Respiratory oscillometry is a simple, non-invasive, and effort-independent lung function test that overlays oscillatory pressure waves onto normal tidal breathing to measure the mechanical properties of the airways and lung parenchyma (1). Compared with spirometry, respiratory oscillometry is more sensitive for assessing the peripheral or small airways, especially in patients with chronic respiratory symptoms and preserved pulmonary function (2–4). Small airways dysfunction is highly prevalent in adults with asthma and is associated with the degree of airflow obstruction, symptom burden, and risk of exacerbations (5–7).

Introducing oscillometric assessment of small airways function into routine clinical practice provides clinicians with a more comprehensive understanding of airway physiology, enabling better assessment of disease activity and the risk of asthma exacerbations. Respiratory oscillometry therefore provides further information to spirometry when assessing current disease control and predicting clinical outcomes (4, 8). For example, patients with asthma who exhibit higher respiratory resistance, indicating worse small airways dysfunction, have been shown to achieve a better response to extra-fine particle inhaled therapy compared with non-extra-fine therapy (9, 10).

A recently published international Delphi study on the interpretation of respiratory oscillometry in adults with asthma or chronic obstructive pulmonary disease (COPD) reported that clinicians who routinely use oscillometry in clinical practice focus on a small number of metrics to guide interpretation, specifically resistance at 5 Hz (R₅), frequency dependence of resistance (R₅–R₂₀ or R₅–R₁₉), reactance at 5 Hz (X₅), and area under the reactance curve (A_X) (11). This expert group agreed that respiratory oscillometry is clinically useful for identifying and grading the severity of lung function impairment, as well as for assessing clinically meaningful changes in lung function over time. The group recommended the use of Z-scores to define abnormal lung function, with cutoffs of >1.64 for R₅ and A_X and <−1.64 for X₅. Because X₅ values are usually negative, more negative values indicate greater impairment in lung function. The severity of abnormal lung function was further defined according to the criteria outlined in Table 1.

Although Z-score cutoffs for impedance parameters provide a statistically robust framework for defining the severity of abnormality, empiric data are required to demonstrate their clinical relevance and validity (11). This evidence gap represents a barrier to some clinicians using this lung function test in clinical practice.

We have previously demonstrated that, among patients with asthma attending our tertiary asthma clinic, R₅–R₂₀, X₅, A_X, and resonant frequency (F_res) are correlated with asthma symptom burden, with the strongest association observed for R₅–R₂₀. Both A_X and R₅–R₂₀ were associated with an increased risk of asthma exacerbations (5). However, this previous analysis was predominantly based on abnormal lung function defined by absolute value cutoffs commonly reported in published studies, rather than Z-scores. As the use of absolute values to define abnormal oscillometry findings was not endorsed by the Delphi study (11) and is subject to inherent limitations (12), the aim of this study was to determine whether the presence and severity of abnormal lung function, defined using Z-scores for oscillometric parameters, are associated with asthma symptoms and exacerbation risk.

This was a single-center, retrospective study of patients with asthma referred to a tertiary respiratory clinic who underwent oscillometry as part of their routine assessment between January 2019 and December 2022. The study was approved by the Human Research Ethics Committee and Research Governance Unit of Fiona Stanley Hospital (RGS5611).

The methods of this study have been previously published (5) and are briefly described here. Eligible patients had a respiratory specialist-confirmed diagnosis of asthma and had completed spirometry and oscillometry, specifically impulse oscillometry (IOS), as part of standard lung function testing. Patients were excluded if they did not have at least one documented IOS measurement performed at our tertiary clinic.

The relevant information was extracted from the medical records of all eligible patients corresponding to their clinic visit at which lung function testing was performed. Collected information included standard demographic data, asthma symptom scores [e.g., asthma control questionnaire (ACQ5) or asthma control test (ACT)], frequency of asthma exacerbations in the 12 months before and after the IOS test, asthma medications, asthma severity (based on GINA criteria), and IOS results. Asthma exacerbations were defined as any worsening of asthma symptoms that required treatment with antibiotics and/or oral corticosteroids or resulted in an unscheduled visit to an accident and emergency department, a hospital, or a general practitioner (13, 14). All exacerbations that occurred were included in the analysis.

Oscillometry was performed in accordance with the manufacturer's recommendations using an impulse oscillometry device (Masterscreen IOS, Jaeger, Germany). Typically, oscillometry was performed on the same day as the respiratory specialist review or within 48 h prior.

Prebronchodilator IOS parameters including R₅, R₂₀, A_X, F_res, and X₅ were analyzed. Normative values for oscillometric parameters were calculated based on data published by Oostveen (15). Z-scores for R₅, R₂₀, X₅, A_X, and F_res were evaluated as continuous and categorical variables to assess their associations with asthma control and exacerbation risk. Other IOS parameters, such as R₅–R₂₀, were not included in this analysis because normative data for these metrics are unavailable; hence, Z-scores could not be calculated. For categorical variables, abnormal lung function was defined as a Z-score > 1.64 for R₅, R₂₀, A_X, and F_res and a Z-score < −1.64 for X₅. The severity of dysfunction was defined a mild, moderate, or severe based on the Z-scores listed in Table 1.

A total of 149 patients were included in this retrospective study. Based on GINA criteria, 69% of patients were classified as having severe asthma (14). Nearly 90% of patients were receiving inhaled corticosteroid-based combination therapy, with equal proportions treated with inhaled corticosteroids (ICS) plus long-acting beta-2 agonists (LABA) and ICS/LABA plus long-acting muscarinic antagonists (LAMA). Based on the FEV₁/FVC ratio, 64% of patients had obstructive airflow. These clinical characteristics are consistent with the typical patient cohort referred to a tertiary asthma clinic. Demographic and clinical characteristics are summarized in Table 2.

Of the 149 patients, 101 (67.8%) had a change in treatment within 12 months after IOS testing. These changes include commencement or switching of a biologic agent (N = 39), ICS dose escalation and/or change to a fine or extra-fine particle ICS formulation (N = 26), or commencement of a LAMA or montelukast (N = 12). In addition, 16 patients underwent treatment “step-down” after optimization of inhaler technique and adherence to original treatments.

The prevalence of abnormal resistance (R₅), defined by Z-scores > 1.64, was high at 70.4%. In contrast, the prevalence of small airways dysfunction was lower, occurring in 40.5% as defined by X₅ Z-score < −1.64 and in 29.5% as defined by A_X Z-scores > 1.64. Abnormal lung function for R₅, X₅, and A_X was associated with poorer symptom control, as assessed by ACQ5 and ACT, compared with patients who had normal oscillometric findings. No significant associations were observed between abnormal F_res and R₂₀ and asthma control (Table 3).

When analyzed as categorical variables, Z-score-defined severity of lung function impairment for R₅, X₅, and A_X was significantly associated with the level of asthma control. Increasing severity of lung function impairment corresponded to poorer asthma control. This relationship was strongest when asthma control was measured using ACQ5 (Figure 1). The strength of these associations with ACT was similar for R₅ but weaker for X₅ and A_X compared with ACQ5 (Supplementary Table S1).

Z-scores for R₅, X₅, and A_X were all significantly correlated with asthma control scores (ACQ5: R₅ r = 0.38, P < 0.001, X₅ r = 0.26, P = 0.001, A_X r = 0.36, P < 0.001; ACT: R₅ r = 0.30, P < 0.001, X₅ r = 0.24, P = 0.003, A_X r = 0.34, P < 0.001). The worse the Z-score, the poorer the asthma control, as assessed by both ACQ5 and ACT (Figures 2A–F).

On multivariate analysis, spirometric indices (FEV₁% predicted) and FeNO were correlated with ACQ5 (P = 0.01 and P = 0.009, respectively), while FEV₁% predicted and FEV₁/FVC were correlated with ACT (r = 0.34, P = 0.001 and r = 0.28, P = 0.02, respectively). The associations between asthma symptom control and Z-scores for R₅ and A_X remained significant after adjusting for spirometry, FeNO, and treatment changes (ACQ5 R₅ r = 0.30, P = 0.005; ACT R₅ r = 0.26, P = 0.05; and ACT A_X r = 0.29, P = 0.04).

Eighty-one patients (54%) experienced at least one documented moderate-to-severe asthma exacerbation in the 12 months before or after the sentinel date. Impaired resistance (R₅) was associated with a significantly increased risk of exacerbation compared with patients with normal R₅ (OR 2.70, 95% CI 1.27–5.17, P = 0.009). The risk of exacerbation increased with greater severity of airway obstruction (abnormal R₅) (Figure 3). Impaired reactance, as reflected by both X₅ and A_X, demonstrated trends toward an elevated risk of exacerbations compared with patients who had normal reactance parameters (X₅: OR 1.24, 95% CI 0.63–2.47, P = 0.53; A_X: OR 1.63, 95% CI 0.75–3.53, P = 0.21). Similarly, there were non-significant trends suggesting that the risk of exacerbations increased with greater severity of impaired reactance. Similar to asthma control, exacerbation risk was not associated with abnormal R₂₀ or F_res (Table 3). The association between R₅ and the risk of exacerbations was not significant on multivariate analysis.

We have shown, in a retrospective study conducted within a tertiary referral clinic for severe asthma, that greater impairment in oscillometric parameters is associated with worse symptoms, i.e., asthma control, and with more frequent severe asthma exacerbations. The relationship between abnormal oscillometry based on Z-scores and poorer asthma control was strengthened by multivariate analysis and is consistent with the post-hoc analysis of the ATLANTIS study, in which small airways dysfunction was identified by A_X, X_5, and R₅–R₂₀ (7). These findings support key recommendations from the recent Delphi study to use Z-scores for defining abnormal lung function as assessed by oscillometry (11).

Significant associations between R₅, X₅, and A_X Z-scores and asthma symptoms used to define control are consistent with our previous analysis, in which abnormal lung function was determined using absolute cutoffs most commonly reported in other studies (5). Similarly, the ATLANTIS study demonstrated that higher absolute values of R₅ and A_X were associated with poorer asthma control (as reflected by lower ACT scores) (6). In addition, Abdo et al. (16) reported significant associations between A_X and asthma control using both the ACT and ACQ7.

In terms of grading the severity of abnormal oscillometry, Liang developed a severity grading system based on categories predicted by FEV₁ Z-scores among patients with asthma (17). For R₅, the classifications of mild, moderate, and severe abnormality closely matched those used in clinical practice (11) and those applied in our study (Table 1). For X₅, moderate impairment was calculated as Z-scores between <−4.5 and ≥−8.5, which differs from the thresholds used in our study (<−2.5 to ≥−4), based on consensus recommendations (11). Liang acknowledged that a major limitation of their methodology was the limited consistency between oscillometry and spirometry and consequently proposed that cutoffs for oscillometry parameters should be guided by clinical practice, as was done in our study (11). However, these discrepancies highlight the need for additional research to further define the severity of abnormal oscillometry.

In our study, the associations between oscillometric parameters and exacerbation risk were strongest for R₅, a measure of airway caliber across the entire airway tree (1). The associations between X₅ and A_X, indicators of small airways dysfunction, and exacerbation risk were weaker. This finding that resistance-based parameters are stronger predictors than reactance is also consistent with the post-hoc analysis of the ATLANTIS study, which focused on patients with mild asthma. In that study, the investigators also used IOS, although Z-scores were derived from 100 healthy individuals included in the study. Thus, our findings indicate that Z-scores for defining severity and abnormality are relatively robust, at least between the two reference equations used (7).

We found no statistically significant associations between X₅ and A_X and asthma exacerbations, whereas a previous analysis from the ATLANTIS study found that X₅, A_X, and R₅–R₂₀ were significantly correlated with asthma exacerbations and that a composite ordinal score based on these three parameters independently predicted the exacerbation risk (18). A retrospective study by Chan and Lipworth found that small airways dysfunction, defined by an A_X ≥ 1.0 kPa/L and R₅–R₂₀ ≥ 0.10 kPa/L/s, was significantly associated with asthma exacerbations (19). Similarly, Gao et al. (20) demonstrated that small airways dysfunction was an important pathological feature among patients with asthma exacerbations and that X₅, A_X, and R₅–R₂₀ were significantly correlated with asthma exacerbations. Measures of reactance, X₅ and A_X, reflect physiologically severe airway narrowing and greater heterogeneity of ventilation. Failure to demonstrate significant relationships with exacerbation risk may be attributable to fewer patients exhibiting moderate to severe impairment when defined using the applied Z-score cutoffs. Greater airway closure and heterogeneity, as measured by single-breath nitrogen washout, has been shown to be associated with an increased risk of exacerbations (21, 22). Whether resistance or reactance parameters relate to exacerbations may differ across populations and may be expected given the marked heterogeneity in underlying pathophysiology among patients.

One of the challenges for clinicians who are less familiar with respiratory oscillometry is the large number of oscillometric parameters and the resulting uncertainty about which ones to use in clinical practice. Similar to spirometry, where interpretation is predominantly based on three core parameters, namely, FEV₁, forced vital capacity (FVC), and FEV₁/FVC (12), the international Delphi study on the interpretation of respiratory oscillometry recommended using three oscillometry indices: R₅, X₅, and A_X (11). Hence, the findings of our study provide further insight and guidance to clinicians on the clinical significance of these parameters. Results from several studies suggest that respiratory oscillometry should be used in conjunction with spirometry (rather than as a replacement for it) (18, 23, 24), as their combination provides a more comprehensive assessment of lung physiology and clinical risks.

The main limitation of this study is its retrospective design and the exclusion of patients with inadequate data, including the presence or absence of exacerbations, as documented in their medical records. There are differences in measurements between oscillometry devices when tested in physical models or healthy participants (15, 25, 26). Differences in measurements between devices, as well as differences in normative values used to calculate Z-scores, may potentially affect the relationships between Z-scores and clinical outcomes (11). Therefore, our findings may not be generalizable to centers that use different oscillometry devices or prediction equations. Robust analysis differentiating the relationship between oscillometry measurements and previous or future exacerbations is limited by relatively few events and by the potential modifying effect of treatment escalations in two-thirds of the cohort in the year following IOS testing.

There are currently no well-established reference equations for the frequency dependence of resistance (R₅–R₂₀); hence, Z-score analysis of this measure of small airways dysfunction could not be performed. Normative data to derive Z-scores for R₅–R₂₀ are needed, as R₅–R₂₀ is a sensitive marker of small airways dysfunction (6), and impairement in R₅–R₂₀ has been associated with an increased risk of asthma exacerbation and poor symptom control (7, 16, 20), including a previously published analysis of this data set, in which abnormality was defined as R₅–R₂₀ > 0.07 kPa/(L/s) (5), as well as the ATLANTIS post-hoc analysis, where abnormal R₅–R₂₀ was defined as a Z-score > 1.645 (7).

One of the strengths of this study is that it was conducted among patients referred to a tertiary severe asthma clinic, a population that matches the patient cohort most likely to have access to respiratory oscillometry in the current real-world clinical setting. As the routine use of oscillometry expands, similar research should be performed in broader adult asthma patient populations managed in primary care. This would help to confirm the clinical utility of this lung function test in this clinical setting.

The findings from our retrospective study provide real-world evidence supporting the use of R₅ Z-scores to define abnormality in preference to other cutoff values, such as absolute values or percentage predicted. In addition, the arbitrary thresholds used to define the severity of abnormality appear to have some relevance in a severe asthma population managed at a tertiary referral clinic.