Abstract
Introduction Noninvasive measurement of respiratory impedance by oscillometry can be used in young children aged from 3 years and those unable to perform forced respiratory manoeuvres. It can discriminate between healthy children and those with respiratory disease. However, its clinical application is limited by the lack of reference data for African paediatric populations. The aim of the present study was to develop reference equations for oscillometry outcomes in South African children and adolescents.
Methods Healthy subjects, enrolled in the Drakenstein Child Health Study, HIV-uninfected adolescents in the Cape Town Adolescent Antiretroviral Cohort and healthy children attending surgical outpatient clinics at Red Cross War Memorial Children's Hospital were measured with conventional spectral (6–32 Hz) and intra-breath (10 Hz) oscillometry. Stepwise linear regression was used to assess the relationship between respiratory variables and anthropometric predictors (height, sex, ancestry) to generate reference equations.
Results A total of 692 subjects, 48.4% female, median age of 5.2 years (range: 3–17 years) were included. The median (interquartile range (IQR)) for weight for age z-score and height for age z-score was −0.42 (−1.11–0.35) and −0.65 (−1.43–0.35), respectively. Stepwise regression demonstrated that all the variables were significantly dependent on height only. Comparison to previous reference data indicated slightly higher resistance and lower compliance values in the smallest children.
Conclusion We established the first respiratory oscillometry reference equations for African children and adolescents, which will facilitate use in early identification and management of respiratory disease. Our results suggest differences in oscillometry measures by ancestry but also highlight the lack of standardisation in methodology.
Abstract
The first respiratory impedance reference equations for African children and adolescents are established to aid in early identification and diagnosis of respiratory impairment https://bit.ly/3HNxKvt
Introduction
Measurement of lung function in early childhood is important for the diagnosis and management of lung disease, to promote optimal lung growth and development. Early life lung function predicts later morbidity and mortality [1, 2]. Spirometry is currently the most commonly performed lung function test, but its use is limited in young children as it requires a forced expiratory manoeuvre, mostly only feasible in children ≥5 years of age. In addition, it is relatively insensitive to detect early lung disease and is a poor measure of peripheral airway function [3].
Oscillometry is an attractive, feasible option in preschool children as it is a simple noninvasive test, requires minimal cooperation and can be used to follow lung function across the life course. Oscillometry measures the response of the respiratory system to an external small-amplitude oscillatory signal of medium (e.g. 4–40 Hz) frequencies which is superimposed on tidal breathing. The oscillatory pressure–flow relationship reflects the mechanical impedance of the respiratory system (Zrs), which consists of two components, namely resistance (Rrs) and reactance (Xrs) [4, 5].
The conventional multifrequency or spectral values of Zrs are generally obtained for a number of consecutive whole breaths (or, more recently, as mean values for the inspiratory and expiratory phases). In contrast, the novel intra-breath measurements, collected with a single-frequency tracking signal, follow the changes in Rrs and Xrs within the breathing cycle [6]. In particular, intra-breath oscillometry focuses on the zero-flow points (end inspiration and expiration); these Zrs values are less influenced by the breathing pattern, which is often variable in young children and reflects less the contribution of the flow-dependent extrapulmonary airways. Owing to the ability to measure Rrs and Xrs at specific points of the respiratory cycle and thus estimate the tidal changes in respiratory mechanics, intra-breath oscillometry has proved more sensitive than standard measures to assess airway obstruction, ventilation inhomogeneity, asthma control and respiratory disease risk [7–10].
Accurate interpretation of lung function measurements depends on the availability of a robust reference standard specific to the population assessed. Population differences in lung function such as anthropometric, sociocultural and environmental characteristics are well recognised [11–13]. Most oscillometry reference standards are specific for Caucasian populations from Europe, North America and Australia between the ages of 2 and 16 years [14–27]. Studies of non-Caucasian participants include Mexican, Thai, Emirati, Korean, Taiwanese, Turkish and Indian population groups with an age range between 3 and 17 years [28–34]. While reference equations derived from Caucasian data may be adequate for Caucasian South Africans, the most recent census describes the South African population as multi-ancestry: 80.7% Black African, 8.8% mixed ancestry (which would include African ancestries, Asian, Caucasian, amongst others) and 2.6% Indian/Asian [35]. Currently, no oscillometry reference equations exist for African populations, despite the high burden of respiratory disease in the region. Additionally, normative data on the novel intra-breath oscillometry measures are scant [8] and are not available for paediatric populations beyond infancy [9, 36]. Recent technical standards for oscillometry equipment and testing, developed by a European Respiratory Society (ERS) task force have highlighted the lack of appropriate paediatric reference standards, especially for underrepresented populations [37].
The aim of this study was to develop reference values for spectral and intra-breath oscillometry measures in healthy South African children and adolescents.
Methods
Participants
Healthy children and adolescents were enrolled from three South African groups: the Drakenstein Child Health Study (DCHS), a birth cohort study [38]; the Cape Town Adolescent Antiretroviral Cohort (CTAAC), including a healthy HIV-uninfected control group [39]; healthy children with no history of respiratory illnesses attending surgical outpatient clinics at Red Cross War Memorial Children's Hospital, Cape Town (HCSOC). Participants from the DCHS birth cohort were tested annually (collected 2015–2020) from 3 to 7 years, with one randomly selected time point per individual included in this study to remove any bias in the sample. Participants from CTAAC (11–15 years) and HCSOC (8–17 years) were tested between 2018 and 2020. All participants were of African ancestry, self-identifying as either Black African or mixed ancestry and from predominantly low socioeconomic communities [38, 39]. Socioeconomic status was determined from questionnaires completed at study visits and was based on household income, including accessed social grants. Household smoking was self-reported.
All children were healthy at the time of testing. Prior to testing they were screened for respiratory symptoms (cough, wheeze, difficult breathing) using a clinical and symptom study questionnaire based on the validated International Study of Asthma and Allergies in Childhood (ISAAC) questionnaire. Those with acute lower respiratory tract illness (LRTI) or any respiratory illnesses in the previous month were excluded from testing. LRTI was defined as per the World Health Organization (WHO) case definition [40]. Children with any chronic respiratory conditions (self-reported or doctor diagnosed) including recurrent or persistent wheeze as well as chronic illnesses such as HIV infection, cardiac or neurological disorders were also excluded.
Ethics
The study was approved by the University of Cape Town Faculty of Health Sciences (048/2020; 082/2018; 423/2012). Parents or legal guardians gave written informed consent in their first language and assent was given by all youth 7 years and older.
Lung function measurements
Oscillometry data were obtained using the same custom-made equipment (INCIRCLE wavetube system, University of Szeged, Hungary) [41, 42] by a trained team of three technologists. Measurements were made with the individual sitting comfortably, breathing through a mouthpiece and filter, nose clip in place and the cheeks firmly supported, in accordance with published consensus guidelines [37]. The oscillometry system operated with either a pseudo-random signal in the 6–32 Hz range (conventional oscillometry) or a 10 Hz intra-breath tracking frequency; the latter corresponds to a 0.1-s temporal resolution allowing identification of the zero-flow Zrs values (see below). Measurements consisted of a maximum of five 16-s epochs of multifrequency oscillations to yield a minimum of three acceptable measurements and one 16-s intra-breath recording, repeated if necessary to obtain a minimum of five regular breaths, i.e. without any vocal cord noise, apnoea, irregular breathing pattern, glottic closure, leak or sighs.
Conventional oscillometry measures included Rrs at 6 Hz (R6), 8 Hz (R8) and 10 Hz (R10), Xrs at 6 Hz (X6), 8 Hz (X8) and 10 Hz (X10), frequency dependence of Rrs (R6–R20), resonance frequency (Fres) and the absolute area of the Xrs versus frequency plot between 6 Hz and Fres (Ax). Additionally, mean respiratory system resistance (R), inertance (I) and compliance (C) were determined from model fitting to the measured Zrs data in the frequency range 10–20 Hz for R and 6–32 Hz for I and C [41–43]. This procedure is illustrated in supplementary figure S1.
The intra-breath measurements were characterised by Rrs at end inspiration (ReI) and at end expiration (ReE), Xrs at end expiration (XeE) and end inspiration (XeI), and their tidal changes ReE–ReI (ΔR) and XeE–XeI (ΔX).
Statistical analysis
Data were analysed using STATA 14.1 (STATA Corporation, College Station, TX, USA) and presented as frequencies, proportions, median and interquartile range (IQR) as appropriate. A natural logarithmic transformation was used for R, R6, R8, R10, C, Fres, Ax, ReE and ReI. The effect of sex on oscillometry outcomes was investigated using Wilcoxon rank-sum test (Mann–Whitney U-test), and the relationship between the oscillometry outcomes and anthropometric covariates (sex, height (Ht) and ancestry) were explored using a backward stepwise linear regression. A reference equation for each outcome was generated and presented with the adjusted R2 and standard error of the estimate (SEE) to allow z-score calculation: z-score=(measured value − predicted value)/SEE.
In order to assess the effect of puberty (particularly as numbers in this age group were relatively low) on the reference equation, backward stepwise regressions with anthropometric data for sex, Ht and ancestry were used to generate a reference equation in children between 3 and 7 years of age from the DCHS cohort.
Bayesian Information Criteria (BIC) models were used to select the best model fit for each of the oscillometry outcomes. In addition, diagnostic checks were done to ensure that the assumptions of linear regression were not violated. This included testing for the presence of multicollinearity using variance inflation factor, normality of residuals using histograms, kernel density and quantile–quantile plots, and homoscedasticity using residual versus fitted plots.
Results
A total of 692 children between the ages of 3 and 17 years were included in the study; 573 (82.8%) were from the DCHS cohort, 38 (5.5%) and 81 (11.7%) from the CTAAC and HCSOC sites, respectively, all representative of the same population group. All were of African ancestry, 361 (52%) were self-identified Black Africans and 331 (48%) of mixed ancestry. Demographic details and anthropometry of cohort including the weight for age z-score and Ht for age z-score data are summarised in table 1 and detailed in supplementary table S1. A total of 13 children (2%) were severely stunted (≤3 standard deviations below the mean) and four children (0.6%) were severely underweight (≤3 standard deviations below the mean). Six children (0.9%) were obese (≥3 standard deviations above the mean). Notably, 29% of mothers self-reported smoking.
Demographics of all children and adolescents (n=692)
The conventional and intra-breath impedance measures are shown for all age groups in supplementary table S2. Values of Fres were available (i.e. fell in the 6–32 Hz range) in 514 (74.3%), less in the youngest and in most of the older children. The R and C data exhibited marked Ht dependences (figure 1); the compensatory parameter I has less physiological importance, and its values are not reported here. In figure 1, regressions on R and C versus Ht established in earlier work using model fitting are also plotted for comparison. The changes in various Zrs measures with Ht are represented in supplementary figures S2a and b, exhibiting a decrease in R6 and increase in X6 with Ht. As shown in supplementary figure S2c–e, Fres, Ax and R6–R20 decreased with increasing Ht. The intra-breath measures are plotted against Ht in supplementary figure S3. ΔR and ΔX exhibited large scatters but were predominantly positive (supplementary figure S3c and f, respectively).
a) Respiratory system resistance (R) and b) compliance (C) versus height in healthy children and adolescents. Black solid and blue dashed lines indicate the regressions on the total population (n=692) and the 3- to 7-year age range (n=573), respectively. R and C versus height regressions from previous work are also plotted for comparison: Hantos et al. [22] (green), Mazurek et al. [25] (red) and Shackleton et al. [41] (pink).
Stepwise regression analysis demonstrated the significant association with Ht for all variables; R6, X8, XeE and XeI were also found to be associated with sex; X8 and X10 with ancestry. However, as demonstrated by the BIC model (supplementary table S3), these additional associations offered negligible contribution to predictive models. Thus, only Ht was included in all regression models. The reference equations are compiled in table 2. An online tool using these equations for z-score calculation is available from supplementary material. The limits of normal are +1.64 z-score for R values, Fres, Ax, R6–20, and −1.64 z-score for X values.
Reference equations for children and adolescents 3 to 17 years of age
To assess the effect of puberty on the reference range equations, stepwise regression in children from the DCHS cohort was done; the coefficients obtained (supplementary table S4) remained very close to that of the reference equations for the entire cohort, with a moderate decrease in adjusted R2 attributable to the narrower Ht range. The consistency of reference equations between the full and reduced ranges in Ht is also illustrated in figure 1 and supplementary figures S2 and S3. Overall, the deviations between the full and reduced Ht range predictions are significant only in ΔR (Figure S2), and mild in C (figure 1), X6 and Ax (supplementary figure S2), ReI, XeE and XeI (supplementary figure S3). Excellent agreements were found for R and R6 between the full and reduced Ht range predictions.
The comparison between R6 predicted with the current equation and other published reference equations for different populations is illustrated in figure 2 [14–26, 30, 34]. Initially, we considered reference data from previous studies only if 1) Rrs values at around 5–6 Hz were analysed, 2) Ht was the single independent variable and 3) higher-order than linear relationship to Ht was assumed. The main features of these studies are summarised in table 3. Our R6 values are similar to the Rrs plots of the other studies at the medium Ht range. Data from nine additional studies that assumed the linear Rrs versus Ht relationship (supplementary table S5) are shown in supplementary figure S4; these reference lines are rather scattered and fall outside the nonlinear regressions and illustrate the inadequacy of the linear Ht dependence, especially in the wide Ht range.
Comparison of respiratory resistance (Rrs) versus height relationships established in the present and previous studies (see table 3 for details).
Summary of reference studies on resistance (R) versus height (Ht) relationships#
Discussion
This is the first study to report oscillometry data in healthy African children and adolescents that includes both conventional and intra-breath measures. Our findings compare favourably with previously published normative data from other populations, suggesting that standardisation of methodology is a key factor accounting for cohort differences, while indicating the role of population differences.
The vast majority of normative data derived since 1972 includes predominantly Caucasian populations, covering various age ranges, and utilises a variety of oscillometry equipment. Additionally, the predictions employed different statistical models and anthropometric variables, further hindering direct comparison. We therefore limited the comparison of the present data to studies that reported Ht as the only independent variable and used a nonlinear Ht dependence of Zrs measures, as appropriate. The inappropriateness of the linear Rrs versus Ht relationship is highlighted in supplementary figure S4.
To our knowledge, figure 2 represents the most comprehensive survey on the Ht dependence of Rrs values in children and adolescents, although the permissive inclusion of the different lowest frequencies (4, 5 or 6 Hz) or frequency ranges for model fitting increases the variability. The roughly inverse relationships between Rrs and Ht exhibit some variability between the normative studies, and our data, which covers one of the widest Ht ranges, is consistent with this (see figure 2). We note that some nonlinear models, such as polynomial regressions, may predict unrealistic inflections in the Rrs versus Ht relationships towards lower Ht [27] or higher Ht [26]. Apart from this, in the lowest Ht range (<120 cm), our 6-Hz Rrs values are among the highest, together with lower-frequency (4 and 5 Hz) measurements expected to result in higher Rrs [16, 21], and that obtained with a special (head generator) device [25] leads to higher values than the uncorrected Rrs. A more rigorous comparison covering only Zrs data at 10 Hz is presented in supplementary figure S5; the relative position of our R10 values remains similar to that shown in figure 2, whereas our X10 data are rather in the middle of the smaller set of available X10 predictions. There appears to be a systematic difference between our predictions and those based on the same oscillometry setup employed in a population of Caucasian children [41]. Comparison of Fres versus Ht regressions reveal a wide scatter between studies, in which our data take a midposition.
Ethnic differences in oscillometry measurements obtained with the same device have been noted [28]; ancestry, environmental and body habitus differences, which influence Ht, were the most likely suggested reasons accounting for this discrepancy. Moreover, differences in Ht between populations appear to be greatest in preschool years [44]. The fact that our cohort had a higher Rrs at Ht <120 cm possibly indicates that the younger children in our study may have smaller lungs for a given Ht compared to other healthy reference populations. We noted a higher Rrs with a predominantly lower Xrs in oscillometry variables in females compared to males, similar to findings by others [9, 14]. However, we found that sex was not independently predictive after adjusting for Ht. Difference in Zrs between females and males in childhood and adolescence may be primarily driven by smaller lung volumes and narrower airways in females compared to males [45, 46].
Many early life factors influence lung growth and development, including environmental smoke exposure and socioeconomic status (SES) [11, 38]. Our study population was from a predominantly low SES community with high smoke exposure; 29% of mothers in our cohort smoked [38, 39]. However, this subtle difference in Rrs at small Ht should be interpreted with care as measurement accuracy has been shown to be rather variable between commercial oscillometry devices at high load impedances, such as Zrs in small children [47]. It is worth noting that the reference equipment in this device comparison study [47] was the wavetube setup [41, 42] employed in the present investigation. Efforts are underway to align and standardise equipment signalling and processing, including the development of consensus guidelines [37, 41, 47, 48].
In addition to the conventionally reported Rrs and Xrs values at the oscillation frequencies, Fres and Ax are increasingly used to determine the elasticity and ventilation inhomogeneity of the respiratory system, whereas R6–20 reflects peripheral inhomogeneity and airway obstruction [47]. With the exception of Fres, these measures are very sensitive to the value of the lowest oscillation frequency, which is rather variable between devices and hence different studies; this is another argument calling for urgent standardisation effort. We have added the mean Rrs (R) and C parameters from model fitting [42, 43] and propose these measures as more robust descriptors of the resistive and reactive behaviour of the respiratory system than the Rrs and Xrs readings at individual frequencies. Reports on R and C in paediatric populations are scant in the literature [22, 25]; the most important comparison with a previous study [41] that employed the same oscillometry device and evaluation procedure (figure 1) reinforces the single-frequency findings on the relatively high resistance and low compliance values in our preschooler population.
This study is one of the first to develop comprehensive reference equations for the novel intra-breath oscillometry measurements in the paediatric population [7–9]. Intra-breath measures have been shown to be a measure of airway obstruction in preschool children with wheezing and altered in children with asthma [7]. We have also previously shown that these measurements were able to predict healthy infants at risk for lower respiratory tract infections [9]. The clinical utility of the intra-breath measures together with standardised conventional spectral variables in children need to be fully established, and ongoing work is recommended in this area to facilitate diagnosis of respiratory disease with more precision.
Strengths of this study include the large sample of healthy children with data collected using the same equipment and methodology. The age range of children extended from preschool to adolescence provides us with a tool useful through childhood and adolescence. The availability of an online tool for calculation of the lower/upper limits of normal and z-score simplifies this further, facilitating its use for users in the field.
A limitation of this study is the relatively small sample size in the 8- to 17-year-old age interval, a time of variable lung growth particularly between sexes, thus assessing the impact of puberty was limited. Since there is a remarkable consistency in the Ht dependencies of the major oscillometry measures between the full (3–17 years) and the lower (3–7 years) age ranges, these reference equations aim to guide clinical practice until they are updated by using more balanced patient cohorts. In addition, these normative values are based only on data from a single province, the Western Cape, of South Africa; therefore this may not necessarily be generalisable to the rest of the Southern Africa region, although recent multi-province data collection in healthy individuals shows concordance in spirometry measurements [45].
In conclusion, we have established the first respiratory impedance reference equations for South African children and adolescents with an online tool to facilitate its use in early identification and management of respiratory disease. While our results reveal differences in oscillometry measures by ancestry, they also highlight the lack of standardisation in methodology.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
FOT z-score calculator 00371-2022.SUPPLEMENT
Supplementary material 00371-2022.SUPPLEMENT
Acknowledgements
We thank the study staff in Paarl and Cape Town, the study data and laboratory teams, and the clinical and administrative staff of the Western Cape Government Health Department at Paarl and Red Cross War Memorial Children's Hospitals. We thank the families and children who participated in this study. We thank Dorottya Czövek (First Department of Paediatrics, Semmelweis University, Budapest, Hungary) for support with DCHS preschool oscillometry measurement setup and Gergely Makan (Department of Technical Informatics, University of Szeged, Szeged, Hungary) for data management support.
Footnotes
Provenance: Submitted article, peer reviewed.
Support statement: The study received funding from the Bill and Melinda Gates Foundation (grant numbers OPP1017641 and OPP1017579); The Wellcome Trust (098479/Z/12/Z and 204755/Z/162); MRC South Africa, The National Research Foundation, South Africa; National Research, Development and Innovation Office (OTKA grant 128701), Hungary; European Respiratory Society (INCIRCLE CRC-2013-02); National Institute of Health R01HD074051; and Harry Crossley Clinical Research Fellowship. Funding information for this article has been deposited with the Crossref Funder Registry.
Conflict of interest: S. Chaya reports support for the present manuscript from Harry Crossley Clinical Research Fellowship.
Conflict of interest: H.J. Zar reports support for the present manuscript from the Bill and Melinda Gates Foundation, the Wellcome Trust, the South African Medical Research Council, the National Research Foundation, SA, and the NIH, USA.
Conflict of interest: Z. Hantos reports grants or contracts from the Hungarian Scientific Research Fund (grant K 128701) and European Respiratory Society Clinical Research Collaboration award CRC_2013-02_INCIRCLE, outside the submitted work; patents planned, issued or pending (2005903034 – A method of diagnosing a respiratory disease or disorder or monitoring treatment of same and a device for use therein; Australian patent issued), disclosure made outside the submitted work.
Conflict of interest: D.M. Gray reports support for the present manuscript from Wellcome Trust and is an executive member of Pan African Thoracic Society, outside the submitted work.
Conflict of interest: The remaining authors have nothing to disclose.
- Received July 25, 2022.
- Accepted February 7, 2023.
- Copyright ©The authors 2023
This version is distributed under the terms of the Creative Commons Attribution Non-Commercial Licence 4.0. For commercial reproduction rights and permissions contact permissions{at}ersnet.org
References
