Article Text
PDF
Abstract
Background Smoking has been considered an important risk factor for idiopathic pulmonary fibrosis (IPF) incidence. However, there are no population-based large-scale studies demonstrating the effects of smoking on the development of IPF. We aimed to evaluate the effect of smoking on IPF development using a nationwide population-based cohort.
Methods Using the Korean National Health Information Database, we enrolled individuals who had participated in the health check-up service between 2009 and 2012. Participants having a prior diagnosis of IPF were excluded. The history of smoking status and quantity was collected by a questionnaire. We identified all cases of incident IPF through 2016 on the basis of ICD-10 codes for IPF and medical claims. Cox proportional hazards models were used to calculate the adjusted HR (aHR) of the development of IPF.
Results A total of 25 113 individuals (0.11%) with incident IPF were identified out of 23 242 836 participants registered in the database. The risk of IPF was significantly higher in current and former smokers than in never smokers, with an aHR of 1.66 (95% CI 1.61 to 1.72) and 1.42 (95% CI 1.37 to 1.48), respectively. Current smokers had a higher risk of IPF than former smokers (aHR 1.17, 95% CI 1.13 to 1.21). The risk of IPF development increased as the smoking intensity and duration increased.
Conclusion Smoking significantly increased the risk of IPF development. Current smokers had a higher risk of IPF than former smokers. A dose–response relationship was observed between smoking and the development of IPF.
- idiopathic pulmonary fibrosis
- clinical epidemiology
- tobacco and the lung
Data availability statement
No data are available. The data that support the findings of this study are available from the National Health Insurance Service in Korea but restrictions apply to the availability of these, which were used under license for the current study, and so are not publicly available.
Statistics from Altmetric.com
Key messages
What is the key question?
How much does smoking actually affect the development of idiopathic pulmonary fibrosis (IPF)?
What is the bottom line?
In this nationwide retrospective cohort study, 23 242 836 participants were included, 25 113 of whom were diagnosed with incidental IPF during the study. The adjusted HRs of IPF development for current and former smokers compared with never smokers were 1.66 and 1.42, respectively.
Why read on?
In this nationwide cohort study, the risk of developing idiopathic pulmonary fibrosis was higher in current and former smokers than in never smokers; furthermore, the risk was higher in current smokers than in former smokers.
Introduction
Idiopathic pulmonary fibrosis (IPF) is the most common type of idiopathic interstitial pneumonia characterised by progressive fibrosis of lung parenchyma and declining lung function. The clinical course of the disease in individual patients is variable; however, IPF generally carries a poor prognosis with median survival of 3 years after diagnosis.1 The most common cause of death in patients with IPF is respiratory failure caused by disease progression.2 Although antifibrotic therapies show significant effects on attenuating the disease progression, the efficacy is limited.3 As yet there is no treatment modality for cure.4 Therefore, given that it is crucial to prevent the development of IPF, identifying and controlling the risk factors are indispensable components of the management of this disease.
The pathogenic process underlying IPF is thought to reflect repetitive injuries to the pulmonary epithelium and subsequent aberrant healing processes.5 Potential extrinsic risk factors in IPF development include chronic exposure to metal or wood dust, viral infections and cigarette smoking.6 7 Among these, cigarette smoking is regarded as one of the most strongly associated risk factors in the development of IPF. Previous studies have reported that the proportion of smokers among patients with IPF ranges from 41% to 83%, and smokers carry a 1.6-fold to 2.9-fold greater risk of developing IPF compared with non-smokers.4 8–10
There are several suggested mechanisms explaining the effect of smoking in the pathogenesis of IPF. Cigarette smoke contains many toxic chemicals and particulate matter which alter the oxidant and antioxidant balance.11 Smoking also increases the permeability of alveolar epithelial cells and endothelium,12 13 which can accelerate the production of reactive oxidants and result in impaired tissue regeneration.14 Oxidants play a role in epithelial damage,15 and increased oxidative stress may affect the gene expression of various profibrotic factors and the activation of transforming growth factor-β1, an important pathway in the pathogenesis of IPF.16 In addition, smoking promotes shortening of telomeres17 and endoplasmic reticulum stress, which are reported to be strongly associated with IPF.18
Despite the robust biological mechanism and association between smoking and IPF incidence, data regarding the impact of specific smoking status and quantity on IPF incidence are relatively limited. In particular, studies on the effect of smoking status (current, former and never smoker) on the development of IPF have reported inconsistent results. In a previous study, former smokers (OR 1.90, 95% CI 1.3 to 2.9) but not current smokers (OR 1.06, 95% CI 0.6 to 1.8) had a significantly higher risk of IPF incidence than never smokers.9 Furthermore, another study reported that former smokers had a significantly higher risk of IPF development than never smokers (OR 3.04, 95% CI 1.72 to 5.38), while current smokers did not (OR 1.29, 95% CI 0.69 to 2.42).19 A recently published case–control study reported that former smoking was significantly associated with an increased risk of IPF (adjusted OR 2.61, 95% CI 2.06 to 3.29); conversely, current smoking was related to a decreased risk of IPF (adjusted OR 0.33, 95% CI 0.17 to 0.62).20 On the other hand, a recent prospective cohort study reported that both current (HR 2.49, 95% CI 1.99 to 3.12) and former smokers (HR 2.03, 95% CI 1.73 to 2.39) had a significantly higher risk of IPF than never smokers.21
Data on the association between smoking quantity and IPF incidence are also scarce.6 9 22 23 Smokers with smoking quantity ranging from 21 to 40 pack-years (PY) had a higher risk of IPF incidence, but those with smoking quantity more than 40 PY had an insignificant OR.9 23 Magnus et al showed that smokers with a higher smoking quantity had a higher risk of IPF; however, the subjects of this study were limited to very patients with severe IPF with long-term oxygen therapy.22 Moreover, so far there are no large-scale cohort studies evaluating these issues. Therefore, we conducted a nationwide population-based cohort analysis to explore the effect of smoking status and quantity on the development of IPF.
Methods
Study data source
The National Health Insurance Service (NHIS) is a mandatory universal health insurance service managed by the South Korean government that covers virtually all of the Korean population.24 The National Health Information Database (NHID) is a population database developed by the NHIS that makes available medical records vital to public health concerns and medical research. The NHID contains personal information, demographics, medical treatment data and medical claims information provided by healthcare providers.25 The other components of the NHID include long-term care insurance, healthcare providers and national health check-up databases.24
Since 1995, the NHIS has provided general national health check-up services for the prevention and early detection of disease to improve the health of the Korean population.26 All insured adults are entitled to a biennial general health-screening programme (annually for manual workers). The health check-up service includes chest X-ray, laboratory tests and detailed questionnaire about medical history and lifestyle habits.
Study population and definition of IPF
We searched the NHID to identify individuals who participated in the health check-up service between 2009 and 2012, then screened their medical claims between January 2007 and December 2016.
According to the Korean government’s policy on enhanced support for rare diseases, patients given an International Classification of Disease, 10th revision (ICD-10) code for IPF (J841 or J8418) benefit from paying only 10% of the total healthcare cost. Therefore, NHI conducts a careful review on individual cases registered as IPF, and the physicians are required to submit eligibility documents and supporting evidence proving the appropriateness of the diagnosis to the Health Insurance Review and Assessment Service.
We applied an operational definition of newly diagnosed IPF using the following criteria. We included participants with at least one ICD-10 of IPF registered by a physician working at a general or referral hospital except a primary care clinic considering recommendation of multidisciplinary discussion for the accurate diagnosis of IPF. In addition, a claim for chest CT should have existed within 1 year prior to registration of an ICD-10 of IPF. An ICD-10 of IPF should not have existed within 1 year prior to the claim for a chest CT scan. Similarly, a claim for a pulmonary function test should have existed within 6 months prior to or after an ICD-10 of IPF was registered. We excluded patients who had been diagnosed with IPF prior to their health check-up. Patients with ICD-10 codes of rheumatic diseases, vasculitis or other pulmonary diseases more than twice in the 2 years prior to IPF registration were also excluded, and patients who were hospitalised more than once because of such diseases were also excluded. The disease codes for rheumatic disease and vasculitis are described in online supplemental table 1. Other lung diseases were defined as any disease with ICD codes from J60 to J709, D86 (sarcoidosis),and J840 (pulmonary alveolar proteinosis).
Supplemental material
Other study variables
Standardised self-reporting questionnaires completed at the time of the health check-up service collected the following information: quantity of cigarette smoking (PY) comprising smoking intensity (defined as the average number of cigarettes smoked in a day) and duration (years), status of smoking (never, former, or current), alcohol consumption, and physical activity. Former smokers were defined as those individuals who had smoked more than 100 cigarettes in their lifetime but who were not smokers at the time of their health check-up. Alongside these variables, income and body mass index (BMI) were also selected as covariates in the multivariable Cox proportional hazards models because these variables could be confounders of smoking, which is the principal variable of interest and could increase the risk of IPF.27–32
Statistical analysis
Normally distributed continuous variables are presented as means±SD and categorical variables as proportions. Student’s t-test and χ2 test were used to evaluate continuous and nominal variables, respectively. Cox proportional hazards models were used to calculate the HR of the development of IPF.
We conducted two sensitivity analyses regarding possible recall bias and reverse causation effect. The first sensitivity analysis excluding the incidence of IPF in the first year of follow-up from the index date was conducted to minimise the reverse causation effect. To minimise recall bias, the second sensitivity analysis included participants who reported the same smoking status in the present health check-up (index date) and the health check-up conducted 2 years previously. Participants in the analysis were divided into the following three groups: (1) never–never group: participants answered ‘never smoker’ in both the present (index date) and 2-years prior (2 years before index date) health check-up; (2) former–former group: participants answered ‘former smoker’ in both the present (index date) and 2-years prior (2 years before index date) health check-up; (3) current–current group: participants answered ‘current smoker’ in both the present (index date) and 2-years prior (2 years before index date) health check-up.
SAS V.9.4 and R V.3.23 (The R Foundation for Statistical Computing, Vienna, Austria; http://www.r-project.org) were used for statistical analyses. We considered a two-sided p value of <0.05 to be statistically significant.
Results
Characteristics of the study population
Our search of the NHID revealed 23 452 862 Koreans who participated in the health check-up service provided by the NHIS between January 2009 and December 2012. We excluded 195 100 individuals younger than 20 years old and 14 926 individuals diagnosed with IPF before their health check-up service. As a result, a total of 23 242 836 participants were included in our analysis. We then assessed the medical claims of this study population using the NHID until December 2016 and identified 25 113 participants who met our operational definition of newly diagnosed IPF (figure 1).
Study flow chart. IPF, idiopathic pulmonary fibrosis.
The baseline characteristics of our study population are shown in table 1 and online supplemental table 2. The mean age of all participants was 47.7 years, 50.7% of whom were male. The participants in the IPF group (those diagnosed with IPF during the follow-up) were significantly older than those in the non-IPF group (64.1±11.3 and 47.7±14.4, respectively). Men accounted for 68.9% of the IPF group, significantly more than in the non-IPF group. Among the patients with IPF, 6842 (27.2%) were current smokers and 5826 (23.2%) were former smokers. Among the participants without IPF, 5 699 543 (24.6%) were current smokers and 3 141 051 (13.5%) were former smokers.
Baseline characteristics of the study population
Development of IPF according to smoking status
The risks of developing IPF were significantly higher in former smokers (HR 1.42, 95% CI 1.37 to 1.48) and current smokers (HR 1.66, 95% CI 1.61 to 1.72), respectively, than never smokers after adjusting for age, sex, alcohol consumption, physical activity, income and BMI (table 2).
Impact of smoking status on the development of IPF
The risk of developing IPF in men who formerly smoked was 1.41-fold higher (HR 1.41, 95% CI 1.36 to 1.46) than in never-smoker men after adjusting for age, alcohol consumption, physical activity, income and BMI. Using the same statistical model, we found that current male smokers had a 1.74-fold higher risk (HR 1.74, 95% CI 1.67 to 1.80) of developing IPF than never-smoker men (table 2). The risks of IPF in women who formerly and currently smoked were 1.62-fold higher (HR 1.62, 95% CI 1.39 to 1.90) and 1.90-fold higher (HR 1.90, 95% CI 1.72 to 2.10), respectively, than in women with no history of smoking after adjusting the same variables (table 2).
The sensitivity analysis excluding the first year of follow-up from the index date revealed consistent results showing that former and current smokers were at higher risk for IPF than never smokers (online supplemental table 3).
We compared the risk of developing IPF between former and current smokers, whereby current smokers had a significantly higher risk of developing IPF than former smokers (adjusted HR (aHR) 1.17, 95% CI 1.13 to 1.21) (online supplemental table 4). Among men, current smokers had a significantly higher risk of IPF than former smokers (aHR 1.23, 95% CI 1.18 to 1.27). However, the difference was not significant among women, with a similar HR but wide CI (aHR 1.16, 95% CI 0.97 to 1.40).
The higher risks of IPF in former and current smokers were shown consistently in the second sensitivity analysis, which included participants who reported the same smoking status in the present health check-up (index date) and check-up conducted 2 years previously. Former–former and current–current smokers had a significantly higher risk of IPF than never–never smokers (online supplemental table 5), while current–current smokers had a higher risk of IPF compared with former–former smokers (online supplemental table 6).
Impact of smoking quantity on the development of IPF
Table 3 shows that both smoking intensity (packs per day) and duration were associated with the development of IPF. The incidence of IPF was higher in individuals with longer smoking durations for equivalent smoking intensity. Individuals with smoking durations longer than 20 years had a higher risk of developing IPF in comparison with participants with no history of smoking, regardless of smoking intensity.
Impact of smoking intensity and duration on the development of IPF
Former and current smokers had higher aHRs than never smokers and, especially in men, the IPF risk increased in line with smoking quantity. Men with smoking history of 30 PY and more had the highest HRs in both former smokers (HR 1.78, 95% CI 1.70 to 1.87) and current smokers (HR 2.12, 95% CI 2.02 to 2.21) (figure 2).
Impact of smoking status on the development of idiopathic pulmonary fibrosis according to smoking quantity. The reference is never smoker. *Adjusted for age, drinking, physical activity, income and body mass index. PY, pack-year.
We compared the IPF risk between current smokers and former smokers according to smoking quantity (online supplemental table 7). Current-smoker men had higher risks of IPF compared with former-smoker men in all smoking quantity groups (<15 PY, 15–29 PY and ≥30 PY). In women, however, we did not find statistically significant differences in IPF risk between current and former smokers. The aHRs of smoking quantity as a continuous variable for IPF development in the total population, men, and women were 1.012 (95% CI 1.011 to 1.013), 1.011 (95% CI 1.011 to 1.012) and 1.025 (95% CI 1.021 to 1.029), respectively.
Impact of smoking on the development of IPF in different age groups
The risk of developing IPF increased according to smoking quantity within the same age group. The risk of developing IPF in individuals older than 55 years with no history of smoking was higher in men than in women. The risk of IPF in participants with no history of smoking increased with increasing age, with the highest risk occurring among those 65–74 years of age (HR 1.52, 1.81, 2.05, and 1.86 in the 45–54-, 55–64-, 65–74-, and >75-year-old age groups, respectively) (figure 3 and online supplemental table 8).
Impact of smoking on the development of idiopathic pulmonary fibrosis in different age groups. *Adjusted for age, drinking, physical activity, income and body mass index. PY, pack-year; Ref., reference.
Discussion
This study demonstrated that smoking significantly increased the risk of IPF development. In addition, current smokers were at higher risk than former smokers, particularly men. These findings were reproduced in the additional sensitivity analyses to minimise recall bias and reverse causation. Both smoking intensity and duration showed a dose–response relationship in the occurrence of IPF. In participants older than 45 years, the risk of the development of IPF increased according to smoking quantity within the same age group, and never smokers had an increased risk of IPF with increasing age.
In this study, we established a nationwide cohort for evaluation of newly diagnosed IPF cases using the NHID. This population-based study enrolled 23 242 836 participants, which covers about 45% of the total population of South Korea. The health check-up data were merged with NHID and allowed us to evaluate the important clinical variables including BMI, alcohol consumption, physical activity and income level, which are considered as confounders to smoking and development of IPF in our analysis. Furthermore, long-term follow-up claims data of individual participants allowed us to perform longitudinal analysis of the association between smoking and IPF development. Therefore, our study demonstrated the causal relationship between smoking and the development of IPF more clearly than previous cross-sectional and case–control studies.9 33
A previous case–control study reported that the OR of the development of IPF in individuals with a history of ever smoking was 1.6 (95% CI 1.1 to 2.4) compared with never smokers.9 Another study published in 1994 reported a 2.94-fold increased risk of IPF in smokers compared with that in normal healthy adults.10 In our study, the aHRs for the development of IPF were 1.42 in former smokers and 1.66 in current smokers, comparable with results from the previous reports. The results of case–control studies that show that former but not current smokers had significantly higher risk of IPF require careful interpretation because case–control studies have innate limitations in proving causality.9 19 In addition, a ‘healthy smoker effect’, whereby former smokers probably quit smoking because they perceived health problems, including alarming respiratory symptoms or significant comorbidities, should be also considered.33 34 A recent prospective cohort study reported that both current (HR 2.49, 95% CI 1.99 to 3.12) and former smoking (HR 2.03, 95% CI 1.73 to 2.39) were associated with a higher risk of IPF than never smoking, which is consistent with the result of our study.21 Prospective cohort studies show more accurate data regarding exposures and outcomes, thus providing strong evidence for causality.
Ekström et al reported a dose–response relationship between smoking and severe pulmonary fibrosis, whereby the dose of smoking was referred to as PY only.22 In the present study, we analysed the impact of daily smoking intensity and duration of smoking, as well as total quantity of smoking, on the risk of IPF. The risk of IPF increased as the smoking intensity increased within an equivalent smoking period, and individuals with the same smoking intensity had a greater risk of IPF if their smoking period was longer. This dose–response relationship between smoking and the development of IPF provides substantial evidence for a causal relationship.
Several studies reported that women were more sensitive to the effect of smoking than men and suggested that the pattern of lung damage caused by smoking may differ according to sex.35 Women tend to be more vulnerable to the consequences of smoking, with a high degree of lung-function decline for a given quantity of cigarette exposure, and develop chronic obstructive pulmonary disease at a young age.36 Although women constitute about one-third of all patients with IPF, studies evaluating the impact of smoking on their risk of IPF are scarce because of the limited number of female smokers. In our study, female ever smokers showed higher HRs of IPF development than male ever smokers. However, the difference in the absolute number of patients with IPF according to sex was large, and the number of female ever smokers was far smaller than that of male ever smokers, with a ratio of 0.085. Therefore, one needs to exercise caution when interpreting female smokers as being more at risk than male smokers in developing IPF.
There are some limitations to our study. First, this cohort study had a 2-year washout period that might not be enough to exclude pre-existing IPF cases. However, considering the known median survival of 3 years after diagnosis1 and that the median time of diagnosis from symptom onset was 13.6 months,37 a 2-year washout period should be appropriate for most IPF cases. Second, we could not evaluate the effects of smoking on IPF severity because of the lack of clinical information such as symptoms, lung function, imaging and treatment, including drugs and oxygen therapy. Third, there could be a competing risk issue: some smokers die of other smoking-related diseases such as lung cancer38 or chronic obstructive pulmonary disease39 before the diagnosis of IPF during the follow-up period. However, this competing risk can lead to an underestimation rather than an overestimation of IPF incidence among smokers and thus would not change the main result that smoking increases the risk of IPF incidence. Fourth, we were unable to evaluate the smoking status of participants at the time of their IPF diagnosis. However, we expect that the majority of participants who were current smokers at the time of their health check-up would have continued to smoke, given that the annual smoking cessation rate in South Korea is only 2.5%.40 Last, our cohort consists of one ethnicity only; our results may not be generalisable to the worldwide population.
Conclusion
This nationwide population-based cohort study demonstrated that smoking significantly increased the risk of IPF development. In addition, current smokers had a higher risk of IPF than former smokers. A dose–response relationship was observed between smoking intensity and duration and the development of IPF.
Data availability statement
No data are available. The data that support the findings of this study are available from the National Health Insurance Service in Korea but restrictions apply to the availability of these, which were used under license for the current study, and so are not publicly available.
Ethics statements
Patient consent for publication
Ethics approval
This study was exempted from review by the Institutional Review Board of Seoul National University Hospital (E-IRB-1804-046-936).
Acknowledgments
We thank Hugh McGonigle from Edanz Group (https://en-author-services.edanz.com/ac) for editing a draft of the manuscript.
References
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.
Footnotes
WB and C-HL contributed equally.
Contributors WB contributed to writing, figures and data interpretation. C-HL contributed to the study design and data interpretation. JL contributed to the literature search and data interpretation. YWK contributed to the literature search. KH contributed to the data collection and data analysis. SMC contributed to the study design, data interpretation and writing.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
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.
Linked Articles
- Airwaves