Smoking status and clinical outcome in idiopathic pulmonary fibrosis: a nationwide study

Abstract

Background: Smoking status has been linked to the development of idiopathic pulmonary fibrosis (IPF). However, the effect of smoking on the prognosis of patients with IPF is unclear. We aimed to investigate the association between smoking status and all-cause mortality or hospitalisation by using national health claims data.

Methods: IPF cases were defined as people who visited medical institutions between January 2002 and December 2018 with IPF and rare incurable disease exempted calculation codes from the National Health Insurance Database. Total 10,182 patients with available data on smoking status were included in this study. Ever-smoking status was assigned to individuals with a history of smoking ≥ 6 pack-years. The multivariable Cox proportional hazard model was used to evaluate the association between smoking status and prognosis.

Results: In the entire cohort, the mean age was 69.4 years, 73.9% were males, and 45.2% were ever smokers (current smokers: 14.2%; former smokers: 31.0%). Current smokers (hazard ratio [HR]: 0.709; 95% confidence interval [CI]: 0.643-0.782) and former smokers (HR: 0.926; 95% CI: 0.862-0.996) were independently associated with all-cause mortality compared with non-smokers. Current smokers (HR: 0.884; 95% CI: 0.827-0.945) and former smokers (HR: 0.909; 95% CI: 0.862-0.959) were also associated with a reduced risk of all-cause hospitalisation compared with non-smokers. A non-linear association between smoking amount and prognosis was found in a spline HR curve and showed increasing risk below 6 pack-years.

Conclusion: Ever-smoking status may be associated with favourable clinical outcomes in patients with IPF.

Keywords: Hospitalization; Idiopathic pulmonary fibrosis; Mortality; Prognosis; Smoking.

Fig. 1

Association between smoking and mortality according to smoking status in patients with IPF. (a) Comparison of the survival curves between never and ever smokers in patients with IPF. (b) Comparison of the survival curves among never, former, and current smokers in patients with IPF. (c) Spline curve analysis of smoking amount for the prognosis The Kaplan–Meier method was used for overall survival estimates, and the log-rank test was used for survival differences by subgroups. The spline curve hazard ratio was computed by adjusting covariates including age, sex, diagnosis year, Charlson comorbidity index, medication (use of steroid and pirfenidone), medical aid, residential address, and low household income. The reference point (6 pack-years) was indicated by a vertical dashed line

Fig. 2

Association between smoking and all-cause hospitalisation according to smoking status in patients with IPF. (a) Comparison of the all-cause hospitalisation-free survival curves between never and ever smokers in patients with IPF. (b) Comparison of the all-cause hospitalisation-free survival curves among never, former, and current smokers in patients with IPF. (c) Spline curve analysis of smoking amount for the all-cause hospitalisation The Kaplan–Meier method was used for all-cause hospitalisation estimates, and the log-rank test was used for all-cause hospitalisation differences by subgroups. The spline curve hazard ratio was computed by adjusting covariates including age, sex, diagnosis year, Charlson comorbidity index, medication (use of steroid and pirfenidone), medical aid, residential address, and low household income. The reference point (6 pack-years) was indicated by a vertical dashed line

Fig. 3

Association between smoking and respiratory hospitalisation according to smoking status in patients with IPF. (a) Comparison of the respiratory hospitalisation-free survival curves between never and ever smokers in patients with IPF. (b) Comparison of the respiratory hospitalisation-free survival curves among never, former, and current smokers in patients with IPF. (c) Spline curve analysis of smoking amount for the respiratory hospitalisation The Kaplan–Meier method was used for respiratory hospitalisation estimates, and the log-rank test was used for respiratory hospitalisation differences by subgroups. Spline curve hazard ratio was computed by adjusting covariates including age, sex, diagnosis year, Charlson comorbidity index, medication (use of steroid, pirfenidone), medical aid, residential address, and low household income. The reference point (6 pack-years) was indicated by a vertical dashed line