Development of reference values and equations for the pulmonary function of Nigerian children aged 6–11 years measured with digital peak flow meter and its validation against local and the Global Lung Function Initiative (GLI) equations

Introduction

Background

Spirometry is used to objectively diagnose and monitor chronic respiratory diseases (CRDs) such as asthma (1-3). Despite an enormous burden of CRDs, resource-constraint sub-Saharan countries like Nigeria suffer poor availability of spirometry services, especially for paediatric care, because of cost-, skill- and maintenance-related challenges (3-6). In such settings, mechanical peak flow meters (mPFM) such as mini-Wright PFM, which are cheaper and easier to use, have limited but valuable usefulness for lung function (LF) measurements (1,6,7). Thus, there are published local reference equations (REs) (8-12) for the peak expiratory flow (PEF) which is the maximum flow of air expired during a forced expiration after a maximal inspiration (2). While PEF can be used in the diagnosis and management of asthma (as percentage of personal best PEF, percentage PEF change from baseline or percentage of predicted PEF), it is highly effort-dependent potentially leading to large intra- and inter-person variability (1,13,14). In contrast, the forced expiratory volume in one second (FEV₁), which is the volume of air that is expired forcefully by the first second of a maximal expiration after maximal inspiration, is a more reliable measure of LF, being less influenced by effort and diurnal variation (2,3,14). Fortunately, there are now digital or electronic peak flow meters (ePFM) which, unlike mPFM, measure both PEF and FEV₁, and also have inbuilt quality checks and memory; these characteristics make ePFM more accurate than mPFMs (15-18).

Rationale and knowledge gap

Accurate interpretation of LF depends on the availability of REs that are most representative of a given population (2,13). The use of ePFMs in clinical settings may be limited by the non-availability of REs of PEF and FEV₁ specifically measured with ePFM; REs derived from measurements made with mPFMs or spirometers may not be valid for LF measured with ePFMs because device-type affects the predictive accuracy of REs, potentially causing mis-diagnosis and over- or under-treatment (19,20). Although Arigliani et al. reported that the Global Lung Function Initiative (GLI)-African-American equations, including for FEV₁, was valid for Nigerian children living in north-central Nigeria (21), this may not extrapolate to other parts of the country because of wide inter-regional socioeconomic, environmental and nutritional variations. There is thus need to further study the association of LF with these factors and anthropometric (measures of body size) or anthropomorphic factors (measures of body proportionality) in order to better account for LF variations within and between populations (22-26).

Objectives

Therefore, we (I) determined the mean and percentile values of ePFM-measured PEF and FEV₁ of 6–11-year-old Nigerian children; (II) determined the association between LF and socio-demographic [age; sex; school type as a proxy for socioeconomic status (SES)], anthropometric [height, weight, lower limb length (LLL), chest circumference (CC)], anthropomorphic [body mass index (BMI), chest-circumference/height ratio (CC/H); upper-segment volume (UV)], nutritional (stunting, overweight/obesity) and environmental (diurnal variation, season) factors; (III) validated measured LF against Nigerian, non-Nigerian and GLI equations; and (IV) developed new PEF and FEV₁ equations for Nigerian school-age children. We present this article in accordance with the reporting TRIPOD checklist for observational studies of the development and validation of prediction equations (available at https://jxym.amegroups.com/article/view/10.21037/jxym-23-29/rc) (27).

Methods

Design/population

This report is a secondary analysis of PEF and FEV₁ data of healthy 6–11-year-old Nigerian school-children, originally enrolled from 14 primary schools into a cross-sectional study of their six-minute walk test (dataset: https://doi.org/10.17632/fcw8vvrxkw.1) (28). The procedural details of all anthropometric and lung-function measurements have been previously published (see supporting document at: https://doi.org/10.1002/ppul.25986) (29) and are summarized below.

Ethical approval and sample selection

The original study was conducted in accordance with the Declaration of Helsinki (as revised in 2013), and was approved by the Lagos State University Teaching Hospital (LASUTH)’s Health Research Ethics Committee (LREC/10/06/486). We also obtained written informed consent and assent from parents/guardians and the children, respectively. Using multi-stage random sampling, we enrolled healthy pupils from primary schools within Ikeja Local Government Area (LGA) of Lagos from 2016–2017. For this report, we excluded children with history/features suggestive of haematologic or cardiopulmonary disorders, recent respiratory tract infections (RTI) or those unable to perform LF maneuvers from our database of 943 children.

Data collection procedures

We used standard procedures (29) to measure weight with digital weight scale (Omron^TM HN283, Omron Ltd., Japan), standing height with stadiometer (Prestige^® Height Stand, Haerdik Medi-Tech, India), and both CC and LLL with an inelastic tape measure (inter-rater reliability for CC, LLL and height were respectively 0.997, 0.826 and 0.992) (29). We used a pocket-size ePFM (Asma-1™, Vitalograph, UK; https://vitalograph.com/product/asma-1-asthma-monitor/) which, according to the manufacturer, measures, and displays PEF in L/min and FEV₁ in L using a stator rotor sensor. The device met the accuracy and precision standards of the American Thoracic Society (ATS)/European Respiratory Society (ERS) for spirometers [accuracy: better than ±3% (FEV₁), ±10% (PEF); flow impedance: better than 0.15 kPa/L/s at 14 L/s; measurement range: FEV₁: 0–9.99 L BTPS, PEF: 0–840 L/min BTPS]. Made for “clinic or personal home monitoring”, it has in-built quality check mechanism that flags and disqualifies measurements when there is any of: (I) ‘bad blows’ resulting from cough; (II) time to PEF >120 s; or (III) extrapolated volume >5% or >100 mL of FEV₆. We conducted FEV₁ maneuvers according to manufacturer’s instructions, in standing position, thus: with the head slightly elevated, the participant breathed in maximally and rapidly, placed the device with an attached disposable filter mouth-piece (SafeTway^TM, Vitalograph, Ireland) into the mouth maintaining a seal with the lips and immediately ‘blowed’ out the air “as hard and fast as possible”. These steps were repeated at least two more times. Nose-clips were not used. We recorded the best PEF and FEV₁ of at least three attempts, using a single ePFM (Model No. 4,000; LOT No. 0107/2014) throughout the study (conducted mostly by P.O.U.). We did not conduct calibration checks on the device because, according to the manufacturer, it required no routine calibration; however, the lead author periodically used it to measure personal LF to ascertain consistent readings.

Data management

Statistical analyses were conducted with JASP Statistics version 0.18.0 (University of Amsterdam, the Netherlands) and Bluesky Statistics version 7.50 (BlueSky Statistics LLC, Chicago, USA), at significant P value of <0.05. We computed BMI [as weight in kg/(height in m)²]; CC/H; upper-segment length (UL) (as height minus LLL); upper-segment/height ratio (as US/H) and UV [as (CC²×US)/4π, assuming the chest to be a cylinder we used equation for cylinder volume; π=3.143] (30). The distribution of continuous variables was assessed with Shapiro-Wilk test and eye-balling of histograms. We used means [standard deviation (SD)] and frequencies to summarise continuous and categorical variables, respectively; and chi-square and student t-test (or ANOVA) to compare categorical and continuous variables among groups, respectively. We used Cohen’s d to quantify effect sizes of significant t-test comparisons: values <0.2, 0.2–0.49, 0.5–0.79 and 0.8 were graded trivial/negligible, small, moderate and large, respectively (31). We assessed linear correlation between LF and anthropometric and anthropomorphic variables with Pearson’s coefficient (r: <0.1, 0.3 and 0.5 implied trivial/small, moderate and strong correlations, respectively) (31). These correlations were also assessed with scatterplots with locally estimated scatterplot smoothing (LOESS) lines, rather than regression lines, to better assess the linearity of PEF and FEV₁ across the range of the determinant variables. We used GLI’s online calculator (http://gli-calculator.ersnet.org/, GLI 2021, Version 2.0) to derive predicted FEV₁ values and Z-scores (zFEV₁) for our sample, and then tested the fit of our sample’s measured FEV₁ to FEV₁ predicted by the GLI equations (comprising four race-specific equations for African-Americans, Caucasians, South-east Asian and North-east Asian, and two race-neutral or race-composite equations, namely “Others” and the new “Global”) (23,32), using GLI’s validation criteria [mean (SD) predicted Z-score <0.5 (1.0) with 5% values below lower limits of normal (<1.645 Z-scores) and 5% above upper limit of normal (ULN) (>1.645 Z-scores)]. Furthermore, Cohen’s kappa (κ) was used to explore the agreement (concordance) of GLI-African American with each of GLI-Global and GLI-Others with respect to classification of participants with low FEV₁ (zFEV₁ <5^th percentile or LLN). Since there are no GLI PEF equations, we calculated PEF Z-scores (zPEF) thus: (measured PEF − predicted PEF)/standard error of estimate (SEE) (33), with predicted PEF and SEE derived using previously published PEF equations for children in Lagos, Nigeria, by Faleti (34). Agreement between measured PEF and PEF predicted by Nigerian and non-Nigerian equations for school-age children was assessed with Bland-Altman analysis [mean bias and limits-of-agreement (LoAs) statistics] (35) based on two validation assumptions: (I) there is no significant mean bias (or error) between measured and predicted PEF when the 95% confidence intervals of the mean bias includes the line of equality (zero line) (36); (II) an acceptable LoA should not exceed 20% of the mean PEF for each sex, given that PEF may vary up to 20% between devices (3). We also developed new PEF and FEV₁ prediction equations using stepwise multivariable linear regression models with outcome variables (OVs) being PEF or FEV₁ and predictor variables being demo-anthropometric variables which show significant correlation with the OVs of at least small effect size (P<0.05; r≥0.2 or Cohen’s d ≥0.2) (37). Complete-case approach was adopted for all bivariate and multivariate analyses. We also used RefCurv^TM (refcurv.com) (38), a free-source R-based software, to generate PEF and FEV₁ smoothed percentile curves. RefCurv^TM uses the General Additive Models for Location Scale and Shape (GAMLSS) add-on packages to generate penalized splines curves based on the Lamda-Mu-Sigma method using three parameters: L (skewness of distribution), M (median) and S (coefficient of variation). The best-fit smoothed percentile models were selected based on sets of hyperparameters (degrees of freedom) with the smallest Bayesian Information Criterion (BIC) (38).

Results

Out of 943 children in our database, 766 were eligible for this analysis (Figure 1).

Figure 1 Flow chart of included participants. RTI, respiratory tract infection; FEV₁, forced expiratory volume in one second.

Characteristics of participants

These 766 children aged 6–11 years (53.1% girls) were from 14 primary schools (9 public) (Table 1). Girls and boys were similar in height, weight, and BMI, but girls were significantly smaller in most of the anthropomorphic variables, compared to boys: US/H [38.6% vs. 39.3%, P<0.001; Cohen’s d =−0.33], UV [15.8 vs. 16.6 L, P=0.014; Cohen’s d =−0.18], CC/H [0.48 vs. 0.49, P=0.042; Cohen’s d (95%) =−0.15], UL [49.8 vs. 50.8 cm, P<0.001; Cohen’s d =0.07] (Table 1).

Determinants of LF

Demographic factors and LF

The mean PEF was 231.1 [95% confidence interval (CI): 225.4–236.7] for girls, 241.0 (95% CI: 235.6–246.3) for boys and 235.8 (95% CI: 231.8–239.7) for the total sample, respectively, being significantly higher in boys [girls versus boys: P=0.013; Cohen’s d (95% CI): −0.19 (−0.34 to −0.04); Table 1]. The mean (95% CI) FEV₁ was 1.36 (1.33–1.39) for girls, 1.44 (1.41–1.48) for boys and 1.40 (1.38–1.42) for the total sample, respectively, also significantly higher in boys [P<0.001; Cohen’s d (95% CI): −0.24 (−0.13 to −0.03)]. Although girls had significantly greater zPEF (PEF corrected for age and height) than boys, boys had greater zFEV₁ (FEV₁ corrected for age and height) than girls.

Table 2 shows age- and sex-based descriptive data including reference percentile values of both PEF and FEV₁; both parameters increased linearly with age in boys and girls (P linear trend <0.001). Figures 2,3 show age- and sex-based PEF and FEV₁ smoothed reference curves, respectively.

Figure 2 Age- and sex-based PEF smoothed centile curves for girls and boys. (A) PEF versus age in girls; (B) PEF versus age in boys; (C) PEF versus height in girls; (D) PEF versus height in boys. PEF, peak expiratory flow.

Figure 3 Age- and sex-based FEV₁ smoothed centile curves for girls and boys. (A) FEV₁ versus age in girls; (B) FEV₁ versus age in boys; (C) FEV₁ versus height in girls; (D) FEV₁ versus height in boys. FEV₁, forced expiratory volume in one second.

Linear correlation of LF with anthropometrics and anthropomorphics

In girls, boys and the total sample, each of PEF and FEV₁ was strongly positively correlated with height (r=0.73–0.86), LLL (r=0.67–0.81), weight (r=0.62–0.76), age (r=0.60–0.66), UV (r=0.58–0.69), UL (r=0.54–0.61) and CC (r=0.52–0.66); generally, r was larger with FEV₁ than PEF (Table 3). While the correlation of both PEF and FEV₁ with BMI was moderate, their correlation with each of height-for-age Z-score (HAZ), weight-for-age Z-score (WAZ), BMI-for-age Z-score (BAZ), US/H and CC/H was weak or negligible. Figures 4,5 show the scatter-plots of these correlations in the total sample. Only age and height showed consistently linear trend with PEF and FEV₁ across the ranges of their values (Figure 4A,4B,4G,4H); other variables showed varying degrees of non-linear relationship with PEF and FEV₁ (Figure 4C-4F,4I,4J; Figure 5A-5J).

Figure 4 Scatterplots of relationship between PEF and FEV₁ with anthropometric variables, with LOESS curves (blue line) and its 95% confidence interval (shaded area). (A) PEF versus height; (B) FEV₁ versus height; (C) PEF versus LLL; (D) FEV₁ versus LLL; (E) PEF versus weight; (F) FEV₁ versus weight; (G) PEF versus age; (H) FEV₁ versus age; (I) PEF versus CC; (J) FEV₁ versus CC. Height and age demonstrated the most consistently linear trend with PEF (A,G) and with FEV₁ (B,H). PEF, peak expiratory flow; FEV₁, forced expiratory volume in one second; LLL, lower limb length; CC, chest circumference; LOESS, locally estimated scatterplot smoothing.

Figure 5 Scatterplots of relationship between PEF and FEV₁ with anthropomorphic variables, with LOESS curves (blue line) and its 95% confidence interval (shaded area). (A) PEF versus BMI; (B) FEV₁ versus BMI; (C) PEF versus upper-segment length; (D) FEV₁ versus upper-segment length; (E) PEF versus upper-segment volume; (F) FEV₁ versus upper-segment volume; (G) PEF versus upper-segment/height ratio; (H) FEV₁ versus upper-segment/height ratio; (I) PEF versus chest circumference/height ratio; (J) FEV₁ versus chest circumference/height ratio. None of the variables demonstrated a consistently linear trend with PEF and FEV₁ across its range. PEF, peak expiratory flow; FEV₁, forced expiratory volume in one second; BMI, body mass index; upper-segment/height ratio, ratio of the upper segment to the standing height; cc_ht_ratio: ratio of the chest circumference to the standing height; LOESS, locally estimated scatterplot smoothing.

LF and socioeconomic, nutritional and environmental factors

Stunting did not affect LF [stunted vs. non-stunted: zPEF: 0.05 (1.17) vs. −0.003 (1.28), P=0.76); zFEV₁: −0.19 (1.05) vs. −0.07 (1.11), P=0.37] (Table 4). Although overweight/obese and their normal-weight peers had similar zPEF [0.059 (1.444) vs. −0.022 (1.191), P=0.48], overweight/obese children had higher zFEV₁ than their normal-weight counterparts [0.160 (1.127) vs. −0.171 (1.084), P<0.001, Cohen’s d =0.302]. Similarly, private school pupils had significantly higher FEV₁ than public school peers (P<0.001). Diurnal variation did not influence either of PEF or FEV₁, but PEF was significantly higher during the rainy season than during the dry season (Table 4).

Validation of measured PEF against local and foreign equations

The acceptable LoAs for girls and boys were ±46 L/min and ±48 L/min, respectively (20% of mean PEF for sex). Table 5 and Figure 6 shows that Nigerian and foreign PEF equations demonstrated significant mean biases from our measured PEF, except the equation by Faleti which showed the lower 95% boundary of the mean bias coincided approximately with the zero line. None of the equations showed LoAs within acceptable limits, including Faleti equation. We thus developed a new equation for the PEF of Nigerian children aged 6 to 11 years (Table 6).

Figure 6 Bland-Altman plots comparing measured PEF with predicted PEF derived from Nigerian and foreign PEF equations. X-axis is the mean of measured and predicted PEF. Y-axis is the difference between measured and predicted PEF. (A) Faleti’s equation in Lagos, south-west Nigeria; (B) Adeniyi’s equation in Abuja, north-central Nigeria; (C) Agaba et al.’s equation in Jos, north-central Nigeria; (D) Ahmed et al.’s equation in India; (E) Lu et al.’s equation in China; (F) Mittal et al.’s equation in Punjab, India. Only the Faleti equations (A: boys and girls) had the mean bias coinciding with the zero line, implying a negligible mean bias; other equations had the mean bias and its 95% CI (shaded area around the middle-dashed line) significantly outside the zero line. None of the equations, including Faleti equations, had the LoAs lines both within our acceptable LoA (defined as ±46 L/min and ±48 L/min for girls and boys, respectively), implying poor agreement. PEF, peak expiratory flow; CI, confidence interval; LoA, limit of agreement.

Validation of FEV₁ against GLI equations

Table 7 shows that, out of the six GLI equations, only the African-American (for boys and girls) and the south-east Asian equations (for girls) fitted our FEV₁; other GLI equations significantly over-estimated our sample’s FEV₁. With respect to classification of participants with zFEV₁ < LLN, GLI-African-American equations had slightly better agreement with GLI-Others [κ (95% CI): 0.50 (0.42–0.58)] than with GLI-Global [κ (95% CI): 0.48 (0.40–0.56)].

Discussion

Key findings

The ePFM-measured mean PEF and FEV₁ of Nigerian school-aged children were 235.8 L/min and 1.40 L, respectively, with height being the most influential determinant. Previously published Nigerian and foreign PEF equations agreed poorly with our sample’s PEF due to their wide LoA and significant bias, although PEF equations of Nigerian children in Lagos derived by Faleti (34) showed small mean bias. For FEV₁, only the GLI-African-American equations fitted both boys and girls; other equations, including the new ‘race-neutral’ GLI-Global (23,24) recently recommended by the ATS for global benchmarking of LF, over-estimated the FEV₁. Our study is one of the first few to explore the applicability of the new GLI-Global equations, especially in African paediatric populations (40,41).

Strengths and limitations

We had a large sample size, resulting in our narrow confidence intervals; this strengthens the precision of our estimates, internal validity and generatability, despite some missing PEF data. Because ePFM does not measure forced vital capacity (FVC), we could not provide data on FEV₁/FVC ratio which is a more precise indicator of airway obstruction than FEV₁ alone (13). However, we aimed specifically to provide ePFM-measured reference values for PEF and FEV₁ for pulmonary evaluation and care in low-resource settings, in the absence of standard spirometers. Also, the device does not have graphical displays for quality assurance (flow-volume or flow-time curves); however, its inbuilt quality checks make ePFM superior to mPFM (13). We studied a narrow age-range (6–11 years) of Nigerian children, implying that our findings and equations may not apply to other paediatric sub-groups. There may be the risk of stitching our equations with others—a practice known to result in misinterpretations (32). Other determinants of LF like maternal smoking, prematurity, low birth-weight, second-hand smoking and exposure to air-pollution were not evaluated or accounted for in our study.

ePFM versus mPFM

The previously highlighted characteristics of ePFM make them superior to mPFM for home self-monitoring of LF and for clinical diagnosis and research. For example, in asthma diagnosis, FEV₁-based bronchodilator reversibility tests conducted with ePFM are preferred to PEF-based tests done with mPFM (1,3). Also, the in-built memory in ePFM allow clinicians, in the hospital, to retrieve and review the patient’s home-measured PEF and FEV₁ from the device, or even remotely, precluding the need for paper-based recording of home measurements (18,42,43). Additionally, patients are unable to falsify LF readings made with ePFM, as may occur with paper-based recordings when using mPFM (17).

Comparison with similar researches and clinical implications

Mean PEF and FEV₁ values of Nigerian children

Online searches did not yield similar study of ePFM-measured LF for comparison; hence, we compared our measurements with those measured with spirometers or mPFMs, cautiously. Of note was that our sample’s mean PEF (235.8 L/min) and FEV₁ values (1.40 L) were strikingly similar to PEF of 230.3 L/min and FEV₁ of 1.48 L reported by Faleti et al. (34,44) among 5–12-year-old children in the same locality of Ikeja, Lagos, Nigeria, using an hand-held digital spirometer (One Flow^TM) [2011–2012]. This perhaps suggests that ePFMs have accuracy similar to spirometers, as previously reported in studies that compared both device types (15,16). When our findings are compared with studies that used mini-Wright mPFM among Nigerian children, our mean PEF exceeded 212 L/min observed in Jos plateau (Northern Nigeria) by Agaba et al. (9) and 216 L/min reported in Abuja (North-central Nigeria) by Adeniyi (8). It was however lower than 257 L/min reported in Kano (Northern Nigeria) by Mohammed et al. (45). These differences may be due to differences in LF device-types, inclusion/exclusion criteria, methodology, study population, ethno-geographic locations or socio-economic status.

Agreement of PEF of Nigerian children with published PEF equations

In asthma management, it is most preferred to compare a person’s LF values with his or her personal best values (3), but this is often not possible because it requires consistent measurements at home for 2 weeks while symptom-free. Thus, normative LF references are still needed for benchmarking measured LF (13). Determining which equation is representative of a given population entails appropriate statistical comparisons with measured values from healthy subjects. Rather than use bivariate statistical tests like t-test to compare measured and predicted PEF values as was done by some authors (10), we used the Bland-Altman analysis to assess true agreement between them; t-test or even linear correlation tests do not provide valid assessments of agreement between variables (35). The significant mean bias (except for Faleti’s equations) and wide LoAs observed with all the PEF equations suggests caution in the use of these equations for our population. Although comparison of our measured PEF with PEF predicted with Faleti’s equation yielded small bias and non-significant paired t-test P value (Table 5), the limit-of-agreement was wide, implying unsatisfactory agreement. This may be due to the use of a relatively small sample size of 100 children Faleti’s study which may have increased the statistical variability of their estimates.

Fit of FEV₁ of Nigerian children to GLI-2012 equations

Recently, ATS (24) recommended the evaluation and use of newly developed race-neutral GLI-Global, over race-specific GLI equations. However, researchers and clinicians, especially in populations not included in the original GLI data, still need to decide which GLI equations are most applicable to their populations (46). Thus, we tested our FEV₁ against GLI FEV₁ equations (32). Masekela et al. (19), in a systematic review of studies that evaluated GLI equations in seven African countries (namely, Algeria, Angola, Benin, Congo, Tunisia, Madagascar, South-Africa) up till November 2018, reported that GLI-African-American equations were valid for sub-Saharan populations, except West Africans. However, the latter was represented only by Beninese children, which may not apply to Nigerian children. Subsequently, Arigliani et al. (21) in a study of Nigerian children and adolescents in northern Nigeria, reported that the GLI-African-American equation fitted them. Similarly, our sample’s FEV₁ fitted the GLI-African-American equation [−0.08 (1.11)], even better than that reported by Arigliani et al. (21) [−0.35 (0.99)]. Compared to adolescence and adulthood, GLI equations appear to fit pre-adolescent age-groups (like our sample) better than adolescent and adult age-groups (47,48), perhaps because the effect of underlying and ongoing determinants of LF become more apparent as the individual grows from childhood to adolescence/adulthood. Thus, whereas GLI-African-American equations fit Nigerian children and adolescents, Fawibe et al. (49) reported that it did not fit Nigerian adult (however the latter study did not use GLI’s validation criteria to determine true fit). Other authors from other African countries have reported that the GLI-African-American FEV₁ equations to fit Zimbabwean (50), Cameroonian (51), Congolese, Madagascan, and Angolan children (52) but not South-African black children (53).

Race-neutral vs. race-based GLI FEV₁ equations

Race-based GLI equations (including GLI-African-American) may reflect and perpetuate structural racism and socioeconomic and health inequities, rather than reflecting true racial differences (23,24). Yet, certain observations still suggests the influence of race, over socioeconomic factors, on LF: despite the fact that African-American persons, and not indigenous Africans, were included in the GLI dataset and that both groups grew in varied environmental and socioeconomic conditions, the GLI-African-American equations fits the FEV₁ of indigenous Nigerian children (21). Similarly, indigenous Nigerian children demonstrated FEV₁ similar to Nigerian children born and living in the UK (54). Contrastingly, the zFEV₁ of indigenous Indian children was significantly lower than that of UK-resident counterparts; however, the zFEV₁ was similar between urban-dwelling indigenous Indian children and their UK-resident peers, while it was lower in rural-dwelling indigenous ones than UK-resident Indian children (55). This suggests that socioeconomic conditions also affect LF and development, additional to racial influence.

The GLI-Others and GLI-Global equations are ‘race-composite’ having been derived as combination of GLI race-based equations; the new GLI-Global equation is a ‘weighted’ derivative of the race-based equations, so as to balance out the effect of differences in the proportion of data contributed into the global dataset from different regions (23). We explored potential impact of these two equations on clinical diagnosis of low FEV₁ (zFEV₁ < LLN). GLI-Others and GLI-Global categorised 19% and 22%, respectively, of our sample as low zFEV₁ while GLI-African-American identified 10% as low zFEV₁. Whether the race-composite equations (GLI-Others and GLI-Global) over-diagnose low LF or that GLI-African-American equations under-diagnose it among our population requires further comparison of their association with clinical outcomes among diseased groups (23,24,40,41). However, our test of concordance suggests that the agreement between GLI-Global and GLI-African-American equations was slightly less than between GLI-African-American and GLI-Others.

Anthro-demographic and anthropomorphic correlation with LF

Well-known determinants of childhood LF include standing height, sitting height, age, weight, CC, LLL and chest volume. In agreement with previous authors (10-12,22,34,39,56-59), standing height showed strongest correlation with PEF and FEV₁. Additionally, two surrogate measures of thoracic volume—UV and UL—also showed correlation with LF, albeit of less magnitude than height. That height, compared to other variables, had the largest correlation with LF (in both bivariate and multivariable analyses) suggests it is a better composite measure of thoracic volume, inspiratory muscle strength, lung distensibility and other less-measurable LF determinants. However, there is still need for variables that may predict LF better than height; few studies suggest that ulna length and arm-span may (46,56,60).

Socio-demographic and environmental determinants of LF

Poverty is associated with stunting which in turn predisposes to lower LF due to smaller lung volume (54). However, we found that stunted and non-stunted children had similar LF in our urban sample. Also, in contrast to Madanhire et al.’s findings among Zimbabwean children (50), LF was affected by school type (a measure of SES) in our study, possibly because children in private-schools were heavier than those in public-schools (data not shown) and we observed that heavier children had FEV₁ that was higher than normal-weight children [similarly observed among children in northern-Nigeria by Mohammed et al. (45)]. However, the higher mean FEV₁ among overweight/obese children may not imply that they have better LF since we did not measure FVC which is required to determine FEV₁/FVC ratio. Due to asymmetric lung growth (dysanapsis), overweight/obese children often have FEV₁ and FVC values that are higher than normal-weight peers, but the FVC is often disproportionately higher than the FEV₁ resulting in reduced FEV₁/FVC (61,62). Like many authors (8,9,11,39,50,63,64), but in contrast to others (45,58), we observed significant sex differences in PEF and FEV₁. The lower FEV₁ noted among girls may be due to their smaller thoracic cavity. African girls, compared to boys, are more exposed to household-air pollutants arising from cooking and cleaning activities, and consequent impaired lung growth (65).

Clinical application of prediction PEF and FEV₁ equations

Simple linear regression equations like ours can estimate the lower-limit-of-normal, percentage of predicted and Z-score values of PEF and FEV₁ measurements (see Table 6 and its footnote) (33). Whereas prediction models derived with advanced statistical tools like GAMLSS account better for LF variability along a wider age range, their differences from linear regression equations may be clinically inconsequential (66), especially when dealing with a narrow age-range such as the school-age group when there is predominantly a linear relationship between height and LF (32). Height-based equations may be more clinically useful than age- or weight-based reference standards in low-resource settings because height is less affected by acute illnesses (like weight), and it is not affected by recall bias (like age).

Conclusions

The GLI-African-American FEV₁ equations are valid for bench-marking FEV₁ of school-aged Nigerian children measured with ePFM especially when there are no spirometers. However, our sample’s PEF did not agree acceptably with published local and foreign PEF equations. Thus, we developed new prediction PEF and FEV₁ (as well as reference curves) for possible clinical use after further external validation in separate local samples. Multi-centre nation-wide studies of the LF of the Nigerian population spanning childhood to adulthood (with the use of cutting-edge advanced statistical tools like GAMLSS) across varied socioeconomic and geographical characteristics are needed for optimal respiratory care. These and those from other sub-Saharan African populations are needed to update the GLI database (67). Additionally, the relative accuracy of GLI-Global, GLI-Others and GLI-African-American equations, and local-derived equations, in predicting relevant clinical outcomes needs further exploration.

Acknowledgments

We express heartfelt gratitude to the school pupils and their parents/guardians, as well as to the Lagos State Universal Basic Education Board (SUBEB), the Lagos State Ministry of Education, school headteachers, proprietors and school staff members for providing support for the study. We also thank Engineer Chris’ Ubuane for helping to procure the peak flow meter and mouthpieces from the UK, and also the research assistants. We also thank Dr Christian Winkler for enormously supporting the use of RefCurv^TM, especially for providing additional capabilities for the 5^th and 95^th centile curves.

Parts of these results were published as conference abstracts of the Nigerian Thoracic Society’s Annual Scientific Conference 2022 and the Pan African Thoracic Society/Respiratory Society of Kenya Congress 2023.

Funding: None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jxym.amegroups.com/article/view/10.21037/jxym-23-29/rc

Data Sharing Statement: Available at https://jxym.amegroups.com/article/view/10.21037/jxym-23-29/dss

Peer Review File: Available at https://jxym.amegroups.com/article/view/10.21037/jxym-23-29/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jxym.amegroups.com/article/view/10.21037/jxym-23-29/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Health Research Ethics Committee of the Lagos State University Teaching Hospital (LASUTH), Ikeja, Lagos, Nigeria (LREC/10/06/486), and we obtained written informed consent and assent from parents/guardians and their children, respectively, prior to enrolment.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Gaillard EA, Kuehni CE, Turner S, et al. European Respiratory Society clinical practice guidelines for the diagnosis of asthma in children aged 5-16 years. Eur Respir J 2021;58:2004173. [Crossref] [PubMed]
2. Graham BL, Steenbruggen I, Miller MR, et al. Standardization of Spirometry 2019 Update. An Official American Thoracic Society and European Respiratory Society Technical Statement. Am J Respir Crit Care Med 2019;200:e70-88. [Crossref] [PubMed]
3. Global Initiative for Asthma (GINA). Global strategy for asthma management and prevention, updated 2023. 2023.
4. Masekela R, Zurba L, Gray D. Dealing with Access to Spirometry in Africa: A Commentary on Challenges and Solutions. Int J Environ Res Public Health 2018;16:62. [Crossref] [PubMed]
5. Plum C, Stolbrink M, Zurba L, et al. Availability of diagnostic services and essential medicines for non-communicable respiratory diseases in African countries. Int J Tuberc Lung Dis 2021;25:120-5. [Crossref] [PubMed]
6. Mortimer K, Masekela R, Ozoh OB, et al. The reality of managing asthma in sub-Saharan Africa – Priorities and strategies for improving care. J Pan Afr Thorac Soc 2022;3:105-20. [Crossref]
7. Ojo O, Ajibare A, Odeyemi A, Fapohunda T, Adeyeye O. Clinical utility of peak flow meter in asthma diagnosis and monitoring in low- and middle-income countries: A narrative review. Int J Med Heal Dev 2023;28:87. [Crossref]
8. Adeniyi FB. Peak expiratory flow rate in healthy urban Nigerian school children in Abuja, Nigeria. Stellenbosch University, 2018.
9. Agaba PA, Thacher TD, Angyo IA, et al. Peak expiratory flow rates in healthy Nigerian children. J Trop Pediatr 2003;49:157-9. [Crossref] [PubMed]
10. Mojiminiyi FBO, Igbokwe UV, Ajagbonna OP, et al. Peak expiratory flow rate in normal Hausa-Fulani children and adolescents of Northern Nigeria. Ann Afr Med 2006;5:10-5.
11. Lu Y, Zheng J, Liu C, et al. Peak expiratory flow among healthy children aged 5-14 years in China. J Thorac Dis 2018;10:1377-85. [Crossref] [PubMed]
12. Ahmed M, Gavhane J, Goel S. Age Specific Peak Expiratory Flow Nomogram in Children. Indian J Pediatr 2017;84:271-5. [Crossref] [PubMed]
13. Stanojevic S, Kaminsky DA, Miller MR, et al. ERS/ATS technical standard on interpretive strategies for routine lung function tests. Eur Respir J 2022;60:2101499. [Crossref] [PubMed]
14. Miller MR, Hankinson J, Brusasco V, et al. ATS/ERS task force. Standardisation of spirometry. Eur Respir J 2005;26:319-38. [Crossref] [PubMed]
15. do Nascimento Maia P, Bazhuni Pombo Sant'Anna MF, Parente AAAI, et al. Correlation of digital flow peak with spirometry in children with and without asthma. J Asthma 2023;60:270-6. [Crossref] [PubMed]
16. Aguilar-Fernández AJ, Villa-Asensi JR, Castro-Codesal M, et al. Concordance between the Piko - 1 portable device and pneumotachography in measuring PEF and FEV(1) in asthmatic children. Allergol Immunopathol (Madr) 2009;37:244-8. [Crossref] [PubMed]
17. Kamps AW, Roorda RJ, Brand PL. Peak flow diaries in childhood asthma are unreliable. Thorax 2001;56:180-2. [Crossref] [PubMed]
18. Sakkatos P, Williams A. Testing the accuracy of a novel digital peak flow meter aligned with a smartphone app compared to a lab spirometer: A pilot work. Digit Health 2021;7:20552076211005959. [Crossref] [PubMed]
19. Masekela R, Hall GL, Stanojevic S, et al. An urgent need for African spirometry reference equations: the Paediatric and Adult African Spirometry study. Int J Tuberc Lung Dis 2019;23:952-8. [Crossref] [PubMed]
20. Milanzi EB, Koppelman GH, Oldenwening M, et al. Considerations in the use of different spirometers in epidemiological studies. Environ Health 2019;18:39. [Crossref] [PubMed]
21. Arigliani M, Castriotta L, Zubair R, et al. Differences in lung function between children with sickle cell anaemia from West Africa and Europe. Thorax 2019;74:1154-60. [Crossref] [PubMed]
22. Lum S, Bountziouka V, Sonnappa S, et al. Lung function in children in relation to ethnicity, physique and socioeconomic factors. Eur Respir J 2015;46:1662-71. [Crossref] [PubMed]
23. Bowerman C, Bhakta NR, Brazzale D, et al. A Race-neutral Approach to the Interpretation of Lung Function Measurements. Am J Respir Crit Care Med 2023;207:768-74. [Crossref] [PubMed]
24. Bhakta NR, Bime C, Kaminsky DA, et al. Race and Ethnicity in Pulmonary Function Test Interpretation: An Official American Thoracic Society Statement. Am J Respir Crit Care Med 2023;207:978-95. [Crossref] [PubMed]
25. Bhakta NR, Kaminsky DA, Bime C, et al. Addressing Race in Pulmonary Function Testing by Aligning Intent and Evidence With Practice and Perception. Chest 2022;161:288-97. [Crossref] [PubMed]
26. Holland R, Bowerman C, Stanojevic S. The Contribution of Anthropometry and Socioeconomic Status to Racial Differences in Measures of Lung Function: A Systematic Review. Chest 2022;162:635-46. [Crossref] [PubMed]
27. Moons KG, Altman DG, Reitsma JB, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 2015;162:W1-73. [Crossref] [PubMed]
28. Ubuane PO, Animasahun BA, Njokanma OF. Dataset: PEF and FEV1 of Nigerian children aged 6-11 years measured with an electronic peak flow meter. Mendeley Data 2023; [Crossref]
29. Ubuane PO, Ajiboye OA, Adekunle MO, et al. Reference values and equations for the 6-minute walk distance of Nigerian children aged 6-11 years: A cross-sectional study. Pediatr Pulmonol 2022;57:2103-15. [Crossref] [PubMed]
30. Whittaker AL, Sutton AJ, Beardsmore CS. Are ethnic differences in lung function explained by chest size? Arch Dis Child Fetal Neonatal Ed 2005;90:F423-8. [Crossref] [PubMed]
31. Goss-Sampson MA. Statistical Analysis in JASP: A Guide for Students v 0.16 (2022). 5th ed. 2022.
32. Quanjer PH, Stanojevic S, Cole TJ, et al. Multi-ethnic reference values for spirometry for the 3-95-yr age range: the global lung function 2012 equations. Eur Respir J 2012;40:1324-43. [Crossref] [PubMed]
33. Neder JA, Berton DC, O'Donnell DE. Calculating the statistical limits of normal and Z-scores for pulmonary function tests. J Bras Pneumol 2022;48:e20220182. [PubMed]
34. Faleti AO. Lung function in children with and without sickle cell anaemia at the Lagos State University Teaching Hospital [dissertation]. Faculty of Paediatric, National Postgraduate Medical College of Nigeria. National Postgraduate Medical College of Nigeria, 2012.
35. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10. [Crossref] [PubMed]
36. Giavarina D. Understanding Bland Altman analysis. Biochem Med (Zagreb) 2015;25:141-51. [Crossref] [PubMed]
37. Barton B, Peat J. Medical Statistics: A Guide to Data Analysis and Critical Appraisal. 2nd editio. West Sussex: John Wiley & Sons Ltd, 2014.
38. Winkler C, Linden K, Mayr A, et al. RefCurv: A software for the construction of pediatric reference curves. Softw Impacts 2020;6:100040. [Crossref]
39. Mittal S, Gupta S, Kumar A, et al. Regression equations for peak expiratory flow in healthy children aged 7 to 14 years from Punjab, India. Lung India 2013;30:183-6. [Crossref] [PubMed]
40. Moffett AT, Bowerman C, Stanojevic S, et al. Global, Race-Neutral Reference Equations and Pulmonary Function Test Interpretation. JAMA Netw Open 2023;6:e2316174. [Crossref] [PubMed]
41. Moaaz M, Badreldin O, El Chazli Y, et al. Race-neutral equations for assessment of lung function in children with thalassemia. Pediatr Pulmonol 2023;58:2543-50. [Crossref] [PubMed]
42. Antalffy T, De Simoni A, Griffiths CJ. Promising peak flow diary compliance with an electronic peak flow meter and linked smartphone app. NPJ Prim Care Respir Med 2020;30:19. [Crossref] [PubMed]
43. Côté J, Cartier A, Malo JL, et al. Compliance with peak expiratory flow monitoring in home management of asthma. Chest 1998;113:968-72. [Crossref] [PubMed]
44. Faleti OA, Akode SO, Disu EA, et al. Pulmonary functions in children with sickle cell anaemia in steady state in Lagos, Nigeria. Ann Heal Res 2017;3:18-25.
45. Mohammed J, Maiwada S, Sumaila F. Relationship between anthropometric variables and lung function parameters among primary school children. Ann Niger Med 2015;9:20. [Crossref]
46. Mannino DM, Townsend MC. Spirometry in 2022: Is a Single Set of Prediction Equations for All the Best Path Forward? Am J Respir Crit Care Med 2023;207:659-61. [Crossref] [PubMed]
47. Hüls A, Krämer U, Gappa M, et al. Age Dependency of GLI Reference Values Compared with Paediatric Lung Function Data in Two German Studies (GINIplus and LUNOKID). PLoS One 2016;11:e0159678. [Crossref] [PubMed]
48. Mozun R, Ardura-Garcia C, Pedersen ESL, et al. Age and body mass index affect fit of spirometry Global Lung Function Initiative references in schoolchildren. ERJ Open Res 2022;8:00618-2021. [Crossref] [PubMed]
49. Fawibe AE, Odeigah LO, Saka MJ. Reference equations for spirometric indices from a sample of the general adult population in Nigeria. BMC Pulm Med 2017;17:48. [Crossref] [PubMed]
50. Madanhire T, Ferrand RA, Attia EF, et al. Validation of the global lung initiative 2012 multi-ethnic spirometric reference equations in healthy urban Zimbabwean 7-13 year-old school children: a cross-sectional observational study. BMC Pulm Med 2020;20:56. [Crossref] [PubMed]
51. Pefura-Yone EW, Balkissou AD, Poka-Mayap V, et al. Spirometric reference equations for Cameroonians aged 4 to 89 years derived using lambda, mu, sigma (LMS) method. BMC Pulm Med 2021;21:344. [Crossref] [PubMed]
52. Arigliani M, Canciani MC, Mottini G, et al. Evaluation of the Global Lung Initiative 2012 Reference Values for Spirometry in African Children. Am J Respir Crit Care Med 2017;195:229-36. [Crossref] [PubMed]
53. Smith SJ, Gray DM, MacGinty RP, et al. Choosing the Better Global Lung Initiative 2012 Equation in South African Population Groups. Am J Respir Crit Care Med 2020;202:1724-7. [Crossref] [PubMed]
54. Arigliani M, Lum S, Zuiani C, et al. Comparison of Lung Function in Healthy Nigerian Children Living in Nigeria and in the United Kingdom. Am J Respir Crit Care Med 2022;206:221-4. [Crossref] [PubMed]
55. Sonnappa S, Lum S, Kirkby J, et al. Disparities in pulmonary function in healthy children across the Indian urban-rural continuum. Am J Respir Crit Care Med 2015;191:79-86. [Crossref] [PubMed]
56. Akodu OS, Faleti OA, Adekanmbi AF, Adetutu T. Correlation of peak expiratory flow rate with anthropometric parameters in apparently healthy children with sickle cell anaemia in Lagos. Sri Lanka J Child Heal 2020;49:329–34.
57. Sharma R, Kaur A, Gupta M, et al. A cross sectional study to assess peak expiratory flow rate in healthy school children for establishment of normative data. J Nepal Paediatr Soc 2020;40:1-6. [Crossref]
58. Oloyede IP, Ekrikpo UE, Ekanem EE. Normative values and anthropometric determinants of lung function indices in rural Nigerian children: A pilot survey. Niger J Paediatr 2013;40:406-11. [Crossref]
59. Akhiwu HO. Spirometric Values in Healthy Nigerian School Children Aged 6-11 Years. Journal of Advances in Medicine and Medical Research 2017;22:1-8. [Crossref]
60. Gauld LM, Kappers J, Carlin JB, et al. Height prediction from ulna length. Dev Med Child Neurol 2004;46:475-80. [Crossref] [PubMed]
61. Park JS, Suh DI, Choi YJ, et al. Pulmonary function of healthy Korean children from three independent birth cohorts: Validation of the Global Lung Function Initiative 2012 equation. Pediatr Pulmonol 2021;56:3310-20. [Crossref] [PubMed]
62. Ferreira MS, Marson FAL, Wolf VLW, et al. Association between Pulmonary Function and Body Composition in Children and Adolescents with and without Obesity. J Clin Med 2022;11:7410. [Crossref] [PubMed]
63. Bouti K, Benamor J, Bourkadi JE. Predictive Regression Equations of Flowmetric and Spirometric Peak Expiratory Flow in Healthy Moroccan Children. J Clin Diagn Res 2017;11:SC01-4. [Crossref] [PubMed]
64. Martins C, Severo M, Silva D, et al. Development and validation of predictive equations for spirometry in Portuguese children. J Allergy Clin Immunol Glob 2023;2:100084. [Crossref] [PubMed]
65. Okello G, Devereux G, Semple S. Women and girls in resource poor countries experience much greater exposure to household air pollutants than men: Results from Uganda and Ethiopia. Environ Int 2018;119:429-37. [Crossref] [PubMed]
66. Martínez-Briseño D, Gochicoa-Rangel L, Torre-Bouscoulet L, et al. Comparing Spirometric Reference Values From Childhood to Old Age Estimated by LMS and Linear Regression Models. Arch Bronconeumol (Engl Ed) 2021;57:172-8. [Crossref] [PubMed]
67. Chang SM, Tsai HJ, Tzeng JY, et al. Reference equations for spirometry in healthy Asian children aged 5 to 18 years in Taiwan. World Allergy Organ J 2019;12:100074. [Crossref] [PubMed]