Reply to: Height seems to be the only variable that determines lung oscillometry indices

Extract

We greatly appreciate the positive comments raised by D. Ghorpade and colleagues regarding our recent publication on reference equations for impulse oscillometry (IOS) in persons aged from 2.7 to 90 years [1], and we fully agree with them about the exciting role that IOS is gaining for the assessment of lung function. In their communication, these renowned researchers made some speculations that induced them to conclude that the subject's height is the variable with the upmost relevance for lung function, as evaluated by IOS, and they postulated that height may be the only variable needed for interpreting the results of IOS parameters. Of course, we agree with the long-standing notion that height is a major determinant of lung mechanics, but the contributing role of sex, age and body weight have always been acknowledged too [2].

Shareable abstract

Age, sex and BMI, as well as height, should be used in IOS reference equations https://bit.ly/3VMUMLz

Reply to D. Ghorpade and colleagues:

We greatly appreciate the positive comments raised by D. Ghorpade and colleagues regarding our recent publication on reference equations for impulse oscillometry (IOS) in persons aged from 2.7 to 90 years [1], and we fully agree with them about the exciting role that IOS is gaining for the assessment of lung function. In their communication, these renowned researchers made some speculations that induced them to conclude that the subject's height is the variable with the upmost relevance for lung function, as evaluated by IOS, and they postulated that height may be the only variable needed for interpreting the results of IOS parameters. Of course, we agree with the long-standing notion that height is a major determinant of lung mechanics, but the contributing role of sex, age and body weight have always been acknowledged too [2]. The preeminent role of height in lung mechanics is quite reasonable because stature determines the lung size, and hence the lung volumes, the airways calibres and the degree of elastic recoil. D. Ghorpade and colleagues suggested that differences observed between males and females in IOS measurements might be due to differences in stature. However, there are sex differences in lung function not explained by height, and probably related to less elastic recoil and/or less respiratory muscle strength in women [3, 4]. On the other hand, ageing is accompanied by diminution of elastic recoil due to a progressive loss of pulmonary elastin [3–5], and an increased body weight is associated with worse lung function, mainly manifested by a lower expiratory reserve volume ascribed to mechanical interference of adipose tissue [6]. In our study [1], the relative contribution of each variable (height, sex, age and body mass index (BMI)) in the prediction of each IOS variable can be assessed by the t-tests and their associated p-values, and as can be seen in tables 2 and 3 of our article, apart from height, in the majority of IOS variables, age, BMI, and less frequently sex were statistically significant contributory variables.

To evaluate how much our reference values calculated by a segmented regression model (including sex, height, age and BMI as independent variables) differ from that calculated by a straight line regression model including height as the only explanatory variable, as proposed by D. Ghorpade and colleagues, we applied both models to our published population of 830 subjects [1] and evaluated the residuals (a residual is the difference between the real IOS value observed in a subject minus the value predicted for that subject by either model). We found that the second model, with height as the only predictor variable, consistently yielded higher errors (residuals), as compared with our segmented model considering height, sex, age and BMI (figure 1). Moreover, for some IOS variables, a clear underestimation/overestimation was observed in some groups of subjects when only height was used for the calculation of predicted values.

FIGURE 1.

Comparison of residuals after applying two reference equations, segmented and linear, to the same population (830 men and women). The segmented model included height, sex, age and body mass index (BMI) as predictor variables, age being adjusted for two inflection points. The linear model only included height as a predictor variable. Resistance and reactance at 5 Hz (R₅ and X₅, respectively) and area of reactance (A_X) are plotted against height, age and BMI. Residuals indicate how much the observed impulse oscillometry system value departs from the value predicted by the respective equation.

It is interesting to note that even with the inclusion of height, sex, age and BMI as explanatory variables, a relatively large dispersion of residuals still persists in most IOS variables, which implies that additional factors not yet identified must also be influencing lung function. In this context, we recently reported that salivary concentrations of granulocyte–macrophage colony-stimulating factor, a cytokine indispensable for surfactant catabolism, and glucagon, a hormone with bronchodilation activity, were correlated with some IOS variables [7]. Likewise, we have unpublished evidence that some IOS variables are also associated with heart rate variability, which might imply that neural influences arriving at smooth muscles, either bronchial or vascular, might also influence the results of lung function tests, including IOS. Therefore, it is possible that in the future, the characterisation and inclusion of such additional influential factors might improve the accuracy of pulmonary function tests reference equations.

Finally, in their letter, D. Ghorpade and colleagues suggested that in some age ranges (e.g. children under ∼18 years old), the only variable needed to evaluate lung function is height. Even if this were true for some age ranges, such an approach would unavoidably lead us again to the frequent problem seen when two or more reference equations are used to cover the full life span, i.e. a sudden change in IOS interpretation due to the change from one equation to the other when a person passes from one age range to another.

In conclusion, we respectfully disagree with D. Ghorpade and colleagues in the sense that we advocate for the inclusion of height, sex, age and BMI in IOS reference equations.