Discussion:
Here we describe the first comparison of bronchodilator assessments performed using oscillometry and spirometry in a preterm-born population. Both oscillometry and spirometry demonstrate that those in the preterm group have a greater response to salbutamol, however the magnitude of the change measured by spirometry and oscillometry was only weakly correlated. Similarly, when a response was defined as ‘significant’ using published thresholds, there was a poor agreement between tests.
Spirometry is the “gold-standard” with which to assess the bronchodilator response, however, we show that oscillometry provides additional information especially in preterm individuals with normal spirometry and respiratory symptoms (wheeze). Spirometry may not detect mild disease that presents as ‘normal’ between exacerbations, for example, an increase in FEV1 ≥12% and 200mL was present in only 17.3% of asthmatics in a meta-analysis of 3 large population studies (n=2,833).35 Spirometry can remain preserved in symptomatic individuals until an advanced stage of lung disease, whilst oscillometry is sensitive to changes in small airway function36 and offers some advantages over spirometry in the identification of individuals with poor asthma control.37,38  Our finding that preterm individuals with an oscillometry BDR only had increased respiratory symptoms, but normal spirometry, suggests that oscillometry has clinical value as a supplement to-, rather than surrogate for- assessing the bronchodilator response in this population.
Oscillometry is not a suitable surrogate for spirometry due to poor agreement between oscillometry and spirometry detected BDRs, and a weak correlation between ΔFEV1 and oscillometry outcome measures. This has been reported previously in a retrospective review of 592 children with asthma or suspected asthma; 18% had a BDR by spirometry only, 9% by oscillometry only, and only 8% had a BDR by both tests.39 Oscillometry and spirometry have different measurement techniques (tidal breathing vs forced manoeuvres) which likely partially explains this discrepancy. Performed during tidal breathing, oscillometry is perceived as a sensitive measure of small airway disease.2 In contrast, spirometry measures flow and volume during a forced manoeuvre, and may be better able to determine the function of larger airways.40 The lack of sensitivity of oscillometry to detect a significant FEV1 change raises concerns surrounding the ability of oscillometry to detect more global changes in airway resistive forces. Notwithstanding, it may be that oscillometry has value in discriminating disease isolated to the small airways in those born preterm.
This poor agreement between oscillometry and spirometry BDR detection is likely exacerbated by the current published definitions of a BDR using both tests. Fixed cut-offs are typically recommended (e.g., ≥12% improvement and 200ml in FEV1) and hence used here, however the response to a bronchodilator is inversely proportional to baseline lung function and therefore also dependant on age, height, and sex.41 Whilst the recently published ATS/ERS guidelines have gone someway to addressing this in spirometry, recommending that the magnitude of the change should be normalised to an individual’s predicted value, rather than their baseline value41, this had little influence on our results (supplementary tables E1 and E2). In oscillometry, there has been debate as to whether a BDR should be expressed as absolute, relative or z-score change, with the latest ERS technical standards advocating for relative change, until there are sufficiently robust healthy data for oscillometry to permit a z-score approach. The published cut-offs (ΔRrs5 ≤-40%, ΔXrs5 ≥50% or ΔAX ≤−80%) were developed from data from healthy children1 and reports are emerging that these values may be too stringent for the adult population.24,42Using a z-score change that incorporates the variability of the reference data set may be a suitable way to address this limitation, however reference values for oscillometry are currently limited and, in part, device specific. There is currently no recommendation for cut-offs for the intrabreath oscillometry measures. Nevertheless, whilst the published cut-offs may be problematic, the weak correlation observed between ΔFEV1% and oscillometry outcomes supports that the poor agreement is more likely reflective of the differences in airway physiology that these tests represent, rather the purely an issue of classification.
Ours is the first study to report within-breath changes with single-frequency oscillometry with R10insp-exp and X10insp-exp measures pre- and post-bronchodilator in preterm-born children. Our findings suggest these within-breath measures may be less useful than spirometry and conventional spectral oscillometry when assessing the bronchodilator response in this population. We observed no difference in the magnitude of the R10insp-exp and X10insp-exp response to a bronchodilator, rather changes were proportional across the breath cycle, and reflected global changes in resistance and reactance at 10 Hz. Recent studies have suggested that intra-breath oscillometry measures may be more useful in detecting wheeze43 and predicting lower respiratory tract infections44 in infants and young children than spectral oscillometry, and in adults with COPD.45 We observed no differences in X10insp-exp measures between preterm and healthy participants at baseline, or in response to a bronchodilator, meaning that airway inhomogeneity is likely not the primary driver of airway obstruction in preterm-born individuals. Indeed, small studies measuring ventilation inhomogeneity using multiple breath washout report no differences between preterm and term-born infants.46-48 Airway obstruction in preterm-born individuals may instead be more attributable to reduced compliance as suggested by our spectral oscillometry outcomes. As the literature around within-breath oscillometry is limited, there are no references for ‘normal’ measures and the physiology behind within-breath outcomes remains somewhat speculative. Further work is needed to explore the physiology of within-breath changes and its implications in individuals born preterm.
Consistent with previous findings we show that, whilst oscillometry is a more feasible test, being born 32 weeks gestation or less did not influence feasibility.40 It should be noted that those with severe impairment were excluded at the time of recruitment, however our results show that for most survivors of preterm birth, similar test success rates should be expected for those born at term in the age range studied. It should be noted that these measurements were made during a research appointment, which is not subject to the same time constraints as a clinical service, however reviews of routine clinical testing reveal similar findings.40 That oscillometry (intrabreath and spectral) is feasible in a preterm born population reinforces its value in both a research and clinical context.