This was a prospective observational cohort study (ECOPD cohort) in Guangdong (Wu et al., 2021). The subjects were continuously recruited and tested in Wengyuan,Lianping countryside and Guangzhou city, Guangdong Province from July 2019 to August 2021.

The subjects in this study must be 40–80 years old and undergo lung function tests, IOS, radiological imaging, and epidemiological investigation of chronic obstructive pulmonary disease. The subjects would be excluded in the following criteria: (I) aged <40 years or >80 years; (II) respiratory infection or exacerbation in the four weeks prior to screening; (III) heart attack (myocardial infarction, malignant arrhythmia) in the past three months; (IV) hospitalized for heart disease within the past one month, (V) chest, abdomen, or eye surgery in the past three months, (VI) previous lobectomy; (VII) malignant tumors newly discovered and being treated; (VIII) receiving anti-tuberculosis drug treatment or active pulmonary tuberculosis; (IX) history of mental disorders, auditory hallucinations, visual hallucinations or taking antipsychotic drugs; (X) history of cognitive disorders, including dementia or cognitive disorders; (XI) history of high paraplegia; and (XII) pregnant or lactating women.

Basic demographic variables included gender, age and body mass index (BMI), smoking index, smoking status, family history of respiratory disease, drug treatment, modified Medical Research Council (mMRC) dyspnoea scale score and COPD assessment test (CAT) score. Chronic obstructive pulmonary disease was diagnosed based on spirometry (criterion for airflow limitation as a post-bronchodilator fixed ratio of FEV₁/FVC <0.70), normal lung function was defined as FEV₁/FVC ≥0.70 and FEV₁ predicted % ≥ 80% after bronchodilation, and abnormal lung function was defined as either FEV₁ predicted % < 80% or FEV₁/FVC <0.70. SAD was assessed by the IOS index using R₅–R₂₀ greater than the upper limit of normal (ULN), which was calculated based on a multicenter, Chinese healthy population impulse oscillometry study (Liang et al., 2021). SAD was diagnosed by spirometry based on at least two of the following three indexes (MMEF, FEF_50%, and FEF_75%), less than 65% predicted, and the predicted values were based on a multicenter study of spirometry reference values in a healthy Chinese population (Jian et al., 2017). All subjects were divided into four groups based on lung function and IOS criteria for assessing SAD: 1) SAD assessed by both spirometry and IOS (concordant SAD); 2) SAD assessed by neither spirometry nor IOS (concordant NO SAD); 3) SAD diagnosed by only spirometry but not IOS (spirometry-only SAD); 4) SAD diagnosed by only IOS but not spirometry (IOS-only SAD).

The Ethics Committee of the First Affiliated Hospital of Guangzhou Medical University approved this study protocol (2018–53). Written informed consent was obtained from all subjects prior to inclusion.

The respiratory resistance and reactance were measured by IOS (MasterScreen IOS, Hochberg Germany). Subjects never received any medication before the IOS test. The operator gently pressed the subject’s cheeks with both hands to avoid cheek vibration affecting the accuracy of the measurement (King et al., 2020). Respiratory system impedance (Z_rs) consists of resistance (R_rs) and reactance (X_rs). R_rs at 5 Hz (R₅) indicates total respiratory resistance and R_rs at 20 Hz (R₂₀) represents central airway resistance; R₅–R₂₀, an index of SAD, reflects peripheral airway resistance; X₅, reactance value at 5Hz, is a measure of the stiffness of the entire system. Resonance frequency (F_res), where E_rs and I_rs make equal and opposite contributions to impedance and reactance, is a sensitive index reflecting increased resistance; AX, the area under X₅ and F_res, is considered to be an important index of early detection and prognosis of COPD (Lipworth and Jabbal, 2018), reflecting the comprehensive index of reactance.

According to the ATS/ERS spirometry guidelines (Miller et al., 2005), the acceptable standard of repeatability was a single test that included no hesitation in the onset of expiration and extrapolation volume <5% FVC or 0.15L. The difference between the two largest values of FEV₁ and FVC was within 0.15 L. Subjects were required to be measured at least three times to ensure reproducibility. For the bronchodilator test, subjects inhaled 400 µg of salbutamol through a nebulizer canister, and a pulmonary function test was performed after 20 min.

Percent emphysema was defined as the total percentage of both lungs with attenuation values less than -950 Hounsfield units on inspiratory images, and percent gas trapping was defined as the total percentage of both lungs with attenuation values less than -856 Hounsfield units on expiratory images (Kim et al., 2014; Lynch et al., 2015). The results of the data were evaluated by at least two clinical radiologists.

Statistical analysis was performed with SPSS statistics version 26.0 (IBM Corp. Armonk, NY, United States). Cohen’s k was used to assess agreement between IOS and spirometry in the diagnosis of SAD. Continuous variables showing a normal distribution were presented as mean (standard deviation), and continuous variables without a normal distribution were presented as median [interquartile range (IQR)]. We compared baseline characteristics between discordant groups using Student’s t-test or Wilcoxon rank-sum test for normal and non-normal continuous variables, respectively, and Fisher’s exact or Chi-squared test for categorical variables and differences among four groups using Analysis of Variance (ANOVA), Kruskal–Wallis test, and chi-squared test. After adjusting for age, sex, BMI, smoking index, and smoking status, we used multivariate regression analysis to analyze factors of discordant results. Line regression adjusting baseline statistical difference variables analyzed the differences between discordant groups in CT imaging. p < 0.05 was considered statistically significant.

Overall, 575 SAD subjects (31.3%) were diagnosed by both spirometry and IOS; 629 subjects (34.3%) had no SAD; 435 SAD subjects (23.7%) were diagnosed by spirometry only but not IOS; 197 SAD subjects (10.7%) were diagnosed by only IOS but not spirometry. There was a slight agreement between spirometry and IOS in the assessment of SAD (kappa 0.322, p < 0.001). Taking spirometry as the standard, the sensitivity and specificity of IOS were 56.9% and 76.2%, respectively. The area under curve (AUC) was 0.665 (0.641–0.690) (Table 1). However, we found that in patients with normal lung function, the number of IOS-only SAD was nearly twice than that of spirometry. Similarly, we used two of the three indicators less than LLN or MMEF less than LLN as abnormal criteria to evaluate SAD. The results showed that the number of IOS-only SAD was nearly four times than that of spirometry (Supplementary Table S1). However, the number of spirometry-only SAD was more than ten times that of spirometry in patients with abnormal lung function (Supplementary Table S2).

Compared with IOS-only SAD, spirometry-only SAD was older (62.34 vs. 60.23), more male (85.1 vs. 70.1%), lower body mass index (22.49 vs. 23.54), higher smoking index (28.00 vs. 6.40), more cough (24.1 vs. 13.7%), phlegm (28.5 vs. 15.2%), more drug treatment (26.7 vs. 7.1%) and family history of respiratory diseases (18.0 vs. 6.1%) (Table 2), and airflow limitation (FEV₁/FVC) (66.37 vs. 79.34) and more severe impaired lung function (FEV₁% predicted) (83.49 vs. 90.88) (Table 3).

After adjusting for age, sex, BMI, smoking index, and smoking status covariates, compared with IOS-only SAD, multivariate regression analysis showed that smoking status (OR = 0.632, 95% CI [0.477–0.837], p = 0.001), phlegm (0.534 95% CI [0.337–0.846], p = 0.008), drug treatment (0.238 95% CI [0.131–0.434], p < 0.001) and family history of respiratory disease (0.281 95% CI [0.147–0.540], p < 0.001) were factors leading to spirometry-only SAD. However, BMI (1.073 95% CI [1.014–1.135], p = 0.014) was the independent factor leading to IOS-only SAD (Figure 1).

In line regression, after adjusting for baseline statistical difference variables, we found that in all subjects and abnormal lung function subjects, spirometry-only SAD was more severe in CT emphysema and gas trapping than IOS-only SAD (Figure 2, Supplementary Figure S1). However, in normal lung function subjects, there was no significant difference in emphysema and gas trapping between the IOS-only SAD and spirometry-only SAD groups (Figure 3).