Race-Specific Spirometry Equations Do Not Improve Models of Dyspnea and Quantitative Chest CT Phenotypes
2023; Elsevier BV; Volume: 164; Issue: 6 Linguagem: Inglês
10.1016/j.chest.2023.07.019
ISSN1931-3543
AutoresAmy L. Non, Barbara Bailey, Surya P. Bhatt, Richard Casaburi, Elizabeth A. Regan, Angela Yee‐Moon Wang, Alfonso Limon, Chantal Rabay, Alejandro A. Díaz, Arianne K. Baldomero, Gregory L. Kinney, Kendra A. Young, Ben Felts, Carol A. Hand, Douglas Conrad,
Tópico(s)Interstitial Lung Diseases and Idiopathic Pulmonary Fibrosis
ResumoBackgroundRace-specific spirometry reference equations are used globally to interpret lung function for clinical, research, and social purposes, but inclusion of race is under scrutiny.Research QuestionDoes including self-identified race in spirometry reference equation formation improve the ability of predicted FEV1 values to explain quantitative chest CT abnormalities, dyspnea, or Global Initiative for Chronic Obstructive Lung Disease (GOLD) classification?Study Design and MethodsUsing data from healthy never-smoking adults in both the National Health and Nutrition Survey (2007-2012) and COPDGene study cohorts, race-neutral, race-free, and race-specific prediction equations were generated for FEV1. Using sensitivity/specificity, multivariable logistic regression, and random forest models, these equations were applied in a cross-sectional analysis to populations of smokers and former smokers to determine how they affected GOLD classification and the fit of models predicting quantitative chest CT phenotypes or dyspnea.ResultsRace-specific equations showed no advantage relative to race-neutral or race-free equations in models of quantitative chest CT phenotypes or dyspnea. Race-neutral reference equations reclassified up to 19% of black participants into more severe GOLD classes, and race-neutral/race-free equations may improve model fit for dyspnea symptoms relative to race-specific equations.InterpretationRace-specific equations offered no advantage over race-neutral/race-free percent predicted FEV1 values in three distinct explanatory models of dyspnea and chest CT scan abnormalities. Race-neutral/race-free reference equations may improve pulmonary disease diagnoses and treatment in populations highly vulnerable to lung disease. Race-specific spirometry reference equations are used globally to interpret lung function for clinical, research, and social purposes, but inclusion of race is under scrutiny. Does including self-identified race in spirometry reference equation formation improve the ability of predicted FEV1 values to explain quantitative chest CT abnormalities, dyspnea, or Global Initiative for Chronic Obstructive Lung Disease (GOLD) classification? Using data from healthy never-smoking adults in both the National Health and Nutrition Survey (2007-2012) and COPDGene study cohorts, race-neutral, race-free, and race-specific prediction equations were generated for FEV1. Using sensitivity/specificity, multivariable logistic regression, and random forest models, these equations were applied in a cross-sectional analysis to populations of smokers and former smokers to determine how they affected GOLD classification and the fit of models predicting quantitative chest CT phenotypes or dyspnea. Race-specific equations showed no advantage relative to race-neutral or race-free equations in models of quantitative chest CT phenotypes or dyspnea. Race-neutral reference equations reclassified up to 19% of black participants into more severe GOLD classes, and race-neutral/race-free equations may improve model fit for dyspnea symptoms relative to race-specific equations. Race-specific equations offered no advantage over race-neutral/race-free percent predicted FEV1 values in three distinct explanatory models of dyspnea and chest CT scan abnormalities. Race-neutral/race-free reference equations may improve pulmonary disease diagnoses and treatment in populations highly vulnerable to lung disease. Take-home PointsStudy Question: Does including self-identified race in the formation of spirometry reference equations improve the ability of predicted FEV1 values to explain quantitative chest CT abnormalities, dyspnea, or GOLD classification?Results: Race-neutral and race-free equations reclassified up to 19% of black smokers to worse GOLD classes in the COPDGene smoking cohort, with the greatest effects seen in individuals with mild smoking-related disease. The generated ppFEV1 values from race-neutral and race-free spirometry equations showed no significant improvement in model fit compared with dyspnea or quantitative chest CT phenotypes (emphysema, air trapping, airway wall thickness).Interpretation: Race-neutral/free reference equations may improve pulmonary disease diagnoses and treatment in populations highly vulnerable to lung disease relative to race-specific equations. Study Question: Does including self-identified race in the formation of spirometry reference equations improve the ability of predicted FEV1 values to explain quantitative chest CT abnormalities, dyspnea, or GOLD classification? Results: Race-neutral and race-free equations reclassified up to 19% of black smokers to worse GOLD classes in the COPDGene smoking cohort, with the greatest effects seen in individuals with mild smoking-related disease. The generated ppFEV1 values from race-neutral and race-free spirometry equations showed no significant improvement in model fit compared with dyspnea or quantitative chest CT phenotypes (emphysema, air trapping, airway wall thickness). Interpretation: Race-neutral/free reference equations may improve pulmonary disease diagnoses and treatment in populations highly vulnerable to lung disease relative to race-specific equations. Interpretation of spirometry results has traditionally relied upon reference equations to provide an estimate of "normal" lung function for an individual's age, gender, height—and controversially—race/ethnicity. These equations are used for clinical, research, and occupational purposes to diagnose pulmonary disease, assess disease progression, and explain radiographic abnormalities, as well as determine disability and evaluate fitness for higher risk jobs, and thus have enormous clinical and financial importance. The inclusion of race in these equations is based on large cross-sectional, population-wide studies that consistently show lower measures of lung function for some racial/ethnic minority groups, specifically up to 10% to 15% lower FEV1 for black individuals.1Hankinson J.L. Odencrantz J.R. Fedan K.B. Spirometric reference values from a sample of the general US population.Am J Respir Crit Care Med. 1999; 159: 179-187Crossref PubMed Scopus (3426) Google Scholar,2Quanjer P.H. Stanojevic S. Cole T.J. 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: 1324-1343Crossref PubMed Scopus (3593) Google Scholar However, the clinical value of race adjustments has increasingly been questioned.3Anderson M.A. Malhotra A. Non A.L. Could routine race-adjustment of spirometers exacerbate racial disparities in COVID-19 recovery?.Lancet Respir Med. 2021; 9: 124-125Abstract Full Text Full Text PDF PubMed Scopus (18) Google Scholar, 4Bhakta N.R. Kaminsky D.A. Bime C. et al.Addressing race in pulmonary function testing by aligning intent and evidence with practice and perception.Chest. 2022; 161: 288-297Abstract Full Text Full Text PDF PubMed Scopus (30) Google Scholar, 5Braun L. Breathing Race into the Machine: The Surprising Career of the Spirometer from Plantation to Genetics. University of Minnesota Press, Minneapolis, MN2014Crossref Google Scholar, 6Beaverson S, Ngo VM, Pahuja M, Dow A, Nana-Sinkam P, Schefft M. Things We Do for No ReasonTM: race adjustments in calculating lung function from spirometry measurements [published online ahead of print October 22, 2022]. J Hosp Med. https://doi.org/10.1002/jhm.12974.Google Scholar Although recent studies found no prognostic benefit of race-specific equations compared with "race-neutral" equations in mortality or respiratory events,7Baugh A.D. Shiboski S. Hansel N.N. et al.Reconsidering the utility of race-specific lung function prediction equations.Am J Respir Crit Care Med. 2022; 205: 819-829Crossref PubMed Scopus (32) Google Scholar, 8Elmaleh-Sachs A. Balte P. Oelsner E.C. et al.Race/ethnicity, spirometry reference equations and prediction of incident clinical events: the Multi-Ethnic Study of Atherosclerosis (MESA) Lung Study.Am J Respir Crit Care Med. 2022; 205: 700-710Crossref PubMed Google Scholar, 9McCormack M.C. Balasubramanian A. Matsui E.C. Peng R.D. Wise R.A. Keet C.A. Race, lung function, and long-term mortality in the National Health and Nutrition Examination Survey III.Am J Respir Crit Care Med. 2022; 205: 723-724Crossref PubMed Google Scholar, 10Liu G.Y. Khan S.S. Colangelo L.A. et al.Comparing racial differences in emphysema prevalence among adults with normal spirometry: a secondary data analysis of the CARDIA Lung Study.Ann Intern Med. 2022; 175: 1118-1125Crossref PubMed Scopus (6) Google Scholar, 11Ekström M. Mannino D. Race-specific reference values and lung function impairment, breathlessness and prognosis: analysis of NHANES 2007-2012.Respir Res. 2022; 23: 271Crossref PubMed Scopus (3) Google Scholar others continue to defend the use of race in prediction equations.12Miller M.R. Graham B.L. Thompson B.R. Race/ethnicity and reference equations for spirometry.Am J Respir Crit Care Med. 2022; 206: 790-792Crossref PubMed Scopus (2) Google Scholar,13Townsend M.C. Cowl C.T. US occupational historical perspective on race and lung function.Am J Respir Crit Care Med. 2022; 206: 789-790Crossref PubMed Scopus (5) Google Scholar Race-specific equations are recommended by the most recent US and European guidelines14Stanojevic S. Kaminsky D.A. Miller M.R. et al.ERS/ATS technical standard on interpretive strategies for routine lung function tests.Eur Respir J. 2022; 602101499Crossref Scopus (150) Google Scholar and are still used in clinical care and pulmonary research worldwide. However, applying race-specific equations may mask developmental or acquired lung damage among minority groups15Barker D.J. Godfrey K.M. Fall C. Osmond C. Winter P.D. Shaheen S.O. Relation of birth weight and childhood respiratory infection to adult lung function and death from chronic obstructive airways disease.BMJ. 1991; 303: 671-675Crossref PubMed Scopus (697) Google Scholar, 16Lovasi G.S. Diez Roux A.V. Hoffman E.A. Kawut S.M. Jacobs D.R. Barr R.G. Association of environmental tobacco smoke exposure in childhood with early emphysema in adulthood among nonsmokers: the MESA-Lung Study.Am J Epidemiol. 2010; 171: 54-62Crossref PubMed Scopus (41) Google Scholar, 17Martinez F.D. Early-life origins of chronic obstructive pulmonary disease.N Engl J Med. 2016; 375: 871-878Crossref PubMed Scopus (321) Google Scholar and risks underdiagnosing damaged lungs in marginalized groups at high risk of respiratory disease,18Fuller-Thomson E. Chisholm R.S. Brennenstuhl S. COPD in a population-based sample of never-smokers: interactions among sex, gender, and race.Int J Chronic Dis. 2016; 20165862026PubMed Google Scholar, 19Rocha V. Soares S. Stringhini S. Fraga S. Socioeconomic circumstances and respiratory function from childhood to early adulthood: a systematic review and meta-analysis.BMJ Open. 2019; 9e027528Crossref Scopus (19) Google Scholar, 20Mamary A.J. Stewart J.I. Kinney G.L. et al.Race and gender disparities are evident in COPD underdiagnoses across all severities of measured airflow obstruction.Chronic Obstr Pulm Dis. 2018; 5: 177-184PubMed Google Scholar, 21Moffett A.T. Eneanya N.D. Halpern S.D. Weissman G.E. The Impact of Race Correction on the Interpretation of Pulmonary Function Testing Among Black Patients.in: Impact of Race, Ethnicity, and Social Determinants on Individuals with Lung Diseases. American Thoracic Society, 2021https://www.atsjournals.org/doi/10.1164/ajrccm-conference.2021.203.1_MeetingAbstracts.A1030Crossref Google Scholar thereby exacerbating racial health inequalities. We examined how the percent predicted FEV1 (ppFEV1) values calculated from race-specific, race-neutral, and race-free reference equations differentially affect pulmonary phenotypes in two large cohorts of smokers. First, using a selected sample of healthy never-smoking adults from both National Health and Nutrition Examination Survey (NHANES) (2007-2012) and COPDGene cohorts (e-Fig 1), we generated new race-free equations that entirely exclude race from model formation and race-specific prediction equations for FEV1 and FVC. Second, these equations were compared with the Global Lung Initiative (GLI) race-specific equations,2Quanjer P.H. Stanojevic S. Cole T.J. 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: 1324-1343Crossref PubMed Scopus (3593) Google Scholar the race-specific equations of Hankinson et al,1Hankinson J.L. Odencrantz J.R. Fedan K.B. Spirometric reference values from a sample of the general US population.Am J Respir Crit Care Med. 1999; 159: 179-187Crossref PubMed Scopus (3426) Google Scholar and the race-neutral GLI-Other (uses a universal race-correction) and the GLI-Global (weights racial groups in the reference population) equations.2Quanjer P.H. Stanojevic S. Cole T.J. 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: 1324-1343Crossref PubMed Scopus (3593) Google Scholar Third, we applied these spirometry prediction equations and determined how they differentially: (1) affect the Global Initiative for Chronic Obstructive Lung Disease (GOLD) severity classification in both the NHANES and COPDGene smoking cohorts; and (2) model quantitative chest CT phenotypes and dyspnea in the COPDGene study participants. Our intent was to compare how the different reference equations model clinically important pulmonary phenotypes. Details on formation and characterization of nonsmoking (asymptomatic) and smoking cohorts are presented in e-Appendix 1, e-Figures 1 to 3, and Table 1.Table 1Characteristics of the NHANES (N = 3,700) and COPDGene (N = 419) Never-Smoker Healthy Cohorts by Race/EthnicityChCharacteristicNHANES,All (N = 3,700)NHANES,White (3)(n = 1,420 [38.4%])NHANES,Black (4)(n = 762 [20.6%])NHANES, Mexican American (1) (n = 682 [18.4%])NHANES,Other Hispanic (2) (n = 473 [12.8%])NHANES,Other/Mixed (5) (n = 363 [9.8%])COPDGene,All (N = 419)COPDGene, White (n = 342 [81.6%])COPDGene, Black (n = 77 [18.4%])Age, y51.0 (19.0)52.0 (21.0)52.0 (18.0)49.0 (11.0)a51.0 (20.0)49.0 (17.0)a59.1 (15.4)61.4 (15.4)55.1 (9.6)aFemale, n (%)2,160 (58.4)803 (56.5)446 (58.5)396 (58.1)302 (63.8)213 (58.7)239 (57.0)194 (56.7)45 (58.4)Height, cm165.1 (14.4)168.4 (15.1)167.7 (13.4)160.9 (12.7)a160.3 (14.4)a162.1 (12.9)a168.6 (14.0)168.7 (14.7)168.0 (12.7)Weight, kg78.8 (25.1)81.8 (26.1)86.5 (27.0)a75.4 (18.9)a75.1 (22.7)a66.1 (19.4)a78.0 (23.1)76.5 (23.9)80.0 (19.2)BMI, kg/m228.7 (7.4)28.6 (7.6)30.4 (8.2)a29.2 (6.0)28.6 (6.5)24.9 (5.5)a27.1 (6.0)26.8 (5.9)28.3 (5.11)FEV1, L2.83 (1.11)3.07 (1.19)2.57 (0.98)a2.89 (1.01)a2.70 (1.03)a2.65 (0.94)a2.83 (1.11)2.92 (1.19)2.62 (0.85)bFVC, L3.55 (1.41)3.90 (1.51)3.16 (1.16)a3.58 (1.32)a3.39 (1.25)a3.30 (1.22)a3.55 (1.38)3.60 (1.45)3.16 (1.20)aFEV1/FVC0.80 (0.07)0.79 (0.07)0.81 (0.07)a0.81 (0.06)a0.80 (0.06)a0.80 (0.07)a0.80 (0.08)0.80 (0.07)0.82 (0.07)appFEV1 GLIdGuideline-based application of GLI race/ethnic-specific reference equations. The GLI equations for white/European individuals were used to estimate ppFEV1 for the NHANES Mexican American and Other Hispanic groups, following other studies (8). The GLI-predicted FEV1 values for the NHANES group "Other/Mixed Race" used the GLI-Other equations.99.0 (17.1)99.8 (16.3)98.0 (18.7)c100.4 (16.2)97.2 (16.4)c96.8 (16.7)c101.0 (18.7)101.6 (18.6)99.9 (16.7)ppFEV1 GLI-OthereThe GLI-Other equation was used to generate race-neutral estimates of ppFEV1 for all racial/ethnic groups.102.61 (20.0)107.1 (17.5)90.5 (17.4)a107.7 (17.4)104.4 (17.6)c96.8 (16.7)a106.6 (20.3)109.0 (20.0)92.3 (16.2)aqCT parametersPi10..................1.77 (0.42)1.75 (0.41)1.94 (0.44)c Air trapping (n = 321)..................7.18 (8.90)7.20 (8.70)6.80 (10.01) Percent emphysema..................0.84 (1.94)0.88 (2.16)0.62 (1.19)c mMRC score (0-4)0.00 (0.00)0.00 (0.00)0.00 (0.00)Values of continuous variables are presented as medians (interquartile range). GLI = Global Lung Initiative; mMRC = modified Medical Research Council; NHANES = National Health and Nutrition Examination Survey; Pi10 = airway wall thickness estimate based on square root of wall area of a 10 mm lumen perimeter (23)46Hoesein F.A.A.M. de Jong P.A. Lammers J.-W.J. et al.Airway wall thickness associated with forced expiratory volume in 1 second decline and development of airflow limitation.Eur Respir J. 2015; 45: 644-651Crossref PubMed Scopus (40) Google Scholar; ppFEV1 = percent predicted FEV1; qCT = quantitative chest CT.Indicates significant difference relative to white group at: aP < .0001, bP < .001, cP < .05, according to the Kruskal-Wallis test for comparison of continuous variables between (non-Hispanic) black and white racial/ethnic groups in COPDGene, and analysis of variance with Tukey ad hoc comparisons for continuous variables between each racial/ethnic group relative to white participants in the NHANES data.d Guideline-based application of GLI race/ethnic-specific reference equations. The GLI equations for white/European individuals were used to estimate ppFEV1 for the NHANES Mexican American and Other Hispanic groups, following other studies (8). The GLI-predicted FEV1 values for the NHANES group "Other/Mixed Race" used the GLI-Other equations.e The GLI-Other equation was used to generate race-neutral estimates of ppFEV1 for all racial/ethnic groups. Open table in a new tab Values of continuous variables are presented as medians (interquartile range). GLI = Global Lung Initiative; mMRC = modified Medical Research Council; NHANES = National Health and Nutrition Examination Survey; Pi10 = airway wall thickness estimate based on square root of wall area of a 10 mm lumen perimeter (23)46Hoesein F.A.A.M. de Jong P.A. Lammers J.-W.J. et al.Airway wall thickness associated with forced expiratory volume in 1 second decline and development of airflow limitation.Eur Respir J. 2015; 45: 644-651Crossref PubMed Scopus (40) Google Scholar; ppFEV1 = percent predicted FEV1; qCT = quantitative chest CT. Indicates significant difference relative to white group at: aP < .0001, bP < .001, cP < .05, according to the Kruskal-Wallis test for comparison of continuous variables between (non-Hispanic) black and white racial/ethnic groups in COPDGene, and analysis of variance with Tukey ad hoc comparisons for continuous variables between each racial/ethnic group relative to white participants in the NHANES data. The predicted FEV1 and lower limit of normal (LLN) values for those that never smoked from the GLI equations were obtained using the GLI website (https://gli-calculator.ersnet.org/index.html, version 2.0, April 2023). Values predicted by Hankinson et al1Hankinson J.L. Odencrantz J.R. Fedan K.B. Spirometric reference values from a sample of the general US population.Am J Respir Crit Care Med. 1999; 159: 179-187Crossref PubMed Scopus (3426) Google Scholar (NHANES III) and LLN values were calculated by using published equations. For both never-smoking data sets (the NHANES data set of 3,700 healthy individuals [nh3700] and the COPDGene data set of 419 healthy individuals), multivariable linear quantile regression was used to generate predicted (median quantile) and LLN (fifth quantile) models and associated R1 values (Table 2). The R1 value is a measure of explained variability of the data in quantile regression and is used to compare models.22Koenker R. Machado J.A.F. Goodness of fit and related inference processes for quantile regression.J Am Stat Assoc. 1999; 94: 1296-1310Crossref Scopus (864) Google Scholar Predictors in the race-specific equations included age (years), height (centimeters), sex (male/female), and self-identified race/ethnicity. Predictors in race-free equations included only age, height, and sex (e-Tables 1-3). A similar approach was used to generate models for predicted and LLN values for log (FVC). This approach generated four race-specific and four race-neutral/race-free models for the predicted log (FEV1) and log (FVC) from different source populations (e-Tables 1-3, Table 2, Table 3). Identity and probability density plots of the differences between the predicted race-specific, race-neutral, and race-free models were used to explore the effect of race in the models.Table 2Models of Predicted log (FEV1) Median and Fifth Percentile Quantile Regression Coefficients for NHANES (N = 3,700) and COPDGene (N = 419)Modelβ0 (Intercept)β1 (Age in Years)β2 (Female Gender)β3 (Height in Centimeters)β4 (Race)R1NHANES: log (FEV1) Median predicted: AGH–0.30931-0.00966–0.160270.01177...0.485 Median predicted: AGHR (white)–0.42116–0.00922–0.149350.01253Reference0.547 AGHR (African American)–0.17048 AGHR (Mexican American)–0.00454 AGHR (Hispanic Other)–0.0219 AGHR (other/mixed)–0.09983 5th percentile: AGH–0.37667–0.01069–0.169760.01091...... 5th percentile: AGHR (white)–0.55486–0.01094–0.155680.01258ref... African American–0.19925 Mexican American–0.00657 Hispanic Other–0.03935 Other/Mixed–0.09986COPDGene: log (FEV1) Median predicted: AGH–0.40813–0.00958–0.170360.01262...0.476 Median predicted: AGHR (black)–0.37885–0.01077–0.154550.01294–0.164830.531 5th percentile: AGH–0.04169–0.01084–0.179100.00946...... 5th percentile: AGHR (black)–0.14623–0.01251–0.155920.01097–0.17822...The equation used to model the predicted log (FEV1) is as follows: predicted (or lower limit of normal) FEV1 (L)= e(β0 + Age∗ β1 + Gender code term ∗ β2 + Height∗ β3 +Race code term∗ β4). In both cohorts, race is coded as black (1) relative to NHW (0) as the reference group. In data from the NHANES data set of 3,700 healthy individuals, race is modeled with NHW (0) as the reference group, and other racial/ethnic groups were coded as 1 if present and multiplied by the corresponding race coefficient (β4). In both cohorts, sex code term for male subjects is 0 and 1 for female subjects. The log of FEV1 (post-bronchodilator values were used when available for NHANES data and always for COPDGene data) in liters was used because it optimized the explained variability compared with modeling raw FEV1 values. AGH = models including age, gender, and height only; AGHR = models including age, sex, height, and race/ethnicity; NHANES = National Health and Nutrition Examination Survey. Open table in a new tab Table 3Summaries of Source Data, Covariates, and Nomenclature for Predicted FEV1 ModelsPredicted FEV1 ModelTypeRaceSource PopulationSource Sample Size by Race/EthnicitySource Age RangeCovariatesReferenceGLI race-specificQR-LinearSpecificInternationalTotal: N = 74,187White: n = 57,395Black: n = 3,545NE Asian: n = 4,992SE Asian: n = 8,2253-95 yAGHR3Anderson M.A. Malhotra A. Non A.L. Could routine race-adjustment of spirometers exacerbate racial disparities in COVID-19 recovery?.Lancet Respir Med. 2021; 9: 124-125Abstract Full Text Full Text PDF PubMed Scopus (18) Google ScholarHankinsonOLS-LinearSpecificNHANES-1999Total n = 7,429White: n = 2,281Black: n = 2,508Mexican American: n = 2,6398-80 yAGHR2Quanjer P.H. Stanojevic S. Cole T.J. 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: 1324-1343Crossref PubMed Scopus (3593) Google Scholarcg419_AGHRQR-LinearSpecificCOPDGene never-smokingTotal: N = 419White: n = 342Black: n = 7745-82 yAGHRTable 2nh3700_AGHRQR-LinearSpecificNHANES 2007-2012 healthy ex-smokersTotal: N = 3,700White: n = 1,420Black: n = 762Mexican American: n = 682Other Hispanic: n = 473Other: n = 36335-79 yAGHRTable 2GLI-OtherGAMLSSNeutralInternationaln = 74,1873-95 yAGHaGLI-Other was calculated by taking "its mean and CoV adjustments the corresponding adjustments for the four main ethnic groups, averaged over group and sex" (3).3Anderson M.A. Malhotra A. Non A.L. Could routine race-adjustment of spirometers exacerbate racial disparities in COVID-19 recovery?.Lancet Respir Med. 2021; 9: 124-125Abstract Full Text Full Text PDF PubMed Scopus (18) Google ScholarGLI-GlobalGAMLSSNeutralInternationaln = 74,1853-95 yAGHbFor GLI-Global, an inverse probability weight was applied for each of the four racial groups included in the data set.35Chan J.Y.C. Stern D.A. Guerra S. Wright A.L. Morgan W.J. Martinez F.D. Pneumonia in childhood and impaired lung function in adults: a longitudinal study.Pediatrics. 2015; 135: 607-616Crossref PubMed Scopus (127) Google Scholarcg419_AGHQR-LinearFreeCOPDGene healthy never-smokingN = 41945-82 yAGHTable 2nh3700_AGHQR-LinearFreeNHANES 2007-2012 healthy never-smokingN = 3,70035-79 yAGHTable 2Details about models and source populations used to develop each of the race-specific, race-neutral, and race-free models used in this study. GLI-Other and GLI-Global equations are race-neutral, but are not race-free, as they averaged race/ethnicity estimates across four major racial/ethnic groups. AGH = age, gender, and height; AGHR = age, gender, height, and race; cg419 = COPDGene data set of 419 healthy individuals; GAMLSS = General Additive Models for Location Scale and Shape; GLI = Global Lung Initiative; NE = northeast; nh3700 = National Health and Nutrition Examination Survey data set of 3,700 healthy individuals; NHANES = National Health and Nutrition Examination Survey; OLS = ordinary least squares regression; QR = quantitative regression; SE = southeast.a GLI-Other was calculated by taking "its mean and CoV adjustments the corresponding adjustments for the four main ethnic groups, averaged over group and sex" (3).b For GLI-Global, an inverse probability weight was applied for each of the four racial groups included in the data set. Open table in a new tab The equation used to model the predicted log (FEV1) is as follows: predicted (or lower limit of normal) FEV1 (L)= e(β0 + Age∗ β1 + Gender code term ∗ β2 + Height∗ β3 +Race code term∗ β4). In both cohorts, race is coded as black (1) relative to NHW (0) as the reference group. In data from the NHANES data set of 3,700 healthy individuals, race is modeled with NHW (0) as the reference group, and other racial/ethnic groups were coded as 1 if present and multiplied by the corresponding race coefficient (β4). In both cohorts, sex code term for male subjects is 0 and 1 for female subjects. The log of FEV1 (post-bronchodilator values were used when available for NHANES data and always for COPDGene data) in liters was used because it optimized the explained variability compared with modeling raw FEV1 values. AGH = models including age, gender, and height only; AGHR = models including age, sex, height, and race/ethnicity; NHANES = National Health and Nutrition Examination Survey. Details about models and source populations used to develop each of the race-specific, race-neutral, and race-free models used in this study. GLI-Other and GLI-Global equations are race-neutral, but are not race-free, as they averaged race/ethnicity estimates across four major racial/ethnic groups. AGH = age, gender, and height; AGHR = age, gender, height, and race; cg419 = COPDGene data set of 419 healthy individuals; GAMLSS = General Additive Models for Location Scale and Shape; GLI = Global Lung Initiative; NE = northeast; nh3700 = National Health and Nutrition Examination Survey data set of 3,700 healthy individuals; NHANES = National Health and Nutrition Examination Survey; OLS = ordinary least squares regression; QR = quantitative regression; SE = southeast. Each individual in both smoking cohorts was assigned a GOLD spirometry class (GOLD 1-4), the preserved ratio impaired spirometry class,23Han M.K. Agusti A. Celli B.R. et al.From GOLD 0 to pre-COPD.Am J Respir Crit Care Med. 2021; 203: 414-423Crossref PubMed Scopus (73) Google Scholar or GOLD 0 (ie, FEV1/FVC ratio > 0.7 and ppFEV1 ≥ 80% using the different race-specific and race-neutral equations). The percentage of individuals who changed GOLD class from the GLI standard (race-specific) was calculated in the total data set and within each racial group for both NHANES former smokers and COPDGene smoking participants. Measured FEV1 values were classified as above or below the LLN to assess the sensitivity and specificity of each ppFEV1 reference equation to model abnormal chest CT phenotypes in COPDGene phase I participants. Chest CT phenotypes were defined as abnormal if: (1) the percent emphysema was > 5%; (2) the percent air trapping was > 15%; or (3) the airway wall thickness estimate based on square root of wall area of a 10 mm lumen perimeter was > 2.5.24Lowe K.E. Regan E.A. Anzueto A. et al.COPDGene® 2019: redefining the diagnosis of chronic obstructive pulmonary disease.Chronic Obstr Pulm Dis. 2019; 6: 384-399Crossref PubMed Scopus (115) Google Scholar The sensitivity, specificity, negative predictive value, positive predictive value, and the area under the curve (AUC) were calculated for each model in the overall population and within each race. A parallel approach assessed the ability of the LLN of each model to predict a modified Medical Research Council dyspnea score > 1. Because complex demographic factors (smoking status and history, sex, FEV1/FVC ratio, and scanner type) and the ppFEV1 influence quantitative chest CT scan metrics,25Zach J.A. Newell J.D. Schroeder J. et al.Quantitative computed tomography of the lungs and airways in healthy nonsmoking adults.Invest Radiol. 2012; 47: 596-602Crossref PubMed Scopus (113) Google Scholar,26Hoffman E.A. Ahmed F.S. Baumhauer H. et al.Variation in the percent of emphysema-like lung in a healthy, nonsmoking multiethnic sample. The MESA Lung Study.Ann Am Thorac Soc. 2014; 11: 898-907Crossref PubMed Scopus (83) Google Scholar multivariable logistic regression models were generated of abnormal chest CT phenotypes using these covar
Referência(s)