Reference Values for Spirometry Derived Using Lambda, Mu, Sigma (LMS) Method in Korean Adults: in Comparison with Previous References

Bum Seak Jo; Jun-Pyo Myong; Chin Kook Rhee; Hyoung Kyu Yoon; Jung Wan Koo; Hyoung Ryoul Kim

doi:10.3346/jkms.2018.33.e16

Journal List > J Korean Med Sci > v.33(3) > 1107680

Go to TopGo to Top Go to BottomGo to Bottom

TOOLS

Jo, Myong, Rhee, Yoon, Koo, and Kim: Reference Values for Spirometry Derived Using Lambda, Mu, Sigma (LMS) Method in Korean Adults: in Comparison with Previous References

Original Article

Occupation & Environmental Medicine

Journal of Korean Medical Science 2017; 33(3): e16.

Published online: 24 November 2017

DOI: https://doi.org/10.3346/jkms.2018.33.e16

Reference Values for Spirometry Derived Using Lambda, Mu, Sigma (LMS) Method in Korean Adults: in Comparison with Previous References

Bum Seak Jo¹

, Jun-Pyo Myong¹

, Chin Kook Rhee²

, Hyoung Kyu Yoon³

, Jung Wan Koo¹

, Hyoung Ryoul Kim¹

¹Department of Occupational and Environmental Medicine, The Catholic University of Korea, Seoul St. Mary's Hospital, Seoul, Korea.

²Department of Internal Medicine, The Catholic University of Korea, Seoul St. Mary's Hospital, Seoul, Korea.

³Department of Internal Medicine, The Catholic University of Korea, Yeouido St. Mary's Hospital, Seoul, Korea.

Address for Correspondence: Hyoung Ryoul Kim, MD. Department of Occupational and Environmental Medicine, The Catholic University of Korea, Seoul St. Mary's Hospital, 222 Banpo-daero, Seocho-gu, Seoul 06591, Korea. cyclor@catholic.ac.kr

Received 27 July 2017 Accepted 14 October 2017

The Korean Academy of Medical Sciences

(open-access):

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Background

The present study aimed to update the prediction equations for spirometry and their lower limits of normal (LLN) by using the lambda, mu, sigma (LMS) method and to compare the outcomes with the values of previous spirometric reference equations.

Methods

Spirometric data of 10,249 healthy non-smokers (8,776 females) were extracted from the fourth and fifth versions of the Korea National Health and Nutrition Examination Survey (KNHANES IV, 2007–2009; V, 2010–2012). Reference equations were derived using the LMS method which allows modeling skewness (lambda [L]), mean (mu [M]), and coefficient of variation (sigma [S]). The outcome equations were compared with previous reference values.

Results

Prediction equations were presented in the following form: predicted value = e^{{a + b × ln(height) + c × ln(age) + M − spline}}. The new predicted values for spirometry and their LLN derived using the LMS method were shown to more accurately reflect transitions in pulmonary function in young adults than previous prediction equations derived using conventional regression analysis in 2013. There were partial discrepancies between the new reference values and the reference values from the Global Lung Function Initiative in 2012.

Conclusion

The results should be interpreted with caution for young adults and elderly males, particularly in terms of the LLN for forced expiratory volume in one second/forced vital capacity in elderly males. Serial spirometry follow-up, together with correlations with other clinical findings, should be emphasized in evaluating the pulmonary function of individuals. Future studies are needed to improve the accuracy of reference data and to develop continuous reference values for spirometry across all ages.

Graphical Abstract

Keywords: Spirometry, Reference Values, Pulmonary Function, LMS Method, Lower Limit of Normal

INTRODUCTION

Spirometric reference values are key in interpreting the findings of pulmonary function tests, which provide essential information to evaluate respiratory diseases.1 2 The principal interest indices in spirometry are forced expiratory volume in one second (FEV₁), forced vital capacity (FVC), and the FEV₁/FVC ratio, in which values below normal are generally of concern. For the evaluation of airflow limitation, a fixed ratio of FEV₁/FVC < 0.7 in a post-bronchodilator test remains standard according to the criteria of the Global Initiative for Chronic Obstructive Lung Disease (GOLD).3 However, it has been suggested that following the fixed ratio criterion may result in misjudging airflow limitations, more frequently in the elderly4 5 and less frequently in young adults.5 To mitigate such errors in diagnosis, the concept of the lower limits of normal (LLN) could be adopted which was formerly introduced in the American Thoracic Society (ATS) and the European Respiratory Society (ERS) articles.2 6

The LLN is defined as a value equal to the fifth percentile of a healthy population. If a variable follows a normal distribution, then the LLN is equivalent to subtracting 1.645 × the standard deviation (SD) from the mean. A variance is expressed as a standard error of the estimate (SEE) or residual standard deviation (RSD) in regressions, thus a normal 95% confidence interval from the 2.5 to 97.5 percentiles could be defined as the mean ± 2 × SEE. However, the FEV₁/FVC ratio is not normally distributed, and the variance (SEE) for the ratio does not vary significantly by age or height. Consequently, calculating the LLN with SEE yields statistical flaws.7

To ensure validity, reference values for spirometry should be updated periodically in consideration of birth-cohort effects and the latest methodology.8 9 Recently, Stanojevic et al.10 11 applied the lamda, mu, sigma (LMS) method to calculate more accurate reference ranges for spirometry. This new approach allowed the prediction of the values of spirometric indices accounting for transition into adulthood from childhood, and provided age-specific values for the LLN.11 In 2008, the Global Lung Function Initiative (GLI) was established in Berlin, sharing spirometric data from various centers and countries around the world. In 2010, the ERS set up the GLI as an ERS Task Force, which was subsequently endorsed by other international societies including the ATS and the Asian Pacific Society for Respirology (APSR). One of the main objectives of the GLI was to develop improved spirometric reference equations that are applicable globally with adjustment for the heterogeneity of between-subject variability according to age, sex, ethnic group, and lung function parameters.12 Spirometric data of the Korean population was also included in the study, but the data was not free from birth-cohort effects since it was collected between November 2001 and January 2002. Besides, Korea was grouped in the North East Asia dataset along with North China because of systematical discrepancies relative to other datasets including Hong Kong, Taiwan, Thailand, China, and the USA.12 It is recommended to adopt reference equations derived from data that most closely reflects the spirometric characteristics of a population, including ethnic, environmental, and socioeconomic properties.2 6 Because Korea differs from other East Asian countries in these traits, and because the data utilized in the study were not updated, reference values from the GLI in 201212 may not properly describe the spirometric characteristics of the Korean population.

The present study was conducted to update the Korean spirometric reference equations and their LLN using an improved statistical approach and to evaluate the reference values by comparing them with values of previously developed prediction equations for spirometry.

METHODS

Data sources and subjects

Raw data from the Korea National Health and Nutrition Examination Survey (KNHANES) IV (2007–2009) and V (2010–2012) were utilized to derive reference equations for spirometry. Conducted by the Korea Centers for Disease Control and Prevention (KCDC), the KNHANES is a national survey that examines the health and nutritional status of Koreans, including health risk factors and major diseases. The population reflected in the KNHANES comprises non-institutionalized citizens enrolled by a multi-stage cluster probability sampling method, securing homogenous and representative sampling.13 The KNHANES IV and V covered 31,705 (response rate: 74.5%) and 31,596 (response rate: 76.5%) people, respectively. Among all the subjects who participated in the health questionnaire and the physical examination survey, the number of adults aged 19 years and above was 38,005, and spirometric measurements were valid for 22,423 participants. The reference population was sampled using the following exclusion criteria: 1) smokers with a lifetime total smoking of 5 packs or more, 2) underlying respiratory diseases, 3) present respiratory symptoms including cough, sputum and dyspnea, 4) abnormal findings on chest radiography, 5) cardiovascular diseases other than hypertension, namely, stroke and ischemic heart diseases, 6) severe renal or liver dysfunction, and 7) any malignancy. In addition, outliers in which spirometric data yielded a z-score < 5.0 or > 5.0 were sought for exclusion (however, no outliers were found). Fig. 1 shows the flowchart for selecting the study population.

Fig. 1

Study population selection process.

KNHANES = Korea National Health and Nutrition Examination Survey, CVD = cardiovascular disease.

Measurements

Spirometry was performed using a dry rolling seal spirometer (Model 2130; Sensor Medics, Yorba Linda, CA, USA). Spirometric indices including FEV₁, FVC, and FEV₁/FVC were measured and evaluated in accordance with the ATS/ERS recommendations for standardized lung function and the criteria of acceptability and repeatability for spirometry.2 In the KNHANES IV, spirometry was performed for adults aged 19 years and over, while adults aged 40 years and over were measured in the KNHANES V.

Age, sex, and height data of all subjects were collected to develop models for outcome variables. Weight was not considered because regression effects make it inappropriate to predict pulmonary function using body mass within mammal species.14

Statistical analysis

The reference population was divided into two groups by sex. The mean, SD, and range of each variables including age, height, FEV₁, FVC, and FEV₁/FVC were calculated.

Separate reference equations for spirometry were derived for males and females using the LMS method. The LMS method was applied utilizing the generalized additive model of location, scale and shape (GAMLSS version 5.0-0) package in statistical software R (version 3.3.0 for Windows 64 bit; R Foundation, Vienna, Austria, http://www.r-project.org). The GAMLSS comprises semi-parametric regression type models, which are suitable for developing models of multiple explanatory variables including age and height.15

The LMS method allows for modeling of expected mean (mu [M]), coefficient of variation (sigma [S]), and skewness (lambda [L]). In addition, a continuous smooth fit over the entire age range is achievable by spline functions, which allow the dependent variable to vary smoothly (non-linearly) as a function of an explanatory variable. Using the Box-Cox-Cole-Green distribution described by Cole and Green,16 the best fit was estimated. To avoid over-fitting, the optimal degrees of freedom for the spline curve were attained by selecting the model with the smallest Schwarz's Bayesian Criterion. Thereby, the reference equations, which are a function of age and height, were derived as the following general form:

M = e^{{a + b × ln(height) + c × ln(age) + M-spline}}

where M is the dependent variable (reference mean value), a, b, and c are coefficients, and M-spline is the age-specific contribution from the spline function. The LLN, equivalent to the fifth percentile, was calculated from L, M, and S as follows:

LLN = e^{{ln(M) + ln(1 – 1.645 × L × S)/L}}

To compare the findings of the present study with other previous reference values, graphs showing changes in values of each spirometric index and their corresponding LLN values across age were illustrated under conditions of fixed subject height. Reference equations in the work of Eom and Kim17 derived using multiple linear regression analyses (2013MRA) and in the work of Quanjer et al.12 from ERS (2012GLI) were chosen as comparison topics.

Ethics statement

The present study protocol was reviewed and approved by the Institutional Review Board of Catholic Medical Center (ID: KC15RIMI0946) with a waiver for informed consent because the data were obtained from a public database.

RESULTS

Anthropometry and spirometry

Spirometric data were analyzed from 10,249 adults (8,776 females) over the age of 19 years. The mean, SD, and range of each variable are presented in Table 1. The demographic distributions are summarized in Supplementary Table 1.

Table 1

Anthropometric and spirometric characteristics of the reference population

Metric parameters	Males (n = 1,473)		Females (n = 8,776)
Metric parameters	Mean (SD)	Range	Mean (SD)	Range
Age, yr	49.3 (15.2)	19–91	51.9 (13.8)	19–90
Height, cm	169.1 (7.0)	124.8–196.0	156.0 (6.1)	124.2–178.3
FEV₁, L	3.39 (0.71)	0.85–5.46	2.42 (0.48)	0.58–4.71
FVC, L	4.28 (0.76)	0.97–7.32	2.99 (0.53)	1.04–5.29
FEV₁/FVC	0.79 (0.08)	0.33–0.99	0.81 (0.06)	0.36–1.00

SD = standard deviation, FEV₁ = forced expiratory volume in one second, FVC = forced vital capacity.

Reference values

Equations for the mean, S, and L values of each spirometric index are presented in Table 2. Age-specific spline values are required for calculation of the mean (M-spline) and sigma (S-spline) for FEV₁, FVC, and FEV₁/FVC. The L-spline value was also required to determine the lambda for the FEV₁/FVC ratio in females. Otherwise, the values for lambda remained fixed. The spline values obtained using penalized B-splines are presented in the Supplementary Materials (see Supplementary Tables 2, 3, 4). Likewise, the LLN for each of the spirometric indices can be calculated by substituting obtained values of M, S, and L on the equation:

LLN = e^{{ln(M) + ln(1 – 1.645 × L × S)/L}}

Table 2

LMS equations for each of the spirometric indices for Korean adults

Indices		Males	Females
FEV₁, L	M	exp{−8.38958 + 2.07792 × ln(H) − 0.27470 × ln(A) + m-s}	exp{−7.58567 + 1.91743 × ln(H) − 0.31150 × ln(A) + m-s}
	S	exp{−3.12394 + 0.27285 × ln(A) + s-s}	exp{−2.77822 + 0.18938 × ln(A) + s-s}
	L	1.6945	1.402355
FVC, L	M	exp{−10.46859 + 2.41303 × ln(H) − 0.12074 × ln(A) + m-s}	exp{−8.997001 + 2.146341 × ln(H) − 0.192659 × ln(A) + m-s}
	S	exp{−3.04184 + 0.23702 × ln(A) + s-s}	exp{−2.81158 + 0.18177 × ln(A) + s-s}
	L	1.2576	0.9978
FEV₁/FVC	M	exp{1.651370 − 0.256827 × ln(H) − 0.145848 × ln(A) + m-s}	exp{1.12152 − 0.168862 × ln(H) − 0.121644 × ln(A) + m-s}
	S	exp{−3.65189 + 0.26804 × ln(A) + s-s}	exp{−3.11659 − 0.09128 × ln(A) + s-s}
	L	3.3771	9.4959 − 1.5297 × ln(A) + l-s

Age-specific spline values for males and females are indicated in the Supplementary Materials (see Supplementary Tables 2, 3, 4).

LMS = lambda, mu, sigma, FEV₁ = forced expiratory volume in one second, FVC = forced vital capacity, M = mu as predicted value, S = sigma as coefficient of variation, L = lambda as index of skewness, H = standing height (cm), A = age (years), m-s = M-spline, s-s = S-spline, l-s = L-spline.

Comparisons between reference equations

The new prediction equations and corresponding LLN values for spirometric indices were compared with previous references equations under conditions of fixed height (Figs. 2 and 3). In comparison with findings from the 2013MRA, the predicted values for FEV₁ and FVC were smaller in young adults and larger in the elderly. Each of the LLN values for FEV₁ and FVC demonstrated similar trends, except for FVC in males, which showed subequal values in young adults. The gap between predicted values for FEV₁/FVC and their LLN values appeared wider in young adults and widened progressively with advancing age. For FEV₁/FVC in males, the LLN showed a steep decrease in subjects older than age 60 in contradistinction to the LLN for FEV₁/FVC in females.

Fig. 2

Comparison of reference values between 2017LMS and 2013MRA. (A) FEV₁ in males. (B) FEV₁ in females. (C) FVC in males. (D) FVC in females. (E) FEV₁/FVC in males. (F) FEV₁/FVC in females.

LLN = lower limits of normal, FEV₁ = forced expiratory volume in one second, FVC = forced vital capacity, LMS = lambda, mu, sigma, 2017LMS = the reference values obtained in the present study, pred val = predicted value for each of the spirometric indices, 2013MRA = the reference values derived using multiple regression analysis by Eom and Kim.17

Fig. 3

Comparison of reference values between 2017LMS and 2012GLI. (A) FEV₁ in males. (B) FEV₁ in females. (C) FVC in males. (D) FVC in females. (E) FEV₁/FVC in males. (F) FEV₁/FVC in females.

LLN = lower limits of normal, FEV₁ = forced expiratory volume in one second, FVC = forced vital capacity, LMS = lambda, mu, sigma, 2017LMS = the reference values obtained in the present study, pred val = predicted value for each of the spirometric indices, 2012GLI = the reference values from the Global Lung Function Initiative in 2012.

The disparity between the reference values for FEV₁ and FVC in young adults was more apparent in comparison to these values in the 2012GLI. For elderly males, the gap between predicted values for FEV₁/FVC and their LLN values appeared narrower in the 2012GLI, leading to a greater difference between the LLN of the present study and that of the 2012GLI.

DISCUSSION

Based on data from versions IV and V of the KNHANES, the present study derived new spirometric reference ranges using the novel LMS approach to properly capture the non-linear relationship of age and height with FEV₁, FVC, and FEV₁/FVC across subjects. Reference equations derived by the LMS method are expressed in the exponential form:

Predicted value = e^{{a + b × ln(height) + c × ln(age) + spline}}

whereas conventional regression models take the form:

Y = β₀ + β₁·x_1i + .... + β_p·x_pi + ε_i

Conventional multiple regression analysis builds a model upon several assumptions where variance around the mean across subject height and age ranges is constant, and residuals are normally distributed. However, these assumptions are barely met for distributions of spirometric measures including FEV₁ and FVC, and particularly for skewed FEV₁/FVC distribution.18 Such statistical limitations of conventional multiple regression analysis are overcome using the LMS method.

In GAMLSS, a distribution-based approach to regression models, the exponential family distribution assumption for a variable of interest (Y) is replaced by a general distribution family, which allows modeling both skewness and kurtosis. The LMS method enables coping with non-normal distributions and facilitates the modeling of complex effects of age and height on the dependent variable using splines, which allow the dependent variable to vary non-linearly, (but smoothly) with age.

Another advantage of utilizing the LMS method is that LLN values are independently determined throughout the age-height spectrum, which is facilitated by the sigma and lambda terms.7 According to the analysis, the normal range of the FEV₁/FVC ratio and its components is wider in younger ages and widens progressively in adults older than 40, demonstrating the inappropriateness of a fixed percent value for LLN.7 In contrast, the GOLD group3 favors a fixed ratio of FEV₁/FVC below 0.70 in post-bronchodilator tests for evaluating airflow limitation. In terms of LLN values for FEV₁/FVC in this study, males cross the cutoff value of 0.70 at about 50 years, while females cross the cutoff value at about 60 years. This finding is similar to findings in the 2013MRA. However, findings in the 2012GLI showed higher LLN for males, falling below the cutoff value of 0.70 at about 60 years. There seems to be ethnic diversity, taking into account findings in Caucasians in which LLN values were shown to cross the cutoff value at about 50 years of age.12 In contrast, a study from Japan reported greater LLN values in both males and females, crossing the cutoff value of 0.70 at about 60 years in males and 70 years in females.9 In the present study, the LLN of FEV₁/FVC falls steeply in male subjects who are older than 60 years. Among the subjects included for analysis, the number of males aged 70 years and above was 152, while the number of females was 1,028 (see Supplementary Table 1 in the Supplementary Materials). Since LLN is independently determined throughout the age-height spectrum with the LMS method, the small number of older males may have led to lower LLN values. This finding needs to be clarified with a greater amount of Korean data in further research.

Small differences were found between the present study and the 2013MRA, particularly in young and old adults. Additional to the KNHANES V, the present study utilized the KNHANES IV to gather spirometric data about adults younger than 40 years of age, because spirometric data were restricted to adults between 40 and 79 years of age in the KNHANES V. By merging data from the KNHANES IV and V, a greater number of older subjects were included in this study than in the 2013MRA. The larger number of subjects herein better captured the decline in lung function with advancing age. However, a relatively small number of younger subjects were included for analysis, resulting in less power to claim representativeness. Additionally, different methods between studies may have also contributed to the incongruity between the current analysis and the 2013MRA.

In comparing the findings of this study with the 2012GLI, partial discrepancies are shown to be more notable in young adults. Eom and Kim17 reported lower predicted values of spirometric indices derived from the KNHANES IV data in comparison to previous reference equations derived from the KNHANES II.19 Results from Eom and Kim17 also showed lower predicted values of FEV₁ and FVC in comparison to these values in the US NHANES III20 and in Morris et al.,21 particularly in young adults. One possible explanation for the differing results refers to the different exclusion criteria between studies. In analysis of the KNHANES II,19 stricter exclusion criteria concerning medical conditions were applied, while individuals with abnormal chest radiography findings were not considered in the analysis of the KNHANES IV.17 To overcome potential bias, individuals with underlying respiratory diseases or symptoms, as well as with positive findings on chest radiography were excluded in the present study. However, the result of seeing lower predicted values for FEV₁ and FVC in young adults was repeated. This may be due to small number of younger subjects, as indicated above.

A more likely explanation is attributable to properties of the 2012GLI reference values regarding North East Asians. Quanjer et al.12 reported significant differences for spirometric data between centers in East Asia. In the 2012GLI, predicted values for FEV₁ and FVC derived from two datasets collected in North China22 and Korea19 were significantly higher than predicted values derived from other datasets collected in Hong Kong,23 24 Taiwan,25 Thailand,26 the USA (according to the Multi-Ethnic Study of Atherosclerosis [MESA] investigators R.G. Barr, P.L. Enright, and J.L. Hankinson wherein the data was personally communicated by Quanjer et al.12) and China.27 The differences between datasets led to the development of separate reference equations for East Asians from the North and the South. The dataset collected in Korea was based on the KNHANES II survey, which was conducted from November 2001 to January 2002. Eom and Kim17 indicated that predicted values derived from the KNHANES II19 were noticeably higher than values derived from the KNHANES IV17 and from the US NHANES III.20 In comparison to another spirometric dataset collected from a region within the Chungbuk province in Korea, the US NHANES III model showed a better fit than the KNHANES II model.28 Based on these findings, Eom et al.17 28 asserted that the KNHANES II equations overestimated the predicted values for FEV₁ and FVC. Since 2012GLI data is partially comprised of KNHANES II data, predicted values from the 2012GLI were expected to be greater than the values of the present study. Nevertheless, the predicted values for FVC in the 2012GLI show a smaller decline than the values of the present study for male subjects between 40 and 50 years of age. This might be due to the mixed effects of including North Chinese along with Korean populations in the 2012GLI dataset for North East Asians. Among several possible explanations, disagreement in measured values between datasets is the most likely reason for the inconsistent findings, but there is no clear proof for such discrepancies in measure values between the data. Quanjer et al.12 reported that no evidence of unrepresentativeness arising from a small sample size was found in the KNHANES II data19 or in the data from North China.22 These ambiguous findings need to be investigated in further research.

According to Quanjer et al.,12 the GLI has produced spirometric reference values that are more generalizable across populations by utilizing many data from various sources. However, more generalizable reference values may not necessarily mean that the values are optimal to describe the spirometric characteristics of an ethnic group. In East Asia alone, there are many subgroups and ethnic minorities differ from each other in physical, social and ecological traits. In addition to age and height, these traits may have correlations with pulmonary function. In Japan, a study reports a sample with greater height driven by increased leg length due to post-war improvements in socioeconomic conditions.29 In sample populations in China, there are considerable differences in socioeconomic conditions between urban and rural areas30 31 and ethnic differences in body builds.32 33 Since certain traits are not shared across ethnic groups and regions, the development of separate spirometric equations for each ethnicity and country is needed. Nevertheless, the work from the GLI has advantages in certain situations. The GLI findings could be adopted to evaluate respiratory diseases in foreigners or adolescents, where representative reference values are absent.

The major weakness of the analysis in this study arise from the properties of the KNHANES data. Because the reference values are calculated independently throughout the age-height spectrum, the findings might be less reliable in ages for which the number of subjects is relatively small. In this study, the number of young subjects was small, and caution is required when interpreting the predicted values for subjects younger than 30 years of age. However, it is clear that the new reference values derived by the LMS method better reflect transitions in pulmonary function in young adults relative to previous reference equations derived by conventional regression analysis.11 Another limitation is that the KNHANES IV data utilized in this study is not absolutely free from birth-cohort effects, since the data collection began in July 2007 and ended in December 2009. Korea is one of the fastest growing countries in terms of the standing height of its population (see Fig. E7 in the online supplementary material of the article by Quanjer et al.12). For example, the average height in subjects aged 40 years in 2008 may differ from the average height in subjects aged 40 years in 2017. In order to develop valid and reliable prediction equations for spirometry, data needs to be periodically updated and collected from a sufficient number of subjects with acceptable representativeness. Furthermore, representative data for Korean adolescents should be collected along with adults to derive continuous reference values for spirometry with transition into adulthood from childhood.

In conclusion, the present study provides new reference values for spirometry and corresponding LLN values using the LMS method, better capturing the transition of pulmonary function in young adults. The results should be interpreted with caution for young adults as well as for the LLN of FEV₁/FVC in elderly males in particular. Considering that clinical evaluation is an art, following spirometry assessment over time and correlating the findings with other exams is important. Finally, future studies should be conducted to improve the accuracy of reference data and to develop continuous reference values for spirometry across all ages.

ACKNOWLEDGMENTS

The authors would like to thank M. Kubota, Kitasato University, Kanagawa, Japan, for his kind advice on the analysis of this study. The authors also would like to thank all subjects and faculties who participated in the KNHANES IV and V.

Notes

^Disclosure The authors have no potential conflicts of interest to disclose.

^{Author Contributions}

Conceptualization: Jo BS, Myong JP, Kim HR.
Data curation: Jo BS.
Formal analysis: Jo BS.
Funding acquisition: Kim HR.
Investigation: Jo BS, Myong JP.
Writing - original draft: Jo BS.
Writing - review & editing: Rhee CK, Yoon HK, Koo JW.

References

1. Miller MR, Hankinson J, Brusasco V, Burgos F, Casaburi R, Coates A, et al. Standardisation of spirometry. Eur Respir J. 2005; 26(2):319–338.

2. Pellegrino R, Viegi G, Brusasco V, Crapo RO, Burgos F, Casaburi R, et al. Interpretative strategies for lung function tests. Eur Respir J. 2005; 26(5):948–968.

3. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease. Updated 2017. Accessed June 20, 2017. http://www.goldcopd.org.

4. van Dijk W, Tan W, Li P, Guo B, Li S, Benedetti A, et al. Clinical relevance of fixed ratio vs lower limit of normal of FEV1/FVC in COPD: patient-reported outcomes from the CanCOLD cohort. Ann Fam Med. 2015; 13(1):41–48.

5. Güder G, Brenner S, Angermann CE, Ertl G, Held M, Sachs AP, et al. GOLD or lower limit of normal definition? A comparison with expert-based diagnosis of chronic obstructive pulmonary disease in a prospective cohort-study. Respir Res. 2012; 13(1):13.

6. Lung function testing: selection of reference values and interpretative strategies. American Thoracic Society. Am Rev Respir Dis. 1991; 144(5):1202–1218.

7. Culver BH. How should the lower limit of the normal range be defined? Respir Care. 2012; 57(1):136–145.

8. Xu X, Laird N, Dockery DW, Schouten JP, Rijcken B, Weiss ST. Age, period, and cohort effects on pulmonary function in a 24-year longitudinal study. Am J Epidemiol. 1995; 141(6):554–566.

9. Kubota M, Kobayashi H, Quanjer PH, Omori H, Tatsumi K, Kanazawa M, et al. Reference values for spirometry, including vital capacity, in Japanese adults calculated with the LMS method and compared with previous values. Respir Investig. 2014; 52(4):242–250.

10. Stanojevic S, Wade A, Cole TJ, Lum S, Custovic A, Silverman M, et al. Spirometry centile charts for young Caucasian children: the Asthma UK Collaborative Initiative. Am J Respir Crit Care Med. 2009; 180(6):547–552.

11. Stanojevic S, Wade A, Stocks J, Hankinson J, Coates AL, Pan H, et al. Reference ranges for spirometry across all ages: a new approach. Am J Respir Crit Care Med. 2008; 177(3):253–260.

12. Quanjer PH, Stanojevic S, Cole TJ, Baur X, Hall GL, Culver BH, 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(6):1324–1343.

13. Kweon S, Kim Y, Jang MJ, Kim Y, Kim K, Choi S, et al. Data resource profile: the Korea National Health and Nutrition Examination Survey (KNHANES). Int J Epidemiol. 2014; 43(1):69–77.

14. McMahon T. Size and shape in biology. Science. 1973; 179(4079):1201–1204.

15. Rigby RA, Stasinopoulos DM. Generalized additive models for location, scale and shape. J R Stat Soc Ser C Appl Stat. 2005; 54(3):507–554.

16. Cole TJ, Green PJ. Smoothing reference centile curves: the LMS method and penalized likelihood. Stat Med. 1992; 11(10):1305–1319.

17. Eom SY, Kim H. Reference values for the pulmonary function of Korean adults using the data of Korea National Health and Nutrition Examination Survey IV (2007–2009). J Korean Med Sci. 2013; 28(3):424–430.

18. Cole TJ, Stanojevic S, Stocks J, Coates AL, Hankinson JL, Wade AM. Age- and size-related reference ranges: a case study of spirometry through childhood and adulthood. Stat Med. 2009; 28(5):880–898.

19. Choi JK, Paek D, Lee JO. Normal predictive values of spirometry in Korean population. Tuberc Respir Dis. 2005; 58(3):230–242.

20. Hankinson JL, Odencrantz JR, Fedan KB. Spirometric reference values from a sample of the general U.S. population. Am J Respir Crit Care Med. 1999; 159(1):179–187.

21. Morris JF, Koski A, Johnson LC. Spirometric standards for healthy nonsmoking adults. Am Rev Respir Dis. 1971; 103(1):57–67.

22. Wu Y, Zhang Z, Gang B, Love EJ. Predictive equations for lung function based on a large occupational population in North China. J Occup Health. 2009; 51(6):471–477.

23. Ip MS, Karlberg EM, Karlberg JP, Luk KD, Leong JC. Lung function reference values in Chinese children and adolescents in Hong Kong. I. Spirometric values and comparison with other populations. Am J Respir Crit Care Med. 2000; 162(2 Pt 1):424–429.

24. Ip MS, Ko FW, Lau AC, Yu WC, Tang KS, Choo K, et al. Updated spirometric reference values for adult Chinese in Hong Kong and implications on clinical utilization. Chest. 2006; 129(2):384–392.

25. Pan WH, Chen JY, Haung SL, Liou TL, Lee TK, Wang LY, et al. Reference spirometric values in healthy Chinese neversmokers in two townships of Taiwan. Chin J Physiol. 1997; 40(3):165–174.

26. Dejsomritrutai W, Nana A, Maranetra KN, Chuaychoo B, Maneechotesuwan K, Wongsurakiat P, et al. Reference spirometric values for healthy lifetime nonsmokers in Thailand. J Med Assoc Thai. 2000; 83(5):457–466.

27. Zhang QL, Zheng JP, Yuan BT, He H, Wang J, An JY, et al. Feasibility and predicted equations of spirometry in Shenzhen preschool children. Zhonghua Er Ke Za Zhi. 2005; 43(11):843–848.

28. Eom SY, Moon SI, Yim DH, Lee CH, Kim GB, Kim YD, et al. Goodness-of-fits of the spirometric reference values for Koreans and USA caucasians to spirometry data from residents of a region within Chungbuk province. Tuberc Respir Dis. 2012; 72(3):302–309.

29. Tanner JM, Hayashi T, Preece MA, Cameron N. Increase in length of leg relative to trunk in Japanese children and adults from 1957 to 1977: comparison with British and with Japanese Americans. Ann Hum Biol. 1982; 9(5):411–423.

30. Morgan SL. Richer and taller: stature and living standards in China, 1979–1995. China J. 2000; (44):1–39.

31. Etler DA. Recent developments in the study of human biology in China: a review. Hum Biol. 1992; 64(4):567–585.

32. Ashizawa K, Tanamachi N, Kato S, Kumakura C, Zhou X, Jin F, et al. Growth of height and leg length of children in Beijing and Xilinhot, China. Anthropol Sci. 2008; 116(1):67–76.

33. Weitz CA, Garruto RM, Chin CT, Liu JC. Morphological growth and thorax dimensions among Tibetan compared to Han children, adolescents and young adults born and raised at high altitude. Ann Hum Biol. 2004; 31(3):292–310.