Impact of spirometry race-correction on pre-adolescent Black and White children

Abstract

Background:

Race-correction for Black patients is standard practice in spirometry testing. History suggests that these corrections are at least partially a result of racist assumptions regarding lung anatomy among Black individuals, which can potentially lead to less frequent diagnoses of pulmonary diseases in this population.

Objective:

The purpose of this study was to evaluate the impact of race-correction in spirometry testing among Black and White pre-adolescents, and examine the frequency of current asthma symptoms in Black children who were differentially classified depending on whether race-corrected or uncorrected reference equations were deployed.

Methods:

Data from Black and White children who completed a clinical exam at age 10 from a Detroit-based unselected birth cohort were analyzed. Global lung initiative (GLI) 2012 reference equations were applied to spirometry data using both race-corrected and race-uncorrected (i.e., population-average) equations. Abnormal results were defined as values less than the fifth percentile. Asthma symptoms were assessed concurrently using the International Study of Asthma and Allergies in Childhood (ISAAC) questionnaire, while asthma control was assessed using the asthma control test (ACT).

Results:

The impact of race-correction on FEV1/FVC ratio was minimal, but abnormal classification of FEV1 results more than doubled among Black children when race-uncorrected equations were used (7% vs. 18.1%) and were almost 8 times greater based on FVC classification (1.5% vs. 11.4%). More than half of Black children differentially classified on FEV1 (whose FEV1 was classified as normal with race-corrected equations but abnormal with race-uncorrected equations) experienced asthma symptoms in the past 12 months (52.6%), which was significantly higher than Black children consistently classified as normal (35.5%, p=0.049), but similar to that of Black children consistently classified as abnormal using both race-corrected and race-uncorrected equations (62.5%, p=0.60). ACT scores were not different based on classification.

Conclusions:

Race-correction had an extensive impact on spirometry classification in Black children, and differentially classified children had a higher rate of asthma symptoms than children consistently classified as normal. Spirometry reference equations should be reevaluated to be aligned with current scientific perspectives on the use of race in medicine.

Introduction

Spirometry is a routinely used pulmonary function test (PFT) in pediatric research and clinical practice for the diagnosis, management, and classification of pulmonary diseases. While obstruction reflected by the forced expiratory volume in 1 second (FEV1) to forced vital capacity (FVC) ratio is of primary interest for the diagnosis of asthma,^{1, 2} FEV1 and FVC alone also guide classification and are used in other applications such as congenital heart disease and idiopathic scoliosis, where restrictive patterns have been found in pediatric populations.^{3, 4} In current practice, reference equations are used to establish predicted values for individuals. Observed lung function results are used to calculate the percent of the predicted value for someone of the same age, sex, height, and race (percent predicted = 100 x observed value/expected value). Take for example two 10-year old males—one Black and one White—that are both 140 cm tall and have an FEV1 of 1.5L. Based on race-corrected spirometry equations (discussed in detail in methods), this would be considered “normal” for the Black child (percent predicted=87.0%, which is greater than the fifth percentile), but “abnormal” for the White child (percent predicted=74.2%, which is less than the fifth percentile), despite equal raw readings.

Treatment guidelines and decisions are based, at least in part, on these results. For example, the National Asthma Education and Prevention Program Expert Panel Report 3 (NAEPP EPR-3) criteria—which has been widely used in clinical practice and research—explicitly uses percent predicted values to guide classification of severity and control.⁵ Race-correction could also have implications in research settings such as impacting eligibility in clinical trials as well as spirometric outcomes. Clinical and research staff as well as patients may not realize that this approach assumes that lung function is inherently lower in Black children and higher in White children, or even that race is used in equations at all. Differences in race-corrected readings could potentially lead to under-diagnosis or treatment in Black children and over-diagnosis in White children. This issue is particularly relevant to pediatric asthma, where the incidence rate in Black children is nearly twice that of White children.⁶

The historical reasons for clinical race-correction is often deeply rooted in racism, and the spirometer is no exception.⁷ Early versions of the spirometer were built by plantation physician and slave-holder Dr. Samuel Cartwright, whose experiments were designed to justify slavery by comparing enslaved persons to White colonists; he concluded that “the deficiency in the negro” was “20 per cent”,⁷ and that forced labor was a way to “vitalize their blood”.⁸ These early ideologies likely unwittingly influenced how modern day reference equations were developed, which were established by performing spirometry on large numbers of “healthy” patients according to age, height, sex, race, and ethnicity.⁹ However, Black patients may be more likely to have undiagnosed lung disease and be disproportionately included in reference equation populations, thereby making lung function appear worse among “healthy” Black populations.¹⁰ Studies have supported associations between both social determinants of health and ancestry and spirometry values, though precise contributions are unclear.¹¹ Regardless, if lower lung volumes observed in Black patients are even partially due to social determinants of health stemming from structural racism (poverty, occupation, lack of access to healthcare, exposures to polluted environments, etc.) rather than fully through genetics or ancestry, it would not be appropriate to apply “racial corrections” to spirometry because this could mask underlying lung pathology.

The underlying assumption behind the adjustment for race in spirometry reference equations is that Black and White lungs are on average biologically different.^{7, 12} However, the medical and scientific community has begun to reevaluate these assumptions and recognize the harmful effects that race-corrected algorithms may have on clinical decision-making.¹³ Social classifications of race have historically been assumed to be tightly associated with genetic variation, but evidence produced during the dawn of the genome era clearly demonstrated that social “race” is an inadequate proxy for genetic variation, as “races” are neither homogeneous nor distinct for most genetic variation.¹⁴ In a 2004 Nature Genetics commentary, Frances Collins wrote that “‘race’ and ‘ethnicity’ are poorly defined terms that serve as flawed surrogates for multiple environmental and genetic factors in disease causation”.¹⁵

The current accepted definition of race is a construct whose categorization depends on several characteristics, including physical appearance, genetic ancestry, culture, language, and religion.^16–18 As a result of a paradigm shift in the definition of race, there is increased awareness in the scientific community that treating race as a biological variable in clinical algorithms is inappropriate and no longer a scientifically valid solution,¹⁹ as it can partially mask differences in lived experiences and exposures that Black Americans have endured for centuries. Besides the issue of treating race as a biological variable, there are additional nuances to race that have yet to be addressed, including that race and ethnicity categories are not universally defined within the U.S., their definitions have changed over time,²⁰ and spirometry reference equations do not allow for multiracial categorization. In this analysis, we use the long-standing Wayne County Health, Environment, Allergy and Asthma Longitudinal Study (WHEALS) birth cohort²¹ to examine how race adjustment in spirometry reference equations impacts PFT results in Detroit pre-adolescents (at 10 years of age), and to describe reported asthma symptoms of Black children who were differentially classified depending on if race-corrected or race-uncorrected equations were used.

Methods

Data from the WHEALS unselected birth cohort were analyzed; cohort details have been previously published.²¹ Briefly, a total of 1,258 pregnant women ages 21 to 49 who were receiving care at Henry Ford Health obstetric clinics were recruited from 2003–2007. Women resided in a predefined geographic area including Detroit and surrounding suburbs. Parents of participants provided written informed consent and children provided assent. The study was performed in accordance with protocols approved by the Institutional Review Board at Henry Ford Health.

Child’s race was obtained by parental-report at the study’s 2-year clinic visit. Only Black/Non-Hispanic/Non-Arabic and White/Non-Hispanic/Non-Arabic children were included in this analysis. Some Other Race (“Other”) and multiracial children were excluded due to small sample sizes and unclear categorization in spirometry reference equations.²² A clinical visit was completed at 10 years of age that included spirometry, which was performed on calibrated JAEGER^® Vyntus CPX (CareFusion, Hoechberg, Germany) equipment in accordance with the American Thoracic Society criteria.²³ Children were in good health at the time of the clinic visit and parents were asked to reschedule if their child became ill beforehand. Height and weight was measured during the visit as described previously.²⁴ Questions regarding pubertal status²⁵ were self-reported. The Global Lung Initiative (GLI) 2012 reference equations were applied to spirometry data using three different equations: 1.) using standard-practice “race-corrected” equations (where Black children were assigned to the “African American” GLI category and White children were assigned to the “Caucasian” GLI category), 2.) using “race-uncorrected” equations by assigning all children to the “Other/mixed” GLI category, which represents a race-composite (i.e., population-averaged) equation by pooling all subjects into one model without any race adjustment²⁶, and 3.) by assigning all children to the “Caucasian” GLI category. Abnormal results for FEV1, FVC, and FEV1/FVC ratio were defined as percent predicted values less than the fifth percentile, which corresponds to a Z-score less than −1.64. This cutoff is referred to as the lower limit of normal (LLN). Preserved ratio impaired spirometry (PRISm) was defined as FEV1/FVC≥LLN and FEV1<LLN.²⁷ In this pattern, the FVC is almost always also <LLN. It is typically called “restrictive”, but true lung restriction is defined by a reduced total lung capacity, which cannot be measured by standard spirometry. The asthma module of the International Study of Asthma and Allergies in Childhood (ISAAC) questionnaire was used to assess frequency of asthma symptoms (wheezing and coughing in the past 12 months),²⁸ while the Asthma Control Test (ACT)²⁹ was used to assess control among asthmatics. Both questionnaires were administered by a study interviewer.

Children were first examined for differences in characteristics by race using ANOVA or Kruskal-Wallis tests for continuous covariates and chi-square or Fisher’s exact tests for categorical covariates. Linear regression was used to examine differences in percent predicted values by race, with mean differences and 95% confidence intervals reported. Because clinicians and researchers often categorize children based on being above or below the cutoff of what is considered “normal”, the kappa statistic and corresponding 95% confidence intervals were used to assess agreement in classification, comparing race-corrected to race-uncorrected equations. Kruskal-Wallis and Fisher’s exact tests were used to compare descriptive factors, asthma symptoms, and asthma control by reference equation classification group.

Results

A total of 343 Black children and 111 White children (454 total) from the WHEALS study underwent a clinical exam with PFT at 10 years of age. All children were free of illness at the time of the visit, but 1 child had bronchitis in the past 4 weeks, 19 children had a cold in the past 4 weeks, and 1 child had both a respiratory infection and a cold in the past 4 weeks. As shown in Table I, the age at spirometry testing did not significantly differ between Black and White children (10.3 vs. 10.5 years; p=0.12), nor did height (145.4 vs. 145.0 cm; p=0.73) or sex (50.4% vs. 49.6% female; p=0.87). However, Black children had significantly shorter sitting heights on average (66.0 vs. 67.9; p<0.001), were significantly more likely to be in mid to late puberty than White children despite no difference in age (62.5% vs. 39.4%; p<0.001), and were significantly more likely to be obese (23.3% vs. 11.7%; p=0.008). Additionally, Black children were more likely to report asthma symptoms in the past 12 months (39.3% vs. 19.1%; p<0.001), were more likely to complete the ACT (which was only administered to asthmatics; 23.3% vs. 10.8%; p=0.004), and typically had lower ACT scores (Median [Q1, Q3]=22 [20, 24] vs. 24 [22, 25.5]; p=0.018), indicating less control in asthmatic Black children (Table I).

Table I.

Descriptive factors, asthma symptoms, and asthma control in Black versus White children.

		Child Race
Characteristic	Category	White N=111	Black N=343	p-value
Descriptive Factors
Age (years), Mean (SD)		10.5 (0.9)	10.3 (0.9)	0.12a
Sex, N (%)	Male	56 (50.4%)	170 (49.6%)	0.87b
	Female	55 (49.6%)	173 (50.4%)
Height (cm), Mean (SD)		145.0 (8.3)	145.4 (9.2)	0.73a
Sitting Height (cm), Mean (SD)		67.9 (4.0)	66.0 (5.4)	<0.001a
Puberty Status, N (%)	Pre-Puberty	27 (24.8%)	42 (12.6%)	<0.001c
	Early Puberty	39 (35.8%)	83 (24.9%)
	Mid Puberty	41 (37.6%)	167 (50.2%)
	Late Puberty	2 (1.8%)	41 (12.3%)
Obese, N (%)	No	98 (88.3%)	263 (76.7%)	0.008b
	Yes	13 (11.7%)	80 (23.3%)
Asthma Symptoms and Control
Any Asthma Symptoms by ISAAC, N (%)	No	89 (80.9%)	207 (60.7%)	<0.001b
	Yes	21 (19.1%)	134 (39.3%)
Completed the ACT, N (%)	No	99 (89.2%)	263 (76.7%)	0.004b
	Yes	12 (10.8%)	80 (23.3%)
ACT Score, Median [Q1, Q3]		24 [22, 25.5]	22 [20, 24]	0.018d
ACT Score Category, N (%)	Well Controlled (ACT>19)	12 (100%)	63 (78.8%)	0.11c
	Not Well Controlled (ACT≤19)	0 (0%)	17 (21.3%)

^aCalculated using ANOVA

^bCalculated using the chi-square test

^cCalculated using Fisher’s exact test

^dCalculated using the Kruskal-Wallis test

Abbreviations: ACT (Asthma Control Test); ISAAC (International Study of Asthma and Allergies in Childhood)

No difference in percent predicted FEV1 was observed between Black and White children using race-corrected equations (β (95% CI)=−0.6 (−3.5, 2.4), p=0.71; Figure I), but Black children had statistically significantly lower percent predicted FEV1 when race-uncorrected equations were used (β (95% CI)=−15.9 (−18.7, −13.1), p<0.001; Figure I). Additionally, no difference in mean percent predicted FVC values were observed by race using race-corrected equations (β (95% CI)=2.1 (−0.7, 4.9), p=0.14; Figure I). A strong association, however, was observed using race-uncorrected equations (β (95% CI)=−14.9 (−17.6, −12.2), p<0.001; Figure I). For FEV1/FVC ratio, Black children had significantly lower mean percent predicted values in both race-corrected (β (95% CI)=−3.2 (−5.0, −1.4), p<0.001; Figure I), and race-uncorrected (β (95% CI)=−2.5 (−4.3, −0.7), p=0.006; Figure I) equations. However, mean differences were similar before and after race-correction and absolute differences were relatively small. For all 3 spirometric measurements, mean differences were similar comparing race-uncorrected equations to when the White reference equation was instead applied to all children (Figure I). When differences by race in race-uncorrected equations were reevaluated after adjusting for sitting height, puberty status, and obesity status, effect sizes were largely unchanged and remained significant for all three spirometric measurements (Table II). When across equation comparisons were made for FEV1 and FVC among Black children, the highest percent predicted values were observed when the race-corrected equation was used, followed by the race-uncorrected equation and the White equation (Figure I).

Figure I.

Difference in percent predicted values using race-corrected, race-uncorrected, and White reference equations, comparing White and Black children (indicated by fill color). Circles and squares over boxplots represent means. Mean differences (Black vs. White children) and 95% confidence intervals are shown with linear regression p-values.

Table II.

Difference in percent predicted values using race-uncorrected equations, before and after adjusting for descriptive factors associated with race. Mean differences (Black vs. White children) and 95% confidence intervals are shown with linear regression p-values.

Adjustment Variable	FEV1		FVC		FEV1/FVC
	β (95% CI)	p-value	β (95% CI)	p-value	β (95% CI)	p-value
Unadjusted	−15.9 (−18.7, −13.1)	<0.001	−14.9 (−17.6, −12.2)	<0.001	−2.5 (−4.3, −0.7)	0.006
Sitting Height	−15.2 (−18, −12.3)	<0.001	−14.1 (−16.8, −11.4)	<0.001	−2.7 (−4.5, −0.9)	0.004
Puberty Status	−16.8 (−19.7, −14)	<0.001	−15.8 (−18.5, −13)	<0.001	−2.6 (−4.5, −0.7)	0.007
Obesity	−16.3 (−19.1, −13.5)	<0.001	−15.4 (−18.1, −12.7)	<0.001	−2.5 (−4.3, −0.7)	0.007

A total of 24 Black children had abnormal (<LLN) FEV1 results using race-corrected equations, but 38 additional Black children were classified as abnormal using race-uncorrected equations, more than doubling the rate (Table III). If Black children were instead assigned to the same equation used in clinical practice for White children, an additional 62 children (for a total of n=124) would have an abnormal FEV1 result. Agreement in FEV1 classification was moderate between race-corrected and race-uncorrected equations among Black children (Table III; kappa (95% CI)=0.51 (0.38, 0.64)). Among the White children, two were classified as abnormal using race-corrected FEV1 while none were similarly classified using race-uncorrected equations.

Table III.

Agreement in abnormal (<LLN) spirometry results comparing race-corrected to race-uncorrected equations, by child’s race.

Child’s Race	Measurement	kappa (95% CI)a	Abnormal (<LLN): race-corrected	Abnormal (<LLN): race-uncorrected	N (%)
Black	FEV1	0.51 (0.38, 0.64)	No	No	281 (81.9%)
			No	Yes	38 (11.1%)
			Yes	No	0 (0.0%)
			Yes	Yes	24 (7.0%)
	FVC	0.21 (0.05, 0.36)	No	No	304 (88.6%)
			No	Yes	34 (9.9%)
			Yes	No	0 (0.0%)
			Yes	Yes	5 (1.5%)
	FEV1/FVC	0.87 (0.80, 0.93)	No	No	259 (75.5%)
			No	Yes	16 (4.7%)
			Yes	No	0 (0.0%)
			Yes	Yes	68 (19.8%)
White	FEV1	N/Ab	No	No	109 (98.2%)
			No	Yes	0 (0.0%)
			Yes	No	2 (1.8%)
			Yes	Yes	0 (0.0%)
	FVC	0.66 (0.04, 1.00)	No	No	109 (98.2%)
			No	Yes	0 (0.0%)
			Yes	No	1 (0.9%)
			Yes	Yes	1 (0.9%)
	FEV1/FVC	0.76 (0.57, 0.94)	No	No	94 (84.7%)
			No	Yes	6 (5.4%)
			Yes	No	0 (0.0%)
			Yes	Yes	11 (9.9%)

^aKappa statistic assesses agreement between race-corrected and race-uncorrected classification of abnormal spirometry results. Values ≤0 indicate no agreement, while 0.01–0.20 indicates none to slight agreement, 0.21–0.40 indicates fair agreement, 0.41– 0.60 indicates moderate agreement, 0.61–0.80 indicates substantial agreement, and 0.81–1.00 indicates almost perfect agreement.

^bUnable to calculate.

For FVC, 34 additional Black children were classified as abnormal using race-uncorrected equations, compared with only 5 using race-corrected equations; 44 additional Black children would be considered abnormal if the White reference was applied. Agreement was fair between FVC classification for race-corrected vs. race-uncorrected in Black children (kappa (95% CI)=0.21 (0.05, 0.36)), but substantial for White children (kappa (95% CI)=0.66 (0.04, 1.00); Table III).

For FEV1/FVC, a total of 68 Black children were considered abnormal using race-corrected equations, and 16 additional Black children were classified as abnormal when race-uncorrected equations were used. Agreement between these equations was high for both Black (kappa (95% CI)=0.87 (0.80, 0.93)), and White children (kappa (95% CI)=0.76 (0.57, 0.94); Table III). When the White reference equation for FEV1/FVC was applied, a total of 63 Black children were classified as abnormal; these children were also classified as abnormal using both the race-corrected and race-uncorrected equations, reflecting the general consistency between the three equations. Additionally, the frequency of PRISm among Black children using race-corrected equations was 1.5% (N=5), while the frequency was 7.3% (N=25) when race-uncorrected equations were used.

We next examined descriptive factors, asthma symptoms, and asthma control by classification group, comparing race-corrected to race-uncorrected equations (both equations classified as normal, both equations classified as abnormal, and differentially classified; Table IV). These comparisons were only able to be performed in Black children due to the small sample sizes by classification group in White children, as shown in Table III. However, data for White vs. Black children consistently classified as normal is shown in Table V for comparison purposes, which demonstrated that Black children consistently classified as normal were significantly more likely to be obese, have shorter sitting heights, and have asthma symptoms in the past 12 months compared to White children consistently classified as normal (Table V). Black children with differentially classified FEV1 and FVC were mostly male, whereas Black children with differentially classified FEV1/FVC ratio were mostly female and roughly a third were classified as obese (Table IV). There was an overall association between FEV1 classification group and frequency of asthma symptoms (overall p=0.007), with 35.5% of those where both equations indicated “normal” has symptoms, 52.6% of those who were “differentially classified” had symptoms, and 62.5% of those who were consistently classified as “abnormal” had symptoms. Both the “abnormal” and “differentially classified” groups had significantly higher rates of symptoms compared to the “normal” group (p=0.049, p=0.014), but there was no significant difference between those that were differentially classified vs. those where both equations indicated abnormal FEV1 (p=0.60). Results were similar for FEV1/FVC, with an overall association between classification group and asthma symptoms (p<0.001). Similar to what was observed for FEV1, differentially classified children again experienced an intermediate frequency of asthma symptoms. However, the rate of symptoms did not significantly differ comparing “both normal” to “differentially classified” on FEV1/FVC ratio (p=0.18) despite a nearly identical effect size to what was observed for FEV1 (FEV1: 35.5% vs. 52.6%; FEV1/FVC: 33.1% vs. 50%). The rate of ACT test completion also differed by FEV1/FVC classification group (p<0.001), but no differences in ACT scores were observed among those who completed the assessment (p=0.41).

Table IV.

Descriptive factors, asthma symptoms, and asthma control by spirometry reference equation classification group among Black children.

FEV1 Classification Group
									p-valuesd
Characteristic	Both Normala N=281			Differentially Classifiedb N=38	Both Abnormalc N=24			Overall	Both Normal vs. Differentially Classified			Both Normal vs. Both Abnormal	Differentially Classified vs. Both Abnormal
Descriptive Factors
Age (years), Mean (SD)	10.4 (0.9)			10.3 (1.0)	10.1 (0.7)			0.24	0.36			0.14	0.61
Female, N (%)	145 (51.6%)			14 (36.8%)	14 (58.3%)			0.17	0.12			0.67	0.12
Height (cm), Mean (SD)	145.6 (9.2)			145.0 (9.5)	143.6 (9.0)			0.69	0.73			0.42	0.64
Sitting Height (cm), Mean (SD)	66.2 (5.4)			65.5 (4.7)	64.8 (6.1)			0.62	0.37			0.64	0.94
Obese, N (%)	69 (24.6%)			6 (15.8%)	5 (20.8%)			0.52	0.31			0.81	0.74
Asthma Symptoms and Control
Any Asthma Symptoms in the Past 12 Months by ISAAC, N (%)	99 (35.5%)			20 (52.6%)	15 (62.5%)			0.007	0.049			0.014	0.60
Completed the ACT, N (%)	53 (18.9%)			14 (36.8%)	13 (54.2%)			<0.001	0.018			<0.001	0.20
ACT Score, Median [Q1,Q3]	22 [21,24]			21.5 [19,23]	22 [20,23]			0.30	0.14			0.42	0.68
Asthma is Not Well Controlled (ACT≤19), N (%)	11 (20.8%)			4 (28.6%)	2 (15.4%)			0.72	0.50			0.99	0.65
FVC Classification Group
									p-valuesd
Characteristic	Both Normala N=304			Differentially Classifiedb N=34	Both Abnormalc N=5			Overall	Both Normal vs. Differentially Classified			Both Normal vs. Both Abnormal	Differentially Classified vs. Both Abnormal
Descriptive Factors
Age (years), Mean (SD)	10.3 (0.9)			10.3 (0.9)	10.6 (1.4)			0.93	0.80			0.78	0.74
Female, N (%)	157 (51.6%)			13 (38.2%)	3 (60%)			0.29	0.15			0.99	0.63
Height (cm), Mean (SD)	145.5 (9.3)			145.1 (8.0)	140.0 (13.2)			0.59	0.91			0.31	0.36
Sitting Height (cm), Mean (SD)	66.2 (5.3)			64.9 (5.5)	62.4 (8.4)			0.39	0.39			0.26	0.56
Obese, N (%)	74 (24.3%)			5 (14.7%)	1 (20%)			0.47	0.29			0.99	0.99
Asthma Symptoms and Control
Any Asthma Symptoms in the Past 12 Months by ISAAC, N (%)	116 (38.4%)			16 (47.1%)	2 (40%)			0.57	0.36			0.99	0.99
Completed the ACT, N (%)	69 (22.7%)			10 (29.4%)	1 (20%)			0.76	0.40			0.99	0.99
ACT Score, Median [Q1,Q3]	22 [20,24]			23 [21,24]	22 [22,22]			0.90	0.66			0.90	0.52
Asthma is Not Well Controlled (ACT≤19), N (%)	15 (21.7%)			2 (20%)	0 (0%)			0.99	0.99			0.99	0.99
FEV1/FVC Classification Group
										p-valuesd
Characteristic		Both Normala N=259	Differentially Classifiedb N=16			Both Abnormalc N=68	Overall			Both Normal vs. Differentially Classified	Both Normal vs. Both Abnormal		Differentially Classified vs. Both Abnormal
Descriptive Factors
Age (years), Mean (SD)		10.4 (0.9)	10.3 (1.1)			10.2 (0.8)	0.17			0.57		0.065	0.65
Female, N (%)		134 (51.7%)	10 (62.5%)			29 (42.6%)	0.26			0.45		0.22	0.17
Height (cm), Mean (SD)		145.5 (9.2)	148.7 (10.6)			144.3 (9.0)	0.35			0.25		0.43	0.18
Sitting Height (cm), Mean (SD)		65.9 (5.4)	68.3 (5.0)			65.9 (5.1)	0.32			0.15		0.82	0.14
Obese, N (%)		60 (23.2%)	5 (31.3%)			15 (22.1%)	0.74			0.54		0.99	0.52
Asthma Symptoms and Control
Any Asthma Symptoms in the Past 12 Months by ISAAC, N (%)		85 (33.1%)	8 (50%)			41 (60.3%)	<0.001			0.18		<0.001	0.58
Completed the ACT, N (%)		47 (18.1%)	4 (25%)			29 (42.6%)	<0.001			0.51		<0.001	0.26
ACT Score, Median [Q1,Q3]		22 [20,24]	22 [19,25]			22 [20,23]	0.41			0.89		0.18	0.74
Asthma is Not Well Controlled (ACT≤19), N (%)		10 (21.3%)	1 (25%)			6 (20.7%)	0.99			0.99		0.99	0.99

^aBlack children with both race-corrected and race-uncorrected values≥LLN

^bBlack children with race-corrected value≥LLN, but race-uncorrected value<LLN

^cBlack children with both race-corrected and race-uncorrected values<LLN

^dCalculated using Kruskal-Wallis for continuous variables and Fisher’s exact test for categorical variables

Abbreviations: ACT (Asthma Control Test); ISAAC (International Study of Asthma and Allergies in Childhood); LLN (lower limit of normal); FEV1 (forced expiratory volume in 1 second); FVC (forced vital capacity)

Table V.

Descriptive factors, asthma symptoms, and asthma control comparing White vs. Black children consistently classified as normal.

FEV1 Classification Group
Characteristic		White: Both Normala N=109	Black: Both Normala N=281		p-valueb
Descriptive Factors
Age (years), Mean (SD)		10.5 (0.9)	10.4 (0.9)		0.19
Female, N (%)		55 (50.5%)	145 (51.6%)		0.91
Height (cm), Mean (SD)		145.0 (8.3)	145.6 (9.2)		0.54
Sitting Height (cm), Mean (SD)		67.8 (4.0)	66.2 (5.4)		0.002
Obese, N (%)		13 (11.9%)	69 (24.6%)		0.006
Asthma Symptoms and Control
Any Asthma Symptoms in the Past 12 Months by ISAAC, N (%)		21 (19.4%)	99 (35.5%)		0.002
Completed the ACT, N (%)		12 (11.0%)	53 (18.9%)		0.070
ACT Score, Median [Q1,Q3]		24 [22, 25.5]	22 [21,24]		0.062
Asthma is Not Well Controlled (ACT≤19), N (%)		0 (0.0%)	11 (20.8%)		0.11
FVC Classification Group
Characteristic		White: Both Normala N=109	Black: Both Normala N=304		p-valueb
Descriptive Factors
Age (years), Mean (SD)		10.5 (0.9)	10.3 (0.9)		0.10
Female, N (%)		54 (49.5%)	157 (51.6%)		0.74
Height (cm), Mean (SD)		145.0 (8.3)	145.5 (9.3)		0.59
Sitting Height (cm), Mean (SD)		67.8 (4.0)	66.2 (5.3)		0.001
Obese, N (%)		13 (11.9%)	74 (24.3%)		0.006
Asthma Symptoms and Control
Any Asthma Symptoms in the Past 12 Months by ISAAC, N (%)		21 (19.4%)	116 (38.4%)		<0.001
Completed the ACT, N (%)		12 (11.0%)	69 (22.7%)		0.008
ACT Score, Median [Q1,Q3]		24 [22, 25.5]	22 [20,24]		0.019
Asthma is Not Well Controlled (ACT≤19), N (%)		0 (0.0%)	15 (21.7%)		0.11
FEV1/FVC Classification Group
Characteristic	White: Both Normala N=94			Black: Both Normala N=259	p-valueb
Descriptive Factors
Age (years), Mean (SD)	10.4 (0.8)			10.4 (0.9)	0.36
Female, N (%)	46 (48.9%)			134 (51.7%)	0.72
Height (cm), Mean (SD)	144.5 (8.2)			145.5 (9.2)	0.36
Sitting Height (cm), Mean (SD)	67.6 (4.0)			65.9 (5.4)	0.004
Obese, N (%)	10 (10.6%)			60 (23.2%)	0.010
Asthma Symptoms and Control
Any Asthma Symptoms in the Past 12 Months by ISAAC, N (%)	18 (19.4%)			85 (33.1%)	0.016
Completed the ACT, N (%)	10 (10.6%)			47 (18.1%)	0.10
ACT Score, Median [Q1,Q3]	24 [22, 25]			22 [20,24]	0.086
Asthma is Not Well Controlled (ACT≤19), N (%)	0 (0.0%)			10 (21.3%)	0.18

^aChildren with both race-corrected and race-uncorrected values≥LLN

^bCalculated using Kruskal-Wallis for continuous variables and Fisher’s exact test for categorical variables

Abbreviations: ACT (Asthma Control Test); ISAAC (International Study of Asthma and Allergies in Childhood); LLN (lower limit of normal); FEV1 (forced expiratory volume in 1 second); FVC (forced vital capacity)

Discussion

In this study, we evaluated the impact of race-norming on spirometry results in pre-adolescent children and assessed the burden of asthma symptoms and control in Black children differentially classified depending on race correction use. Our findings indicated that race-correction had a large impact on FEV1 and FVC separately, but did not have a large impact of FEV1/FVC ratio (consistent with previous GLI reports³⁰), with the rate of obstruction in Black children increasing by 4.5% when race correction was removed. Though this is a small percentage, it translates to a large number of differentially classified Black children on the population level. Additionally, although not a statistically significant difference, half of Black children differentially classified on FEV1/FVC (8/16) experienced recent asthma symptoms compared to one third of those normally classified (85/259). No differences in ACT scores were observed. However, children on the cusp of abnormal spirometry classification may be those with the most opportunities for intervention (as opposed to severely impaired children) yet may be overlooked. Further, for both FEV1 and FEV1/FVC, differentially classified Black children appeared to have a rate of asthma symptoms most similar to “abnormally” classified Black children and distinct from that of “normal” Black children, in which they would typically be classified with using race-corrected equations.

Though we were limited in power to detect differences by FEV1/FVC classification group with only 16 differentially classified Black children, the considerable effect sizes observed warrant further exploration in larger diverse studies. Another potential limitation of the study is that some children were diagnosed with asthma at age 10 by a study physician, and asthma diagnosis as well as medication access, use, and adherence may impact spirometry results. However, because physicians diagnosed asthma using clinical evaluation (including history, physical exam, and objective data), race-corrected spirometry results may have impacted their decision-making, making casual relationships difficult to disentangle. However, the goal of this study was not to develop a new set of reference equations from a “healthy” population, but rather to examine how race-correction impacts results and examine the frequency of asthma symptoms in misclassified Black children (regardless of if they had an “asthma diagnosis” or not).

The impact of race-correction on FEV1 and FVC was seen both in the distribution of continuous percent predicted values as well as the classification of abnormal values. Specifically, the rate of abnormal FEV1 results more than doubled among Black children when race-uncorrected equations were used instead of race-corrected equations (7% vs. 18.1%) and was almost 8 times larger based on FVC (1.5% vs. 11.4%). Further, the frequency of PRISm was almost 5 times higher in Black children when race-uncorrected equations were used (1.5% vs. 7.3%), reflecting the fact that most of the differently classified results involved both FEV1 and FVC. Given that PRISm is associated with increased mortality and is clinically heterogenous,²⁷ whether it is being underdiagnosed in Black children warrants further exploration. Given that spirometric measurements are used jointly to make clinical decisions, race-norming in reference equations has serious implications for Black children. In a previous study comparing racial disparities in allergic outcomes at age 10 in these same children, we found that Black children were at higher risk of nearly all outcomes examined, including eczema, asthma, and allergic sensitization.³¹ However, PFTs—in which race-corrected reference equations were applied—did not significantly differ or were in the opposite direction as hypothesized. This re-analysis provides clear evidence that race-norming explains this contradictory result.

While there are associations between race and ancestry or genetics, race is not a valid or reliable proxy for these markers. For these reasons, some studies have examined if biological variables presumed to be related to race are associated with differences in lung function. Previous studies examining the association between percent African ancestry and lung function have found an inverse association with FEV1 and FVC.^{11, 32} However, ancestry informative markers are not fully independent from the social components of race, as ancestry is correlated with skin tone³³ and darker skin has been associated with heightened experiences of racism³⁴ and poverty.³⁵ Though social variables such as socioeconomic status have been tied to lung function³⁶ and have been adjusted for in some of these analyses, the full impacts of intergenerational exposure to structural racism are difficult to capture and are an ongoing area of research.³⁷

The role of various anthropometric variables have also been investigated. A study of school-aged White and Asian children was unable to explain the effect of ethnicity on lung function using measures of chest dimensions.³⁸ Sitting height has also been examined in the literature, as a smaller trunk:leg ratio—which is often observed in Black children—is thought to correspond to lower lung volumes.³⁹ A study examining a wide range of factors contributing to racial differences in childhood lung function found that sitting height explained 42–53% of the difference by race,³⁹ while socioeconomic factors and dietary markers accounted for an additional 7–10%.³⁹ Though we observed a strong association between race and sitting height, it failed to explain racial differences in race-uncorrected lung function measurements. Given these findings, unobserved indicators such as low-quality care or inadequate access to care, chronic stress, and discrimination cannot be ruled out as contributing factors.

We also observed that Black children were in later stages of puberty at the time of lung function measurement despite no difference in age. Racial differences in pubertal development have been previously shown,⁴⁰ and it is plausible that pubertal status impacts lung function.⁴¹ The causes of racial differences in pubertal development are not well understood, but previous studies have suggested that complex interactions with genetic, nutritional, and environmental factors play a role in determining pubertal timing.⁴² A growing literature on environmental toxicants (endocrine disrupting chemicals)—which members of non-White race groups tend to have higher exposures to due to environmental racism—suggests that exposure may induce early puberty,⁴³ so pubertal status is not fully attributable to biology. Obesity was also more common in Black children and weight-related anthropometrics have been associated with lung function in children.⁴⁴ However, the causes of childhood obesity are multifactorial, with environmental factors, lifestyle, and genetics playing a role.⁴⁵ However, similar to sitting height, adjusting for puberty status and obesity failed to negate racial differences. If there is sufficient high-quality scientific evidence that additional biological variables impact spirometry results similar to the impact of age, sex, and height, then these quantities should be measured directly and adjusted for in reference equations rather than race itself.

To conclude, the inclusion of race in spirometry reference equations should be seriously and thoughtfully reconsidered as its use has not been fully justified, is rooted in structural racism, and can mask the effects of environment and lived experiences which may lead to inequitable allocation of resources. At minimum, Black children who may be differentially classified depending on reference equation used should be provided additional clinical consideration to ensure appropriate care, as they are more likely to report asthma symptoms than children consistently classified as normal. Further, the issues with spirometry reference equations highlights the importance of a comprehensive and multifactorial approach to diagnosis and management and should not solely rely on PFTs, particularly for pediatric asthma, which is a complex and heterogeneous disease. Additionally, the fact that regardless of reference equation used, Black children consistently classified as “normal” lung function appeared to still have a significantly higher frequency of asthma symptoms may call into question the diagnostic accuracy of spirometry in Black pediatric populations; though not the focus of this manuscript, this should be explored in future studies.

A paradigm shift is occurring in the scientific community at large such that race-norming in medical algorithms is no longer an acceptable practice;¹³ therefore, there is a critical need for the pulmonary community to identify the variables that underly “race” and develop valid reference equation algorithms.¹² Recently, the American Thoracic Society (ATS) published an official statement on the use of race and ethnicity in pulmonary function test interpretation, which recommended “to replace race and ethnicity-specific equations with race-neutral average reference equations, which must be accompanied with a broader re-evaluation of how PFTs are used to make clinical, employment, and insurance decisions”.¹⁸ The statement was also endorsed by the European Respiratory Society. The statement also called to engage additional key stakeholders and continue research to understand the impact of various reference equations on different populations. Clinicians and researchers seeking immediate change in their clinical practice and/or research data may wish to use the GLI reference equations and set race to “Other/mixed” for all patients. However, consistent with the statement of caution regarding uncertain effects of changes and potential harms noted in the official ATS statement,¹⁸ it should be noted that this approach will increase variability and could potentially result in over-testing and diagnosis.¹² New promising solutions for developing race-neutral reference equations have recently been published,¹⁰ though large datasets with chest dimensions and other characteristics of lung size may still be needed. The clinical standard in PFTs remained unchanged for nearly twenty years despite several studies and commentaries discussing the issues with race-correction in PFT dating as far back as 2005.^{46,12, 18} However, the new guidelines from ATS indicate a paradigm shift may be occurring in pulmonary medicine. By allowing race-norming to be included in PFT, we may be masking abnormal lung function in Black children and missing opportunities for appropriate intervention in an already vulnerable population which has been profoundly impacted by the effects of explicit and structural racism.

Highlights Box.

What is already known about this topic?

The inclusion of race correction in spirometry reference equations is based on the assumption of biological differences. However, this practice is increasingly being brought into question, as social determinants of health strongly differ by race.

What does this article add to our knowledge?

This study demonstrates that race-correction in pulmonary function testing leads to underestimation of likely abnormal results in Black children, and that a larger proportion of differentially classified Black children experience asthma symptoms compared to Black children consistently classified as normal.

How does this study impact current management guidelines?

This study provides evidence that race-correction in pre-adolescents has important implications for the interpretation of spirometry results and should be reconsidered, as race is a flawed surrogate for biological markers.

Abbreviations

ACT

Asthma Control Test

ATS

American Thoracic Society

FEV1

forced expiratory volume in 1 second

FVC

forced vital capacity

GLI

Global Lung Initiative

ISAAC

The International Study of Asthma and Allergies in Childhood

LLN

lower limit of normal

NAEPP EPR-3

National Asthma Education and Prevention Program Expert Panel Report 3

PFT

Pulmonary Function Test

PRISm

preserved ratio impaired spirometry

WHEALS

Wayne County Health, Environment, Allergy, and Asthma Longitudinal Study