Image: NASA Key points • Arbitrary differences in the way lung function is expressed and interpreted may result in mismanagement of patients as well as hindering our understanding of the global burden of lung disease. • Currently, international and regional boundaries, together with individual preferences, may have as much impact on estimates of disease prevalence and treatment decisions as does the true pathophysiological heterogeneity of disease. • Use of the all-age (3–95 years), multiethnic Global Lung Function Initiative (GLI) spirometry equations, which provide well defined lower limits of normal, will allow global standardisation of how spirometry results are interpreted. This will also avoid errors that have occurred in the past due to overdependence on fixed thresholds to diagnose lung disease or extrapolation of prediction equations in either very young or elderly patients
Transcript
Image: NASA
Key points
• Arbitrary differences in the way lung function is expressed and interpreted may result in mismanagement of patients as well as hindering our understanding of the global burden of lung disease.
• Currently, international and regional boundaries, together with individual preferences, may have as much impact on estimates of disease prevalence and treatment decisions as does the true pathophysiological heterogeneity of disease.
• Use of the all-age (3–95 years), multiethnic Global Lung Function Initiative (GLI) spirometry equations, which provide well defined lower limits of normal, will allow global standardisation of how spirometry results are interpreted. This will also avoid errors that have occurred in the past due to overdependence on fixed thresholds to diagnose lung disease or extrapolation of prediction equations in either very young or elderly patients
The Global Lung FunctionInitiative: dispelling somemyths of lung function testinterpretation
Educational Aims
N To summarise limitations and implications associated with using outdatedspirometry reference equations to interpret lung function.
N To describe the Quanjer et al., 2012 ‘‘Global Lung Function Initiative’’ (GLI)spirometry equations and the advantages of using these in both clinical practice andresearch studies.
N To discuss the necessary steps and challenges when switching to the GLI, includingadjustment for ethnicity, re-calculation of previous results for accurate trend reportsand education of both patients and professionals.
SummaryLung function results can help with establishing a diagnosis, with assessment oftreatment effects and with making a prognosis. However, arbitrary differences inthe way lung function is expressed and interpreted may result in mismanagementof patients as well as hindering our understanding of the global burden of lungdisease. In this article, we summarise the Global Lung Function Initiativespirometry reference equations and dispel some common myths related to theuse and interpretation of spirometry results.
Statement of InterestNone declared.
HERMES syllabus linkmodule: D.1
ERS 2013
Breathe | December 2013 | Volume 9 | No 6 463DOI: 10.1183/20734735.012113
Sanja Stanojevic1,
Philip Quanjer2,
Martin R. Miller3,
Janet Stocks4
1Division of Respiratory Medicine, Hospital for SickChildren, Toronto, Canada2Department of Pulmonary Diseases and Department ofPaediatrics, Erasmus Medical Centre, Erasmus University,Rotterdam, The Netherlands3Institute of Occupational and Environmental Medicine,University of Birmingham, Birmingham4Portex Respiratory Unit, Institute of Child Health,University College London, London, United Kingdom
Sanja Stanojevic:Respiratory Medicine,Hospital for SickChildren, Toronto,Canada
Every patient with, or suspected of having, arespiratory condition should routinely havepulmonary function tests (PFTs) performedto assess the type and severity of the diseaseand/or to monitor disease progress andtreatment effectiveness. Patients with respir-atory disease only account for ,60% ofreferrals to most PFT laboratories [1]; patientsreceiving general anaesthetics, transplantsand treatment for cancer or other chronicconditions also have lung function measuredroutinely. PFTs are also used to estimate theglobal burden of respiratory disease [2]. Withsuch a high volume and demand for thisphysiological measurement, why is it that thesame patient seen in a different country, adifferent hospital in the same country or evenby different physicians in the same hospitalon the same occasion, can have markedlydifferent interpretation of lung functionresults?
Correct interpretation of lung functiondepends on knowing how big a patient’slungs should be. However, arbitrary differ-ences in the way in which lung function iscurrently expressed and interpreted [3, 4] notonly results in potential confusion andmismanagement of patients, but also hinderour understanding of the global burden oflung disease. Recently, the Global LungFunction Initiative developed a unified andglobal approach for the interpretation ofspirometry results, such that, for the first
time in history, there is the potential tostandardise how lung function is interpretedaround the world [5].
Global Lung FunctionInitiative ReferenceEquations for Spirometry
The objective of the Global Lung FunctionInitiative (GLI) Task Force (www.lungfunction.org) was to derive ‘‘all-age’’reference equations for spirometry from pre-school children to the elderly that covered asmany ethnic groups as possible. As a result ofunprecedented international cooperation,tens of thousands of spirometric measure-ments from healthy, non-smoking males andfemales have been made available by some70 centres and organisations worldwide.These data were collated and analysed withmodern statistical techniques (fig. 1), andhave led to the derivation of the Quanjer‘‘GLI-2012’’ prediction equations [5].
The GLI-2012 equations have beenendorsed by all major respiratory societiesand, for the first time, provide a unifiedapproach to the interpretation of PFTs thathas the potential to streamline presentationand interpretation across laboratories, citiesand countries worldwide. Within less than ayear of publication, the GLI equations havealready been validated in several populations[6–8], widely cited in the literature [9–17],and implemented into numerous lung func-tion devices (see www.lungfunction.org/
manufacturers.html). While the GLI representa huge step forward, the lack of awarenessand logistical roadblocks for their appropriateimplementation can hamper widespread clin-ical use. In this article we will describe anddispel some common myths related to thepresentation and interpretation of lung func-tion results and demonstrate how a unifiedglobal approach to interpret spirometryresults can benefit patients worldwide.
Myths
Myth 1: our current PFT reports arereliable and appropriate for our localneeds
Several jurisdictions (e.g. ATS/ERS, EuropeanDrivers’ License, ARTP) [18, 19] have produced
1
2
3
4
5
0
Pred
icte
d FE
V1 L
Age years0 20 40 60
Caucasian
Males
S East AsianAfrican American
N East Asian
80 100
Figure 1Predicted values for FEV1 in males for the four ethnic groups considered within the GLI.A further equation (‘‘other/mixed’’) has also been derived for those of mixed ethnicity orwho are not represented by the groups shown above.
The Global Lung Function Initiative
Breathe | December 2013 | Volume 9 | No 6464
evidence-based recommendations to facilitateaccurate data collection, quality control andinterpretation of spirometry results. Thechoice of how PFTs are presented andinterpreted is, however, largely at the discre-tion of the individual laboratory or physician.Consequently, in many centres, there remainsan enormous gap between these evidence-based guidelines and the actual clinicalinterpretation of pulmonary function testresults [3]. Furthermore, there is generally anaive trust placed in reports automaticallyproduced by software, with little (if any)thought given as to how tests were performedor how the numbers were derived.
There are several aspects of the PFTreport that ought to be questioned.
Are the patient details recorded correctly?
The major determinants of spirometric lungfunction are height, age, sex and ethnicity [5].Despite the dependence of predicted valueson height, it is not uncommon (particularly inadult centres) for the patient’s height to beself-reported rather than measured. Men tendto over-report their actual height [20], andadults tend to ‘‘shrink’’ as they age, but oftenreport their height as the maximum achievedduring adulthood. Since height is a majordeterminant of expected lung function, dis-crepancies in height measurements, includ-ing those resulting from a poorly calibratedstadiometer, can lead to misinterpretation ofresults [11].
Age is also an important determinant oflung function throughout the life span, suchthat accurate documentation (in years to onedecimal point) is essential, particularly duringchildhood when growth and development areso rapid [11]. Current practice in manycommercial devices of either truncating or‘‘rounding’’ age to the nearest year, ordependence on self-reported age (rather thanthat based on difference between date of testand date of birth) can also lead to seriousmisinterpretation [11].
Finally, PFT reports should also considera patient’s ethnicity. Although it is wellestablished that there are ethnic differencesin lung volumes after correcting for age,height and sex [21–24], PFT results in non-Caucasians are typically compared to healthyCaucasian subjects of white European descent,which can lead to significant under-estimationof lung function in such subjects.
Myth 2: it doesn’t really matter whichreference equation is used
There are more than 300 published referenceequations for spirometry, not to mention thenumerous unpublished equations availableon PFT equipment. It is important toappreciate that not all reference equationsare created equal. Within each spirometer,the user has the option to select a referenceequation that they believe is appropriate forthe local population. Oftentimes the defaultequations provided by the manufacturer arenever changed and, even if these are changedat the request of the lab manager at time ofequipment delivery, it is not uncommon thatautomatic re-booting of computer systems orinstallation of new software restores defaultreference equations without the knowledge ofthe user.
It is also common practice for referenceequations to be stitched together such that awider age range can be tested without the needfor the user to manually switch betweenequations when testing different patients.Often these prediction sets or modules arenot evidence-based but simply derived forconvenience, or to meet demands of userswho do not fully appreciate the consequences
It is recommended that height is measuredat each visit to one decimal point using anaccurate and regularly calibrated stadi-ometer.
It is recommended that age be calculatedaccurately in years to one decimal pointusing the patient’s date of birth and thedate of test.
In addition to clearly stating exactly whichreference equations have been used, PFTreports should display a patient’s ethnicgroup, as well as the ethnic group of thereference population used to derive thepredicted values.
The Global Lung Function Initiative
Breathe | December 2013 | Volume 9 | No 6 465
of developing such prediction modules. As aresult, out-dated or inappropriate referenceequations are often used to interpret spirometryresults, with arbitrary break points betweenspecific age groups (e.g. preschool to school-age children, adolescents to adults), which canlead to serious misinterpretation of results [25].
Equally concerning is the lack of trans-parency to the general user as to whichreference equations have been used in anyselected module. Commercial devices willsometimes allow extrapolation of predictionequations beyond the age range they werederived for (e.g. interpretation of lung func-tion in a 4-year-old from equations derivedfrom children aged above 6– 8 years, extra-polation of European Community for Steeland Coal (ECSC) and the National Health andNutrition Examination Survey (NHANES) tothose aged above 70 and 80 years respect-ively). Furthermore, within any age range,different spirometric outcomes may be inter-preted using entirely different equations, withpotentially serious impact on interpretation ofthe relative sensitivities of different variables.
The potential misinterpretation of spiro-metry results is greatest during the transition
from paediatric to adult care in patients withchronic respiratory conditions [25]. Forexample, until recently in the UK, it has beenrecommended that the Rosenthal equations[26] be used during childhood, and the ECSCequations [27] during adulthood. Manylaboratories use a prediction module whichjoins these equations at 18 years of age. Infigures 2 and 3 we highlight a few exampleswhere interpretation of results is influencedby the inconsistencies both within andbetween reference equations from the same‘‘prediction module’’.
Even if the investigator/clinician is aware ofexactly which reference equation has beenselected, there are numerous factors, includ-ing the inclusion/exclusion criteria used toselect the reference population, how well-nourished the reference population was, andthe age span which will inevitably affect whatthe predicted value will be at any given age andheight. Equations derived from 100 individualswill be far less representative than equationsderived from 10,000 individuals [30]. Further-more, equations that are derived for the entireage range, will be far more robust thanseparate equations that are artificially stitchedtogether to cover the entire age range.
Awareness of the limitations of outdatedreference equations has led to an increasingnumber of attempts to construct population-specific or even centre-specific referenceequations. These are often derived for apopulation over a limited age range andusing a limited sample of patients such thatcomparison with larger, more rigorouslyderived equations will inevitably identifydifferences, all of which may largely beattributed to the phenomenon of samplingvariability [11]. Recent evidence suggests thatvalidation of reference equations for local userequires a minimum of 300 subjects (150males and 150 females) [11].
90
80
70
60
FEV
1 % p
redi
cted
Age years13.5 14.0 14.5 15.0 15.5 16.0
Figure 216 serial measurements between 13 and 16 years of age in a boy with Cystic Fibrosis withheight measurements at the 2nd centile [28]. % predicted values according to Rosenthalequations (blue line) [26] are much higher than those calculated from GLI (black line).This is likely due to the fact that the Rosenthal equations do not take age into account,and thus result in marked over-estimation of % predicted in growth restricted children.Furthermore, there was a sudden fall in % predicted results by 14% at around 15.5 yearsof age when expressed according to Rosenthal even though there had been no change inclinical status and the absolute FEV1 had continued to increase as expected with growth.This was caused by the fact that, when using the Rosenthal equations, this adolescent’sheight reached a critical break point between two consecutive visits resulting inautomatic transfer by the software from the male child’s equations (where predictedFEV1 for a height of 162.4 cm was 2.77 L) to the male ‘‘post-pubertal’’ referenceequation for which the predicted FEV1 for a height of 163.9 cm was 3.43 L [29].
PFT technologists, respiratory clinicians andreferring physicians should be aware ofwhich reference equations are used tointerpret PFTs, and whether these areappropriate for their patient population,paying particular attention to whether pre-diction modules are used and the potentialimpact of any break-points on interpretationof results.
Clinicians and patients both want to knowwhether the PFT result falls within the normalor expected range. The interpretation of PFTsthus hinges around knowing what ‘‘normal’’is. Results from clinical chemistry, haemo-globin, lipids and so forth are compared witha reference range, which summarises mea-surements made in a group of healthyindividuals in the absence of disease. Thereference range is derived from the upper andlower values that contain 95% of the healthyreference population.
Similarly, pulmonary function measure-ments are compared with a group of healthyindividuals, although traditionally interpretationhas been slightly different from that observed inother disciplines since lung size is stronglyassociated with body size, dimensions of thethoracic cavity, sex and age. During childhoodand adolescence, growth is particularly rapidwith lung function increasing 20-fold during thefirst 10 years of life [31]. Furthermore, duringchildhood, forced vital capacity (FVC) outgrowsforced expiratory volume in 1 second (FEV1),leading to falls in FEV1/FVC; these trends arereversed during adolescence [32]. The correctdescription of the range of normal valuesrequires that these physiological factors aretaken into account. This has rarely been possiblewhen analysing relatively small datasets overlimited age ranges or when using simple linearregression models and is responsible for manyof the discrepancies seen when comparingresults from children and adolescents that havebeen interpreted according to GLI-2012 withthose obtained from previously publishedequations (see figs 2 and 3).
Respiratory clinicians, technicians andtheir patients are accustomed to the practiceof expressing and interpreting PFT results as% predicted without realising the inherenterrors in this methodology. The % predictedis calculated by taking the observed meas-urement (absolute values of FEV1 and FVC inL) and dividing it by a predicted valuemultiplied by 100 (% predicted 5 (observed/
predicted)6100). The predicted values areobtained from a group of healthy individuals,such that 100% predicted reflects the averagevalue expected in a healthy individual of anygiven size, sex and age. Although 80%predicted is commonly used as the cut-off foridentifying abnormal results, the use of this
fixed cut-off comes from unsubstantiatedclaims in the 1960s that suggested this wasa good ‘‘rule of thumb’’ [33]. The majorlimitation of using % predicted (and 80%predicted as a fixed threshold) is that it doesnot take into account the fact that the naturalvariability of spirometry outcomes in health ishighly age and outcome dependent. Thepractical implication being that the ‘‘normalrange’’ for FVC or FEV1 is considerably widerthan the frequently quoted ‘‘80–120% pre-dicted’’ both for young children and forsubjects older than 30 years [5, 34]. This leadsto a high percentage of false-positive findingsparticularly in the elderly (fig. 4a).
The use of the fixed cut-off for the FEV1/
FVC ratio has similar consequences. Severallarge population studies have shown that theFEV1/FVC has a strong negative agedepen-dency, such that the frequently used fixedthreshold of 0.7 for FEV1/FVC isnot attaineduntil about 50 years of age inmen and later inwomen. Consequently, airway obstruction inyounger subjects is missed [5, 34–38],whereas the 0.7 cut-off falsely identifies alarge number of olderhealthy subjects ashaving lung disease (fig. 4b) [10, 39–41].
Myth 4: I can’t understand Z-scores orSRS or explain them to my patient
For many years, both the ERS and ATS haverecommended that results are presented and
130
120
110
100
90
80
70
FEV
1 % p
redi
cted
Age years13 14 15 16 17 18
Figure 3Eight serial measurements are presented from a white Caucasian male below the 2ndcentile for height [28] who was tested between his 12th and 18th birthday. There isincreasing discrepancy over time between the GLI (black line) and Rosenthal equations(blue line) [26]. The difference between the two equations is as large as 25% at 17 years.At 18 years, the software automatically switches from Rosenthal to the ECSC equationsand the % predicted value plummets from 114% to 83%. By contrast when the GLIequations are used, the patient’s results tracked seamlessly across the pubertal period.
The Global Lung Function Initiative
Breathe | December 2013 | Volume 9 | No 6 467
interpreted in the context of the normal rangewherein the lower limit of normal (LLN) isage specific. The normal range can berepresented as a pictogram (fig. 5), or as Z-scores (otherwise known as standardisedresidual scores or SRS) in which 95% ofhealthy subjects will have Z-scores (or SRS)values within ¡2 z-scores, and 90% within¡1.64 z-scores. The Z-score indicates howmany standard deviations a measured valueis from predicted ((observed-predicted)/standard deviation). Although it is customaryto classify the severity of lung functionimpairment using FEV1 as a percentage ofpredicted (ATS/ERS 2005) [18], the use of Z-scores removes the age-related bias [15].Further work is necessary to better under-stand what Z-score values represent clinicallymeaningful outcomes, as was done for %predicted FEV1 in the past.
Myth 5: the Z-score cannot replace theclinically established % predicted
The use of % predicted is associated withage- and height-related bias and shouldtherefore be abandoned. A valid alternativemethod of reporting lung function is toexpress results as Z-scores. Use of Z-scoressolves many potential problems by takinginto account age, height, sex and ethnicgroup, as well as the age-dependent referencerange. Unlike % predicted, it is therefore freeof any bias. An obvious advantage is that anygiven Z-score indicates comparable lung
function between individuals, irrespective oftheir sex, height, age or ethnicity. The Z-scorealso facilitates bias-free interpretation ofserial measurements within a person duringgrowth and ageing, and direct comparisonbetween different lung function outcomes.
Because lung function tests are notapplied indiscriminately to the population,when patients have symptoms or known riskfactors (e.g. smoking history), it is usualpractice to use the -1.64 Z-scores cut-off toidentify subjects outside the normal range. Iflung function tests are used for untargetedscreening, then -1.96 Z-scores should beused. Both these cut-offs indicate the prob-ability of a false positive (5% and 2.5%respectively for -1.64 and -1.96 Z-scores), orthe proportion of completely normal subjectswith values below these cut-offs.
35
FemalesMales
a) b)
30
25
20
15
10
5
0
% w
ith F
EV1
<80
% p
redi
cted
Age years0 20 40 60 80 100
35
30
25
20
15
10
5
0
% w
ith F
EV1/
FVC
<0
.7
Age years0 20 40 60 80 100
Figure 4The percentage of healthy subjects in whom FEV1 is less than 80% predicted (left), or in whom the FEV1/FVCratio is less than 0.7 (right) [5]. As can be seen, use of 80% as a fixed threshold for FEV1 leads to a highpercentage of false positives in the elderly, while use of ,0.7 as the threshold for abnormal FEV1/FVC will lead tounder-diagnosis of airway obstruction in the young and over-estimation in the elderly.
When interpreting results, it is important toremember that there will always be a degreeof within-person variability, so that bychance a measurement may be just outsidethe normal range on one occasion, but justwithin it on the next. It is also essential totake other clinical information into account,and to weigh the consequences of anerroneous false positive against that of amissed diagnosis. Particular caution isrequired when interpreting results whichlie close to the somewhat arbitrary cut offsbetween health and suspected disease,especially when results are limited to asingle test occasion.
The Global Lung Function Initiative
Breathe | December 2013 | Volume 9 | No 6468
However, both % predicted and Z-scoreshave limitations when classifying the severityof obstructive lung disease or definingprognosis. This can be illustrated as follows.Both indices indicate how far a measurementis from an average value in a healthy subject;however, mortality relates more directly tohow close a measured value is from theminimum that is compatible with life. Formales and females aged .50 years this isan FEV1 of ,500 and 400 mL, respectively[42, 43]. At age 50 years, for a man of averageheight this represents 13.2% predicted(Z-score -5.77), for a woman (165 cm) 14.0%(Z-score -5.92). At age 85 years, correspond-ing findings are 18.9% (Z-score -3.73) and21.4% (Z-score -3.88), respectively. Hence, inclinical assessment both the Z-score andremaining ventilatory reserves needs to betaken into account. Further investigations todefine disease severity and how this trans-lates to Z-scores using the GLI are necessary.
Myth 6: I don’t need to worry aboutreference equations as I only look at theabsolute values
Physicians that care for patients with chronicrespiratory conditions are particularly inter-ested in identifying changes and deterior-ations over time that are outside the normalvariability for an individual patient. Whenserial measures of absolute values of lungfunction (i.e. FEV1 and FVC in L) are used totrack progress in young and middle-agedadults, these can be referred internally to thesubject’s own ‘‘best’’ values and interpretedwithout the need for any reference equation.This approach cannot, however, be usedwhen attempting to compare results betweenpatients or centres. Even within an individual,interpretation of change becomes much morecomplicated during periods of growth orageing. Under such circumstances, it will benecessary to express results in relation tosome accepted reference.
Myth 7: if I follow patients over time, itdoesn’t matter which reference equationis used, as long as it is the same one
Accurate interpretation of how much changecan be attributed to disease progression orresponse to therapy needs to be made incontext of how much change is likely to occuras a result of within-subject, between-testvariability, as well as in relation to thechanges attributed to the natural process ofgrowth, development and ageing.
In contrast to popular belief, interpret-ation of changes in lung function dependsmarkedly on which reference equation isselected. In contrast to the GLI, olderreference equations were often derived usingsimple linear regression based on relativelyfew subjects over a limited age range, withage not being included as a determinant inmany of the paediatric equations still incommon use. Consequently, the predictedvalue at any given age can vary markedly
0.20
0.15
0.10
0.05
0.00
FEV1
FEV1/FVC
FVC
-1.64 +1.64
-1.96 +1.96
Z-score
Freq
uenc
y
-6 -4
0.1 25
50
9598 99.9
Percentiles
-2 0 2 4
Figure 5Illustration of the normal distribution and corresponding Z-scores and percentiles. Thepictogram (horizontal bars) demonstrates the normal range (white region) with arrowsindicating how far from the normal range an observation is (Z-scores) [5]. The 50thcentile (0 Z-scores) is equivalent to 100% predicted.
A lung function test must never be used inisolation to define disease severity andprognosis; a number of factors, includingquality of life, are likely to contribute, andthe ideal approach remains to be deter-mined. Neither % predicted nor Z-scoresused in isolation can answer those fun-damental questions.
The Global Lung Function Initiative
Breathe | December 2013 | Volume 9 | No 6 469
between and within equations, with markedsubsequent impact on the way in which asubject’s results will be interpreted either ona single occasion or over time (see Myth 2and figure 6).
In practice, the interpretation of serialmeasurements is further complicated by thefact that, with the exception of the GLI-2012equations, no spirometric reference equationencompasses the entire age span. Con-sequently when following a patient over time,results are interpreted either by extrapolatingresults beyond the age they were derived for(a practice that should be discouragedstrongly due to the major errors it canintroduce) or by stitching reference equationsderived from different populations togetheras part of ‘‘prediction modules’’ (see Myth 2).With increasing use of spirometry in pre-school children as young as 3 years of age [44]these issues are now of greater relevance andcan also lead to serious misinterpretationparticularly when patients transition frompaediatric to adult care (fig. 3) [25].
It is also not uncommon for patients toacquire disease and suffer from a significantreduction in lung function, even though theirlung function values remain within the
normal range. Understanding the subject’sbaseline lung function is critical for correctinterpretation of the changes observed due todisease. When tracking a patient over time ormonitoring treatments, it is important toconsider the patient’s ‘‘best lung function’’ asa reference point and how much change hasbeen observed. A major limitation of anyreference equation is that it is based on across-sectional snap-shot of a populationcomprised of individuals at different ages,which does not reflect how individual patientschange over time.
Myth 8: the GLI might make patients’results that were normal abnormal andvice versa. How can we explain this to thepatient/their doctor?
For reasons discussed above, there willinevitably be some changes in interpretationof results when they are re-analysed using GLI2012, and this will need to be dealt withsensitively when discussing with the patientand their family [10]. When examined inclinical populations, there were minor, clin-ically unimportant changes in the prevalencerate of airways obstruction upon adopting theGLI equations, whereas the prevalence rate ofa ‘‘spirometric restrictive pattern’’ (low FVCbut no obstruction) was higher [6, 9, 10]. Asa matter of principle, the fact that newimproved methods of interpreting resultshave been introduced should be discussedprior to any patient receiving a copy of resultsbased on the new equations to allay anyanxiety, accompanied by simple explanatoryleaflets. Similar communication will be neces-sary with all relevant health professionals incontact with the patient. Appropriate edu-cational material is being developed currentlyand will be made widely available free ofcharge in the near future.
In the meantime, it is important toremember that no reference equation iscapable of determining whether a patienthas normal or abnormal lung function inisolation (see Myth 5). Reference equations(and the lower limit of normal) give anindication of how an individual patient’sresult compares to that of a group of healthyindividuals. The further away a result liesfrom the normal range, the more certain wecan be that the results are not compatiblewith health. Thus results from a patientwhose results fall on or near the lower limit
15
10
5
0
-5
-10
Diff
eren
ce fr
om p
redi
cted
FEV
1 %
Age years
6 8 10 12 14
Girls
16 18
Zapletal
Quanjer
Polgar
Knudson
Hankinson
GLI 2013
Wang
Stanojevic
Rosenthal
Figure 6Differences from GLI in predicted values compared to observed values for 9 differentreference equations used for children [11].
Most commercial PFT devices facilitate there-calculation of trend reports based on asingle reference. Therefore the transition toGLI equations should be accompanied witha re-calculation of trend reports to facilitateinterpretation of results in context ofprevious results.
The Global Lung Function Initiative
Breathe | December 2013 | Volume 9 | No 6470
of normal and who has evidence of respir-atory symptoms and other clinical indicationsof disease should be interpreted quite differ-ently from an otherwise healthy individualwith similar predicted values but no apparentsigns or symptoms of disease. By definition,when using a LLN of -1.64 Z-scores or (the 5thcentile) 5% of healthy individuals (one in 20PFTs) will have results below the lower limitof normal (fig. 5); therefore simply relying ona number to indicate whether lung function isnormal or abnormal will inevitably lead to amisdiagnosis. One should treat the patient,not the numbers.
Myth 9: It is better to wait until the GLIequations have included ALL ethnicgroups before we switch.
The GLI represent a huge step forward byproviding a unified approach to interpretinglung function in several ethnic groups.Importantly the GLI analyses were able toshow that a wide range of countries andethnicities have lung function that is consist-ent with a white ‘‘Caucasian’’, minimising theneed for multiple reference equations wherethe differences are not physiologically orclinically meaningful [45]. Second, the ethnicdifferences identified were proportional suchthat the general growth and decline of lungfunction was systematic across ethnic groups.
While the GLI do include several ethnicgroups, they are by no means comprehensiveof all ethnic groups. Notably missing fromthe GLI are data from the African continent,South Asia (Indian sub-continent) and LatinAmerica. Several on-going efforts are under-way to collect data from these regions for thenext instalment of the GLI in the future [7].Since ethnic differences are proportional,interim ethnic-specific correction factors canbe derived for new ethnic groups currentlynot represented within the GLI.
However, even these efforts pale in com-parison to the breadth of ethnic diversityrepresented in the world. The growing numberof migrants and bi-racial children pose twosignificant limitations to any attempt to
develop reference equations for every ethnicgroup. Migration (both within and betweencountries), adoption of Western lifestyles andsecular changes in anthropometric character-istics in resource poor nations will all need tobe investigated in the future.
Myth 10: it is better to wait until GLIequations are available for all pulmonaryfunction tests before we switch
Whenever reference equations are devel-oped, the outcomes included and age rangeencompassed is usually at the discretion ofthe investigators conducting the study. Noset of reference equation contains all possiblespirometric outcomes, and very few measuremultiple outcomes (e.g. from spirometry,plethysmography and gas transfer) on thesame subjects. It is therefore unrealistic tohope that a single reference population formultiple pulmonary function tests, over theentire age range and for multiple ethnicgroups will ever be a reality. Consequently,reports which present predicted results for arange of PFTs generally include predictedvalues from multiple prediction equationsderived from different healthy populations.Furthermore, many PFT devices allow forprediction modules to be developed, suchthat the user only needs to select onereference, although in fact this representsmultiple reference equations arbitrarilystitched together. Of course, if the predictedspirometric FVC is derived from one referencepopulation and the plethysmographic VCfrom a different population, then it is highlyunlikely that these two values will match.Rather than producing all-inclusive predictionsets that are, at best, arbitrary, efforts shouldbe put towards investing in appropriatestudies that collect high quality data that willfurther help to improve the interpretation ofPFTs. For this reason, a GLI working party iscurrently trying to establish predicted valuesfor transfer factor.
Currently, results from each PFT need tobe interpreted independently, and in thecontext of the patient. While report of a FVCvalue within the normal range and VC valueoutside the normal range does present aconundrum, reference equations and pre-sentation of results in relation to what isexpected in a subject of that age, height, sexand ethnicity is only one indication of apatient’s health status. These values should
When new improved methods of interpret-ing results are introduced, these should bediscussed prior to any patient receiving acopy of results based on the new equationsto allay any anxiety, accompanied by simpleexplanatory leaflets.
The Global Lung Function Initiative
Breathe | December 2013 | Volume 9 | No 6 471
be interpreted in the context of the patient’sother clinical findings. Until a suitablealternative to these discrepancies becomesavailable, the interpretation of discordantresults, such as these, requires careful clinicaljudgement, rather than inappropriate applica-tion of out-dated reference equations.
Myth 11: we cannot use GLI because itdoes not include our favourite outcomes
While it is possible to extract hundreds ofoutcomes from a spirometry test, and almostall commercial devices will allow you to displayyour favourite selection of these on reports,the clinical utility of these additional outcomesis highly questionable [46]. Similarly, thereference populations used to derive thepredicted values for many of these outcomesshould be examined more closely before theyare used as a basis for any clinical decisions.
The GLI spirometry equations weredeveloped to include physiologically relevantoutcomes for which there were sufficient datato develop robust reference equations. TheGLI focussed on outcomes recommended bythe ATS/ERS guidelines (i.e. FEV1, FVC andFEV1/FVC) [47], and thus do not include allpossible outcomes, although the facility tointerpret FEV0.75 in younger subjects has beenincluded for white children (data on thisoutcome not being available for other ethnicgroups) together with forced expiratory flowbetween 25 and 75% of FVC (FEF25-75) and at75% of FVC (FEF75) for various ethnic groups.
Myth 12: it is impossible to interpretresults easily using GLI because it doesnot present a typical reference flow–volume loop with which to compare mypatient’s results
Many physicians are familiar with the presenta-tion of an ‘‘expected’’ flow–volume loop super-imposed on top of a patient’s flow–volumeloop commonly presented on reports. To createthis it is necessary to have accurate recordingsof peak expiratory flow and other flow out-comes; data that were not available in many ofthe spirometric datasets included in GLI.Furthermore, these ‘‘typical curves’’ on whichclinicians place so much reliance when inter-preting their patient’s results, are fundament-ally flawed. Despite popular belief, flow–volumecurves do not actually contain any informationthat is not available from FEV1, FVC and FEV1/
FVC. The myth that flows detect small airwaysdisease which goes undetected when usingconventional spirometric indices arose becauseflows were expressed as % predicted, usinginappropriate prediction equations and disre-garding the true LLN. The ‘‘expected’’ flow–volume loop is drawn from predicted valuesfrom a mixture of sources, which do not alwayscorrespond with the predicted values and %predicted values presented on the report. Inaddition the typical curves are inappropriate forchildren. Generally, whenever FEV1, FVC andFEV1/FVC are within the normal range, so areFEF25–75% and FEF75%, making flow indicesredundant [46].
Understanding whether a patient’s resultsare within the normal range, and the natureand severity of any abnormality is muchbetter assessed using a pictogram (fig. 5),which clearly delineates the range of valuescompatible with normal and how an indivi-dual’s results compare to this range.Assessment of the shape of the flow–volumeloop should be done independently of anypredicted diagram and ideally in the contextof pre–post bronchodilator results.
Myth 13: changing to GLI will justconfuse everyone
The reality of the current situation is that mostpeople are confused when it comes to usingreference equations, though few are aware ofthe magnitude of error that can occur bymisusing such equations. Bearing in mind thata large proportion of pulmonary functionlaboratories [3, 4, 48], let alone general practice
It is inadvisable to include predicted valuesfor spirometry outcomes not included inthe GLI from external reference equationsin the form of prediction sets.
When switching to GLI-2012 equations forspirometry, laboratories should investigatewhich reference equations are being usedfor other PFTs and decide whether theseremain the most appropriate for theirpatient population at the present time.There are obvious limitations to usingreference equations for different PFTsderived from different populations, andthese should be considered when interpret-ing discordant findings from different PFTs.
The Global Lung Function Initiative
Breathe | December 2013 | Volume 9 | No 6472
offices, are not aware of which referenceequations are being used to interpret theresults, ignorance or fear of the effects ofmaking the change to GLI-2012 is not anacceptable reason to continue the misinter-pretation of PFT results. Patients deserve morefrom the respiratory community. As high-lighted above, the key to successful transitionlies in appropriate education and ensuringthat, once the switch is made, all previousresults from each patient are re-analysed usingthe same standards, so that reliable trendreports are obtained.
Conclusions
The use of inappropriate reference equations,and misinterpretation even when usingpotentially appropriate equations, can leadto serious errors in both under- and over-
diagnosis, with the associated burden interms of financial and human costs. It is nolonger acceptable to continue applying blindfaith in results produced by equipment wheninterpreting lung function tests. Any algo-rithm can produce a result, but that does notmean it is the correct result. Having endorsedthese equations, national and international
respiratory societies now need to follow theexample of the Association of RespiratoryTechnology and Physiology (ARTP) and form-
ally recommend usage of the GLI equations toall their members. Similarly, researchers, clin-icians and technicians are all equally respons-
ible for enforcing pressure to request the latestATS/ERS recommendations are available forthe interpretation of patient results. Finally,
manufacturers of lung function testing equip-ment need to be proactive and activelyparticipate in facilitating the switch to GLI.
References
1. Pretto JJ, Brazzale DJ, Guy PA, et al. Reasons forreferral for pulmonary function testing: an audit of 4adult lung function laboratories. Respir Care 2013; 58:507–510.
2. Buist AS, McBurnie MA, Vollmer WM, et al.International variation in the prevalence of COPD(the BOLD Study): a population-based prevalencestudy. Lancet 2007; 370: 741–750.
3. Mohanka MR, McCarthy K, Xu M, et al. A survey ofpractices of pulmonary function interpretation inlaboratories in Northeast Ohio. Chest 2012; 141:1040–1046.
4. Pattishall EN. Pulmonary function testing referencevalues and interpretations in pediatric trainingprograms. Pediatrics 1990; 85: 768–773.
5. Quanjer PH, Stanojevic S, Cole TJ, et al. Multi-ethnicreference values for spirometry for the 3-95-yr agerange: the global lung function 2012 equations. EurRespir J 2012; 40: 1324–1343.
6. Quanjer PH, Weiner DJ. Interpretative consequencesof adopting the global lungs 2012 reference equa-tions for spirometry for children and adolescents.Pediatric Pulmonology 2013: [In press; doi: 10.1002/ppul.2287].
7. Bonner R, Lum S, Stocks J, et al. Applicability of theglobal lung function spirometry equations incontemporary multiethnic children. Am J Respir CritCare Med 2013; 188: 515–516.
8. Hall GL, Thompson BR, Stanojevic S, et al. TheGlobal Lung Initiative 2012 reference values reflect
Suggested reading:
N A review of how reference equations are derived and used: Stanojevic S, Wade A, Stocks J.Reference values for lung function: past, present and future. Eur Respir J 2010; 36: 12–19.
N A detailed description of how the GLI-2012 were derived: Quanjer PH, Stanojevic S, ColeTJ, et al. Multi-ethnic reference values for spirometry for the 3-95-yr age range: the globallung function 2012 equations. Eur Respir J 2012; 40: 1324–1343.
N A comparison of several commonly used reference equations, and how interpretationchanges when the GLI-2012 are applied: Quanjer PH, Brazzale DJ, Boros PW, et al.Implications of adopting the Global Lungs Initiative 2012 all-age reference equations forspirometry. Eur Respir J 2013; 42: 1046–1054.
N Evidence supporting the misuse of fix cut-offs to interpret spirometery results: Miller MR,Quanjer PH, Swanney MP, et al. Interpreting lung function data using 80% predicted andfixed thresholds misclassifies more than 20% of patients. Chest 2011; 139: 52–59.
N A summary of the clinical benefits of adopting the GLI-2012: Swanney MP, Miller MR.Adopting universal lung function reference equations. Eur Respir J 2013; 42: 901–903.
9. Brazzale DJ, Hall GL, Pretto JJ. Effects of Adoptingthe New Global Lung Function Initiative 2012Reference Equations on the Interpretation ofSpirometry. Respiration 2013; 86: 183–189.
10. Quanjer PH, Brazzale DJ, Boros PW, et al.Implications of adopting the Global Lungs Initiative2012 all-age reference equations for spirometry. EurRespir J 2013; 42: 1046–1054.
11. Quanjer PH, Hall GL, Stanojevic S, et al. Age- andheight-based prediction bias in spirometry referenceequations. Eur Respir J 2012; 40: 190–197.
12. Vaz Fragoso CA, McAvay G, Gill TM, et al. Ethnicdifferences in respiratory impairment. Thorax 2013;[In press; doi: 10.1136/thoraxjnl-2013-203631].
13. Schultz ES, Gruzieva O, Bellander T, et al. Traffic-related air pollution and lung function in children at 8years of age: a birth cohort study. Am J Respir CritCare Med 2012; 186: 1286–1291.
14. Coates AL, Graham BL, McFadden RG, et al.Spirometry in primary care. Can Respir J 2013; 20:13–21.
15. Quanjer PH, Pretto JJ, Brazzale DJ, et al. Grading theseverity of airways obstruction: new wine in newbottles. Eur Respir J 2013; [In press; doi: 10.1183/09031936.00086313].
16. Vaz Fragoso CA, Gill TM, McAvay G, et al.Respiratory impairment in older persons: when lessmeans more. Am J Med 2013; 126: 49–57.
17. Vandevanter DR, Pasta DJ. Evidence of diminishedFEV and FVC in 6-year-olds followed in the Europeancystic fibrosis patient registry, 2007–2009. J CystFibros 2013; [In press; doi: 10.1016/j.jcf.2013.04.005].
18. Pellegrino R, Viegi G, Brusasco V, et al. Interpretativestrategies for lung function tests. Eur Respir J 2005;26: 948–968.
19. Steenbruggen I, Mitchell S, Severin T, et al.Harmonising spirometry education with HERMES:training a new generation of qualified spirometrypractitioners across Europe. Eur Respir J 2011; 37:479–481.
20. Merrill RM, Richardson JS. Validity of self-reportedheight, weight, and body mass index: findings fromthe National Health and Nutrition ExaminationSurvey, 2001–2006. Prev Chronic Dis 2009; 6: A121.
21. Greenough A, Hird MF, Everett L, et al. Importanceof using lung function regression equations appro-priate for ethnic origin. Pediatr Pulmonol 1991; 11:207–211.
22. Harik-Khan RI, Muller DC, Wise RA. Racial differencein lung function in African-American and Whitechildren: effect of anthropometric, socioeconomic,nutritional, and environmental factors.Am J Epidemiol 2004; 160: 893–900.
23. Ip MS, Karlberg EM, Karlberg JP, et al. Lung functionreference values in Chinese children and adolescentsin Hong Kong. I. Spirometric values and comparisonwith other populations. Am J Respir Crit Care Med2000; 162: 424–429.
24. Wang X, Dockery DW, Wypij D, et al. Pulmonaryfunction between 6 and 18 years of age. PediatrPulmonol 1993; 15: 75–88.
25. Kirkby J, Aurora P, Spencer H, et al. Stitching andswitching: the impact of discontinuous lung functionreference equations. Eur Respir J 2012; 39: 1256–1257.
26. Rosenthal M, Bain SH, Cramer D, et al. Lung functionin white children aged 4 to 19 years: I--Spirometry.Thorax 1993; 48: 794–802.
27. Quanjer PH, Tammeling GJ, Cotes JE, et al. Lungvolumes and forced ventilatory flows. ReportWorking Party Standardization of Lung FunctionTests, European Community for Steel and Coal.
Official Statement of the European RespiratorySociety. Eur Respir J 1993; 6: Suppl. 16, 5–40.
28. Cole TJ, Freeman JV, Preece MA. British 1990 growthreference centiles for weight, height, body mass indexand head circumference fitted by maximum pena-lized likelihood. Stat Med 1998; 17: 407–429.
29. Rees S, Kirkby J, Fettes E, et al. Appropriateinterpretation of longitudinal spirometry assess-ments. Eur Respir J 2013; Suppl. 348s – 349s.
30. Quanjer PH, Stocks J, Cole TJ, et al. Influence ofsecular trends and sample size on reference equationsfor lung function tests. Eur Respir J 2010; 37: 658–664.
31. Quanjer PH, Stocks J, Polgar G, et al. Compilation ofreference values for lung function measurements inchildren. Eur Respir J 1989; 4: Suppl. 184S–261S.
32. Quanjer PH, Stanojevic S, Stocks J, et al. Changes inthe FEV/FVC ratio during childhood and adolescence:an intercontinental study. Eur Respir J 2010; 36:1391–1399.
34. Stanojevic S, Wade A, Stocks J, et al. Reference rangesfor spirometry across all ages: a new approach.Am J Respir Crit Care Med 2008; 177: 253–260.
35. Brandli O, Schindler C, Kunzli N, et al. Lung functionin healthy never smoking adults: reference values andlower limits of normal of a Swiss population. Thorax1996; 51: 277–283.
36. Falaschetti E, Laiho J, Primatesta P, et al. Predictionequations for normal and low lung function from theHealth Survey for England. Eur Respir J 2004; 23:456–463.
37. Hankinson JL, Odencrantz JR, Fedan KB. Spirometricreference values from a sample of the general U.S.population. Am J Respir Crit Care Med 1999; 159:179–187.
38. Kuster SP, Kuster D, Schindler C, et al. Referenceequations for lung function screening of healthynever-smoking adults aged 18-80 years. Eur Respir J2008; 31: 860–868.
39. Miller MR, Quanjer PH, Swanney MP, et al.Interpreting lung function data using 80% predictedand fixed thresholds misclassifies more than 20% ofpatients. Chest 2011; 139: 52–59.
40. Quanjer PH, Enright PL, Miller MR, et al. The need tochange the method for defining mild airwayobstruction. Eur Respir J 2011; 37: 720–722.
41. Swanney MP, Miller MR. Adopting universal lungfunction reference equations. Eur Respir J 2013; 42:901–903.
42. Miller MR, Pedersen OF. New concepts for expres-sing forced expiratory volume in 1 s arising fromsurvival analysis. Eur Respir J 2010; 35: 873–882.
43. Ward H, Cooper B, Miller MR. Validation of lungfunction prediction equations from patient survivaldata. Eur Respir J 2012; 39: 1181–1187.
44. Beydon N, Davis SD, Lombardi E, et al. An officialAmerican Thoracic Society/European RespiratorySociety statement: pulmonary function testing inpreschool children. Am J Respir Crit Care Med 2007;175: 1304–1345.
45. Kiefer EM, Hankinson JL, Barr RG. Similar relation ofage and height to lung function among Whites,African Americans, and Hispanics. Am J Epidemiol2011; 173: 376–387.
46. Quanjer PH, Weiner DJ, Pretto JJ, et al. Measurementof FEF25–75% and FEF75% does not contribute toclinical decision making. Eur Respir J 2013; [In press;DOI: 10.1183/09031936.00128113].
47. Miller MR, Hankinson J, Brusasco V, et al.Standardisation of spirometry. Eur Respir J 2005; 26:319–338.
48. Basson E, Stewart RI. The standard of spirometry inthe RSA. S Afr Med J 1991; 79: 361–363.
