We measured long-term variability of IOS parameters in stable COPD and asthma in this study. According to the demographic data and the baseline lung function test results of the COPD and asthma patients, although asthma [(49.86 ± 14.26) years] patients were younger than COPD [(62.68 ± 9.57) years] patients, the spirometry parameters in COPD were lower compare to asthma (P < 0.0001), the COPD patients had higher R5, R5-R20, AX and more negative X5, except for R20 and Freq, compared to the asthma patients (P < 0.0001). These findings suggest that there were differences in pulmonary ventilation function and airway obstruction between COPD and asthma.

As airway obstruction increases with disease progression and airway remodeling, characteristic changes in resistance and reactance are more pronounced. Particularly, resistance spectrum increases in direction of low frequencies. At the same time overall resistance increases but more so at low frequencies than at high frequencies [16, 17]. Thus, when we compared the long-term variability of spirometry and IOS parameters in COPD and asthma, we found that SDbv of FVC, FEV1, MEF 50, MEF 25, MMEF75/25, R5, R5-R20, AX, Freq and X5 were statistically significantly different in COPD and asthma, SDbv of FVC, FEV1, R5, R5-R20, AX, Freq and X5 in COPD were higher compare to asthma, SDbv of MEF50, MEF25 and MMEF75/25 in COPD were lower compare to asthma. SDbv of R20 had no difference between COPD and asthma. COV for FVC, FEV1, MEF50, MMEF75/25, R5-R20 and X5 were statistically significantly different between COPD and asthma. However, R5, R20, AX and Freq were not. COV of FVC, FEV1, MEF50, MMEF75/25, and X5 in COPD were higher than those in asthma. SDbv and COV of FVC, FEV1, MEF 50, MEF 25, MMEF75/25 in IOS parameters were higher compare to spirometry. These results indicate that long-term variability of IOS parameters in COPD was higher than in asthma, and long-term variability in IOS parameters was higher than spirometry in COPD and asthma.

COR values were less than 66% in COPD and asthma, except for Freq (69%) and X5(80%) in asthma with the relative CORs for IOS being more variable than for spirometry. However, the repeatability of IOS parameters, apart from Freq and X5, were high (> 0.80). We are not sure if this has any relationship to the difference of distal airway distensibility in the two conditions. These results suggest that despite there being higher long-term variability in IOS measurements than spirometry, IOS is still highly repeatable and stable. The high variability may be due to different baseline characteristics, the airway caliber and elastic characteristics of respiratory system, which fluctuated with time [18, 19].

The current gold standard to assess airway limitations is spirometry. However, performing an optimal spirometry always requires good patient cooperation. Additionally, repeated forced breathing causes changes in bronchial motor tension, false positive results occur often. Studies on the quality of spirometry in elderly patients have shown that only 30% of patients are able to perform a spirometry that meets the quality standards of the European Respiratory Society/American Thoracic Society[20, 21]. Furthermore, FEV1 cannot fully assess small airway abnormalities. Thus, MEF50, MEF25 and MMEF75/25 have been studied as markers of small airway obstruction but are highly variable due to atmospheric airway obstruction [22]. We found that between-visit variability (SDbv) of FVC, R5, AX, X5 were statistical difference between GOLD1-2, GOLD3 and GOLD4 groups. Small airways assessed by IOS parameters including X5 and AX correlate more strongly with clinical symptoms than with spirometry [23]. Between-visit variability relative to the mean COV for FVC, FEV1, R5, R5-R20 and X5 were statistically significantly different in GOLD1-2, GOLD3 and GOLD4 patients. The values of SDbv and COV for IOS parameters were higher than for spirometry in different stages of COPD. The higher the COPD stage, the lower ICC values of spirometry parameters (FVC, FEV1, MEF50, MEF25 and MEF75/25), the ICC values were less than 0.8 in GOLD3 and GOLD4. The lower ICC values of FVC, FEV1 perhaps were because COPD patients with poorer lung function are less able to cooperate. Nevertheless, the ICC values (> 0.8) of IOS resistance parameters were relatively stable and high. Similarly, the higher the COPD stage, the higher COR values of spirometry parameters, but the COR values of IOS parameters were relatively stable. The results show that long-term variability of spirometry parameters (FVC, FEV1, MEF50, MEF25 and MEF75/25) was higher, repeatability was lower than IOS parameters in different GOLD stages. Additionally, the higher the stage, the worse the repeatability. In COPD it is well established that the changes in R5, R20 correlate well with GOLD1 to GOLD4 severity and our results tend to corroborate this. IOS resistance parameters remain relative stable and reproducible over time compare to IOS reactance parameters. Reactance is comprised of both inertance and elastance. Diseases (for example: interstitial lung diseases) that influence the elasticity of the lung will increase capacitance negatively and X5 will be more negative. Values of reactance parameters are affected by age and weight, increase of age and weight will determine a less negative reactance. Moreover, reactance is affected by the heterogeneous distribution of airway calibres and lung compliances [2, 24]. These factors may lead to high variability and poor repeatability of reactance. So, IOS resistance parameters can be used as a routine adjunct to lung function test.

The short-term variability in IOS parameters is known, particularly within-day, day-to-day or week-to-week repeatability of resistance (Rrs) and reactance (Xrs) with high ICC values (> 0.80) [25,26,27]. It is the first study to relate measurement of long-term variability to airflow obstruction. Previous studies have largely assessed the within session repeatability and variability of resistance [28, 29]. However, these studies may not be applicable to clinical settings, where patients in clinically stable stage of disease are often examined several months apart. In this study, repeatability of IOS measurements between clinical visits was a representation of the real-world behavior of these parameters. The median (IQR) time between first and third visit was 6.1 (3.4–16.6) months in COPD and 4.9 (4.0–12.6) months in asthma (Table 1). Only one study conducted a long-term variability analysis of IOS parameters in stable COPD and asthma [10]. They only performed a long-term variability analysis at consecutive three follow-ups. Based on what is already known, with this study our research reveals: (1) Our sample size was much larger; (2) Long-term variability between IOS parameters and spirometry in COPD, COPD-GOLD stages, and asthma was further compared; (3) It was also the first study to determine the relationship between long-term variability and airflow obstruction in COPD.

Significant correlations at baseline between FEV1 and R5 (P < 0.05) but not with R20 has been reported and during 1-year follow up. The changes in R5 and R20 did not significantly correlate with FEV1. Additionally, the R20 is unrelated to the severity of airflow obstruction in patients with COPD [30]. We found that during the first three visits, SDbv of R5 correlates with %FEV1 in GOLD4 (rs = 0.61, P < 0.05). Additionally, multiple regression analyses indicated that %FEV1 is potential predictor for long-term variability of R5 and R5-R20 (t = 2.90, P = 0.005; t = 2.44, P = 0.017). Their mean values are potential predictor for long-term variability of R5, R20 and R5-R20 (P < 0.05). The severity of airflow obstruction significantly was correlate with long-term variability of R5 and R5-R20, not with long-term variability of R20.

One potential limitation of this study was that the groups were not matched by sex and age in different diseases and GOLD stages. There was one female in GOLD1, 20 male and 2 female in GOLD2, 25 male and 6 female in GOLD3, 18 male and 1 female in GOLD4, 58 male and 61 female in asthma. There is long-term disease progression in stable diseases [31], our maximum follow-up period was 22 months. Longer follow-up period to assess long-term variability would perhaps lend more credibility to future studies.