Clinical significance of self-reported cough intensity and frequency in patients with interstitial lung disease: a cross-sectional study

To minimise selection bias, consecutive new patients with ILD, i.e., idiopathic interstitial pneumonias (IIPs), connective tissue disease-associated interstitial lung disease (CTD-ILD), or CHP were invited to participate in this cross-sectional study. The study was performed at Kyoto University Hospital between August 2015 and March 2018. IIPs were diagnosed as previously described [14‐16]. IPF was diagnosed by either surgical lung biopsy or HRCT according to the guidelines [17]. CHP was diagnosed in multidisciplinary team discussions according to the established criteria [18]. Patients who had had a respiratory tract infection in the preceding month, those who had lung cancer, postnasal drip, rhinitis, catarrhal symptoms, or a history of adult asthma, and those younger than 20 years of age were excluded.

Data were obtained from patient-completed questionnaires and electronic patient records. The intensity and frequency of cough were evaluated using a 100-mm VAS (0, no cough; 100, unbearable), which is the most commonly used tool to subjectively assess cough severity [19]. The VAS was used in the standardised manner recommended by the CHEST Expert Cough Panel in 2015 [20]. Patients were asked to place a vertical mark on the scale reflecting the intensity of their cough, and another vertical mark indicating the frequency of their cough. To explore the features underlying any difference between the VAS scores for cough intensity and cough frequency, we stratified patients into three categories using the following formula: (cough intensity) − (cough frequency), defining a cough frequency-dominant group, ≤ − 10 mm; an equal cough severity group, − 9 mm to 9 mm; and a cough intensity-dominant group, ≥10 mm. We assessed cough-specific HRQoL using the Japanese version of the LCQ [21]. The total score on the LCQ ranges from 3 to 21, with lower scores indicating greater impairment of health status as a result of cough. The LCQ has been previously shown to have appropriate feasibility and sensitivity; thus, it can be considered an appropriate clinical outcome tool for use in clinical trials that include patients with ILD [22]. Permission to use the LCQ was obtained from the developer (Dr Surinder Birring) and translators (Drs Akio Niimi and Haruhiko Ogawa). Dyspnoea was assessed using the Medical Research Council (MRC) chronic dyspnoea scale. The symptoms of gastro-oesophageal reflux disease (GORD) were assessed using the frequency scale for symptoms of GORD (FSSG) developed by Kusano et al. [23], which has been previously used in studies of cough [24, 25]. This questionnaire consists of 12 questions; the total score on the FSSG ranges from 0 to 48, with higher scores indicating more severe symptoms of GORD. We simultaneously collected data on the intensity and frequency of cough, LCQ, MRC chronic dyspnea scale, and FSSG scores. Additional data, including for age, sex, body mass index, smoking history, anamnesis, medication, and pulmonary function tests performed within 3 months of the assessment of cough intensity and frequency, were obtained from the clinical records. The pulmonary function tests were performed using the CHESTAC system (Chest M.I. Inc., Tokyo, Japan), and the diffusing capacity of the lung for carbon monoxide (D_Lco) was determined by the single-breath technique. The Composite Physiologic Index (CPI) was used to predict the extent of fibrosis on HRCT. The formula used to calculate the CPI is as follows: CPI = 91.0 − (0.65 × percent predicted D_LCO) − (0.53 × percent predicted forced vital capacity [FVC]) + (0.34 × percentage of predicted forced expiratory volume in 1 s [FEV₁]) [26].

The sample size was determined by the correlation between the severity of cough and %FVC in patients with IPF based on previous data [3]. However, there were few reports available on the correlation between severity of cough and %FVC in patients with ILD until the present study. Therefore, we hypothesised a mild effect size of 0.3 with an alpha level of 5% and a power of 90% and calculated that a minimum sample size of 113 would be required for this study.

The data were compared between the three groups using either the chi-square test or Fisher’s exact test for categorical variables, and either one-way analysis of variance or the Kruskal–Wallis test for continuous variables. If the one-way analysis of variance or Kruskal–Wallis test result was significant, the three groups were compared using either Tukey’s test or a Bonferroni-corrected significance level of 1.6%. The correlations of cough intensity and frequency with potential predictor variables were tested using either the Mann–Whitney U test or Spearman’s rank correlation coefficient as appropriate. After the cough intensity and frequency data were converted to a dichotomous variable using the median value, multiple logistic regression analysis was performed to identify independent variables predicting cough intensity and frequency. Only one variable in a set of variables with a correlation coefficient > 0.5 was used in the multivariate logistic regression analysis because of multicollinearity. Potential predictive variables were age, sex, pack-years of smoking, body mass index, FSSG, MRC chronic dyspnoea scale and CPI scores, use of glucocorticoid medication, antifibrotic therapy and proton pump inhibitors, and type of ILD. Variables were selected using a forward selection method based on the likelihood ratio test. The goodness of fit of the model was assessed using the Hosmer–Lemeshow test. Missing data were not included in the analyses. The data are expressed as the number and percentage, mean and standard deviation, or median with the interquartile range. All statistical analyses were performed using SPSS software (version 25; IBM Corp., Armonk, NY, USA). A p-value < 0.05 for a two-tailed test was considered statistically significant.

Twenty-one of 150 patients examined for eligibility were excluded (9 with a respiratory tract infection in the preceding month, 1 with lung cancer, 6 with rhinitis, 3 with a history of adult asthma, 1 with an acute exacerbation of IPF in the previous month, and 1 with pneumoconiosis). Seventy patients with IIPs, 49 with CTD-ILD, and 10 with CHP were included in the study. The 70 patients with IIPs included 36 (51.4%) with IPF, 2 (2.9%) with idiopathic nonspecific interstitial pneumonia, 2 (2.9%) with respiratory bronchiolitis-associated ILD, 1 (1.4%) with desquamative interstitial pneumonia, 2 (2.9%) with cryptogenic organising pneumonia, 5 (7.1%) with idiopathic pleuroparenchymal fibroelastosis, and 22 (31.4%) with unclassifiable idiopathic interstitial pneumonia. The 49 patients with CTD-ILD included 15 (30.6%) with systemic sclerosis (SSc), 10 (20.4%) with polymyositis or dermatomyositis, 8 (16.3%) with Sjögren’s syndrome, 11 (22.4%) with rheumatoid arthritis, 1 (2.0%) with systemic lupus erythematosus, and 4 (8.2%) with microscopic polyangiitis. None of the patients were using angiotensin-converting enzyme inhibitors.

Table 1 shows the characteristics and intensity and frequency of cough in patients with IIPs, CTD-ILD, and CHP. Patients in the IIP group were more likely to be male and using an antifibrotic agent, and less likely to be using either glucocorticoid or proton pump inhibitor therapy than those in the CTD-ILD and CHP groups. The patients with CTD-ILD had a less extensive pack-year smoking history, higher FSSG scores, and better pulmonary function test results (i.e., FEV₁, total lung capacity, and D_Lco) than those with IIPs. The patients with CTD-ILD also had a greater total lung capacity than those with CHP. Among the three subtypes of ILD, the patients with the IIPs had the greatest intensity of cough but not frequency of cough. Patient age, current smoking, body mass index, MRC chronic dyspnoea scale, CPI, and LCQ scores were not affected by the subtype of ILD.

The unadjusted factors associated with intensity and frequency of cough in ILD are shown in Tables 2 and 3, respectively. In patients with IIPs, both the intensity and frequency of cough were negatively associated with the D_Lco and positively with the CPI. For patients with CTD-ILD, the D_Lco and CPI were associated with cough frequency, and both the intensity and frequency of cough were associated with FSSG and the MRC chronic dyspnoea scale score. However, the association between cough frequency and either D_Lco or MRC chronic dyspnoea scale score was rather weak (Spearman rho = − 0.287 for D_Lco and rho = 0.291 for the MRC chronic dyspnoea scale score) in patients with CTD-ILD. Of note is that the association of cough intensity and frequency with CPI (Fig. 1) and FSSG score (Fig. 2) was numerically weaker and stronger, respectively, in patients with CTD-ILD than in those with IIPs. There was no significant association of either the intensity or frequency of cough with any variable in patients with CHP.

Table 4 shows the results of the multiple logistic regression analysis of factors associated with the intensity and frequency of cough in all patients with ILD. IIPs and the FSSG score were independently associated with both the intensity and frequency of cough. Furthermore, CPI tended to be independently associated only with the frequency of cough (p = 0.052).

Finally, we examined the features of the differences between cough intensity and frequency in all patients with ILD (n = 129). There was a positive correlation between cough intensity and cough frequency (Spearman rho = 0.801, p < 0.001), but notably, the cough frequency-dominant group had significantly lower total and subdomain LCQ scores than either the equal cough severity group or cough intensity-dominant group (Table 5). The Spearman’s correlation coefficients between total LCQ scores and intensity and frequency of cough were − 0.675 and − 0.762, respectively (both p < 0.001).