IBD cohort, a diagnosis of OLD was more frequent among females (8.5%) than males (7.2%). When looking at the distribution of OLD in relation to age group, most persons with OLD were found in the oldest age groups ([41–64, 11.8%], [≥65, 19.9%]). The proportion of persons with OLD was comparable between those with a diagnosis of UC and CD (11.3% and 11.9%, respectively). Still, it was higher compared to their corresponding controls (7.7% UC controls, and 7.1% CD controls). This was the case across all other covariates, that is, a higher proportion of OLD was evident among the IBD group compared to the corresponding group of controls.
The overall adjusted odds ratio (aOR) for OLD prior to IBD diagnosis/index date was higher among persons with IBD compared to non‐IBD individuals (aOR: 1.60, 95% CI: 1.53–1.67) (Table 2, Tables S2 and S3). Males had 16% lower odds for OLD than females (aOR: 0.84, 95% CI: 0.81–0.86). Compared to the young adults ([17–40] years of age), the odds for OLD were approximately 40% higher among children and older adults (aOR: 1.40, 95% CI: 1.31–1.48 for ≤16 years of age; aOR: 1.43, 95% CI: 1.38–1.49 for [41–64] years of age). However, odds were markedly higher in the oldest age group (aOR: 2.70, 95% CI: 2.61–2.80 for ≥65 years of age). Discrete differences were also noted in the year of cohort entry, with aORs slightly increasing over time.
After excluding the 21,103 individuals who had OLD already diagnosed prior to their IBD diagnosis/index date, a total of 245,515 persons were eligible for follow‐up (Table S4). The median (interquartile IQR) follow‐up time in the entire cohort (N = 245,515) was 9.7 (7.1–12.6) years. Among these persons, a total of 20,998 (8.6%) were identified with a diagnosis of OLD, out of whom 2384 (11.1%) were in the IBD group and the remaining 18,614 (8.3%) in the control group. The characteristics of this cohort were comparable to those of the initial study population. Slight differences were only evident concerning OLD across age groups among persons with IBD. In particular, the proportion of OLD diagnoses among persons with IBD decreased from 11.2% to 6.5% among children (≤16 years of age), while it increased from 11.8% to 14.3% among persons between 41 and 64 years of age.
The cumulative incidence of OLD among persons with IBD was consistently higher over time compared to the group of non‐IBD individuals, reaching 18.1% (95% CI: 16.7%–19.2%) and 13.6% (95% CI: 13.2%–13.9%) after 16 years of follow‐up, respectively (Figure 1). Females with IBD appeared to have the highest cumulative incidence of OLD when stratifying by sex (19.2%, 95% CI: 17.4%–20.6% after 16 years of follow‐up). However, older age appeared to play an important role, especially in the presence of IBD (35.4%, 95% CI: 29.1%–39.9% for ≥65 years of age with IBD; 26.9%, 95% CI: 24.4%–28.2% for ≥65 years of age without IBD).
FIGURE 1 Kaplan–Meier survival analysis curves for cumulative incidence of obstructive lung disease. CD, Crohn's disease; IBD, inflammatory bowel disease; UC, ulcerative colitis.
The adjusted hazard ratio (aHR) for OLD among persons with IBD was 1.43 (95% CI: 1.37–1.49) compared to persons without IBD (Figure 2, Table S5). For IBD subtypes, the aHRs were 1.42 (95% CI: 1.35–1.49) and 1.46 (95% CI: 1.35–1.49) for UC and CD, respectively. No statistically significant difference in the aHR between males and females was noted (aHR: 1.46, 95% CI: 1.38–1.54 for females; aHR: 1.39, 95% CI: 1.30–1.48 for males). Significant differences were observed regarding aHRs across age groups, with the aHR 1.08 (95% CI: 0.87–1.33) among children (≤16 years of age) to 1.54 (95% CI: 1.41–1.67) for the oldest age group (≥65 years of age).
FIGURE 2 Risk of obstructive lung disease based on both ICD‐10 and anatomical therapeutic chemical codes among persons with IBD compared to matched persons without IBD in Denmark 2003–2018: Results of Cox regression analyses. IBD, inflammatory bowel disease.
Out of the 20,998 diagnoses of OLD captured during the follow‐up period, 9292 (44.3%) based on ICD‐10 codes were considered in the sensitivity analyses (Figure 3, Table S6). Results of this analysis were comparable with those of the primary analysis, although all aHRs were somewhat elevated. Thus, the aHR among persons with IBD increased to 1.63 (95% CI: 1.53–1.73). Interestingly, the difference between aHR for UC and CD also increased, while differences in aHRs across age groups did not reach statistical significance.
FIGURE 3 Risk of obstructive lung disease based on ICD‐10 codes only among persons with IBD compared to matched persons without IBD in Denmark 2003–2018: Results of Cox regression analyses. IBD, inflammatory bowel disease.
Analysis of the separate components (COPD, asthma, and bronchiectasis) revealed increased IRs per 1000 person years (PYs) (IRs per 1000 PY) (3.6 vs. 2.4 for COPD; 2.3 vs. 1.4 for asthma; 0.4 vs. 0.2 for bronchiectasis) and, in turn, elevated risk for all three diagnoses among persons with IBD compared to non‐IBD individuals (aHR: 1.61, 95% CI: 1.49–1.74 for COPD; aHR: 1.62, 95% CI: 1.47–1.78 for asthma; aHR: 2.44, 95% CI: 1.91–3.11 for bronchiectasis) (Table 3, Figure S1).
In this nationwide population‐based cohort study of 24,238 individuals with IBD, the odds of an OLD diagnosis increased by 60% in the 4 years before IBD onset compared to matched non‐IBD individuals from the general population. Furthermore, persons with IBD had a more than 40% increased risk of OLD after their IBD diagnosis compared to individuals without IBD. Additionally, when using ICD‐10 codes exclusively, this risk appeared even higher, reaching more than 60%. Similar results were found when considering COPD and asthma separately, whereas the risk of bronchiectasis increased by more than 2‐fold.
Previous studies suggest that subclinical pulmonary manifestations in persons with IBD are detectable on imaging and in pulmonary function tests. ^18^ , ^19^ However, evidence of OLD development after IBD onset is scarce, and incidence studies are non‐existing. Two studies have investigated the prevalence of OLD in IBD cohorts. In a Canadian population‐based study from Manitoba, the prevalence ratio (PR) for asthma in UC patients was 1.53, with one health contact increasing to 1.66 with at least 5 health contacts; for bronchitis, the PR was 1.33, rising to 2.10. In CD, the PR for asthma was estimated between 1.34 and 1.43, and for bronchitis between 1.36 and 1.86, depending on health contacts. ^7^ However, in a recent prevalence study from the United States, using the Nationwide Readmission Database, a 46% increased prevalence of bronchiectasis was reported among individuals with IBD. Notably, merely a 5.5% higher rate of asthma and no difference in COPD was reported. ^11^ Estimates from the former study using a period prevalence over approximately 16 years seemed more consistent with our findings compared to the latter study, which used a point prevalence during hospitalization.
Other studies have investigated the prevalence of IBD in individuals with OLD. A Swedish study, using data from an inpatient register, reported an approximately two‐fold increased risk of IBD among patients who later developed COPD. ^10^ In a British prevalence study, estimates were even higher for COPD (OR: 3.87, 95%CI: 1.19–12.62), asthma (OR: 2.54, 95%CI: 0.78–8.26), and bronchiectasis (OR: 8.38, 95% CI: 2.43–28.89). ^8^ Interestingly, although a high degree of uncertainty was inherent in their estimates, the most significant association was between IBD and bronchiectasis, in line with our results. Furthermore, the most prominent finding in a study on IBD patients without overt respiratory symptoms undergoing high‐resolution computed tomography imaging was bronchial wall thickening apart from centrilobular nodules. ^20^ Additionally, in an older review of case series covering thoracic manifestations of IBD, bronchiectasis was the most frequently reported IBD‐associated lung disease. ^21^ However, whether these notable associations between IBD and bronchiectasis reflect an underlying pathophysiological mechanism is yet to be untangled.
Despite the heterogeneity in the methods and varying estimates, most previous evidence points toward an association between OLD and IBD. However, prevalence studies do not provide insight into the temporal sequence of disease development. To our knowledge, only two incidence studies in this area exist. Hemminki et al., using the Swedish nationwide Hospital Discharge Register, reported an increased risk of several autoimmune diseases in individuals already diagnosed with asthma. ^5^ In the study by Brassard et al., cases were defined based on administrative prescription data and labeled as either asthma or COPD, depending on the age at the time of diagnosis. ^6^ Both studies reported an increased risk of IBD except for the risk of UC in individuals with asthma, which was not significant in the latter study. Importantly, these studies investigated the incidence of IBD in individuals already diagnosed with COPD and asthma, contrary to our study.
In the present study, we found no significant differences between IBD subtypes, though CD appeared to be associated with a slightly elevated risk of OLD. While sex differences in OLD risk were not statistically significant, females with IBD, possibly due to a higher CD incidence in the inclusion period, showed a tendency towards higher OLD risk both before and after IBD onset. ^22^ Age was an important factor for OLD in our study. OLD exhibited a higher likelihood of manifesting before IBD onset in the youngest age group. Considering the age‐related peak incidence patterns of different OLD subtypes, this association is most likely attributed to asthma. Nevertheless, a diagnosis of OLD was most predominant both before and after an IBD diagnosis within the oldest age groups. Notably, this difference between age groups was no longer significant when using ICD‐10 diagnosis codes exclusively, restricting the cohort to individuals with in‐ or outpatient hospital contacts. This observation might suggest that patients with IBD requiring hospitalization for OLD experience an earlier onset of IBD. In the study by Ekbom et al., using data from their inpatient register, patients with UC had a higher prevalence of COPD in the age group below 60 years compared to the older age groups. ^10^
This study possesses several strengths. Firstly, the population‐based study design allowed the inclusion of a large, unselected cohort. Secondly, the substantial study population ensured highly reliable estimates, and the length of the follow‐up period (up to 16 years) enabled sufficient time to capture OLD occurrence. Thirdly, the Danish registries offer complete coverage of the entire country's population since 1968. Lastly, the positive predictive value (PPV) of IBD (PPV: 95%) ^17^ and COPD (PPV: 92%) ^23^ related ICD‐10 codes in DNPR has previously been assessed and found to be high. A study validating asthma in the DNPR reported a specificity of 98% and a sensitivity of 44%. ^24^ The latter underlined the fact that many individuals with asthma and COPD in Denmark are managed by a general practitioner or a private consultant pulmonologist without being referred to a hospital. To mitigate selection bias by only using hospital contacts (ICD‐10 codes), we included prescription data from the Danish National Prescription Registry to identify individuals with OLD not already verified through a diagnosis in the DNPR. Importantly, our definition of OLD did not include ATC codes for systemic corticosteroids, but only inhaled medications (see Supporting Information S1).
Limitations mainly related to the use of administrative data in this study should also be considered. Evidence on the use of ATC codes to classify OLD cases is limited. While inhalation medications prescribed for OLD could also be used for other transient pulmonary conditions, the criterion of two prescriptions on two different dates within 1 year was established based on the available but limited evidence, considering both sensitivity and specificity. One study, using the same definitions as the present study, reported a PPV of 30%–97% (average: 65%) increasing with age. ^25^ Another study on pediatric asthmatic patients reported a specificity of 86% and sensitivity of 63% using only one claimed prescription. ^26^ These studies illustrate that the validity of COPD and asthma based on ATC codes depends significantly on the age of the specific population. COPD is more prevalent in older populations, whereas asthma is more prevalent in pediatric populations. Because of a significant overlap between drugs administered to patients with OLD across the age spectrum, we did not use ATC codes to label the different OLD subtypes. Instead, a sensitivity analysis was conducted exclusively using ICD‐10 codes to estimate the separate risks associated with specific OLD subtypes. This approach acknowledges that COPD, asthma, and bronchiectasis, despite falling under the category of OLD, are distinct conditions.
The potential influence of surveillance bias should be addressed. When assessing the probability of OLD prior to IBD, surveillance may have less impact since the manifestation of IBD symptoms often prompts a gastroenterology referral, irrespective of whether an individual is already under surveillance for OLD at the general practitioner or outpatient clinic. In contrast individuals with OLD might remain undiagnosed, especially when presenting with subtle symptoms. ^27^ While regular follow‐up could potentially lead to surveillance bias, screening for pulmonary symptoms in IBD outpatient clinics is not a routine practice. Immortal time bias related to our case definition of IBD requiring two registrations should also be addressed. To mitigate the time during which individuals with IBD were not at risk, the interval between the two contacts was limited to 2 years. In practice, among the 24,238 eligible persons with IBD included in this study, the median time between these two registrations was only 27 days (IQR: 13–56 days). Furthermore, matching was done using the second IBD registration date to ensure balance between the two comparator groups.
Another limitation pertains to the recording of tobacco consumption. Smoking is a recognized risk factor for CD and COPD. ^28^ Conversely, it is thought to protect against the development of UC. ^29^ Regrettably, the documentation of smoking habits in the administrative data is insufficient, which prevented us from adjusting for smoking in our analyses. ^30^ This limitation is a widespread challenge even in studies related to this topic, as longitudinal data on tobacco use is not readily available at the population level and is typically derived from survey data.
However, the observed associations in the current study are not believed to be explained solely by the potential confounding effects of tobacco, and there are several reasons supporting this assertion. The marginal increased risk of OLD in CD when compared to UC does not align with the protective role typically attributed to tobacco use in UC. Additionally, the significantly increased odds of OLD preceding the onset of IBD in the youngest age group cannot reasonably be attributed solely to tobacco use. Furthermore, despite the potential confounding influence of smoking from an etiological perspective, our results remain pertinent for clinicians involved in the care of these patients, especially since early diagnosis might improve long‐term outcomes and reduce the increased mortality reported in individuals diagnosed with both CD and COPD. ^31^ , ^32^
In conclusion, in a large nationwide population‐based cohort study, we demonstrate an increased prevalence OR of OLD prior to IBD onset compared to a matched population without IBD. Furthermore, unprecedentedly, we demonstrate an increased incidence and risk of OLD including asthma, COPD, and bronchiectasis in individuals with IBD.
Henrik Albæk Jacobsen, Tine Jess, and Lone Larsen equally took part in conceptualization. All authors contributed equally to the design of the methodology. Anastasia Karachalia Sandri was responsible for data analysis and visualization. All authors took part in the interpretation of the results. Henrik Albæk Jacobsen was responsible for writing the original draft. All authors participated in the review process and approved the final version of the article, including the authorship list.
The corresponding author confirms on behalf of all authors that there are no conflicts of interest or financial disclosures to report.
Study permission was obtained from the Danish Data Protection Agency. Ethical approval is not required for registry‐based research in Denmark.
The present study was supported by a National Center of Excellence Grant (DNRF148) from the Danish National Research Foundation and a grant from the Danish Colitis and Crohn's patients' organization.
Jacobsen HA, Sandri AK, Weinreich UM, Jess T, Larsen L. Increased risk of obstructive lung disease in inflammatory bowel a population‐based cohort study. United European Gastroenterol J. 2024;12(4):477–86. 10.1002/ueg2.12527
The study is based on data from the Danish nationwide (https://sundhedsdatastyrelsen.dk). The register data are protected by the Danish Act on Processing of Personal Data. Access can be attained through the Danish Data Protection Agency and the Danish Health Data Authority.
The study is based on data from the Danish nationwide (https://sundhedsdatastyrelsen.dk). The register data are protected by the Danish Act on Processing of Personal Data. Access can be attained through the Danish Data Protection Agency and the Danish Health Data Authority.