Spirometry Centile Charts for Young Caucasian Children: The Asthma UK Collaborative Initiative

Spirometry centile charts for young children 1

Spirometry Centile Charts for Young Caucasian Children: The

Asthma UK Collaborative Initiative

Sanja Stanojevic1,2

, Angie Wade2, Tim J Cole

2, Sooky Lum

1, Adnan Custovic

3, Mike

Silverman4, Graham L Hall

5, Liam Welsh

1, Jane Kirkby

1, Wenche Nystad

6, Monique

Badier7, Stephanie Davis

8, Steven Turner

9, Pavilio Piccioni

10, Daphna Vilozni

11,

Howard Eigen12

, Helen Vlachos-Mayer13

, Jinping Zheng14

, Waldemar Tomalak15

,

Marcus Jones,16

John L Hankinson17

and Janet Stocks1 on behalf of the Asthma UK

collaborative group

1 Portex Unit: Respiratory Physiology and Medicine, University College London:

Institute of Child Health, London, UK 2 Medical Research Council: Center of Epidemiology for Child Health, University

College London: Institute of Child Health, London UK 3University of Manchester and National Institute of Health Research Translational

Research Facility in Respiratory Medicine, University Hospital of South Manchester,

Manchester, UK 4 Child Health, Institute for Lung Health, University of Leicester, Leicester, UK

5Respiratory Medicine, Princess Margaret Hospital and School of Paediatric and Child

Health, University of Western Australia, Perth, Australia. 6 Norwegian Institute of Public Health, Division of Epidemiology, Oslo, Norway

7 Lung Function Laboratory, Hôpital Sainte Marguerite, Marseille, France

8Department of Pediatrics; North Carolina Children's Hospital; University of North

Carolina at Chapel Hill; Chapel Hill, North Carolina 9Academic Child Health, University of Aberdeen, Aberdeen Scotland

10National Health Service, Pneumology, Turin, Italy

11 Edmond and Lili Safra Children's Hospital, Sheba Medical Center, Tel Aviv, Isreal

12 Section of Pediatric Pulmonology, Riley Hospital for Children, Indianapolis,

Indiana, USA 13

Division of Respiratory Medicine, Department of Pediatrics, Center Hospitalier

Universitaire de Sherbrooke-Hôpital Fleurimont, University of Sherbrooke,

Sherbrooke, Canada 14

State Key Lab of Respiratory Disease, 1st Affiliated Hospital of Guangzhou Medical

University, Guangzhou, China 15

Department Physiopathology of Respiratory System, Institute for TBC and Lung

Diseases, Rabka Branch, Poland 16

Division of Pediatric Pulmonary, Hospital São Lucas Pontifícia Universidade

Católica do RGS Porto Alegre, Brazil

Page 1 of 35 AJRCCM Articles in Press. Published on July 2, 2009 as doi:10.1164/rccm.200903-0323OC

Copyright (C) 2009 by the American Thoracic Society.


17 Hankinson Consulting, Inc., Valdosta, Georgia

Running Title: Spirometry centile charts for young children

Descriptor # 43 Pulmonary Function Testing

Word Count: 3525

Scientific Knowledge: Preschool lung function testing is relatively novel and clinical

use of spirometry in this age group is further limited by the lack of appropriate

reference data.

Adds to knowledge: The Asthma UK collaborative initiative has collated all

available reference data for spirometry in young children. These data have been used

to produce reference equations which are smoothly joined to existing all-age reference

equations and may improve interpretation of spirometry in early childhood.

Funding: SS, LW and JK funded by Asthma UK, TJC was funded by UK MRC grant

number G9827821, JK and SL funded by UK MRC grant number G0401525, the

Australian data collection was funded by NHMRC (Australia), MHJ was funded by

CAPES/CNPq, data provided by SD funded Cystic Fibrosis Foundation (Grant

Support), and Cystic Fibrosis Foundation Therapeutics, Incorporated (Grant Support)

and Cystic Fibrosis Therapeutic Development Network.

Corresponding Author:

Dr Sanja Stanojevic

Portex Unit: Respiratory Physiology and Medicine

UCL, Institute of Child Health

30 Guilford Street

London, WC1N 1EH

[email protected]

Telephone: +44 (0) 20 7905 2759

Fax: +44 (0) 20 7829 2793

Page 2 of 35

mailto:[email protected]


Abstract:

Rationale: Advances in spirometry measurement techniques have made it possible to

obtain measurements in children as young as 3 years of age; however, in practice,

application remains limited by the lack of appropriate reference data for young

children, which are often based on limited population-specific samples.

Objectives: We aimed to build on previous models by collating existing reference

data in young children (aged 3-7 years), to produce updated prediction equations that

span the preschool years and that are also linked to established reference equations for

older children and adults.

Methods: The Asthma UK Collaborative initiative was established to collate lung

function data from healthy young children aged 3-7 years. Collaborators included

researchers with access to pulmonary function test data in healthy preschool children.

Spirometry centiles were created using the LMS (Lambda-Mu-Sigma) method and

extend previously published equations down to 3 years of age.

Main Results: The Asthma UK centiles charts for spirometry are based on the largest

sample of healthy young Caucasian children aged 3-7 years (n=3777) from 15 centers

across 11 countries and provide a continuous reference with a smooth transition into

adolescence and adulthood. These equations improve existing pediatric equations by

considering the between-subject variability to define a more appropriate age-

dependent lower limit of normal. The collated dataset reflects a variety of equipment,

measurement protocols and population characteristics and may be generalizable

across different populations.

Conclusions: We present prediction equations for spirometry for preschool children

and provide a foundation which will facilitate continued updating.

Page 3 of 35


Abstract word count: 251

Keywords: spirometry, pulmonary function tests, reference values, preschool

Page 4 of 35


Introduction:

Advances in spirometry measurement techniques have made it possible to obtain

measurements in children as young as 3 years of age.1-3

These advances make it

possible to assess lung function continuously from early childhood through adulthood.

In practice, clinical application remains limited by the lack of appropriate reference

data for young children <5 years of age. Where datasets do exist, there is often a lack

of detail regarding the sample population and measurement or analysis techniques

and/or the use of inappropriate statistical methods.4 Furthermore, arbitrary break

points between preschool and school age equations often lead to shifts in predicted

values and confusion when it comes to clinical management as preschool children

move into the school age range.

Ideally, spirometry reference values would be derived from a prospective multi-

center, multi-national study which measured spirometry under standard conditions

and protocols. However, such a comprehensive study would require an extensive

collaborative network and would be logistically and financially difficult to conduct.

Furthermore, this scenario may not reflect the diverse nature either of the equipment

and software used, or of the extent to which adherence to recommended procedures

and quality control occurs in clinical practice.

The Third National Health and Nutrition Examination Survey (NHANES III) dataset

is one of the few large and representative reference populations spanning different

ethnic groups over a wide age range; however the original publication was limited to

subjects ≥8 years of age.5 It was recently demonstrated that it is possible to collate

reference data from different sources to produce a single smoothly changing model

Page 5 of 35


from 4-80 years.6, 7

These “original all-age” equations explained the complex age

related changes and variability to improve interpretation during childhood and at the

transition to early adulthood. Nevertheless these were still limited to <300 subjects <

8 years of age and hence only provided an interim solution. The aim of this study was

to build on the previous models by collating reference data from young children (aged

3-7 years), to produce updated prediction equations that not only span the preschool

years, but are also linked to established reference equations for older children and

adults. Preliminary results have been presented at the European Respiratory Society

Congress and the British Thoracic Society.8-10

Methods:

Recruitment:

The Asthma UK Collaborative initiative was established to collate lung function data

from healthy young children (www.growinglungs.org.uk) for three techniques:

Spirometry, Interrupter Resistance Technique and Specific Airway Resistance. This

paper will focus on spirometry. Initially the collaborative group comprised members

of the ATS/ERS pediatric pulmonary function test task force. Subsequently,

collaborators were identified by: systematically searching PubMed, advertising at

international respiratory conferences, through membership bulletins, word of mouth

and by hand searching relevant respiratory periodicals.

Data Collection:

Spirometry data were collected in healthy children aged 3-7years. All participating

centers were asked to provide detailed information about reasons for data collection

(i.e. hospital controls, population cohort etc.), recruitment details, population

Page 6 of 35

http://www.growinglungs.org.uk/


characteristics, and equipment and measurement protocols. All data were anonymized

prior to submission to this initiative and came from research studies where full local

ethics approval and informed parental consent had been obtained.

Quality Control:

Quality control (QC) criteria for preschool children were not formally published until

20073; prior to this individual centers applied their own adaptations of adult criteria.

11

To evaluate QC in this retrospective dataset every attempt was therefore made to visit

each collaborating center to conduct inter-lab comparisons and obtain a random

sample of original flow-volume curves. Where original data were not available, a

random sample of flow-volume curves from recent clinical patients was obtained. We

developed modified age-specific QC criteria based upon published recommendations2,

3 (Table 1) and applied these to the random sample flow-volume curves . Each of the

flow-volume curves collected was assessed by a respiratory physiologist (LW/JK) and

graded out of five using a standardized over-read sheet. Each curve was scored with a

1 or 0 for the five criteria: 1) rapid onset of expiration, 2) clearly defined peak

expiratory flow, 3) no cough or glottic closure, 4) end expiratory plateau, and 5)

repeatability. The over-read sheet can be found at www.growinglungs.org.uk

Exclusions:

Initially the Asthma UK population consisted of 7,983 observations spanning 3-7

years of age but the centers differed with respect to inclusion criteria, population

characteristics, equipment and measurement protocol. In order to make the collated

dataset more homogeneous, 4206 exclusions were made based on predefined criteria

(Figure 1); details can be found in the On-Line supplement (OLS).

Page 7 of 35



Outcomes measured:

Data were collated for all available spirometry outcomes, but centile charts were only

developed for the outcomes collected in the NHANES III survey (FEV1, FVC, FEF25-

75 and FEV1/FVC) to facilitate interpretation across all ages. As part of the original

all-age study,6 FEV0.75 was calculated by one of the authors (JH) but was not included

in the statistical modelling as it was limited to subjects ≥ 8 years of age. These data

were combined with the preschool FEV0.75 data from the current study to construct

FEV0.75 and FEV0.75/FVC centiles from 3-17 years. It should be noted that not all

centers recorded the same outcomes; hence the number of observations varied for

each outcome and are not directly comparable.

Statistical Analysis:

The centile charts were constructed as described previously6 using the LMS (Lambda,

Mu, Sigma) method;12

details can also be found in the OLS.

Prediction Models:

The fitted models provide sex-specific, height-and-age adjusted values for the median,

CV and skewness. The median (M) is the predicted value for the individual, which

together with (S), the co-efficient of variation (CV) and (L), the skewness allows the

individual’s measurement to be converted to a z-score or percent predicted. The

prediction equations and look-up tables are available in the OLS. Individual

spirometry results or retrospective analysis of an entire dataset can be interpreted

using a user-friendly Excel add-in program, available at www.growinglungs.org.uk.

The equations can also be installed into commercially available equipment upon

request from any manufacturer (see also www.lungfunction.org).

Page 8 of 35



Results:

The final pre-school population consisted of 3777 observations in children aged 3-7

years from 15 centers and 11 countries, collected between 1993 and 2008 (Figure E1,

Table 2). There were 562 children <four years of age, 1289 four year olds, 1143 five

year olds, 614 six year olds and 169 children older than seven years but less than

eight. These were combined with the original all-age dataset, described previously,6 to

create spirometry centiles from 3-80 years based on 7209 Caucasian subjects (n=3298

male, n=3500 female). The results presented here focus on the updated pediatric

reference equations as the equations remain virtually the same above 10 years of age

(Table E2 OLS).6 Comparison of the between-center differences can be found in the

OLS.

Median predicted value:

A worked example of the equations can be found in the OLS. Height was the

strongest predictor of all outcomes and the relationship between height and

spirometric lung function was described as allometric, such that each outcome is

expressed as height raised to a power (Figure 1). The power ranged from 2-3

depending on outcome, which arose from the proportional model where both the

outcome and height variables were logarithmically transformed. There were minimal

differences between males and females until early puberty.

After taking height into account, there was an independent and complex relationship

with age, which emerged in early adolescence. The age relationship for FEV0.75 was

the simplest, but was based on a relatively small sample (n=1066 for girls and n=998

for boys) from 9 centers and limited age range (3-17 years). Girls and boys had

Page 9 of 35


similar predicted values for FEV0.75 until early puberty (13 years of age), after which

boys had greater FEV0.75 for a given height. Similarly for FEV1 male and female

curves diverged during early puberty, when males demonstrated a sharper and more

complex transition. The median FVC curves were very similar to those observed for

FEV1 in both sexes. FEF25-75 was relatively greater in males and also demonstrated

greater changes during puberty in males.

Between-subject variability:

The between-subject variability (% CV) was expressed as a function of age. In males

the %CV for FEV0.75 decreased linearly with age from 13% at 3 years of age to 11%

at 17y in males; the %CV was similar in females but declined at a faster rate. Thus the

normal range for FEV0.75 in males was from 74-126% predicted for a 3 year old and

from 78-122% predicted in a 17 year old. The %CV was greater for FEV1 but similar

in both sexes, ranging from 17% in the youngest children to 11% at 20 years (Figure

2) (i.e. Lower Limit of Normal (LLN) 66% predicted at 3 years and 78% predicted by

20 years). The variability for FVC was similar to that for FEV1 but slightly greater in

younger children. The CV was greatest for FEF25-75, decreasing from 27 to 24%

across the age range; thus the normal range for FEF25-75 was between 46-154%

predicted in a 3 year old and 52-148% predicted in a 25 year old.

Skewness:

Although there was no evidence of skewness for FEV0.75, FEV1 or FVC, even after

logarithmic transformation of FEF25-75 it was necessary to incorporate skewness (L) to

define a more accurate LLN.

Page 10 of 35


FEV1/FVC:

The highly skewed and truncated nature of the FEV1/FVC ratio was accounted for in

the modelling. The median predicted value for FEV1/FVC and the LLN is illustrated

in Figure 3 and demonstrates that in younger and shorter children the median ratio is

greater than 0.90, and virtually 1.0 at three years of age. Females have a greater

predicted ratio compared to males, the difference being greatest after 10 years of age.

Quality Control:

A random sample of 126 flow-volume curves was obtained from nine of the 19

original centers; samples were not available in the other centers because equipment

was no longer available or results were stored numerically. Of the 126 results, fifteen

(11.9%) were flagged as questionable, of which nine (of 20 collected) were from one

center. This center had also been identified as an outlier on initial examination of the

data and was excluded. The average QC score (Table 1) for the sample of subjects

was 4.1/5, which suggests that the centers which provided flow-volume curves

generally contributed good quality data; however, since less than 50% centers

provided flow-volume curves these findings cannot be generalized to the entire study

population. The success rate also varied for each of the criteria; 95% success for

‘rapid onset of expiration’, 89% for ‘clearly defined peak expiratory flow’, 81%

success for ‘no cough or glottic closure’ and 71% for repeatability.

Analysis of the biological control data demonstrated that even for experienced adult

subjects there was considerable variability between the centers; the within-subject

difference between FEV1 measured at the Institute of Child Health and the

collaborating site ranged from -20% to 6%. The greatest difference was observed in

Page 11 of 35


an asthmatic control who was not known to have an exacerbation at time of test. After

excluding this control, differences ranged from -6% to 6%. Greater differences

between the two sites were observed for FVC, even when the asthmatic subject was

excluded (range -3% to 11.4%).

Discussion:

The Asthma UK centiles charts for spirometry, which are based on the largest sample

of healthy children <7 years of age, also link seamlessly with an all-age reference,

which will facilitate interpretation of spirometry results from 3–80 years. These

equations improve existing pediatric equations by considering the increased between-

subject variability in younger children to define a more appropriate age-dependent

LLN. The collated dataset reflects a variety of equipment, measurement protocols and

population characteristics and may be generalizable across different populations.

This study updates previously published equations6 to improve interpretation in young

children but does not change the interpretation of results in subjects above 10

years.(see OLS) Since there is little differences between the two equations after the

age of 10 years, it could be argued that the original all-age reference would suffice,

especially for centers who rarely study younger subjects. However, since preschool

spirometry is becoming increasingly popular and the original all-age equations

included relatively few children <6 years of age, we would strongly advise the

implementation of these updated equations to facilitate interpretation of results across

the entire age range. Furthermore, the availability of reference equations for FEV0.75

may facilitate more appropriate assessment of lung function in young children who

cannot produce an FEV1. The FEV0.75 sample was, however, considerably smaller

than that for the other outcomes which may have limited our ability to exposure age

Page 12 of 35


and sex differences. In centers that measured both outcomes, we did not find a

significant difference in FEV0.75 between those who did and did not obtain an FEV1;

however these comparisons may have been underpowered (data not shown). The

centiles developed for the FEV1/FVC and FEV0.75/FVC ratios complement the age

and sex related changes observed previously13, 14

and adjust for the highly skewed and

truncated distribution to provide a more accurate description of the LLN.

Given the variability in spirometry due to equipment, measurement technique and/or

population differences, it is unrealistic that any population-specific reference will fit

all centers perfectly. The current study provides generalizable results, which reflect

this variability; however generalizability may come at the cost of sensitivity.

Interpretation of results within the clinical setting should not be affected as long as

emphasis is placed on how a subject’s lung function tracks with time, rather than

simply whether a result falls just below or above the LLN. In research studies

misinterpretation can be avoided if a control group is measured. Nonetheless, the

observed between-center differences highlight the need for more standardized

equipment and measurement protocols.

When available, quality control data suggested that most centers provided high quality

data for young children, with an average score of 4.1/5; however some flow-volume

curves were classified as “unacceptable” emphasizing the need for a standard

pediatric quality scoring system.15

More specifically the low scores for cough or

glottic closure are worrying since this may indicate poor patient technique and

unreliable results. Furthermore, despite using a more liberal criterion of 10% for

repeatability, as recommended for younger children,3 only 71% of traces reviewed

Page 13 of 35


met this criterion, reflecting the difficulties in obtaining repeatable results in this age

group. Given the difficulties associated with repeatability in preschool children, test

results should not be excluded solely on the basis of poor repeatability.

Study Limitations:

While the current study improves on previous work it is not without its limitations.

There were insufficient data to present equations for FEV0.5 and the smaller sample

size for FEV0.75 may limit identification and interpretation of any sex differences for

this variable. Like most other reference ranges, this study was restricted to cross-

sectional data, which limits interpretation of longitudinal changes in individuals. The

current dataset also reflects the limited data for non-Caucasian subjects in this age

group. This serious deficit requires urgent attention in the near future, ideally by

studying multi-ethnic populations with identical equipment and protocols, and a more

sophisticated approach to characterizing body shape and size so that global equations

applicable across a wide range of ethnic groups can be produced, rather than further

population-specific equations.

The retrospective approach made it difficult to distinguish differences attributed to

equipment and measurement techniques from real population differences. In the

future, use of ATS/ERS guidelines for preschool spirometry 3 may reduce some of

the inter-center variability observed in this study. The retrospective nature of the data

collection also limited our ability to define “ideal” lung function as we had no control

over original exclusion/inclusion criteria. Selection of subjects to comprise the

reference population is an important step in defining ‘normal’ lung function; however,

health is difficult to define and the choice of inclusion/exclusion criteria relies on the

Page 14 of 35


intended purpose of the reference population. Hence there will inevitably be

differences between predicted values derived from studies that have selected from the

general population and those with more refined exclusion criteria. In this study, we

were unable to ascertain whether any potential differences between centers that

explicitly excluded asthmatics and those that did not could actually be attributed to

asthma or were due to some other factors common to those centers. While we did find

a slightly lower mean zFEV1 in those centers that did not explicitly exclude

asthmatics, (mean (95% CI): 0.17z-scores (-0.11; -0.22)); this difference is not

clinically relevant (equivalent to a mean difference of < 2-3%). A balance of

appropriate inclusion/exclusion criteria is necessary to select an unbiased and

generalizable reference population which is representative of health and can

discriminate between health and disease.

Future Directions:

The current study provides a foundation and a significant step towards an

internationally representative all-age reference population. Future efforts should

include data from non-Caucasian and elderly subjects >80 years of age, other

pulmonary function tests (i.e. static lung volumes, diffusion capacity, multiple breath

washout etc.) and longitudinal data (ideally measured over both short-term [2-3w] and

longer [6-12 month] intervals). The pediatric equations could also be expanded to

include outcomes such as FEV0.5, FEF50 and FEF75 and Area Under the Curve (Aex).16

.

Future studies should aim to record the characteristics of the study population (OLS

Table E4) to avoid exclusion of too many subjects, which may lead to over-sensitive

reference ranges. Current efforts are underway to establish an ERS task force for

Page 15 of 35


reference ranges, which would provide a forum to address these issues

(www.lungfunction.org).

Conclusions:

This study presents prediction equations for Caucasian preschool children which are

also linked seamlessly to existing adult equations and provide a single reference with

which to interpret spirometry results across all ages. There is however still a need to

expand these equations to encompass other ethnic groups and other pulmonary

function tests, such as lung volumes and transfer factor.

Acknowledgements:

The authors would like to thank the children and families who participated in the

original studies, and the investigators involved in the original data collection: Mark

Rosenthal (UK), Mary Corey (Canada), Patrick Lebecque (Belgium), M Bugiani

(Italy), Margaret Rosenfeld (USA) and Gwendolyn S. Kerby (USA), Meeghan Hart,

(USA), Robert Castile (USA), Oscar Mayer (USA), Peter Hiatt (USA), Clement Ren

(USA), Lyle Palmer (Australia), Peter Rye (Australia) and Neil Gibson (Australia),

Jack Goldblatt (Australia), Peter LeSouef (Australia), Garth Kendall (Australia),

Anne Callaghan (Australia), Raewyn Mutch (Australia), Jennifer Kent (Australia),

Merci Kusel (Australia), Mario Eddy Dumas (Canada), Jean-Paul Praud (Canada), the

Sherbrooke Pediatric Spirometry Work Group (Canada), Leslie Lowe (UK), Caroline

Beardsmore (UK), Viviane Viegas Rech (Brazil), Paula Cristina Vasconcellos Vidal

(Brazil) Hilário Teixeira de Melo Júnior (Brazil), Chantal Guillot (France) and Alois

Zapletal (Czech Republic).

Page 16 of 35


References: 1. Eigen H, Bieler H, Grant D, Christoph K, Terrill D, Heilman DK, Ambrosius

WT, Tepper RS. Spirometric pulmonary function in healthy preschool children. Am J

Respir Crit Care Med 2001;163(3):619-23.

2. Aurora P, Stocks J, Oliver C, Saunders C, Castle R, Chaziparasidis G, Bush A.

Quality control for spirometry in preschool children with and without lung disease.

Am J Respir Crit Care Med 2004;169(10):1152-9.

3. Beydon N, Davis SD, Lombardi E, Allen JL, Arets HG, Aurora P, Bisgaard H,

Davis GM, Ducharme FM, Eigen H, Gappa M, Gaultier C, Gustafsson PM, Hall GL,

Hantos Z, Healy MJ, Jones MH, Klug B, Lodrup Carlsen KC, McKenzie SA, Marchal

F, Mayer OH, Merkus PJ, Morris MG, Oostveen E, Pillow JJ, Seddon PC, Silverman

M, Sly PD, Stocks J, Tepper RS, Vilozni D, Wilson NM. An Official American

Thoracic Society/European Respiratory Society Statement: Pulmonary Function

Testing in Preschool Children. Am J Respir Crit Care Med 2007;175(12):1304-45.

4. Stanojevic S, Wade A, Lum S, Stocks J. Reference equations for pulmonary

function tests in preschool children: A review. Pediatr Pulmonol 2007;42(10):962-72.

5. 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-

87.

6. Stanojevic S, Wade A, Stocks J, Hankinson J, Coates AL, Pan H, Rosenthal

M, Corey M, Lebecque P, Cole TJ. Reference ranges for spirometry across all ages: a

new approach. Am J Respir Crit Care Med 2008;177(3):253-60.

7. 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):880-98.

8. Stanojevic S, Lum S, Wade A, Cole TJ, Custovic A, Silverman M, Stocks J.

An International Collaborative Initiative to Develop Reliable Reference Ranges for

Pulmonary Function in Young Children. Thorax 2006;61(Supplement II):ii94.

9. Stanojevic S, Lum S, Wade A, Cole TJ, Custovic A, Silverman M, Stocks J.

Collating International Reference Data for Lung Function in Young Children. Eur

Respir J 2006;28 (Supplement 50):834s.

10. Stanojevic S, Wade A, Lum S, Cole TJ, Custovic A, Silverman M, Hall GL,

Badier M, Nystad W, Stocks J. International reference ranges for spirometry for

young children Eur Respir J 2007;30 (Supplement 31):723S.

11. Pellegrino R, Decramer M, van Schayck CP, Dekhuijzen PN, Troosters T, van

Herwaarden C, Olivieri D, Del Donno M, De Backer W, Lankhorst I, Ardia A.

Page 17 of 35


Quality control of spirometry: a lesson from the BRONCUS trial 1. Eur Respir J

2005;26(6):1104-9.

12. Cole TJ, Green PJ. Smoothing reference centile curves: the LMS method and

penalized likelihood. Stat Med 1992;11(10):1305-19.

13. Becklake MR. Gender Differences in Airway Behaviour (physiology) over the

human lifespan. European Respiratory Monograph 2003:8-25.

14. Swanney MP, Ruppel G, Enright PL, Pedersen OF, Crapo RO, Miller MR,

Jensen RL, Falaschetti E, Schouten JP, Hankinson JL, Stocks J, Quanjer PH. Using

the lower limit of normal for the FEV1/FVC ratio reduces the misclassification of

airway obstruction. Thorax 2008;63(12):1046-51.

15. Kirkby J, Welsh L, Lum S, Fawke J, Rowell V, Thomas S, Marlow N, Stocks

J. The EPICure study: comparison of pediatric spirometry in community and

laboratory settings. Pediatr Pulmonol 2008;43(12):1233-41.

16. Zapletal A, Hladikova M, Chalupova J, Svobodova T, Vavrova V. Area under

the maximum expiratory flow-volume curve--a sensitive parameter in the evaluation

of airway patency. Respiration 2008;75(1):40-7.

17. Turner SW, Craig LC, Harbour PJ, Forbes SH, McNeill G, Seaton A,

Devereux G, Helms PJ. Spirometry in 5-year-olds--validation of current guidelines

and the relation with asthma. Pediatr Pulmonol 2007;42(12):1144-51.

18. Neve V, Edme JL, Devos P, Deschildre A, Thumerelle C, Santos C, Methlin

CM, Matran M, Matran R. Spirometry in 3-5-year-old children with asthma. Pediatr

Pulmonol 2006.

19. Nystad W, Samuelsen SO, Nafstad P, Edvardsen E, Stensrud T, Jaakkola JJ.

Feasibility of measuring lung function in preschool children. Thorax

2002;57(12):1021-7.

20. Silverman RA, Flaster E, Enright PL, Simonson SG. FEV1 performance

among patients with acute asthma: results from a multicenter clinical trial. Chest

2007;131(1):164-71.

21. Crenesse D, Berlioz M, Bourrier T, Albertini M. Spirometry in children aged 3

to 5 years: reliability of forced expiratory maneuvers. Pediatr Pulmonol

2001;32(1):56-61.

Page 18 of 35


Figure Legends

Figure 1. Predicted values and Lower Limit of Normal (LLN), adjusted for the median

age at each height. The lower limit of normal (LLN) is defined as the 5th

centile (-

1.645 z-scores).

Figure 2. Between-subject variability expressed as the % CV. The CV is higher than

the frequently quoted 10% (corresponding to a LLN of 80% predicted, see text for

details) in children <20 years for FEV1 and FVC and at all ages for FEF25-75.

Figure 3. Predicted values for FEV0.75/FVC and FEV1/FVC, adjusted for the median

age at each height. Both ratios are greater in females at all ages and for FEV1/FVC the

differences between males and females starts to diverge during early puberty.

Page 19 of 35


Table 1 Modified Quality Control Cr iteria for preschool spirometry Category Description Evidence

Start of Test 1) Back Extrapolated Volume (BEV) <12.5%

FVC or 80ml

2) A clearly determined peak flow

2, 3

Within manoeuvre 3) Visual inspection of the flow-volume trace.

Maneuver is free from artifact, cough, glottic

closure, leak

2, 3, 11, 17

End of Test 4) Clear end expiratory plateau on volume-time

trace with no sharp drop or cessation of flow.

No specification for a minimum FET

2, 3, 17, 18

Repeatability 5) Two largest FEV1 and FVC values were within

100mLs or 10%. A test session was deemed

adequate if two acceptable manoeuvre are

obtained

2, 3, 19-21

Page 20 of 35


Table 2. Summary of population characteristics for the 15 centers used to develop the prediction models; the four excluded centers are presented in Table E1 in the OLS.

Center Year Studied

N < 8 years

Age range (years)

Exclusion Criteria

1 1990 166 4.4-8 Non-Caucasian, recent respiratory tract infection, chronic pulmonary

disease

2 2001 109 2.5-6.5 Previous symptoms or treatment for asthma (child or family), upper

respiratory tract infection, history of RSV, bronchitis, preterm birth,

oxygen admin or mechanical ventilation, chronic pulmonary disease,

environmental tobacco exposure, allergic rhinitis, atopic dermatitis,

3 1995 193 2.9-6.9 <34 weeks gestational age and <5lb at birth, hospitalization with

respiratory condition, physician diagnosis of asthma, congenital heart

disease, serious chest problems, chronic bronchitis

4 2002 152 3.1-5.9 Gestational age <35wks, asthma, chronic pulmonary disease, chronic

respiratory tract infection (wheeze, cough, dyspnoea), upper respiratory

tract signs or symptoms in the past 2 wks, neuromuscular disease,

debilitating chronic disease, non ambulatory children, mentally disabled

children

5 2001-06 189 4.2-8 Current respiratory tract infection, history of past respiratory disease

6 2001-02 173 3.0-70 Asthma, recurrent bronchitis, other significant respiratory disease

7 2000-05 848 2.5-7 Upper respiratory tract infection and chronic pulmonary disease. All

children were suspected of having asthma but were subsequently

diagnosed as healthy using a series of pulmonary function tests and clinical

evaluation.

8 1996 81 3.4-7.5 Congenital abnormalities, preterm birth

9 1999 483 3-8 Major respiratory tract infection

10 1993-97 79 4.2-8 Upper respiratory tract infection in the past 2 weeks

11 2001-06 65 6.5-8 Cardiopulmonary disease, chest wall deformity, preterm birth, asthma

requiring daily medication, body mass index >30kg/m2

12 Not

recorded

496 4.6-6.3 Not recorded

13 2002 556 3-6 Subjects with only one acceptable result, skeletal anomalies or chronic

pulmonary disease other than asthma

14 2002-05 149 2.7-7 Congenital abnormalities, ventilatory support following birth, hospital

admission for respiratory conditions, current asthma

15 2007-08 38 3-5 <34 weeks gestational age, confirmed diagnosis of bronchopulmonary

dysplasia, acute respiratory tract infection in the past 3 weeks, temperature

>38.5C, prior hospitalization for respiratory illness, diagnosis by a

physician of asthma, reactive airway disease, bronchitis, chronic

pulmonary disease, asthma medications ever, congenital heart disease,

chest problems, recurrent wheeze (more than 2 occasions ever), chronic

cough not due to a cold, neuromuscular or bone disease

Page 21 of 35


Figure 1

Page 22 of 35


Figure 2

Page 23 of 35


Figure 3

Page 24 of 35


Spirometry Centile Charts for Young Children: The Asthma UK

Collaborative Initiative

Sanja Stanojevic, Angie Wade, Tim J Cole, Sooky Lum1, Adnan Custovic, Mike

Silverman, Graham L Hall, Liam Welsh, Jane Kirkby, Wenche Nystad, Monique Badier,

Stephanie Davis, Steven Turner, Pavilio Piccioni, Daphna Vilozni, Howard Eigen, Helen

Vlachos-Mayer, Jinping Zheng, Waldemar Tomalak, Marcus Jones, John L Hankinson

and Janet Stocks on behalf of the Asthma UK collaborative group

Online Data Supplement

Page 25 of 35


OLS Figure Legend: Figure E1 Summary of exclusions for the preschool dataset.

Figure E2. An example (x) of an outlying center, where results were significantly

different from the rest of the reference population.

Figure E3. Fitted z-scores for each center contributing pediatric data demonstrates good

fit overall but small off-sets in nearly all centers.

Inclusion/Exclusion Criteria: Exclusions were made based on predefined criteria (Figure E1). For example, although

most studies did not record gestational age they generally excluded children born

prematurely; amongst studies which did record gestational age 111/2725 subjects born

<36 weeks were excluded. Subsequently, during preliminary modelling, the limited

sample of non-Caucasian subjects (Afro/Caribbean, Oriental, “other”) was found to have

significantly lower lung function compared to Caucasian subjects. These differences

remained significant after adjustment for center and equipment; therefore the final sample

was limited to Caucasian subjects (resulting in the exclusion of 574 subjects from two

centers). Two centers were excluded because they were comprised of all non-Caucasian

subjects and two centers because they were outliers. The first exclusion was attributed to

equipment differences where subjects measured using hand-made equipment had

significantly lower results compared to the rest of the collated dataset. The second center

had higher results than the rest (Figure E2); this center used a modified face-mask rather

than a mouth-piece, which may have introduced some turbulence or alinearity, but we

were not able to systematically evaluate this effect. The excluded centers are described in

Table E1.

Page 26 of 35


Table E1. Characteristics of the excluded centers Center Year Studied N <8 Age range

(years) Exclusion Criteria

16 Not Recorded 1443 5.1-7.3 Not Recorded

17 1995-1997 1430 2.0-7.0 Gestational age <36 weeks, twins

18 2004 277 3.0-6.0 Abnormal weight, physician diagnosed asthma, recurrent bronchitis,

upper respiratory tract infection in last 2 weeks, other significant

respiratory diseases and other system diseases, prenatal environmental

tobacco smoke, chronic cough, phlegm, and shortness of breath.

19 2005-2006 79 5-12 History of respiratory disease, asthma, chronic pulmonary disease,

preterm birth

Description of statistical techniques

LMS is an extension of regression analysis which includes three components: 1) the

skewness (Lambda), which models the departure of the variables from normality using a

Box-Cox transformation, 2) the median (Mu), and 3) the coefficient of variation (Sigma),

which models the spread of values around the median and adjusts for any non-uniform

dispersion, hence LMS. The coefficient of variation (CV) is defined as (100 x

SD/median). The three quantities (LMS) are allowed to change with height and/or age, to

reflect changes in the distribution as children grow. We applied the LMS method using

the Generalized Additive Models of Location Scale and Shape (GAMLSS) package22

in

the statistical program R (Version 2.6.1; R Foundation, http://www.r-project.org).

Separate models were developed for males and females. Cubic splines were used to

produce smoothly changing curves, which explain the age related changes. The goodness

of fit was assessed using the Schwarz Bayesian Criterion (SBC)23

which compares

consecutive models directly while adjusting for the increased complexity to determine the

simplest model with best fit. A detailed description of the GAMLSS technique as it

pertains to spirometry can be found in Cole et al. (2009).7

Comparison with initial ‘A ll-age’ reference equation

Table E2. A comparison of the current equations to the previously published “all-age”

equations6 for a hypothetical subject who lies on the median for FEV1 using the current

Page 27 of 35

http://www.r-project.org/


equations. This comparison demonstrates that after 8 years of age the two equations are

virtually the same; however, the current equations are better in children <8 years as the

previous equations overestimate FEV1 and the other indices (data not shown).

Age

(yrs)

Height

(cm)

Measured

FEV1

Current equations Stanojevic AJRCCM

2008

% predicted z-score % predicted z-score

3 95 0.72 99.12 -0.05 N/A N/A

4 102 0.89 99.02 -0.06 107.85 0.50

5 110 1.10 100.71 0.05 106.79 0.47

6 116 1.27 99.76 -0.02 102.71 0.20

7 123 1.45 100.21 0.2 100.98 0.08

8 128 1.63 100.28 0.2 100.08 0.01

9 133 1.81 100.26 0.2 99.87 -0.01

10 138 2.00 101.74 0.2 101.59 0.14

20 177 4.53 100.07 0.01 100.06 0.01

30 180 4.58 99.62 -0.03 99.55 -0.04

Page 28 of 35


Spirometry Prediction Equations Table E3. Prediction Equations

The models take on the general form:

M = exp[a + b.ln(height) + c.age^p + s]

S = exp[d + e.age^p + t]

L = h

Value a, b, c, d, e, p and h can be found in the table below; s and t can be found in the

spirometry look-up table. Where p = 0, this indicates a logarithmic transformation (i.e.

ln(age)) Ln = natural logarithm; Height in cm; Age in Years

a b c p d e h

FEV0.75 M -9.60340 2.02223 0.02676 1 -2.031934 -0.008781 1

FEV0.75 F -10.15756 2.13863 0.02140 1 -1.87805 -0.02477 1

FEV1 M -11.1048 2.3395 0.1489 0.25 -1.8009 -0.1249 1

FEV1 F -10.27934 2.19871 0.03408 0.25 -1.84730 -0.08156 1

FVC M -11.6897 2.4365 0.2796 0.25 -1.5854 -0.2471 1

FVC F -11.0342 2.3341 0.1453 0.25 -1.7676 -0.1231 1

FEF25-75 M -7.9540 1.8070 -0.0049 1 -1.425505 0.004887 0.4538

FEF25-75 F -6.583 1.613 -0.268 0.25 -1.9445 0.2871 0.5266

FEV0.75/FVC M 0.20915 -0.03716 -0.10277 0 -2.40933 -0.0449 2.227

FEV0.75/FVC F 0.58617 -0.13217 -0.04524 0 -2.592108 0.007362 3.273

FEV1/FVC M 0.31380 -0.03323 -0.15041 0.25 -3.3908 0.3697 4.048

FEV1/FVC F 0.49976 -0.07939 -0.11855 0.25 -3.3947 0.3267 4.088

Worked example:

If the observed FEV0.5 for a 122cm boy at age 5 is 1.0L, the predicted M (median) can be

calculated as:

M FEV1 = exp(-11.1048 + 2.3395 x ln(122) + 0.1489(50.25) + -0.0177557 = 1.40

The S (CV) value for this subject can be calculated as:

S FEV0.5 = exp(-1.8009 -0.1249(50.25

) + 0.048156 = 0.14

The L (skewness) value is fixed at 1.

The 3 parameters L, M & S can be combined algebraically to calculate z-scores or percent

predicted:

Z-score = [(Measurement/M)L - 1] / [L x S]

% predicted = (Measurement/M)*100

Page 29 of 35


Applying these estimates to the z-score equation above gives a z-score of -2.0, a percent

predicted value of 71.2% as calculated below and a LLN of 1.1:

Z-score = [(1.0/1.4)1

- 1] / [1 x 0.14] = -2.0

% predicted = (1.0/1.4)x100 = 71.2%

LLN = (1.4)x(-1.645x(0.14) +1)1) = 1.1

Between-center differences:

These analyses were limited since not all centers measured the same outcomes,

nonetheless significant between-center differences were observed. While the majority of

individual observations fell within +/-2 z-scores derived from these new equations, some

centers, were on average, above zero z-scores while others were below (Figure E3). In

most cases the offset was not statistically significant and would not affect the

interpretation of results.

No consistent patterns were observed between the different equipment types, and it was

difficult to separate equipment differences from other center differences. Jaeger was the

most commonly used equipment (6 centers, 40% of observations); in total 10 different

equipment types were used. Ten centers used nose clips, 10 measured children in the

seated position and 10 used incentive spirometry; however it is important to note that

these were not the same 10 centers. After adjustment for equipment there were no

differences in z-scores between subjects measured seated compared to standing or

between those who did and did not use incentive software. The use of incentives did not

appear to influence results, which is reassuring since these can be useful both in

increasing success rates and ensuring that the spirometry manoeuvre is performed

Page 30 of 35


correctly in young children.24, 25

The “candle” incentive can help to encourage good peak

expiratory flow, since achieving a rapid onset of expiration is crucial for obtaining

accurate measures of FEV0.5 and FEV0.75.26

Similarly, the “bowling” or “toaster”

incentives may help ensure end expiratory plateau is reached, which is especially difficult

to obtain in younger children,2, 27

as demonstrated by the low success rate in this study

(56%).

The use of nose clips produced lower results by an average of 0.35 z-scores (Range 0.27-

0.44). Further investigations into whether equipment or methodology influences

spirometry are necessary but need to be conducted within systematic validation studies

within the same individuals.

Recommendations:

Future studies should aim to record the information provided in Table E4 to facilitate

further collation of normative data.

Page 31 of 35


Table E4 Recommendations for collection of spirometry data in healthy children.1

Information and suggested criteria should be reviewed every 5-10 years by appropriate

professional bodies (e.g ERS/ATS Task force for Lung function in children) to ascertain

whether modifications required to reflect significant changes in evidence, equipment or

clinical practice

Variable Recommendation

Demographics Measured height (standing &

sitting)

Recorded to nearest 0.1 cm

Calculated age (Date of test) – (Date of birth)

Measured weight Recorded to nearest 0.1 kg

Sex 1 “M”; 2 “F”

Gestational age Recorded as completed weeks

Birth weight Recorded to nearest 0.1 kg

Ethnicity Consensus required for definition

History of wheeze* Wheeze in the past 12 months/ever

Environmental Tobacco exposure* Prenatal/Current (objectively measured)

Exclusion Criteria Current respiratory tract infections symptoms (cough, wheeze) in the past 3

weeks

Chronic respiratory disease Cystic fibrosis, doctor diagnosis of

asthma, skeletal/congenital abnormalities,

bronchiolitis, pneumonia, tuberculosis

Outcomes FEV0.5, FEV0.75, FEV1, FVC,

FEF25-75, FEF75, FEF50, FET, PEF,

BEV, Aex, maximum number of

attempts

Selected from the maneuver with the

largest sum of FEV1 + FVC (as

recommended by the ATS/ERS)

Equipment make, model, software version All equipment to meet ATS/ERS

standards 28

Protocol Measurements made using

ATS/ERS protocols

2007 ATS/ERS for preschool children

2005 ATS/ERS for older children and

adults

Position Seated (otherwise recorded)

Nose clips Yes (otherwise recorded)

Incentives (children <10 years) Yes (software recorded), which

incentives recorded

Pre/Post Bronchodilator

Quality control ATS/ERS recommendations

applied

2007 ATS/ERS for preschool children3

2005 ATS/ERS for older children and

adults28

Random selection of flow-volume

curves

To be centrally inspected and scored

using accepted over-read sheet

Biological variability Site visits conducted at all contributing

centers

Automated export program Every effort should be made to minimize

transcription errors

Depending on the purpose of the data collection these may or may not be exclusion criteria,

but should always be recorded.

Page 32 of 35


Figure E1

Page 33 of 35


Figure E2

Page 34 of 35


Figure E3

Page 35 of 35

Spirometry Centile Charts for Young Caucasian Children: The Asthma UK Collaborative Initiative

Documents