s American Thoracic SocietyMEDICAL SECTION OF THE AMERICAN LUNG ASSOCIATION

LUNG FUNCTION TESTING: SELECTION OF REFERENCE VALUES ANDINTERPRETATIVE STRATEGIES

Tiers OFRCIAL STATEMENT OF THE AM E R I C A N

THORACIC S0crEl-Y WAS ADoPTED BY THE ATSBOARD OF DIRECTORS, MARCH 1991.

ContentsIntroduction

BackgroundFocus

Sources of Variation in Lung Function TestingConceptual issues pertinent to the

interpretation of lung function testsTechnical sources of variationProcedural sources of variationBiologic sources of variation

Statistical Considerations in theDerivation of Prediction Equations

General commentsCharacterizing the distribution and

determinants of lung function inreference populations

Evaluating prediction equationsDistributions and “lower limits of normal”

Sources, Uses, and Selection of Referencevalues

General commentsSources of reference equationsDetermination of the “normal range”Smoking as an independent variableCross-sectional and longitudinal

predictionsCriteria for selection of reference valuesPublished reference equationsLimitations of currently available equations

Interpretative StrategiesConceptual issues concerning normality

and the limits of normalObstructive and restrictive vemilatory

defectsBronchodilator responseinterpretation of lung function tests in

clinical practiceRecommendations

OverallSelecting reference valuesRecommendations for interpretation

IntroductionBackground

During the last 3 decades lung function testshave evolved from tools for physiologic studyto clinical tools widely used in assessing re-spiratory status. In addition to their use in

1202

clinical case management, they have becomea part of routine health examinations in re-spiratory, occupational, and sports medicineand in public health screening. It is commonpractice for the results of lung function teststo be interpreted in relation to reference values,and in terms of whether or not they are con-sidered to be within the “normal” range (l-6).A wide selection of published reference valuesand “lower limits of normal” is available (4).Computerized equipment adds a new dimen-sion with preselected or menus of referencevalues and interpretation algorithms whoseorigin and justification may be unclear.

To maximize the clinical value of lung func-tion tests and to assist those managing clini-cal lung function testing laboratories, theAmerican Thoracic Society (ATS) (7-12), theEuropean Community for Coal and Steel(ECCS) (4), and the European Society forClinical Respiratory Physiology (13) have pub-lished guidelines, focusing primarily on spi-rometry as the most widely used lung func-tion test. The 1987 ATS statement in spirom-etry (8) outlined the steps necessary to achievestandardization: (1) equipment performance,validation, and quality control; (2) subjectperformance; (3) measurement procedures todetermine acceptability and reproducibility;(4) reference values and interpretation. Thefmt three have been addressed in official state-ments or position papers of the ATS (7-12).This statement addresses the fourth.

FocusThe charge by the ATS was to prepare a com-prehensive and practical document dealingwith conceptual issues and their scientific ba-sis and providing guidelines for daily use intwo areas: (I) selecting reference values and(2) interpretative strategies The statement wasto address the concerns of those who gener-ate lung function reports and those who uselung function reports to assist in clinical casemanagement. Epidemiologic and publichealth issues are not addressed though epi-demiologic studies provide the scientific ba-sis for many of the concepts used in inter-preting lung function results. The ATS haspublished standardization procedures forepidemiologic studies (14). The focus of thisstatement is spirometry, but reference is madeto other lung function tests when pertinent.

Although the statement deals primarily withadults, the conceptual issues apply to chil-dren as well. Terms and abbreviations followthe American College of Chest Physicians(ACCP)-ATS joint committee on pulmonarynomenclature recommendations (15). The nextfour sections deal with conceptual issues andtheir scientific basis; the last section deals withpractical considerations and recommendations.

Sources of Variation in LungFunction Testing

Conceptual Issues Pertinent to theInterpretation of Lung

Function TestsAll clinical measurements, including pulmo-nary function tests, are subject to (I) techni-cal variation related to instrument, procedure,observer, subject, and their interactions; (2)biologic variation, the focus of interest ofmost of the nonclinical biological sciences;(3) variation caused by dysfunction or dis-ease, the focus of clinical medicine (5). In clin-ical pulmonary function testing, it is impor-tant to minimize the variation caused by tech-nical factors and to take biologic variationinto account so that variations caused by dis-ease can be properly interpreted. Sources oftechnical and biologic variation and the esti-mated magnitude of their effects are listedin tables 1 and 2.

Interpretation of pulmonary function testsdepends upon establishing the variation ofinterest (the signal) and its relation to all oth-er sources of variation (the noise) (5). Whichsources of variation constitute signal andwhich noise will depend on the question be-ing asked. For instance, in a physiologic studyof the effects of posture on FEV,, variationcaused by posture would constitute the sig-nal and all other sources of within-individualvariation, the noise. Similarly, in an epidemi-ologic study of the effects of an occupation-al exposure on a work force, variation causedby exposure will constitute the signal, and allother sources of between population varia-tion, the noise In the clinical context, signaland noise will vary according to the clinicalquestion. For instance, when assessing the out-come of a treatment, the signal would be thechange after treatment, and the noise wouldbe within-individual variation in the absence

AM REV RESPIR DIS 1991; 1449202-1218

AMERICAN THORACIC SOCIETY 1203

, TABLE 1

SOURCES OF VARIATION IN LUNG FUNCTION’

Source

Technical

BiologicWithin individual

Between individual

Between population

Determinants

Instrument, subject, posture, observer, procedure (including number of tests),software: temperature; altitude

All of the aboveDiurnal (circadian) and seasonal effects, endocrinologic effectsAll of the abovePersonal factors, including size, age, sex, physical activity, muscularity,

race, and other genetic characteristics and past and present healthEnvironmental factors, including tobacco smoke (personal and environmental),

occupation, residence (urban or rural), air pollution (home, environmental),and socioeconomic status

All of the aboveSelection factors which determine inclusion or exclusion of certain subjects

from study populations

* Baaed on table m reference 5 and reproduced with permissron.

of treatment. When lung function tests areused as an aid in diagnosis, the signal is usu-ally the patient’s results compared with theexpected result for subjects without diseasebut similar in the personal characteristics thatdetermine lung function such as sex, size, age,and, possibly, race (table 1).

Technical Sources of VariationIN~TR~~~ENTATI~N

Detection of instrument problems is an inte-gral part of interpretation. Readers shouldconsult ATS recommendations on spirome-try and D~00, which give practical limits ofacceptable instrument variability (7-9). In-struments and procedures used in develop-ing of reference values and those used to evalu-ate patients should meet, and preferably ex-ceed, current ATS recommendations.

PRECUON AND ACCURACY

In considering the variability of a test, a dis-tinction must be made between precision andaccuracy. Precision refers to the repeatabilityof the measurements, even if the values ob-tained are not accurate (16). Accuracy, whichis not easy to establish, refers to how closethe measurements made by an instrument areto the “true” value Because most instrumentshave better precision than accuracy, between-instrument variation usually contributes moreto total measurement variability than within-instrument variation.

COMPVrZR SOFTWABE AND HARDWARE

Overall, the use of computers in spirometrysystems has reduced technical variability;nevertheless errors associated with computersoccur. Even small differences in the techniquesused to calculate flow can produce relativelylarge differences in derived flow measure-ments (17,18). It is imperative that spimme-try systems using computers be validated ini-tially and each time changes are made in soft-ware or hardware One simple method ofvalidating computer computations is to com-pare manual calculations of spirometricvalues with computer-calculated values. Thevalues should be close, -c 2 to 3%, but they

will not be identical. Computers can also pro-vide immediate feedback on the success ofa subject’s performance and improve overalltest quality. Quality control algorithms thatdetect coughs, late peak flows, premature ter-mination of effort, excessive extrapolatedvolumes using the back extrapolation tech-nique, and excessive variation between maneu-vers can be programmed to provide immedi-ate feedback to the technician.

SPECLAL CONSIDERATIONS FOR

TESTING CHILDREN

Equipment for testing children should havean accuracy for volume of + 50 ml to below0.5 L. The output for the hard copy displayshould be scaled to the size of the signal witha variable attenuation to a minimum of30 mm/L. There should be a visible real-timedisplay to encourage both the child and thetechnician and to ensure that effort is sus-tained over a sufficient time Equipment, in-cluding mouthpieces and noseclips, shouldbe adjustable and comfortable for childrenwith heights as low as 12O cm. Children shouldbe tested in a laboratory where personnel arefamiliar with clinical testing of children andwhere interpretations can be made by personsfamiliar with pulmonary function testing inchildren. Detailed recommendations for pedi-atric testing have recently been issued by theEuropean Society of Clinical RespiratoryPhysiology (13).

Procedural Sources of VariationThe largest single source of within-subjectvariability is improper performance of thetest. Therefore, interpretations of spirometryshould include a statement about test qualitybefore any other interpretation is rendered.The AT’S (7-12), the ECCS (4), the Califor-ma Thoracic Society (2), the IntermountainThoracic Society (19), the European Societyfor Clinical Respiratory Physiology (13), andseveral texts (1,3,20-23) have all recognizedthe importance of procedure in reducing mea-surement variability. Readers should consultthese references for detailed recommendations.

TABLE 2

ESTIMATES OF THE PROPORTION OFMEASURED BETWEEN-INDIVIDUAL

VARIATION IN FEV, OR FVC IN ADULTSATTRIBUTABLE TO IDENTIFIED FACTORS’

Factor

Proportion ofVariation

Attributable

Sex

AgeHeightWeightEthnic

0.08up to 0.30

0.20 up to 0.300.02 l-

differences 0.10Technical 0.03Unexplainedt 0.27Total 1 .oo

. Reproduced with permission from reference 5.t Includes all other determkwtta of biologic variation dia-

cussed in SOURCES OF V ARIATION IN LUNG F UNCTION Tssr-ING Mether environmental (e.g., smoking, active, and passive.occupational expoaurea. resfdential p-allutfon. socioeconomicstatus) or hoat. (e.g., genetic, allergic, paat and present rea-piratory heafth status). The latter two are usuafly the focus ofintereat to the clinical pulmonary function laboratory.

Biologic Sources of VariationWrrmN-INDIVIDUAL (INlX4rNDMDuAL)

bRlATION

This section addresses short-term intrain-dividual variations in lung function that donot originate with instrumentation and arenot related to disease, environment, the in-take of drugs, smoking, or failure of the sub-ject to inspire or expire maximally duringspirometric maneuvers. The main residualsources of variation are: (I) body position,(2) head position, (3) effort dependence ofmaximal flows, and (4) circadian rhythms.

(I) Body position. Body position affectsspirometric volumes, particularly FVC andVC, which are 7 to 8% lower in the supinethan in the standing position and 1 to 2%lower in the sitting than in the standing posi-tion (24-27). Body position should be keptconstant in comparison studies. The stand-ing position may be particularly.advantageousfor obese subjects (28).

(2) Headposition. Systematic increases inmaximal expiratory flows have been docu-mented during neck hyperextension (29).These increases are believed to be related toelongation and stiffening of the trachea andrange from minimal to 35% of baseline valuesfor lung volumes above FRC (6O to 80% ofVC). Corresponding changes in FEV, havenot been documented. Conversely, neck flex-ion may decrease peak expiratory flow rate(29) and increase airway resistance (30).Avoiding hyperextension and flexion of theneck seems sufficient to eliminate this sourceof variability. The effect of neck position isusually less than that of body position, butit may be important for patients tested in bed.

(3) Effort dependency of maximal flows.The imperative for standardization is one rea-son for the recommendations that the expi-ratory maneuver be performed with maximaleffort. Nevertheless, FEV, may be 100 to

1204

200 ml lower when the effort is maximal com-pared with submaximal efforts because theairways are narrower with respect to the ex-haled volume (31-34). Variable expiratory ef-fort may thus be a confounding factor whenassessing small changes in maximal flows ortimed volumes such as those resulting frombronchodilator response, therapy, or aging.When a flow-volume curve is available, peakexpiratory flow may be an index of maximalexpiratory effort (31). In some subjects,repeated maximal efforts may trigger bron-chospasm, resulting in a progressive decreasein FVC and FEV, (35). This may also accountfor a subject’s inability to achieve the reprodu-cibility standard recommended by the ATS.It is of interest that failure to meet thesereproducibility standards may itself be a mea-sure of less than perfect health (36, 37).

(4) Circadian rhythms Variations in lungfunction tests with a period of approximate-ly 24 h are well documentedQ8-40). For max-imal expiratory flows, the lowest values areusually seen in the early morning (4 to 6 A.M.),and the largest values are seen around noon(38). In healthy subjects, FEV, has beenshown to increase by about 0.15 L in the mom-ing and decrease by 0.05 L in the afternoon(39); for peak expiratory flow rate (PEFR),the peak-to-trough amplitude is on the orderof 8% (40). Circadian variations have alsobeen documented for airway resistance, spe-cific airway conductance, functional residu-al capacity, total lung capacity, and residualvolume (41-44). The mechanisms responsi-ble for these diurnal variations in lung func-tion have not yet been elucidated (45, 46).Much larger diurnal changes are seen in asth-matic patients who often exhibit a severe“morning dip” in pulmonary functionparameters with decreases of 50% or morein PEFR (40,41,47,48). As with healthy sub-jects, the largest values are usually seen aroundnoon, but this pattern may be substantiallyshifted by the timing of treatment (49). Ex-aggerated circadian variations have also beenobserved in patients with chronic bronchitis(50, 51). Seasonal variations of respiratoryfunction have also been recorded (49).

BETWEEN-INDMDUAL (INTERINDMDUAL)VARIABILITY: HOST FAC’IDRS

The most important host factors responsiblefor interindividual variation in lung functionare (I) sex and size, and (2) aging, which ac-count for approximately 30, 22, and 8Q0,respectively, of the variation in adults (5) (ta-ble 2). Other sources of interindividual vari-ation are (3) race and (4) past and presenthealth. Approximately 27% of interindividualvariation remains unexplained (5) (table 2).

(Z) Size and sex. Size is usually measuredas standing height (6,52). Sitting height, notas easy to measure as standing height, gener-ally explains less of the variability (53), butit may be a useful predictor in certain circum-stances (cg., when dealing with a populationof mixed ethnic ori& see below). Arm spanmeasurements provide a practical substitutefor standing height in subjects unable to stand

or those with a skeletal deformity such askyphoscoliosis (19,54). Lung function is de-creased at both extremes of weight (55, 56).Including measurements of chest circumfer-ence only slightly improves the prediction oflung function (57-60). Variations in airwayand air-space dimensions and geometry alsocontribute to interindividual variation in lungfunction (61,62). Accurate methods of mea-suring airway and air-space geometry are notwidely available, and the contribution thesemeasurements will make to increasing predic-tion accuracy is unknown. After correctingfor body size, girls appear to have higher ex-piratory flows than do boys, whereas adultmen have larger volumes and flows than dowomen (6, 63, 64).

(2) Aging. An appropriate model for lungfunction changes caused by aging during theadult years includes a period after adult heightis attained in which there is either an increase(usual in young men) or little or no decreasein function (usual in young women), afterwhich the function decreases at an accelerat-ing rate with increasing age (6, 65) (see alsoGROWTH section below). These accelerated ag-ing effects are typically found in longitudinalstudies and not in studies based on cross-sectional data. The differences between cross-sectional and longitudinal studies are ex-plained by both statistical issues (66-69) andcohort effects (5, 6, 52, 55, 70).

(3) Race. Race has been consistently shownto be an important determinant of lung func-tion (20, 55, 58, 63, 64,71-83). When com-pared with Caucasians of European descent,values for most other races usually showsmaller static and dynamic lung volumes andlower forced expiratory flow rates but simi-lar or higher FEVJFVC ratios. In some popu-lation groups diffusing capacity (transfer fac-tor) is also lower (71). Regression equationsderived from white populations using stand-ing height as the measure of size usually over-predict values measured in black subjects byabout 12% for TLC, FEV,, and FVC and byapproximately 7% for FRC and RV (20). Peo-ple of mixed race usually have intermediatevalues. These differences persist after al-lowances are made for age., stature, smoking,air pollution, habitual activity, and altitude.The reason for the differences between theraces is unclear. Differences may be due inpart to differences in body build (58,63,64,72-76). Blacks, on average, have a smallertrunk:leg ratio than do whites (77). The useof sitting height as an index of body size inprediction equations reduces but does not ful-ly eliminate the observed differences betweenwhites and blacks (58,63,64,77) or the differ-ences between Europeans, Indians, and Asians.Environmental differences, perhaps relatingto nutrition, physical activity, community airpollution, and socioeconomic factors are alsothought to contribute to these differences(78-85).

(4) Past andpresent health. Lung functionat any one point in time reflects not only thepresent health of the individual but also thesum of all the insults and injuries the lung

AMERICAN THORACIC SOUEW

has sustained in the past including those fromthe prenatal and immediate postnatal peri-ods (86-88).

BETWEEN-INDMDUAL (INTERINDIVIDUAL)VARIATION: ENMRONMENTAL FACTORS

The effects of exposure to tobacco smoke, byfar the most important environmental factorknown to alter lung function, are well docu-mented elsewhere (89). In this section con-sideration is given to other environmental fac-tors that account for between-individualdifferences in lung function.

(I) Geographic factors. Altitudes as highas 1,500 m do not appear to cause measur-able changes in lung volumes, though mea-surement of some flow rates may be affectedby changes in air density even at these alti-tudes (90-92). FEV, and forced expiratoryflows are slightly increased at high altitudes,mainly because of the decreased density ofair (93, 94). During acute exposures to alti-tude there may be slight reductions in VC,TLC, and FRC, most likely because of in-creased thoracic fluid (95). Those residing athigh altitudes probably have larger lungvolumes than do residents at low altitudes.The reasons are unclear because of the con-founding effects of variables such as nutri-tion (%, 97).

(2) .&posure to environmental and occupa-tional pollution. Exposure to airborne ini-tams such as ozone, nitrogen dioxide, sulfurdioxide, and sulfuric acid may produce mea-surable transient changes in pulmonary func-tion tests in controlled human exposure ex-periments and epidemiologic studies (98-102).Those who are exercising and sensitive sub-groups of the general population have in-creased responses. For example, short-termexposures (minutes) to high concentrationsof SO, can trigger transient bronchoconstric-tion in exercising asthmatics (103). Reducedlung function levels and an increased rate ofdecline in lung function have been associat-ed with long-term exposures to sulfur oxides,inhalable particles, and photochemical ox-idants (100).

EnvironmentaI exposure to tobacco smokeappears to affect the lung function of chil-dren (104,105) and, possibly, adults (106-108).More recent observations also show an effecton bronchial reactivity in children (109). Thehealth effects of other indoor pollutants havenot yet been conclusively established (110,111).Exposure to occupational pollutants, includ-ing dusts, chemicals, gas, em, may induceacute and chronic changes in lung function(112-114).

(3) Socioeconomic status. Adverse effectsof low socioeconomic status on lung func-tion are well documented and detectable evenin industrialized countries (85,115,116). Lowsocioeconomic status is often associated withunfavorable environmental conditions suchas living in polluted urban-industrial areas,increased environmental and occupationalexposures, increased indoor air pollution,increased rates of respiratory illness, anddecreased access to health care. Moreover, dif-

AMERICAN THORACIC SOCIETY 1205

, ferences in lung function attributed to genet-ic factors may be partly or even largely at-tributable to differences in socioeconomicstatus (84).

G R O W T H

Growth affects the relationships between in-dices of body size and spirometric measure-ments in children and adolescents. Some ofthe determinants of lung volumes and ven-tilatory flows are therefore briefly reviewedhere.

(I) Relationship to height. The relationshipof ventilatory function to height from child-hood through late adolescence to adulthoodis not linear. Prediction equations for chil-dren are usually based on power or exponen-tial functions of height, both of which seemto fit the data equally well (63,64, 117-120).

(2) Age-dependence. Growth in standingheight, measured in cross-sectional or longitu-dinal population studies, is not in phase withlung growth during the adolescent growthspurt (120-125). Growth in chest dimensionslags behind that of the legs (60,122,124,125).In boys, standing height and VC are often notmaximal by 17 yr of age (123). VC continuesto increase after growth in height ceases andmay not be maximal until after 25 yr of age.Girls, however, seem to attain their maximalvalues at about 16 yr of age (120, 122, 123).In younger subjects, FVC and FEV, seem totrack constant percentiles over time (126).Ideally, developmental rather than chrono-logic age should be included in predictionequations for children and adolescents, butsuch equations are not available or practical.

(3) Respiratory muscies. The opposing ef-fects of increasing muscularity and obesityhave been invoked to explain the observed in-crease in ventilator-y function that parallelsincrease in body mass and the decline in lungfunction beyond an optimal weight (55). Like-wise, an increase in lung volumes and bodymass when growth in height had stopped hasbeen attributed to an increase in muscle massand the consequent increase in respiratorymuscle force (124, 127, 128). However, dataon maximal inspiratory and expiratory pres-sures generated at different ages are incon-clusive No differences were observed betweenrespiratory pressures in adolescents and adults(129). In adolescents there is evidence of onlya small increase in maximal respiratory pres-sure with growth of the lung and thorax(130-N). The average maximal respiratorypressures of boys are larger than those of girls(130-133). Although there is a large varlabili-ty in maximal inspiratory and expiratory pres-sures between individuals of the same sex, re-spiratory force accounts for only a small por-tion of the differences in ventllatory function(134, 135).

(4) Elastic properties. From the neonatalperiod to old age, the thoracic cage growsstiffer (136). Lung recoil increases from birthto adulthood and then decreases with aging(136-143). The relatively constant FRC/TLCratio (120) and the measurements of respira-tory system mechanical properties (136) sug-

gest that changes in lung and chest recoil arewell balanced during growth.

(5) Lung volumes and ventilatory flows.From childhood to adulthood the FEV,/FVCratio and the ratio of maximal expiratory flow(derived from flow-volume curves) to the FVCare almost constant. Girls generate larger ex-piratory flows than do boys of the same ageand stature (120, 127, 135, 144, 145). This isdue in part to the fact that girls have a smallerVC for the same TLC than do boys, but itmay also reflect both the smaller muscle massand the smaller number of alveoli found ingirls (146). Airway tone appears to decreasein girls but not in boys after a deep inspira-tion (147). Finally, in children between 2 and12 yr of age airway resistance is less in girlsthan in boys (148). These observations war-rant using different prediction equations forboys and girls at all ages.

Statistical Considerationsin the Derivation of

Prediction EquationsGeneral Comments

Reference equations provide a context forevaluating the pulmonary function values ofan individual patient or subject in compari-son to the distribution of measurements ina reference population. The clinicians requestfor tests often contains the implicit question:Are these results below the “lower limit of nor-mal?” This section deals with statisticalaspects and limitations of this concept.

Characterizing the Distribution andDeterminants of Lung Function

in Reference PopulationsSubjects with similar characteristics for thevariables that affect lung function (sex, age,height, race) can be grouped together in a stra-tum or a cell. Comparing the performanceof an individual subject with the values gener-ated from a reference population requires oneto know something about the data in the ap-propriate cell, specifically: (0 the number inthe cell, (2) measures of central tendency suchas the mean value, (3) estimates of dispersionsuch as variance or standard deviation (SD),and (4) information about the symmetry ofthe distribution. If the number of subjectsin each cell is sufficient, lung function canbe described by providing descriptors of thedistribution such as mean and SD. Such tabu-lations are infrequently used for lung func-tion because there are too many possible cells(consider all possible combmations of age andheight). Regression equations are an econom-ical and efficient alternative method to de-scribe expected values as a function of sex,height, and age Regression techniques assumethat pulmonary function varies in a symmet-ric fashion about the mean value in each celland that the variance about the mean is con-stant from one cell to another. The closer thedistribution of pulmonary function valuescomes to symmetry or, better still, to a Gaus-sian distribution within cells, the more it is

possible to take advantage of the simplifica-tions possible with Gaussian data.

Evaluating Prediction EquationsLinear regression is the most common but notthe only model used to describe pulmonaryfunction data in adults. Such equations per-form less well at the edges of the data distri-bution and in those cells where there are fewdata. Estimates are likely to be misleading ifthey go beyond the range of the independentvariables used to create the equation. Regres-sion analyses are often simplified by restrict-ing the range of possible values to cells (rangesof height and age) in which reasonable predic-tions are possible. One approach to regres-sion analysis is to use separate simple regres-sion equations for several different age groups(149,150). This approach may introduce con-flicting estimates at the points of transitionbetween equations.

Complex equations may provide more bio-logically plausible models and reduce the av-erage differences between observed and pre-dicted values for every cell (eg., age andheight) in comparison with simple linear equa-tions. The improved predictions, however,usually come at the cost of increased com-plexity of computation.

The most commonly reported measures ofhow well regression equations fit the data theydescribe are the square of the correlationcoefficient (rz) and the standard error of theestimate (SEE). The proportion of variationin the observed data explained by the inde-pendent variables is measured by r2. The SEEis the average SD of the data around the regres-sion line SEE will decrease and 9 will increaseas regression methods diminish the differencesbetween predicted and observed pulmonaryfunction values in the reference population.When the same equations are used to describea different population, SEE will invariablybe larger, and r2 wilI be smaller. In addition,since these statistics reflect average charac-teristics of the regression, r2 and SEE maynot reflect the ability of the equation to de-scribe the tails of the distribution or the limitsof “normal,” and therefore are not sufficientcriteria on which to choose the best equationsto evaluate a clinical population.

Distributions and “LowerLimits of Normal”

Distributions of FEV, and FVC in popula-tion studies am usually found to be close toGaussian in the middle age range, but not atthe extremes. Distributions of flow measure-ments and ratio measures (eg., PEVJFVC)are usually not symmetric (149). Tiansforma-tion or age statification of the data may helpproduce symmetric distributions about themean. Ideally, publications describing refer-ence populations should include not only theprediction equations but also a means ofdefining their lower limits. In the absence ofexplicit recommendations, a lower limit canbe estimated from a regression model. Forspirometry, values below the fifth percentileare taken as below the expected range (below

1206

the”lower limit of normal”), and those abovethe fifth percentile are taken as within the ex-pected range (149,150). Percentiles can be cal-culated directly from the data if there are suf-ficient measurements within each category(56,149,150). If individual observations havea distribution close to Gaussian, the value ofthe fifth percentile can be roughly estimatedas: Lower limit of normal = Predicted value- 1.645 x SEE. Ideally, the SD of the residu-als should be constant for all cells. This istrue for some equations for adults (149). Inother studies, the estimated SD for the loga-rithm of FVC and FEV, among preadoles-cent children, and for height-adjusted FVCand FEV, among adults, appears to be con-stant for each sex and race (56). If SD isproportional to the predicted mean value, asit may sometimes be in children (126), the fifthpercentile can be estimated as a constantproportion of the predicted.mean, i.e, a per-cent of predicted. A comparison of severalprediction equations for spirometry hasshown substantial agreement using the fifthpercentile criterion but not using the - 1.645x SEE criterion (151).

Sources, Uses, and Selectionof Reference Values

General CommentsNormal ventilatory function has come tomean the average spirometric values of a rep-resentative sample of healthy subjects drawnfrom the general population. Various crite-ria for excluding study subjects have been sug-gested based on (r) past and present medicalhistory (eg., presence of respiratory symp-toms such as cough, sputum production, andwheezing; presence of physician-diagnosed re-spiratory disease such as asthma, bronchitis,emphysema, or tuberculosis; hospitalizationfor lung or chest conditions; the presence ofheart disease; employment exposures; and cig-arette smoking); (2) physical examination; and(3) chest radiographic findings. The most im-portant selection criteria are those based ona history of past disease and respiratory symptoms. A reference population should, ideal-ly, be representative of the general popula-tion from which the clientele of the laborato-ry comes. Although a random sample of apopulation is ideal, one report found that oncehospital patients were excluded, the methodfor selecting the study sample used to gener-ate reference vaiues had relatively little effecton either the mean value or the range of valuesobtained (152).

Sources of Reference EquationsIn the 196Os, a number of reference equationswere published based on data gathered in spe-cific population groups such as laboratorypersonnel, workers in a particular industry,school populations, subjects attending a spe-cific clinic, volunteers, and general industrialworkers (153-157). Some are derived frompopulation-based data gathered in epidemi-ologic studies carried out for other purposes;in these studies reference equations are a

byproduct (56,63, 126, 149, 150). Others arebased on data gathered specifically for thecreation of reference equations (91, 158).

Determination of the “Normal Range”FIXED PERCENT OF PREDICTED VALUES

The practice in many clinical laboratories hasbeen to classify values of FVC and FEV, lessthan 80% of predicted as abnormal. Thisfixed value has no statistical basis in adults(91, 159-162). Although some studies haveshown that for adults of average age andheight, 80% of predicted FVC and FEV, isclose to the fifth percentile, use of a fried valuewill result in shorter, older subjects being morereadily classified as “abnormal” (159, 162),whereas taller, younger adult subjects aremore likely to be erroneously classified as“normal.” The practice of using 80% ofpredicted as the lower limit of normal forFEF25-71% or the instantaneous flows willalso cause important errors since, for theseflows, the lower limits of normal are closerto 50% of predicted (149, 150). The practiceof using a fiied percent of predicted as a low-er limit of normal may be acceptable in chil-dren (163) (see section on DISTRIB~ONS AND

bXVER bMIIS OF NORMAL).

FEV JFVC RATIO

Defining a fried FEVJFVC ratio as a lowerlimit of normal is not recommended in adultsbecause FEVJFVC is inversely related to ageand height (91, 149, 150). The use of a friedratio will therefore result in an apparent in-crease in the prevalence of impairment as-sociated with aging or with age-confoundedfactors such as cigarette smoking or occupa-tional exposures. In addition, some athleteshave values for FVC that are relatively largerthan those for FEV,, resulting in a lowerFEVJFVC. This may also be true of work-ers in some physically demanding occupationssuch as mining and deepsea diving.

PERCENTILES AS TBE “LOWER LIMITOF Nom”

One statistically acceptable approach for es-tablishing lower limits for any spirometricmeasure is to define the lowest 5% of the ref-erence population as below the lower limitof normal (see section on DISTIUBUTIONS AND

LOWER LQarrs OF NORMAL). This implies a 5%false positive misclassification, a rate gener-ally considered acceptable.

Smoking as an Independent VariableSubjects who smoke cigarettes usually havelower values for spirometry and forced expi-ratory flows even if they meet the same healthcriteria for “normal” as nonsmokers (164).Smoking has both biologic and technical ef-fects on Dwo (9,165). A clear choice for themost appropriate method of adjusting spiro-metric indices for the effect of smoking is notreadily evident from published data in whichany of the following have been used: smok-ing status (current smoker or exsmoker),amount currently smoked, duration of smok-

AMERICAN THORACX SOclETY

ing, and pack-years of smoking. Neglectingthe correlation of some of these factors (e.g.,pack-years) with age can introduce errors inanalyzing the effect of smoking. In one study,the lifetime loss of FEV, for the average malesmoker was 7.4 ml/pack-year, and for the av-erage female smoker it was 4.4 ml/pack-year(164). Current smoking also adds an acute def-icit in FEV, of approximately 150 ml over andabove the cumulative effect of lifetime smok-ing (164, 166).

The distribution of a smoking variable inthe reference population and its relation toother health indicators will affect the regres-sion term calculated for smoking. For exam-ple, in one study a twofold greater deficit inspirometric measurements in relation to pack-years was found in subjects with chroniccough compared with those without chroniccough (167). The mean spirometric value maynot be the best index for determining lungfunction deficit caused by smoking since theeffect on the susceptible minority tends to beoverwhelmed by the unaffected majority(168). Whether the effects of smoking aresimilar across other independent variablessuch as sex and age is unknown. Some of thesex differences in smoking-associated pulmo-nary dysfunction may be related to differencesin smoking behavior (169). The effect ofsmoking also increases with age (166). Theeffect of smoking on the developing lung islikely to be different from the effect of smok-ing on the adult lung.

Finally, the effects of smoking cessationon pulmonary function are inconsistent. Ex-smokers are found to have both reversible andirreversible ventilatory decrements (164,166,170). Most cross-sectional studies in older sub-jects have found older exsmokers to havevalues intermediate between those who con-tinue to smoke and those who have neversmoked. Young exsmokers may exhibit high-er spirometric values than never smokers,probably as a result of health selection effect(134, 171). Whether the pulmonary functionof ex-smokers is better or worse than that ofcurrent smokers probaby depends on the ageof the subjects, how long they have smoked,and on why they abandoned smoking.

Cross-sectional and LongitudinalPredictions

Cross-sectional data are subject to a biascalled “cohort” effect. A person who is 40yr of age today is different from one who be-came 40 two decades ago because of a varie-ty of host and environmental factors (6,52).The age-related lung function deficit predictedfrom cross-sectional data tends to be greaterthan that predicted from longitudinal pulmo-nary function data in adults (67-70) and chil-dren (172-174). Prediction equations basedon cross-sectional data are appropriate fordetermining the prevalence of pulmonaryfunction impairment in defined populations.They are less well-suited to determine age-related events including the incidence orprogression of impairment. Percentiles of ad-

AMERICAN THORACIC SOCIEI'Y 1207

, justed lung function (similar to those used portant issues in the selection of referenceby pediatricians to assess growth) have been values. However, neither is as important asadvocated by several investigators for assess- the choice of a reference population that (I)ment of both growth and decline of pulmo- provides an appropriate comparison for thenary function (56, 63, 126). A person would subjects to be evaluated, and (2) is based onbe expected to track along the same percen- measurements made with instruments andtile as he or she ages if the loss (gain) in func- methods comparable to those used in the lab-tion was at a rate comparable to that of the oratory for which reference values are beingreference population. selected (2, 5).

Criteria for Selection ofReference Values

Pubiished Reference Equations

Criteria for selecting reference values to beused in the clinical or in the epidemiologiccontext fall into three categories: methodo-logic, epidemiologic, and statistical (5).

(I) Methodologic criteria. If possible, ref-erence values should be based on data ob-tained by trained operators using equipmentand techniques that meet ATS criteria (7-12).In contrast with the use of the FVC in Ameri-ca, predictions of VC from Europe are usual-ly based on inspiratory vital capacity (IVC)or slow expiratory vital capacity (EVC). TheIVC and EVC are, on average, somewhat larg-er than FVC in healthy subjects; in subjectswith airflow limitation, the differences aremore pronounced (4, 175).

(2) Epidemiologic criteria. The populationfrom which the subjects are drawn should besimilar with respect to age, height, sex, andethnic composition to the population towhom the prediction values are to be applied.Prediction equations should use age, height,sex, and, probably, ethnic group as indepen-dent variables. For most clinical uses theyshould be based on cross-sectional studies oflifetime nonsmokers.

For the convenience of readers, selected pub-lished reference equations for adult whites andblacks and scaling factors for blacks current-ly in use are listed in tables 3 to 9. A compre-hensive listing up to 1983 was published bythe ECCS (4). The results of a survey of ref-erence equations used in North American pul-monary teaching centers is shown in table 10.Equations for children and adolescents aredetailed elsewhere (13,63, 117-119, 131, 149,176, 177). Laboratories should use the pub-lished reference equations that most closelydescribe the populations tested in their labora-tories. This may also be assessed empiricallyby comparing the results for a group of 20to 40 local reference subjects with thoseprovided by the intended reference equations.The local reference subjects should be ap-propriately selected by age, ethnic group, andsex, to match the clientele of the laboratoryand should meet the selection criteria listedin section CRITERIA FOR SELECTION OF REF-ERENCE %LUES.

Limitations of CurrentlyAvailable Equations

(3) Statistical criteria. These are discussedin STATISTICAL CONSIDERATIONS IN THE DERI-VATION OF PREDICTION EQUATIONS . Both bio-logic plausibility and simplicity in the modelused to develop prediction equations are im-

Reference equations now available include rel-atively few results for adolescents and theelderly. Even fewer equations span the agesfrom grade school through adulthood and,with few exceptions, they are discontinuousfor children and adults (55, 178). Older sub-

jects reflect their lifetime experiences with re-spect to nutrition, health status, and otherfactors and are therefore subject to a cohorteffect. Most equations in current use are basedon linear statistical models. All these aspectsare subject to change. For this reason, refer-ence equations should be reviewed regularly.

interpretative StrategiesConceptual Issues Concerning Normality

and the Limits of NormalThe word “normal” is used in a number ofways (5,6, 13, 179). In popular use it meansideal, conventional, or usual. It is used bystatisticians to describe a specific distributionabout a central tendency and by biologists inways that vary according to their focus of in-terest. Anatomists, for instance, use it to de-scribe structural variations consistent withgood function; physiologists use it to describevariations that preserve the “internal milieu,”and clinicians use it to describe variation with-in the limits of “good health” and exclusiveof “disease” (5). Issues of biologic “normali-ty” are discussed in greater detail elsewhere,and interested readers are referred to thosereviews (5, 6, 179-181).

Because most laboratory tests are quantita-tive variables with overlap between measure-ments in healthy and diseased subjects, theidea of a range of values defining biological-ly “normal” is, in the view of its critics, mis-leading (5,6, 182). For instance, in interpret-ing laboratory test results where there is anoverlap between healthy and diseased popu-lations, the “normal” range should theoreti-cally change with different disease processesand with the clinical questions being asked(181). It has also been pointed out that select-ing a normal range “requires careful evalua-tion of benefit in terms of morbidity or mor-tality, inconvenience, and distress caused to

TABLE 3

PREDICTED VALUES FOR FE’.‘, AND FVC DERIVED FROM SELECTED STUDIES OFNONSMOKING CAUCASIAN MEN’

First Author, Age RangeYear (Ref) on) Studied

Morris, 1971 (224) 20-84 517 3.63Cherniack, 1972 (225) 15-79 870 3.74Quanjer. 1977 (4) 21-64 189 3.59Crapo. 1981 (91) 15-91 125 3.96*Knudson, 1983 (149) 25-84 86 3.61Dockery. 1985 (56) 25-74 624 3.78

Rota. 1986 (226) 20-70 443 3.95Paoletti. 1986 (150) 29-64 59 3.83Miller, 1986 (158) 18-85 176 3.94

FEV,t forHt 1.75 m.Age 45 yr

Coeficient

Ht AgeRSD or

SEE

3.62 - 0.0323.59 - 0.0234.05 - 0.0314.14 - 0.0246.65 - 0.029

EquationnonlineaP

4.99 -0.0214.94 -0.0275.66 - 0.023

0.55 4.84

NR 4.520.43 4.510.49 4.89s0.62 4.640.40 4.72

0.440.480.41

WC? for

Ht 1.75 m,Age 45 Y

5.155.064.84

RegressionCoefficient

Ht Age

5.83 - 0.0254.76 -0.0146.11 - 0.0326.00 - 0.0218.44 - 0.030

Equationnonlinear5

6.78 - 0.0157.24 - 0.0277.74 - 0.021

RSD orSEE

0.74N R

0.560.640.640.47

0.530.580.51

Detinilion of abbraviations: RSD = re&ual standard daviatiin; SEE = standard error of ttz8 estimate; NR = not raponad.* To ba included studies had to (I) induda man and women; (2) adequately dascdba the mathods used: (3) anal-e spiromatric values in terms of aga and height. lnstr~mants of measurement

ware: water spirometer (56. 91. 224); dry or wedge spiromater (158. 225); pneumotamograph (4, 149. 150. 226). Equation to pradii f%V, or NC using this table:

Predicted FEV, or WC = Prediiad v&at for Ht 1.75 tn. Age US + Ht Coefficient x (Ht - 1.75) l Age Coefficient x (Age - 45)

t Predicted value for Hr = 1.75 m. Age I 45.*Studies carried out at en alttude of 1.400 m.OFEV, I Ht’(1.541 - 4.06 x lO-" Age - 6.14 x 101 Age’): FVC = HF (1.75 - 1.35 x lo-’ Age - 1.01 x lo-’ Age’).

1208 AMERICAN THoRAac SoaalY

,TABLE 4

PREDICTED VALUES FOR FEV, AND FVC DERIVED FROM SELECTED STUDIES OFNONSMOKING CAUCASIAN WOMEN’

Firat Author,Year (Ref)

FEV,T forRegression Regression

Age RangeCoefficient FVCT for

Number f-6 1.65 m, RSD orCoefficient

Ht 1.65 m. RSD or0 Studied Age45Yr Ht Age SEE Age 45 yr Ht Age SEE

Morris, 1971 (224) 20-w 471 2.72 3.56 - 0.025 0.47 3.54 4.53 - 0.024 0.52Cherniack, 1972 (225) 15-79 452 2.67 2.37 -0.019 NR 3.36 3.08 -0.015 NRQuanjer, 1977 (4) 21-64 514 2.71 3.17 - 0.031 0.35 3.39 4.64 - 0.027 0.42crapo, 1981 (91) 15-64 126 2.92$ 3.42 - 0.026 0.33 3.54$ 4.91 -0.022 0.39Knudson. 1983 (149) 20-87 264 2.79 3.09 - 0.020 0.39 3.36 4.27 - 0.017 0.49Dockery, 1965 (56) 25-74 1,630 2.79 Equation 0.40 3.41 Equation 0.47

nonlinear§ nonlinear5Aoca. 1986 (226) 20-70 427 2.67 3.17 - 0.025 0.31 3.72 4.54 - 0.021 0.40Paoletti, 1986 (150) 21-64 313 2.64 2.43 - 0.020 0.29 3.76 4.12 -0.015 0.39Miller, 1986 (158) 18-62 193 2.91 2.66 - 0.025 0.33 3.59 4.14 - 0.023 0.45

* To be induded studii had to (7) include men and women; Q adequatety describe the memods used: (3) snslyze spimmetnc values in terms of age and height. Instruments of measurementwere: wtef SWometer (66, 91. 224); dry or Wedge spiromater (156. 225); pneumotachograph (4, 149, 150. 226). Equation to predict FEV, or n/C using this table:

Predicted FEV, or FVC = Predicted value* for Ht 1.65 tn. Age 46 + Ht Coefficient x (Ht - 1.65) + Age Coefficient x (Age - 45)

7 Predicted value for Ht = 1.65 m. Age = 4.5 yr.*StUdbScarriedaRat~aitih&of1,400lll.0 = -FEV, Ht* (1.332 4.06 x lO-’ Age - 6.14 x lO-’ m; FVC = HP (1.463 - 1.36 x 10-‘ Age - 1.01 x lo- Age’).

TABLE 5

PREDfCTED VALUES FOR FEV, AND FVC DERIVED FROM SELECTED STUDIES OFBLACK MEN AND WOMEN’

First Author, Age Mean Number FEV, forYear (Ref) or Range Studied Ht and AgeT

RegressionCoefficients

Ht AgeRSD or

SEENC for

Ht and AgeT

RegressionCoefficients

Ht AgeRSD or

SEE

MenJohannaen. 1966 (227)Miller, 1970 (229)Oscbtwitz, 1972 (61)Roaaiter. 1974 (229)Lapp. 1974 (236)Cookaon, 1976 (231)Patrick, 1976 (232)

20-50 12035-a 96

50.3 * 6.6 11021-70 147

34.9 * 11.9 7943.6 -c 15.1 141

16-65 213

Ht 1.75 mAge 45 yr

2.96*3.052.943.043.533.123.11

2.673.402.994.513.542.204.23

Ht 1.75 mAge 45 yr

- 0.017 0.46 4.07$ 4.09 - 0.52- 0.024 0.37 3.79 4.44 - 0.024 0.46- 0.031 0.64 3.76 3.70 - 0.027 0.66- 0.027 0.521 3.64 5.i7 - 0.019 0.596- 0.025 0.23 4.11 3.94 - 0.021 0.32- 0.024 0.50 3.74 3.90 - 0.017 0.65- 0.023 NR 3.72 3.51 - 0.025 NR

WomenHt 1.65 m Ht 1.65 m

Age45yrJohannaen. 1966 (227) 20-50 100 Ag;.:*yr 2.18 - 0.013 0.34 2.74* 2.51 - 0.015 0.35Miller, 1970 (228) 35-54 109 2.19 2.45 - 0.018 0.31 2.74 3.15 - 0.020 0.36Cookaon, 1976 (231) 36.7 -c 11.6 102 2.35 2.40 - 0.026 0.41 2.86 3.00 -0.019 0.42Patrick, 1976 (232) 16-65 117 2.10 1.49 - 0.014 N R 2.64 3.17 - 0.020 NR

’ ln.struments of mdiPurement used were: water spirometer (227.223.231). a dry or bellows spirometer (226.230). and various others (81,232). Predicted values for men and woman are calculat-ed as shown in footnotes to tables 3 and 4.

t Predicted value for a 45yrold man 1.75 m tall. and a 45yr-old woman 1.66 m tall.* Corrected from ATPS to ETPS conditions, assuming a spimmeter temperature of 22O C.5 Inch&s caucasian subjects.

subjects by further investigation and tmat-ment, and the costs of making the wrong de-cision” (182). The “normal” range only givesinformation about the distribution of testresults in the healthy population from whichthey were derived. It says nothing about thetrue positive rate, the false negative rate, orthe predictive power of a positive test.

To draw inferences about the presence ofdisease from a test, one should, ideally, knowthe prior probability that the patient has thedisease and the distributions of test valuesfor subjects with and without the disease inquestion. Although this ideal is rarely met,

clinicians must use their understanding of theclinical situation to put an interpretation inproper perspective

Obstructive and RestrictiveVentilatory Defects

DEFINITION OF AN O~~TIWCTNE DEFECT

An obstructive ventilatory defect may be de-fmed as a disproportionate reduction of max-imal airflow from the lung with respect to themaximal volume (VC) that can be displacedfrom the lung. It indicates airflow limitationand implies airway narrowing during expira-tion. The earliest change associated with flow

limitation in small airways is thought to beslowing in the terminal portion of the spiro-gram,even when the initial part of the spiro-gram is unaffected (1, 21-23). This slowingis reflected in a proportionally greater reduc-tion in the instantaneous flow measured af-ter 75% of the FVC has been exhaled (FEFIS)or in FEFSS-,5a than in FEV,. Abnormalitiesin these midrange flow measurements duringa forced exhalation are, however, not specificfor small airway disease and, though sugges-tive, should not be used to diagnose small air-way disease in individual patients (183). Asairway disease becomes more advanced and/

AYERIcAN THORACIC SOCIETY 1209

TABLE 6

PREDICTED VALUES FOR FEV,/FVC% DERIVED FROM SELECTED STUDIES OF CAUCASIANAND BLACK MEN AND WOMEN’

First Author, Age RangeYear (Ret) (vr)

RegressionFEV,/FVC%T f o r Coefficients FEV,/FVC%T for

RegressionCoefficients

Number Ht 1.75 m and RSD or Number Ht 1.65 m and RSD orStudied Age 45 yr Ht Age SEE Studied Age 45 yr Ht Age SEE

Quanjer. 1977 (4) 21-64 189Crapo, 1981 (91) 15-91 125Knudson, 1983 (149) 25-85 86Paoletti. 1986 (150) 8-64 263Miller, 1986 (158) 18-85 176

Johannsen, 1968 (227)Oscherwitz. 1972 (81)

Rossiter, 1974 (229)Cookson, 1976 (231)

20-5050.3

(* 6.6)21-7043.6

(2 15.1)

120110

147 77.2 0.62 -0.34 7.26141 81.4 - -0.25 10.7

Caucasian Men

78.4 -80.9* - 13.082.0 -75.9 - 5.380.5 - 13.1

Black Men75.0 -i 7 . 7 4.2

-0.16 5.3 514-0.15 4.8 126-0.11 6.3 204- 0.23 6.1 538-0.15 5.6 193

-0.29 8.6- 0.32 10.2

Caucasian Women80.2 - 0.24 6.481.9* - 20.2 -0.25 5.382.8 -18.5 -0.19 7.670.5 - 4.311 - 0.31 5.882.3 -21.5 -0.15 6.8

Black Women

102 82.3 -0.38 11.7

* Table comprttes studies cited in tables 3 to 5. which also reported values for FEV,/FVC% analyzed in relation to height and age. For the instruments of measurement used, see footnotesto tables 3 to 5. Note: studies of Caucasian subjects were confined to nonsmokers: studies of black sub)ects included all smoking categories. Predicted values for FEv,/FVC are calculated asshown in footnotes to tables 3 and 4. Only one study gives equations for black women.

t Predicted vatue for a 45-ywld man 1.75 m tall, and a 45yrold woman 1.65 m tall.* Studies carried out at an attitude of 1.400 tn.5 Includes Caucasian subjects.Ii Coetficient not significant.

TABLE 7

PREDICTED VALUES FOR DIFFUSING CAPACITY (DLCO) AND Kco (DL&VA) DERIVED FROM SELECTED STUDIES OF MEN AND WOMEN’

First Author, Age Mean Number DLcoT forYear (Ref) = SD or Range Studied Ht and Age

RegressionCoefficients

Ht Age

RSD or DL&VAT forSEE Ht and Age

RegressionCoefficients

Ht Age

RSD orSEE

MenBilliet, 1963 (233)Cotes. 1965 (20)Teculescu. 1970 (234)Van Ganse. 1972 (235)Frans, 1975 (236)Marcq, 1976 (237)Satorinne, 1976 (238)Crapo. 1981 (239)Miller, 1983 (165)Paoletti, 1985 (240)Knudson, 1987 (241)Rota. 1990 (242)

WomenSilliet, 1963 (233)Van Ganse. 1972 (235)Salorinne, 1976 (238)Hall, 1979 (243)crapo. 1981 (239)Miller. 1983 (165)Paoletti, 1985 (240)Knudson, 1987 (241)

20-75 5719-72 12719-67 4725-79 70

39 f 12 6417-79 6420-69 6915-91 123

43 f 16 7419-64 8025-64 7120-70 194

20-68 4124-76 7220-69 10127-74 11317-84 122

43 f 15 13018-64 29120-86 99

Ht 1.75 m.Age 45 yr

35.330.332.629.333.329.930.736.6631.437.11138.41133.6

Ht 1.65 m,Age 45 yr

25.220.325.03 0 . 1 ”27.4523.727.91128.21)

57.6 - 0.24 4.232.5 - 0.20 5.133.3 - 0.30 4.216.4 - 0.20 3.828.5 -0.14 4.210.4 - 0.20 3.914.2 - 0.23 3.641.6 - 0.22 4.816.4 -0.23 4.844.1 -0.19 5.835.5 -0.27 4.636.7 - 0.20 4.4

21.9 -0.1616.8 -0.1621.9 -0.1228.3 -0.1925.6 -0.1416.0 -0.1115.7 -0.0718.7 -0.15

3.63.62.84.13.64.04.34.5

Ht 1.75 m,

Age45Yr4.964.835.17*5.68NR

4.595.025.4554.774.81115.6111

- 0.04 0.92- 0 . 0 4 0.81- 0.04 0.73

- 0.90 -0.03 1.07

- 0.03-3.53 - 0.03

-0.03- 2.24 - 0.03-0.12tt - 0.02- 2.35w - 0.04

Equation nonstandard~

0.650.630.840.730.710.80

Ht 1.65 m.Age 45 yr

. 5.55 -0.035.61 -0.17 - 0.015.27 - 3.96 -0.015 .66 ” - 0.025.469 -0.034.62 -1.81 - 0.024.8511 - 2.51 -0.025.3711 - 2.78w - 0.03

0.850.990.740.740.780.800.850.85

* Tabie refers to 0~~0 and includes prediied vatues from published reports in which the number of subjeots studied and their age were given and in which equations for DL~ were deSCribeClin tarrhs of height and age according to AT.6 recommendations (s). All but one study (20) refer to nonsmokers. Residual volume or FRC was measured as foflows: single-breath helium dilution(165.234.236-242). muttipte-breath helium dilution (20.233,243). open cirouit N, washout (235). Predicted values for DLCO and DUVn are calculated as Shown in fwtnotes to tables 3 and 4.

t Predtted vaiue for a 45yr-old man 1.75 m tatl. and a 45yroki woman 1.65 m tall.* Resuits adjusted to 1 alps.g Measurements made at an aftiiude of 1.4X tn.0 Correctiin for breathholding time as in the Epidemiology Standardiation Project (240. 241). Note that calculated DL is sensitiie to the methods used to calculate breathhofd time.1 Form of the equation not that recommended by the AT6** Results caicuiated for atl smoking categories and adjusted for smoking effect.ft Coefficient not significant.

1210 AMERICAN THORACIC SOCIEN

TABLE 8

PREDICTED VALUES FOR TOTAL LUNG CAPACITY (TLC) AND RESIDUAL VOLUME (RV)DERIVED FROM SELECTED STUDIES OF MEN AND WOMEN-

Reareseion &ore&on

First Author, Age Mean NumberCoefficients

RSD orCoefficients

RSD orYear (Ref) or Range Studied Ht and Age Ht Age SEE Ht and Age Ht Age SEE

MenGoldman, 1959 (92)Cotes, 1965 (20)Boren, 1966 (155)Black, 1974 (244)Crapo, 1982 (245)

44 f 17 4419-72 12720-62 4221659 8315-91 123

Ht 1.75 m, Ht 1.75 m,Age 45 yr Age 45 yr

6.61 9.40 -0.015 0.65 2.04 2.70 0.017 0.396.68 8.67 - 0.91 Not reported6.35 7.80 - 0.87 1.62 1.90 0.012 0.536.84 7.80 - 0.68 2.15 3.80 0.034 0.576.72 7.95 0.003 0.79 1.67 2.16 0.021 0.37

Ht 1.65 m, Ht 1.65 m.Women Age 45 yr Age 45 yr

Goldman, 1959 (92) 38 f 16 50 5.10 7.90 - 0.008 0.53 1.78 3.20 0.009 0.37Grimby. 1963 (246) 18-72 58 5.05 7.31 - 0.016 0.52 1.44 2.92 0.008 0.35Black. 1974 (244) 16-59 110 5.20 6.40 - 0.62 1.76 2.30 0.021 0.46Hall, 1979 (243) 27-74 113 5.30 7.46 - 0.013 0.51 1.80 2.80 0.016 0.31Crapo, 1982 (245) 17-04 122 5.20 5.90 - 0.54 1.73 1.97 0.020 0.38

' Only one (245) Of thesa studies Conforms strictly to the AT.9 rscommandations for splmmetry (8): references 20. 155. 243. 244 included all smoking categories. and in two (92. 246) smokingStatus was not defined. R&dual volume was fttaawred as follows: halium rebreathing (20,92, 243.246). whole-body plethysmograph (244). single-breath helium dilution (245). and helium rebreath-ing on open circuit N, washout III one study (155). Predicted values for TLC and RV are calculated as shown in fwtnwss to tables 3 and 4.

7 Predicted value for a 45yr-old man 1.75 m tall. and a 45yrold woman 1.65 m tall.

or more proximal airways become involved,timed segments of the spirogram such as theFEV, will become reduced out of proportionto the reduction in VC.

DEFINITION OF A RESTRICTIVE DEFECT

A restrictive ventilatory defect is character-ized physiologically by a reduction in TLC.One may infer the presence of a restrictiveventilatory defect when VC is reduced andFEVJFVC is normal or increased. Severe air-flow limitation is another common cause ofa reduced VC either because airflow is so slowthe subject cannot continue to exhale longenough to complete emptying or because air-ways collapse Occasionally, patients will havea small VC, a normal FEVJFVC, and a nor-mal TLC. If there is a contradiction betweenVC and TLC in defining restriction the clas-sification should be based on TLC.

TABLE 9

FACTORS FOR ADJUSTING REFERENCEVALUES FOR CAUCASIANS WFH A

VIM TO THEIR BEING USEDFOR BLACK AMERICANS’

FEV,W CFEV,lFVCTLCRVRVKLCDiising Capecity (transfer fector)TINA (BTPS)

0.66t0.68t00.880.93$1.050.931.05

* 6wrce: Rossiter and Weill with annotatipn (229). Althoughthe average Caucasian admixture in studies of 6&k Anwiicans varies. a raasonable average is 22% (247).

t Also apply to women younger than 55 yr of age; in Oldersubjack. the cofracticn may bs larger (approximately 0.80;Dcckery ef al. [56]).

* A larger correction (approximately 0.88) was pmpnsad byLapp et al. (230).

Bronchodilator ResponseBronchial responsiveness is an integratedphysiologic mechanism involving airway ep-ithelium, nerves, mediators, and bronchialsmooth muscle. Because the within-individualdifference in response to a series of differentbronchodilators is variable, and as many as20 to 30% of responsive subjects will respondto one type of agent but not to another (l&t),the assumption that a single test of bronchodi-lator response is adequate to assess both theunderlying airway responsiveness and thepotential for therapeutic benefits of bron-chodilator therapy is overly simplistic (185).The correlation between bronchoconstrlctionand bronchodilator responses is imperfect,and it is not possible to infer with certaintythe presence of one from the other.

Data on the percent change in FVC, FEVI,and FEF,,S,, after bronchodilator adminis-tration in general population studies as wellas in patient populations are summarized in

table 11. These studies showed a tendency forthe calculated bronchodilator response to in-crease with decreasing baseline VC or FEV,,whether response was considered as an abso-lute change or as a percent of the initial val-ue Bronchodilator responses in patient-basedstudies are, not surprisingly, somewhat high-er than those in general population studies(table 11).

Interpretation of change after a bronchodi-lator should be made in light of the clinicalquestion. If the question is whether a patienthas an increased bronchodhator response, theappropriate reference is probably one of thepopulation-based studies. If the question iswhether the patient is different from otherpatients or from previous visits, patient groupsmay provide the most appropriate referencedata.

There is no clear consensus on what con-stitutes reversibility in subjects with airflowobstruction (192). In part, this is because there

TABLE 10

SURVEY OF SPIROMETRY REFERENCE EQUATIONS USED INNORTH AMERICAN PULMONARY TRAINING CENTERS’

NC or VC FEV, FEV,lFVCT

M F M F M F

Morris ef al. (224) 65 65 65 65 58 60Crap0 et al. (91) 27 27 27 27 29 29Knudson et al. (149) 24 24 25 25KoryeteL(153) 7 8Koty et al. (249) 7 8Chemiack et d. (225) 3 3 4 4Miller et a/. (180) 2 2 2 2 2 2Other studies* 11 11 6 8 11 9

- &sad on a questionnaire survey of adutt raspiratoty diiass training pmgrams in the United Statasand Canada. Responsas from 139 of 160 institutions are summarizad (248).

7 Thiny-nine centers predictad FEV,IFvC by dividing predicted FtZV, by pradictad FVC.* Studies cited only once.

AMERICAN THORACIC SOCIETY 1211

c TABLE 11

RESPONSE TO BRONCHODILATOR: RESULTS FROM SELECTED POPULATION STUDIES

Population

1,063 subjects 8-75 yr of ageGeneral population samplefrom Tucson, AZ (186)

Agent/Mode of Delivery W C

Two inhalations of isoproterenol via 10.7%metered-dose inhaler (403 ml)

FEV,

7.7%(315 ml)

FEF,,.,,,or FEF,

20%

Comments

95th percentile for percentage change frombaseline (absolute value in parentheses)

2.609 subjects; random sampleof three areas in Alberta,Canada (187)

75 selected normal subjects

(188)

500 pg terbutaline administeredvia spacer

Two inhalations from aBronkometer”’ metered-doseinhaler

Females 9% - 95th percentile for percentage change from(224 ml) baseline in asymptomatic never smokers

Males 9% with FEV, > 89% predicted (absolute(338 ml) value in parentheses)

5.1% 10.1% 48.3% Upper 95% confidence limits (two-tailed)(231 ml) (365 ml) for percentage change from baseline

RESPONSE TO BRONCHODILATOR: RESULTS FROM SELECTED PATIENT STUDIES

40 patients referred topulmonary function lab (189)

985 patients with COPDparticipating in the IPPBtrial (190)

150 patients with airwayobstruction (191)

Placebo

256 pg isoproterenol aircompressor nebufiier

200 pg salbutamol or 500 ugterbutaline via metered-doseinhaler

14.9%(340 ml)

15%

(330 mf)

12.3%(178 ml)

15%

10%(160 ml)

45.1% Upper 95% confidence interval changeafter placebo inhalation. Absolutevalues in parentheses.

Average increase as percent of initial FEV,(5% as percent of predicted normalFEV,)

95% confidence interval for absolutechange: absolute rather than relativechange preferred measure ofbronchcdilator response

is no consensus on how a bronchodilator re-sponse should be expressed. The three mostcommon methods are: percent of the initialspirometric value, percent of the initialpredicted baseline value, and absolute changeExpressing the change in FEV, as a percentof predicted FEV, deserves further study asit has been reported to have advantages overcurrent methods (193). When using the per-cent change from the initial values as thecriterion, most authorities would require atleast a 12 to 15% increase in FEV, from thebaseline value as necessary to define ameaningful response Increments of less than8% (or of less than 150 ml) are likely to bewithin measurement variability (191,192). Oneshould interpret improvement in an individualsubject only if the percent change and abso-lute change in FEV, or VC are clearly beyondthe expected variability of the measurementduring a single testing session. A patient mayrespond to long-term bronchodilator thera-py even though a bronchodilator response isnot seen in a single laboratory testing session.

The FEF21_,Sk is a highly variable spiro-metric test, in part because of its dependenceon FVC, which increases with expiratory timewith obstruction. If FVC changes, postbron-chodilator FEF,,,,, is not comparable withthat’measured prebronchodilator. Volume ad-justment of FEF2,_,5s. has been used to dealwith this issue (194,195). At least two studieshave assessed the utility of FEF,+,,,. Theresults were disappointing, with only 8% ofasthmatics (195) and 7% of patients withchronic obstructive pulmonary disease(COPD) (196) identified by FEF2,,S, criteriaalone as outside the expected range Tests suchas the FEVJVC ratio and flow rates mea-

sured at some fraction of the VC may alsobe misleading in assessing bronchodilator re-sponse if expiratory time changes are not con-sidered and if flows are not measured at thesame volume below TLC.

Current published criteria and the Work-shop recommendations for determining bron-chodilator response are given in table 12.

Interpretation of Lung Function Testsin Clinical Practice

Pulmonary function tests may be used to ad-dress major issues in clinical case manage-ment. These include describing dysfunctionand assessing its severity, explaining it in termsof diagnosis, establishing prognosis, planningmanagement, and assessing trends over time,including changes after treatment. pulmonaryfunction tests may also be used to identifyan abnormality in subjects without a knownpulmonary disorder, as in preoperative assess-ments, in routine health status evaluations,and in clinical screening. Finally, pulmonaryfunction tests are increasingly requested as

part of health assessment on behalf of a thirdparty (cg., an insurance company or a gov-ernmental agency) where the clinician is notin his or her usual patient advocacy role andthe subject or patient is, consequently, wary.In each of these situations, the question askedof the puhnonary function laboratory is quitedifferent. Ideally, interpretations of pulmo-nary function tests should depend on the pur-pose of the tests and, when performed on pa-tients with known disease, should be orient-ed to answering the specific question of theclinician ordering the procedure Tests inter-preted without clinical information will belimited in their clinical utility and the interpre-tation will usually represent only a refineddescription of the data obtained.

The first step in interpreting a lung func-tion test is to evaluate the quality of the study.If there are reasons to suspect the quality ofthe test, avoid specific diagnostic statements.Dysfunction discovered under these circum-stances should indicate only the need for moredefinitive testing.

TABLE 12

RECOMMENDED CRlTERlA FOR RESPONSE TOA BRONCHODILATOR IN ADULTS

FVC FEV, FEF,7,~Organiaation (%) (%) (%) Comments

American College of Chest Physicians (197) 15-25 15-25 15-25 % of baseline in at leasttwo of three tests

Intermountain Thoracic Society (19) 15 12 45 % of baseline

ATS (current document) 12 12 % of baseline and anabsolute change of 200 ml

1212

PATTERNS OF DYSFUNCTION

Certain patterns of physiologic abnormali-ties can be recognized, and although they areseldom if ever pathognomonic for a specificdisease entity, the types of clinical illnessesmost likely to produce the observed set ofphysiologic disturbances can be pointed out.Regardless of the extent of testing, the mostimportant point with regard to pattern rec-ognition is the need to be conservative withrespect to suggesting a specific diagnosis forthe underlying disease process based only onpulmonary function abnormalities. Recogni-tion of characteristic patterns of dysfunctiondepends a great deal the comprehensivenessof the lung function evaluation. However,even with only spirometric results, one candetermine whether the pattern is compatiblewith obstruction with or without a reductionin VC. A reduced VC without evidence of ex-piratory slowing is a nonspecific finding.There was controversy among Workshop par-ticipants about using the term “restrictive”when VC is low. The majority thought it wasacceptable to interpret the finding as indicat-ing a “restrictive type of ventilatory impair-ment,” or a “restrictive ventilatory defect”while recognizing that it does not necessarilyindicate restrictive lung disease Others arguedthe interpretation should be descriptive only,in, simply noted as “reduced vital capacity”or “nonobstructive defect,” and call for fur-ther testing, including lung volumes, to clarifyits nature.

The VC, FEV,, and FEVJVC ratio are thebasic parameters used to interpret spirome-try. Although FVC is often used in place ofVC it is preferable to use the largest VC,whether obtained on inspiration (IVC), slowexpiration (EVC), or forced expiration (FVC),for clinical testing. The FVC is usually reducedmore than IVC or EVC in airflow obstruc-tion. Limiting primary interpretation of spiro-grams to three variables avoids the problemof simultaneously examining a multitude ofmeasurements to see if any abnormalities arepresent, a procedure that will lead to an inor-dinate number of “abnormal” tests among thehealthiest groups in a population (198, 199).Even when the rate of abnormality for anysingle test is only 5%, the frequency of at leastone abnormal test was shown to be 10% in251 healthy subjects when FEV,, FVC, andFEVJFVC ratio were examined and increasedto 24% when a battery of 14 different mea-surements were analyzed (198).

The FEV,/VC ratio is the most importantmeasurement for distinguishing an obstruc-tive impairment. Expiratory flow measure-ments other than the FEV, and FEVJVCshould be considered only after determiningthe presence and clinical severity of obstruc-tive impairment using the basic values men-tioned above When FEV, and the FEVJVCratio are within the expected range, abnor-malities in flow occurring late in the maxi-mal expiratory flow-volume (MEFV) curveshould not be graded as to severity, and, ifmentioned, interpretation of their clinical sig-nificance should be guarded. In the presence

of a borderline value for FEVJFVC, how-ever, they may help confirm the presence ofairway obstruction. The same is true for av-erage flows such as FEF25-7501.. Even whenused in this limited way, the wide variabilityof these tests in healthy subjects must be tak-en into account in their interpretation.

One should be cautious in interpreting ob-structive dysfunction when the FEV, and VCare both above predicted even when theFEV,/VC ratio is below the lower limit of nor-mal since this pattern is sometimes seen inhealthy subjects, including athletes. Tests oth-er than spirometry, including lung volumes,diffusing capacity, and blood gas determina-tions allow amplifying statements on the over-all pattern of the dysfunction observed dur-ing spirometry.

UWER LIMITS OF “NORMAL”m Cumx~ INTERPRETATION

Lower limits of normal are often used in clin-ical practice without thoughtful reflectionabout their inherent variability (5, 180, 181,200-207) or their implications (5, 182). (Seealso sections DIST~I~JT~ON AND I_ownn LamsOF Nom, DETERM~ATION OF rrr~ NORMALRANGE, and CONCEPTUAL ISSUES CONCERN-ING NORMALITY AND THE Lr&irrS OF NORMAL.)Although clinical interpretation is usuallystraightforward when a pulmonary functionresult is well above or below a “lower limitof normal,” this is not so when a measuredvalue falls close to the “lower limit of nor-mal.” Predicting the presence or absence ofdisease requires knowledge about the distri-bution of dysfunction in various disease statesand the prior probability of disease. For ex-ample, consider the meaning of a spiromet-ric study that shows FEV, values and otherexpiratory flow mtes to be just above the lowerlimit of normal. If the patient were a healthymale who sought medical assistance becausehe was disqualified for life insurance on thebasis of his spirometry, it would be appropri-ate to interpret his spirometry as within nor-mal limits. If, in contrast, the same data wereobtained from a smoker with complaints ofintermittent coughing and occasional wheez-ing, it would be appropriate to suggest thatthe study is consistent with mild obstructivedysfunction, although it could also representa variant of normal. In both of these instances,computer printouts, or robotic physician in-terpretation that simplistically declare theresults to be “normal” or “abnormal” on thebasis of whether the observed values fall toone side or the other of a single number, couldgive information that does not perform a use-ful service to the patient. One suggestion forminimizing the problems of overly simplisticuse of the lower limits of “normal” in the in-terpretation of lung function tests is use ofterms such as ‘Unusually low” rather than “ab-normal” for tests close to the lower limit ofnormal.

ASSESS~G S~nnrr~Severity scores are most appropriately derivedfrom studies that relate pulmonary function

AMERICAN THORAUC SOCIETY

test values to independent indices of perfor-mance such as ability to work and functionin daily life, morbidity, and prognosis (208-212). For instance, in general, ability to workand to function in daily life relates to one’spulmonary function level. FVC and/or FEV,,which also relate to maximal 90~ and workeffort, are used in several published systemsto rate impairment (208, 209). Pulmonaryfunction level is also associated with morbidi-ty; those with lower function having more re-spiratory complaints (212). Lung function lev-el is also associated with prognosis, includ-ing a fatal outcome from heart as well as lungdisease (213, 214) even in patients who havenever smoked (215). In the Framingham study,vital capacity was a major independent predic-tor of cardiovascular morbidity and mortali-ty (213,214). In several occupational cohortsFEV, and FEVJFVC were independentpredictors of all cause or respiratory diseasemortality (216-218). In addition, a meta-analysis of mortality in six surveys in variousU.K. working populations showed that therisk of dying of COPD was related to FEV,level. In comparison to those whose FEV, atinitial examination was within 1 SD of aver-age, those whose FEV, was more than 2 SDbelow average were 12 times more likely todie of COPD, over 10 times more likely todie of non-neoplastic respiratory disease, andmore than twice as likely to die of vasculardisease over a 20-yr follow-up period (219).A reduced FEV, also carries a 4- to S-foldexcess risk of lung cancer mortality (adjust-ed for cigarette smoking) (220,221). Althoughthere is good evidence that FEV, correlateswith the severity of symptoms and prognosisin many circumstances (208. 211, 212, 219),the correlations do not allow one to accuratelypredict symptoms or prognosis for individu-al patients.

In clinical practice, predicted values are alsoused to grade severity. The severity of thespirometric abnormality is usually based onthe actual or percent predicted FEV, in thecase of obstructive disorders or on VC innonobstructive disorders. An example of analgorithm sometimes employed for gradingseverity when nothing is known about the clin-ical question being asked is shown in table13. It is intended only as an example and notas a standard. Its approach is based as muchon clinical impression as on objective data.Although clinical experience has alwaysplayed a major role in assessing severity, itcan be enhanced by more exact methods, andphysicians should probably view arbitraryseverity scoring systems with caution.

Comments on the severity or significanceof any abnormality depend on the circum-stances under which a test is obtained. Forexample the assessment of severity of obstruc-tion illustrated in table 13 may be relevant toCOPD, but it would not be applicable to apatient with tracheal stenosis whose obstruc-tion could be life-threatening and yet classi-fied as only mildly reduced by this scheme

The VC has some relationship to the ex-tent of loss of functioning lung parenchyma

AMERICAN THORAUC SOCIETY 1213

. TABLE 13

EXAMPLE OF CRITERIA FOR ASSESSING THE SEVERITY OF ABNORMALITIES’

A. Normal: The test is interpreted as “within normal limits” if both the VC and the FEV,NC ratio are inthe normal range.

B. Obstructive abnormality: This is interpreted when the FEV,NC ratio is below the normal range. Theseverity of the abnormality might be graded as follows:

“May be a physiological variant”‘Mild”“Moderate”‘Moderately severe”‘Severe”“Very severe”

% Pred FEV, > 100% Pred FEV, < 100 and >, 70% Pred FEV, < 70 and > 60% Pred FEV, < 60 and 2 50% Pred FEV, < 50 and 2 34% Pred FEV, < 34

C. Restrictive abnormality: This is most reliably interpreted on the basis ef TLC. If this is not available.one may interpret a reduction in the VC without a reduction of the FEV,NC ratio as a “restriction of thevolume excursion of the lung.” The severity of the abnormality might be graded as follows:

Based on the TLC“Mild” % Pred TLC < LLN but B 70“Moderate” % Pred TLC < 70 and B 60“Moderately severe” % Pred TLC < 60

Based on spirometry“Mild” % Pred VC < LLN but 3 70“Moderate” % Pred VC < 70 and B 60“Moderately severe” % Pred VC < 60 and 3 50Severe” % Pred VC < 50 and 2 34“Very severe” % Pred VC < 34

Definition of abbreviation: LLN = lower limit of normal._ This schema was contributed by Burrows and Lebowitz. It has been in use in the lung function laboratory at the Health

Sciences Center in Tucson. Ariuona for clinical purposes. It is intended only as an example of a transparent schema for assessingseverity. Other schema may be acceptable as well. More work is required before any schema can be adopted as a standard. Note:All statements regardmg severity should be accompanied by a disclaimer such as “as assessed by spirometry” or “physiologicassessments of seventy may differ from clinical assessments.”

TABLE 14

CHANGE IN SPIROMETRIC INDICES OVER TIME

Percent Chanaes Reauired to be Sianificant

FVC FEV, FEFz-7,

Within a day (222)Normal subjectsPatients with COPD

Week to week (222)Nonal subjectsPatients with COPD

25 35 2 13>, 11 3 13 > 23

> 11 > 12 >, 21a 2 0 220 2 30

Year to year (69) > 15 >, 15

in many nonobstructive lung disorders. It isalso of some use in assessing respiratory mus-cle involvement in certain neuromuscular dis-eases. Here again, however, the VC may beonly slightly impaired in diffuse interstitialdiseases of sufficient severity to lead tomarked loss of diffusing capacity and severeblood gas abnormalities, and a relatively smalldecrement in VC may indicate the onset ofa severe respiratory problem in patients witha rapidly progressive neuromuscular disease.

The FEVJVC ratio should not be used inisolation to determine the severity of an ob-structive disorder. Both the FEV, and VC maydecline with progression of disease, and anFEVJVC of OS/l.0 indicates more impair-ment than one of 2-O/4.0, though both yielda ratio of 50%. Systems that use FEV,/FVCto grade the severity of obstruction must dealwith the effect of total expiratory time on FVCand FEVJFVC (19).

C H A N G E S I N SPIROMEIRY OVER TYMEReliance should be placed on FEV, and VCfor examining changes over time as they arethe only spirometric variables that will con-sistently and correctly reflect the direction ofthe change in overall ventilatory function.Even using these simple tests, it is never easyto determine whether a change is “real” oronly a result of test variability. All lung func-tion measurements tend to be more variablewhen made weeks to months apart than whenrepeated at the same test session or even daily(222, 223). Changes should therefore be in-terpreted cautiously. It is more likely that areal change ha.s occurred when there are a se-ries of tests that show a consistent trend. Asshown in table 14 significant changes, whetherstatistical or biologic, vary by parameter, timeperiod, and the type of patient. For FVC andVC in healthy subjects, within-day change of5% or more, between-weeks changes of 11 to

12% or more, and yearly change of 15% ormore were generally thought by the Workshopto be clinically important.

The clinician seeing the patient can ofteninterpret results of serial tests in a useful man-ner, not reproducible by any simple algorithm.For example, seemingly stable tests may provevery reassuring in a patient receiving therapyfor a disease that is otherwise rapidly progres-sive. The same tests may be very disappoint-ing if one is treating a disorder that is expect-ed to improve dramatically with the therapyprescribed. Depending on the clinical situa-tion, statistically insignificant trends in func-tion may be very meaningful to the clinician.The greatest errors occur when one attemptsto interpret serial changes in subjects with-out disease because test variability will usu-ally far exceed the true annual decline, andreliable rates of change for an individual sub-ject cannot be calculated without prolongedfollow-up (69). Thus, in subjects with “nor-mal” lung function, changes in VC or FEV,over 1 yr should probably exceed 15% (table14) before any confidence can be given to theopinion that a meaningful year-to-year changehas occurred.

RecommendationsOverall

TECHNICAL ISSUES

Although technical sources of variation inspirometry have been fully dealt with in oth-er documents, it was considered importantto reemphasize their key role, particularly inrelation to the following points.1. Laboratory directors should be constantly

on guard to maintain the precision and ac-curacy of the measurements made in theirlaboratories and should be aware of thepotential sources of technical variation.Quality control includes strict adherenceto ATS guidelines for equipment perfor-mance and calibration.

2. Attention should be given to the spirome-ter temperature where the tests are per-formed. Temperature-related errors will bereduced when the spirometer temperatureis between 17’ and 40’ C.

3. Computer calculations should be validat-ed at the time equipment is purchased andafter any changes are made in software orhardware.

BIOUXZIC VARIAnoN ANDSTATLSTrCAL ISSUES

1. Laboratory directors should be aware ofthe biologic Sources of within- and between-individual variation in order to optimizethe application of lung function tests to aparticular patient. A number of within-individual sources of variation fall withinthe domain and control of the laboratory,whereas between-individual sources of vari-ation are important in selecting appropri-ate reference values.

2. Environmental sources of variation perti-nent to a given patient are more likely tobe known to the referring clinician than to

1214

the laboratory director and should be usedin evaluating the clinical pertinence of agiven lung function report. Laboratorydirectors should request this informationfrom clinicians.

3. Those who generate and report lung func-tion tests should be aware of the strengthsand weaknesses of the statistical techniquesused to generate the prediction values usedfor interpretation. Laboratory directors andchest physicians should also be aware ofthe strengths and limitations of the statisti-cal concepts of normality.

Selecting Reference ValuesGENERAL CONSIDERATIONS

1. Because of unexplained differences be-tween published reference values, no oneset of reference values is likely to be ap-plicable to all laboratories and all clienteleunder all circumstances. The choice of ref-erence values should be a matter of carefulconsideration by laboratory directors. Itshould not be left to the judgment ofmanufacturers of automated equipment.

2. Laboratories should indicate the source ofreference values on their reports.

3. Ideally, reference values should be basedon data obtained using equipment and pro-cedures that conform to current ATS recom-mendations. The prediction equations listedin tables 3 and 4 and published since 1981conform to current ATS recommendations.

EPIDEMIOLOGIC CONS~DEXM~ONS1. Reference values should not come from

studies based on hospital patients.2. Reference values for most clinical applica-

tions should be based on cross-sectionalstudies.

3. Subjects used to generate reference valuesshould be free of respiratory symptoms anddisease. It is preferable to choose referencevalues for men and women from the samepopulation source.

4. Reference equations based on nonsmok-ers should be used for most clinical appli-cations. The problems in making adjust-ments for the biologic effects of smokinglead to the recommendation that such ad-justments should not be part of routine clin-ical interpretation. Such adjustments may,occasionally, be made to address specificquestions.

5. Altitude may be important in the selectionof reference values for flow rates and Droo.

STATISTICAL CONSIDERATIONS

1. Prediction equations for adults shouldinclude age and height as independent vari-ables. Usually, separate equations are usedfor men and women.

2. Linear equations perform adequately foradults though they may overpredict inyoung adults and underpredict in theelderly.

3. Prediction equations should come fromstudies that present lower limits of normalor present information from which suchlower limits can be calculated.

4. Reference equations should, in general, not

be extrapolated for ages or heights beyondthose covered by the data that generatedthem. If, for example, one calculates apredicted FEV, for an 85-yr-old personfrom prediction equations based on a popu-lation younger than 65 yr of age, the re-port should contain a cautionary statement.

5. The choice of reference values should con-sider the ethnic origins of the clientele ofthe laboratory. Although it is preferable touse equations based on the ethnic originsof the subject being tested, this is not al-ways possible or practical. For instance, ifa laboratory only occasionally serves sub-jects of a particular ethnic group, it is ac-ceptable to adjust for ethnic differences byusing a scaling factor as suggested in table 9.

LLWER LIMITS OF NORMAL1. Normal ranges should be based on calcu-

lated fifth percentiles. Estimates of fifthpercentiles based on the SEE are accept-able for indices with distributions that areclose to Gaussian.

2. Lower limits of normal are variable and,therefore, should not be considered as ar-bitrary limits that correctly classify all pa-tients into normal and abnormal groups.Patient values that lie close to lower limitsshould be interpreted with caution.

3. The use of 80% of predicted for a lowerlimit of normal for adult pulmonary func-tion parameters is not recommended. Thiscriterion works only for average personsand for a limited number of parameters.It creates major errors when applied toFEFzs_,s, and the instantaneous flows.Fixed percent of predicted values may beacceptable in children.

4. In adults, it is not acceptable to used a fixedFEVJFVC ratio as a lower lit of normal.

OTHER CON~IDERAX-~ONS1. It is preferable for North American labora-

tories to select reference value studies basedon North American populations and Eu-ropean laboratories studies based on Eu-ropean populations because an importantportion of the variation between popula-tion studies remains unexplained.

2. To assist in the choice of reference values,it may be useful to make an empirical as-sessment of how different equations relateto measurements made in 20 to 40 healthysubjects typical of the laboratory’s clien-tele. If the distribution of these measure-ments is, on the whole, within the rangepredicted, the choice is probably suitableIf this is not the case, the differences maybe due to the laboratory (apparatus, tech-nician, procedure) or it may be that the ref-erence values are inappropriate for thelaboratory’s clientele. Both possibilitiesshould be considered.

Recommendations for InterpretationOvERAu CLW~~AL IN~ERPR~~A~~ON

1. Because interpretation of the lung func-tion tests of an individual patient is bestmade in light of the clinical question askedof the tests, the clinician requesting the test

AMERICAN THORACIC SOCIETY

should frame this question as precisely aspossible. Likewise, the laboratory directorresponsible for seeing that the tests are car-ried out should insist that the clinical ques-tion be included in the requisition.

2. Interpreters of lung function tests shouldbe conservative in suggesting a specific di-agnosis based only on pulmonary functionabnormalities.

3. Borderline “normal” values should be in-terpreted with caution. Such interpretationsshould, when possible, use clinical infor-mation in the decisions as to what is nor-mal and what is abnormal.

4. The first step in interpretation is to evalu-ate and comment on the quality of the tests.

5. The number of test indices (e.g., FVC,FEV,, etc.) used in interpretation shouldbe limited to avoid an excessive number offalse positive results.

6. The primary guides for spirometry interpre-tation should be VC (slow or forced), FEV,,and FEVJVC.

7. Tests performed on children are best inter-preted by those familar with pulmonaryfunction in children.

CONCERNING AIRWAY OBSTRUCTION1. FEVJVC should be the primary guide for

distinguishing obstructive from nonob-structive patterns.

2. Instantaneous and mid flows may be usedto confirm the presence of airway obstruc-tion in the presence of a borderlineFEV,/VC.

3. FEF,,-,5a and the instantaneous flowsshould not be used to diagnose small air-way disease in individual patients.

4. The pattern of a low FEVJVC ratio andgreater than average VC and FEV, shouldbe recognized as one that may occur inhealthy individuals.

5. The severity of airway obstruction shouldbe based on FEV, rather than FEVJVC.

6. Abnormalities in instantaneous flows andFEF,,_,sr should not be graded as toseverity when FEV, and FEVJVC arewithin the normal range

CONCERNING BRONCHODTLATORR~spoNsn

1. VC (forced or slow) and FEV, should bethe primary indices used to judge bron-chodilator response. Total expiratory timeshould be considered when using FVC toassess bronchodilator response since FVCincreases in obstructed patients as expira-tory time increases.

2. A 12% increase, calculated from theprebronchodilator value, and a 200-ml in-crease in either FVC or FEV, are reason-able criteria for a positive bronchodilatorresponse in adults.

3. FEFlr,,a and the instantaneous flowsshould be considered secondarily in evalu-ating bronchodilator response If used, theymust be volume-adjusted or the effect orchanging FVC must be dealt with in theinterpretation.

4. Ratios such as FEVJVC should not be usedto judge bronchodilator response

.

AUERNXN THORlIClC SoclETy

n 5. Patients may respond to bronchodilatortherapy even though a bronchodilator re-sponse is absent in a laboratory test.

CONCERNING RESTRICTION1. The diagnosis of a restrictive lung abnor-

mality is based on a reduced TLC. A re-duced VC in the presence of a normalFEV,/VC may be used to suggest but notdiagnose the presence of restriction.

2. The severity of restriction should be basedon TLC. If VC is used to infer the presenceof restriction, severity may be based on VC.

This Statement was prepared by the par-ticipants of a Workshop on Lung FunctionTesting: Selection of Reference Values and In-terpretative Strategies. The workshop partic-ipants were: MARGARET BECKLAKE, M.D., andROBERT 0. m M.D., Co-Chairpersons, A.SONIA Buts-r, M.D., BENJAMIN BURROWS,M.D., JACK L. CLAUSEN, M.D., ALLAN L.COATES, M.D., JOHN COTES, D.M., Douu~sW. DOCICERY , PH.D., REED M. GARDNER,PH.D., JOHN L. H-SON, PH.D., JA M E S

HANLEY, PH.D., ROBERT L. JO H N S O N, JR . ,M.D., MICHAEL D. LEBOWITZ, PH.D., PAOL~PAOLETTI, M.D., RENE PESLIN, M.D., GEORGEPOLGAR, M.D., Prrnlp H. QUANJER, M.D.,MELVYN S. TOCKMAN, M.D., Seen-r T. WEISS,

.M.D., MS., MARY ELLEN B. WOHL, M.D.

References1. Bates DV. Respiratory function in disease. 3rded. Philadelphia: WB Saunders, 1989.2. Clausen JL, ed. Pulmonary function testing:guidelines and controversies. New York: Academ-ic Press, 1982.3. Miier A, ed. Pulmonary function tests: a guidefor the student and house officer. Orlando, FL:Grune & Srratton, 1987.4. Quanjer PhH, ed. Standardized lung functiontesting: report of the working party. Bull Eur Phys-iopathol Respir 1983; 19(Suppl 5:1-95).5. Be&lake MR. Concepts of normality appliedto the measurement of lung function. Am J Med1986; 80:1158-63.6. Buist AS. Evaluation of lung function: Con-cepts of normality. In: Simmons DH, ed. Currentpulmonology. Vol 4. New York: John Wiley andSons, 1982; 141-65.7. American Thoracic Society. Snowbird work-shop on standardization of spirometry. Am RevRespir Dis 1979; 119831-8.8. American Thoracic Society. Standardization ofspirometry: 1987 update Am Rev Respir Dis 1987;1361285-98.9. American Thoracic Society. Single-breath car-bon monoxide diffusing capacity (transfer factor).Recommendations for a standard technique AmRev Respir Dis 1987; 136:1299-307.10. Gardner RM, Clausen JL, Epler G, Hankin-son JL, Permutt S. Plummer AL. Pulmonary func-tion laboratory personnel qualifications. Am RevRespir Dis 1986; 134~623-4.11. Gardner RM, Clausen JL, Crap0 RO, ef al.Quality assurance in pulmonary function labora-tories Am Rev Respir Dis 1986; 134~625-7.12. Gardner RM, Clausen JL, Cotton DJ, Crap0RO, Hankinson JL, Johnson RL. Computer guide-lines for pulmonary laboratories. Am Rev RespirDis 1986; 134628-g.13. Quanjer PhH, Helms P, Bjure J, Gaultier C.Standardimnon of lung function tests in paediatrics.Em Respir J 1989; 2(Suppl 4:121S-264s).

14. Ferris BG Jr, ed. Epidemiology standardiza-tion project. Part 2. Am Rev Respir Dis 1978;118:1-120.15. Pulmonary terms and symbols: a report of theACCP-ATS Joint Committee on Pulmonary No-menclature. Chest 1975; 67:583-93.16. Jay F, ed. IEEE standard dictionary of elec-trical and electronic terms. 3rd ed. New York: In-stitute of Electrical and Electronics Engineers, Inc,1984.17. Nelson SB, Gardner RM, Crap0 RO, JensenRL. Performance evaluation of contemporaryspirometers. Chest 19%. 92288-97.18. Pistelli G, Carmignani G, Paoletti P, et al.Comparison of algorithms for determining the endpoint of the forced vital capacity maneuver. Chest1987; 91:100-5.19. Morris AH, Kanner RE, Crap0 RO, GardnerRM. Clinical pulmonary function testing: a man-ual of uniform laboratory procedures. 2nd ed. SaltLake City, UT: Intermountain Thoracic Society,1984.20. Cones JE. Lung function: assessment and ap-plication in medicine 4th ed. Oxford: BlackwellScientific Publications, 1979.21. Wilson AF, ed. Pulmonary function testing,indications and interpretations Orlando, FL; Grune& Stratton, 1985.22. Burrows B, Knudson RJ, Quan SF, Kettel LJ.Respiratory disorders: a pathophysiologic approach.Chicago: Year Book Medical Publishers, 1983.23. Enright PL. Hyatt RE. Office spirometry: apractical guide to the selection and use of spirome-ters. Philadelphia: Lea & Febiger, 1987.24. Allen SM, Hunt B, Green M. Fall in vital ca-pacity with posture Br J Dis Chest 1985; 79267-71.25. Pierson DJ, Dick NP. Petty TL. A compari-son of spirometric values with subjects in standingand sitting positions. Chest 1976; 70~17-20.26. Blair E, Hickam JB. The effect of change inbody position on lung volume and intrapulmonarygas mixing in normal subjects. J Clin Invest 1955;34383-g.27. Townsend MC. Spirometric forced expirato-ry volumes measured in the standing versus the sit-ting posture. Am Rev Respir Dis 1984; 130~123-4.28. Glindmeyer H. Predictable confusion. J Oc-cup Med 1981; 23845-g.29. Melissinos CG, Mead J. Maximum expirato-ry flow changes induced by longitudinal tensionon trachea in normal subjects. J Appl Physiol1977;43z537-44.30. Liistro G, Stanescu D, Dooms G. RodensteinD, Veriter C. Head position modifies upper airwayresistance in men. J Appl Physiol1988: 64~1285-8.31. Krowka MJ, Enright PL, Rodarte JR, HyattRE. Effect of effort on measurement of forced ex-piratory volume in one second. Am Rev Respir Dis1987; 136:829-33.32. Ingram RH Jr, Schilder DP. Effect of thorac-ic gas compression on the flow-volume curve ofthe forced vital capacity. Am Rev Respir Dis 1966,9456-63.33. Sadoul P. Mesure de la capacite vitale et desdebits maximaux. In: Denolin H, Sadoul P, OrieNGM, eds. L’exploration fonctionnelle pulmonaire2nd partie Paris: Flammarion, 1971; 389.34. Van de Woestijne KP, Afschrift M. Airwaydynamics during forced expiration in patients withchronic obstructive lung disease. In: Orie NGM,Van der Lende R, eds. Bronchitis III. Assen: RoyalVan Gorcium, 1970; 195-206.35. Gimeno F, Berg WC, Sluiter HJ, TammelingGJ. Spiromeuy-induced bronchial obstruction. AmRev Respir Dis 1972; 105:68-74.36. Eisen E, Dockery DW, Speizer FE, Fay ME,Ferris BJ Jr. The association between health statusand the performance of excessively variable spirom-etry tests in a population-based study in six U.S.

12l5

cities. Am Rev Respir Dis 1987; 136:1371-6.37. Becklake MR. Enidemioloav of test failureBr J Ind Med 1990, 47~73-4. -_38. Hetzel MR. The pulmonary clock. Thorax1981; 36481-6.39. Guberan E, Williams MK, Walford J, SmithMM. Circadian variation of FEV, in shift work-ers. Br J Ind Med 1969; 26:121-5.40. Hetzel MR, Clark TJH. Comparison of nor-mal and asthmatic circadian rhythms in peak expi-ratory flow rate Thorax 1980; 35~732-8.41. Clark TJH, Hetzel MR. Diurnal variation ofasthma. Br J Dis Chest 1977; 71:87-92.42. McDermott M. Diurnal and weekly cyclicalchanges in lung airways resistance J Physiol (Lond)1966; 18690.43. De Millas H, Ulmer WT. Circadian rhythmsof airways resistance in healthy subjects and pa-tients with obstructive airways disease Pneumologie1971; 144237-52.44. Kerr HD. Diurnal variation of respiratory func-tion independent of air quality. Arch EnvironHealth 1973; 26144-52.45. Barnes PJ. Circadian variation in airway func-tion. Am J Med 1985; 79(Suppl 6A:5-9).46. Barnes P, Fitzgerald G, Brown M, Dollery C.Nocturnal asthma and changes in circulating epi-nephrine, histamine and cortisol. N Engl J Med1980; 303:263-7.47. Hetzel MR, Clark TJH, Branthwaite MA.Asthma: analysis of sudden deaths and ventilato-ry arrests in hospital. Br Med J 1977; 1:808-H.48. Bass LR. Hushes DTD. Diurnal variation inpeak ex&atory flow in asthmatics. Eur J RespirDis 1980; 61:298-302.49. Halberg F. Chronobiology and the lung: im-plications and applications. Bull Eur PhysiopatholRespir 1987; 23:529-31.50. Lewinsohn HC, Cape1 LH, Smart J. Changesin forced expiratory volumes throughout the day.Br Med J l%o, 1:462-4.51. Minette A. Contribution to the chronobiolo-gy of lung function: changes of baseline values offour lung function indices between 8h and 17h inpatients with broncbitic complaints without asth-matic components. Bull Eur Physiopathol Respir1987; 23:541-3.52. Buist AS, Vollmer WM. The use of lung func-tion tests in identifying factors that affect lunggrowth and aging. Stat Med 1988; 7:11-8.53. Ferris BG Jr, Stoudt HW. Correlation of an-thropometry and simple tests of pulmonary func-tion. Arch Environ Health 1971; 22~672-6.54. Hibbert ME, Lanigan A, Raven J, Phelan PD.Relation of armspan to height and the predictionof lung function. Thorax 1988; 43:657-g.55. Schoenberg JB, Beck GJ, Bouhuys A. Growthand decay of pulmonary function in healthy blacksand whites. Resuir Physiol 1978; 33:367-93.56. Dockery DW, Ware JH, Ferris BG Jr, ef ui.Distribution of forced expiratory volume in onesecond and forced vital capacity in healthy, white,adult never-smokers in six U.S. cities. Am Rev RespirDis 1985; 131:511-20.57. Howatt WF, DeMuth GR. II. Configurationof the chest. Pediatrics 1%5; 35:177-84.58. Damon A. Negro-white differences in pulmo-nary function (vital capacity, timed vital capacity,and expiratory flow rate). Hum Biol 1966; 38:381-93.59. Marcus EB, Buist AS, Curb JD, et al. Corre-lates of FEV, and prevalence of pulmonary condi-tions in Japanese-American men. Am Rev RespirDis 1988; 138r1398-404.60. DeGroodt EG, van Pelt W, Quanjer PhH, vanZomeren BC, Borsboom GJJM. Growth of lungand thorax dimensions during the pubertal growthspurt. Eur Respir J 1988; 1:102-8.61. Green M, Mead J, Turner JM. Variability of

1216 AMERICAN THORACIC SOCIETY

maximum expiratory flow-volume curves. J ApplPhysiol 1974; 3267-74.62. Mead J. Dysanapsis in normal lungs assessedby the relationship between maximal flow, staticrecoil, and vital capacity. Am Rev Respir Dis 1980;121:339-42.63. Schwartz JD, Katz SA, Fegley RW, TockmanMS. Analysis of spirometric data from a nationalsample of healthy 6- to 24-year-olds (NHANES II).Am Rev Respir Dis 1988; 138:1405-14.64. Schwartz DJ. Katz SA, Fe&y RW, TockmanMS. Sex and race differences ii the developmentof lung function. Am Rev Respir Dis 1988;138:1415-21.65. Becklake MR, Permutt S. Evaluation of testsof lung function for “screening” for early detec-tion of chronic obstructive lung disease. In: Mack-lem PT, Permutt S, eds. The lung in the transitionbetween health and disease New York: Marcel Dek-ker, 1979; 345-88.66. Vollmer WM, Johnson LR, McCamant LE,Buist AS. Methodological issues in the analysis oflung function data. J Chronic Dis 1987; 40~1013-23.67. Vollmer WM, Johnson LR, McCamant LE,Buist AS. Longitudinal versus cross-sectional esti-mation of lung function decline: further insights.Stat Med 1988; 7:685-%.68. Louis TA, Robins J, Dockery DW, Spiro AIII, Ware JH. Explaining discrepancies between lon-gitudinal and cross-sectional models. J Chronic Dis1986; 39831-g.69. Burrows B, Lebowitz MD, Camllli AE, Knud-son RJ. Longitudinal changes in forced expiratoryvolume in one second in adults. Methodologic con-siderations and fiidings in healthy nonsmokers.Am Rev Resdr Dis 1986: 133:974-80.70. Van Pelt W, Quanjer.PhH, Borboom G. Co-hort effects in ventilatory function and chronic re-spiratory symptoms in a 12 year follow-up study(abstract). Eur Respir 3 1989; 2(Suppl 5:366s).71. Weisrnan IM, Zeballos RH. Lower single breathcarbon monoxide dlffuslng capacity (Dt& in blacksubjects compared with Caucasians. Chest 1987;92(Suppl:142S).72. National Center for Health Statistics. Bodyweight, stature, and sitting height: white and Ne-gro youths 12-17 years. Washington, DC: HealthServices and Mental Health Administration, 1973.(Vital and Health Statistics. Series 11, no. 126)(DHEW publication no. mRAS]74-1608).73. National Center for Health Statistics. Bodydimensions and proporrions, white and Negro chil-dren 6-11 years Washington, DC: Health ResourcesAdministration, 1974. (Vital and Health Statistics.Series 11, no. 143) (DHEW publication no. [HRAS]75-1625).74. National Center for Health Statistics. Forcedvital capacity of children 6-11 years. Washington,DC: Public Health Service, 1978. (Vital and HealthStatistics. Series 11, no. 164) (DHEW publicationno. pHS]78-1651).75. National Center for Health Statistics. Skinfolds, body girths, bicromial diameter and select-ed anthropometric indices of adults. Washington,DC: Public Health service, 1970. (Vital and HealthStatistics. Series 11, no. 35) (DHEW publicationno. [PHS] 100-11-35).76. National Center for Health Statistics. Select-ed body measurements of children 6-11 years.Washington, DC: Health Services and MentalHealth Administration (Vital and Health Statis-tics. Series 11, no. 123) (DHEW publication no.[HSM]73-1605).77. Van de Wal BW, Emsmus LD, Hechter R. Stemand standing heights in Bantu and white South Afri-cans: their significance in relation to pulmonaryfunction values. S Afr Med J 1971; 45(Suppl:568-70).78. Woolcock AJ, Coleman MH, Blackbum CRB.

Factors affecting normaI values for ventilatory lungfunction. Am Rev Respir Dis 1972; 106:692-709.79. Cotes JE, Anderson HR, Patrick JM. Lungfunction and the response to exercise in New Guin-cans: role of genetic and environmental factors. Phi-10s Trans R Sot Lond [Viol] 1974; 268349-61.80. Miller GJ, Saunders MJ, Gin RJ, et UL Lungfunction of healthy boys and girls in Jamaica inrelation to ethnic composition, test exercise per-formance, and habitual physical activity. Thorax1977; 32~486-96.81. Oscherwitz M, Edlavitch SA, Baker TR, ef al.Differences in pulmonary functions in various ra-cial groups. Am J Epidemiol 172; %:319-27.82. Mueller WH, Yen F, Soto P, et al. A multina-tional Andean genetic and health program. VIII.Lung function changes with migration between al-titudes. Am J Phys Anthropol 1979; 51:183-95.83. Brown P, Sadowsky D, Gajdusek DC. Ven-tilatory lung functions studies in Pacific IslandMicronesians. Am J Epidemiol 1978; 108:259-65.84. Myers J. Differential ethnic standards for lungfunction or onestandard for all? S Afr Med J 1984;65~768-72.85. Steinberg M, Becklake MR. Socio-environ-mental factors and lung function. S Afr Med J 1986;70~270-4.86. Burrows B, Taussig LM. As the twig is bent,the treeinclines (perhaps). Am Rev Respir Dis 1980;122:813-6.87. Burrows B, Knudson RJ, Lebowitt MD. Therelationship of childhood respiratory illness to adultobstructive airway disease Am Rev Respir Dii 1977;115:751-60.88. Bertrand JM, Riley P, Popkin JA, Coates A.The long term pulmonary sequelae of prematuri-ty: the role of familial airway hyperreactivity andthe respiratory distress syndrome. N Engl J Med1985; 312~742-5.89. U.S. Department of Health and Human Ser-vices. The health consequences of smoking: chronicobstructive lung disease, a report of the SurgeonGeneral. Rockville, MD: Public Health Service Of-fice on Smoking and Health, 1984.90. Kryger M, Aldrich F, Reeves JT, Grover RF.Diagnosis of airflow obstruction at high altitudeAm Rev Respir Dls 1978; 117:1055-8.91. Crap0 RO, Morris AH, Gardner RM. Refer-ence spirometric values using techniques and equipment that meet ATS recommendations. Am RevRespir Dis 1981; 123:659-64.92. Goldman HI, Becklake MR. Respiratory func-tion tests. Normal values at median altitudes andthe prediction of normal results. Am Rev Tuberc1959; 79457-67.93. Gautier H, Peslin R, Grassino A, et al. Me-chanical properties of the lungs during acclimati-zation to altitude. J Appl Physiol1982; 52:1407-15.94. Manse11 A, Powles A, Sutton J. Changes inpulmonary PV characteristics of human subjectsat an altitude of 5366 m. J Appl Physiol1980; 49:79-83.95. Jaeger JJ, Sylvester JT, Cymerman A, et (II.Evidence for increased intrathoracic fluid volumein man at high altitude. J Appl Physiol 1979;47:670-6.%. Frisancho AR. Functional adaptation to highaltitude hypoxia. Science 1975; 187:313-g.97. Heath D. Williams DR. The lung at hiah alti-tude. Invest Cell Path01 1979; 2147-56. -98. Lippmann M. Health significance of pulmo-nary function responses to irritants. J Air PollutCont Assoc 1988; 38:881-7.99. Goldsmith JR. Effects on human health. In:Stern A, ed. Air pollution. Vol VI. New York: Aca-demic Press, 1986; 391-463.100. Committee on the Epidemiology of Air Pol-lutants. Epidemiology and air pollution. Washing-ton, DC: National Academy Press, 1985.

101. Lippmann M. Health effects of ozone: acrit-ical review. J Air Pollut Cont Assoc 1989; 39672-95.102. Shy CM. Health effects of acid aerosols. Re-view, discussion and summary of epidemiologicalstudies. Environ Health Perspect 1989; 79187-90.103. Linn WS, Theodore G, Venet TG, et al. Re-spiratory effects of sulfur dioxide in heavily exer-cisina asthmatics. Am Rev Resoir Dis 1983: 127:278-83.104. Department of Health, Education and Wel-fare The health consequences of involuntary smok-ing: a report from the Surgeon General. Washing-ton, DC: USGPO, 1986. (DHHS publication no.[CDC]87-8398).105. Fielding JE, Phenow KJ. Health effects ofinvoluntary smoking. N Engl J Med 1988; 319:1452-60.106. Spitaer WO, Lawrence V, Dales R, et aL Linksbetween passive smoking and disease: a best evi-dence synthesis. Clin Invest Med 1990; 13:17-42.107. Lebowitz MD. Ouackenboss JJ. The effectsof environmental tobacco smoke on pulmonaryfunction. Int Arch Environ Health 1989; 44(Suppl:147-52).108. Masi M, Hanley JA, Ernst P, Becklake MR.Environmental exposure to tobacco smoke and lungfunction in young adults. Am Rev Respir Dis 1988;138:296-g.109. Martinez FD, Antognoni G, Macri F, et al.Parental smoking enhances bronchial responsive-ness in nine-year-old children. Am Rev Respir Dis1988; 138:518-23.110. Sarnet JM, Marbury MC, Spengler JD.Health effects and sources of indoor air pollution.Part I. Am Rev Respir Dis 1987; 136:1486-508.111. Sarnet JM, Marbury MC, Spengler JD. Healtheffects and sources of indoor air pollution. PartII. Am Rev Respir Dis 1988; 137~221-42.112. Cotes JE, Steel J. Work related lung disord-ers. oxford Blackwell Scientific Publications, 1987;436.113. Becklake MR. Occupational exposures: evi-dence of a causal association with chronic obstruc-tive pulmonary disease. Am Rev Respir Dis 1989;H&585-91.114. Chart-Yeung M, Lam S. Occupational asth-ma. Am Rev Respir Dis 1986; 133:686-703.115. Lebowitz MD. The relationship of socio-environmental factors to the prevalence of obstruc-tive lung diseases and other chronic conditions. JChronic Dis 1977; 303599-611.116. Viegi G, Paoletti P, Prediletto R, et al. Prev-alence of respiratory symptoms in an unpollutedarea of northern Italy. Eur Respir J 1988; 1:311-8.117. Polgar G, Weng TR. The functional devel-opment of the respiratory system: from the periodof gestation to adulthood. Am Rev Respir Dis 1979;120:625-95.118. Polgar G, Promodhat V. Pulmonary func-tion testing in children: techniques and standards.Philadelphia: W B Saunders, 1971; 272.119. Quanjer PH, Stocks J, Polgar G, Wise ME,Karlberg J, Borsboom G. Compilation of referencevalues for lung function measurements in children.Eur Respir J-1989; Z(Supp1 4:184S-261s).120. DeGroodt EG, Ouanier PH. Wise ME. vanZomeren BC. Changing rel&onships betwkstat-ure and lung volumes during puberty. Respir Physiol1986; 65:139-53.121. Cole TJ. Linear and proportional regressionmodels in the prediction of ventilatory function.J R Stat Sot [A] 1975; 138:297-338.122. Schrader PC, Quanjer PhH, van ZomerenBC, Wise ME. Changes in the FEV,-height rela-tionship during pubertal growth. Bull Eur Phys-iopathol Respir 1984; 20~381-8.123. Seely JE, Guzman CA, Becklake MR. Heartand lung function at rest and during exercise inadolescence. J Appl Physiol 1974; 36:34-40.

AMERICAN THORACIC SOCIETY 1217

* 124. Tanner JM. Growth at adolescence, with ageneral consideration of the effects of hereditaryand environmental factors upon growth and matu-ration from birth to maturity. 2nd ed. Oxford:Blackwell, 1978.125. Simon G, Reid L, Tanner JM, Goldstein H,Benjamin B. Growth of radiologically determinedheart diameter, lung width, and lung length from5-19 years, with standards for clinical use ArchDis child 1972; 47:373-81.126. Dockery DW, Berkey CS, Ware JH, SpeizerFE, Ferris BG Jr. Distribution of forced vital ca-pacity and forced expiratory volume in one secondin children 6 to 11 years of age. Am Rev RespirDis 1983; 128405-12.127. Rea H, Becklake MR, Ghezzo H. Lung func-tion changes as a reflection of tissue aging in-youngadults. Bull Eur Physiopathol Respir 1982; 18:5-19.128. Black LF, Hyatt RE. Maximal respiratorypressures: normal values and relationship to ageand sex. Am Rev Resnir Dis 1969: 99:696-702.129. Smyth RJ, Chapman KR, Rebuck AS. Max-imal inspiratory and expiratory pressures in adoles-cents: normal values. Chest 1984; 86568-72.130. Wagner JS, Hibbert ME, Landau LI. Maxi-mal respiratory pressures in children. Am Rev RespirDis 1984; 129873-5.131. Wilson SH, Cooke NT, Edwards RHT, SpiroSG. Predicted normal values for maximal respira-tory pressures in Caucasian adults and children.Thorax 1984, 39535-8.132. Cook CD, Mead J, Orzalesi MM. Staticvolume-pressure characteristics of the respiratorysystem during maximal efforts. J Appl Physiol1964,191016-U.133. Gaultier C, Zinman R. Maximal static pres-sures in healthy children. Respir Physiol 1983;51:45-61.134. Leech JA. Ghezzo H, Stevens D, BecklakeMR. Respiratory pressures and function in youngadults. Am Rev Respir Dis 1983; 128:17-23.135. Schrader PC, Quanjer PhH, Olievier ICW.Respiratory muscle force and ventilators function inadolescents. Eur Respir J 1988; 1:368-75.136. Sham JT. Druz WS. Balaaot RC. BandelinVR, Danon J. Total respiratory compliance in in-fants and children. J Appl Physiol1970; 29:775-g.137. Fagan DG. Shape changes in static V-P loopsfrom children’s lungs related to growth. Thorax1977; 32:198-202.138. Zapletal A, Paul T, Samanek M. Pulmonaryelasticity in children and adolescents. J Appl Physiol1976; 0953-61.139. Baran D, Yernault JC, Paiva M, Englert M.Static mechanical lung properties in healthy chil-dren. Stand J Respir Dis 1976; 57:139-47.-140. ManselI AL. Brvan AC. Levison H. Relation-ship of lung recoil to-lung volume and maximumexpiratory flow in normal children. J Appl Physiol1977; 42:817-23.141. De Troyer A, Yemault JC, Englert M, BaranD, Paiva M. Evolution of intrathoracic airwaymechanics during lung growth. J Appl Physiol1978;44:521-7.142. Turner JM, Mead J, Wohl ME. Elasticity ofhuman lungs in relation toage J Appl Physiol1968;25664-71.143. Knudson RJ, Clark DF, Kennedy TC, Knud-son DL. Effect of aging alone on mechanical prop-etties of the normal adult human lung J Appl Phys-iol 1977; 43:1054-62.144. Cotes JE, Dabbs JM, Hall AM, HeywoodC, Laurence KM. Sitting height, fat free mass andbody fat as reference variables for lung functionin healthy British children: comparison with stat-ure Ann Hum Biol 1979; 6307-14.145. Leeder SR, Swan AV, Peat JK, Woolcock AJ,Blackbum CRB. Maximum expiratory flow-volumecurves in children: changes with growth and in-

dividual variability. Bull Eur Physiopathol Respir1977; 13:249-60.146. Thurlbeck WM. Postnatal human lunggrowth. Thorax 1982; 37:564-71.147. Taussig LM, Cota K, Kaltenborn W. Differ-ent mechanical properties of the lung in boys andgirls. Am Rev Respir Dis 1981; 123640-3.148. Duiverman EJ, Clement J, van de WoestijneKP, Neijens HP, van der Bergh ACM, KerrebijnKF. Forced oscillation technique. Reference valuesfor resistance and reactance over a frequency spec-trum of 2-26 Hz in healthy children aged 2.3-12.5years. BullEur Physiopathol Respir 1985; 21:171-8.149. Knudson RJ, Lebowitz MD, Holberg CJ,Burrows B. Changes in the normal maximal expi-ratory flow-volume curve with growth and aging.Am Rev Respir Dis 1983; 127:725-34.150. Paoletti P, Pistelli G, Fazzi P, et al. Refer-ence values for vital capacity and flow-volumecurves from a general population study. Bull EurPhysiopathol Respir 1986; 22:451-g.151. Lebowitz MD, Holbem CJ. Comoarisons ofspirometric reference values and the proportionsof abnormals among male smokers and those symptomatic in a community population. Am Rev RespirDis 1990, 141:1491-6.152. Van Ganse W, Biiiet L, Ferris BG. Medicalcriteria for the selection of normal subjects. In: Ar-cangeli P, Cotes JG, Cournand A, et al, eds. Nor-mal values for respiratory function in man. A&hero:Pamninerva Medica, 1970; 15-27.153. Kory RC, Callahan R, Boren HG, Syner JC.The Veterans Administration-Army cooperativestudy of pulmonary function. 1. Clinical spirome-try of normal men. Am J Med 1961; 30:243-58.154. De Kroon JPM, Joosting PE, Visser BE Lesvaleurs not-males de la capacite vitale et du volumeexpiratoire maximum seconde Recherche chez lesouvriers d’une acierie aux Pays-Bas. Arch Mal Prof1964, 25:17-30.155. Boren HG, Kory RC, Syner JC. The Veter-ans Administratiott-Army cooperative study of pul-monary function. II. The lung volume and its sub-divisions in normal men. Am J Med 1966; 41:96-114.156. Amrein R, Keller R. Joos H, et al. Valeurstheoriques nouvelles de l’exploration de la fonc-tion ventilatoire de poumon. Bull PhysiopatholRespir 1970; ti317-49.157. Jouasset D. Nom&&ion des epreuves fonc-tionnelles respirator& dans les pays de la Com-mtmaute Europe&me du Charbon et de l’Acier. Pou-mon Coeur 1960; 161145-59.158. Miier A, Thornton JC, Warshaw R, Bem-stein J, Selikoff IJ, Teirstein AS. Meanand instan-taneous expiratory flows, FVC and FEV,: predic-tion equations from a probability sample of Michi-gan, a large industrial state. Bull Eur PhysiopatholRespir 1986; U:589-97.159. Miller MR, Pincock AC. Predicted values:how should we use them? Thorax 1988; 43:265-7.160. Mler A. Prediction equations and “normalvalues.” In: Miller A, ed. Pulmonary function testsin clinical and occupational lung disease New York:Grune & Stratton, 1986; 197-213.161. Oldham PD. Percent of predicted as the lim-it of normal in pulmonary function testing: astatistically valid approach. Thorax 1979; 34569.162. Sobol BJ, Weinheimer B. Assessment of ven-tilatory abnormality in the asymptomatic subject:an exercise in futility. Thorax 1%; 21:445-g.163. Quanjer PhH. Predictedvalues: how shouldwe use them (letter). Thorax 1988; 43663-4.164. Docket-y DW, Speizer FE, Ferris BG Jr, WareJH, Louis TA, Spiro A III. Cumulative and revers-ible effects of lifetime smoking on simple tests oflung function in adults. Am Rev Respir Dis 1988;137:286-92.165. Miller A, Thornton JC, Warshaw R, Ander-son H, Teirstein AS, Selikoff IJ. Single breath

diffusing capacity in a representative sample of thepopulation of Michigan, a large industrial state.Am Rev Respir Dis 1983; 127:270-7.166. Beck GJ, Doyle CA, Schachter EN. Smok-ing and lung function. Am Rev Respir Dis 1981;123:149-55.167. Burrows B, Knudson RJ, Cline MG, LebowitzMD. Quantitative relationships between cigarettesmoking and ventilatory function. Am Rev RespirDis 1977; 115:195-205.168. Fletcher C, Peto R, Tinker C, Speizer FE.The natural history of chronic bronchitis and em-physema. London: Oxford University Press, 1976.169. Enjeti S, Hazelwood B, Permutt S, MenkesH, Terry P. Pulmonary function in young smok-ers: male-female differences. Am Rev Respir Dis1978; 118:667-76.170. Higenbottam T, Clark TJH, Shipley MJ, RoseG. Lung function and symptoms of cigarette smok-ers related to tar yield and number of cigarettessmoked. Lancet 1980; 1:409-11.171. Becklake MR, Lallo U. The “healthy smok-er” effect: a phenomenon of health selection. Respi-ration 1990, 57:137-44.172. Strope GL, Helms RW. A longitudinal studyof spirometty in young black and young white chil-dren. Am Rev Respir Dis 1984, 130:1100-7.173. Pattishall EN, Helms RW, Strope GL. Non-comparability of cross-sectional and longitudinalestimates of lung growth in children. Pediatr Pul-mono1 1989; 7~22-8.174. Hibbert ME, Lannigan A, Landau LI, Phe-lan PD. Lung function values from a longitudinalstudy of healthy children and adolescents. PediatrPulmonol 1989; 7:101-9.175. Bencowitz HC. Inspiratory and expiratotyvital capacity. Chest 1985; 85:834-5.176. Taussig LM, Chemiack V, Wood R, FarrellPM, Mellins RB. Standardization of lung functiontesting in children. Recommendations of the GAPConference Committee, Cystic Fibrosis Foundation.J Pediatr 1980; 97:668-76.177. Zapletal A, Samanek TP. Lung function inchildren and adolescents: Methods and referencevalues. Basel-Munchen: Karger, 1987.178. Burrows B, Cline MG, Knudson RJ, TaussigLM, Lebowtiz MD. A descriptive analysis of thegrowth and decline of the FVC and FEV,. Chest1983; 83:717-24.179. Galen RS. The normal range: a concept intransition. Arch Path01 Lab Med 1977; 101:561-5.180. Zweig MH, Beck JR, Collinsworth WL.Assessment of clinical sensitivity and specificityof laboratory tests. Proposed guideline Pbiladel-phia: National Committee for Clinical Laborato-ry Standards, 1987 (NCCLS Document GP 10-Pvol. 7 no. 6).181. Henry RJ, Reed AH. Normal values and theuse of laboratory results for the detection of dis-ease. In: Hem-y RJ, Cannon DC, Winkelman JW,eds. Clinical chemistry: principles and techniques.New York: Harper & Row, 1974; 343-70.182. Cochrane AL, Elwood PC. Laboratory dataand diagnosis (letter). Lancet 1%9; 1:420.183. Flenley DC. Chronic obstructive puhnonarydisease. Dis Mon 1988; 34537-99.184. Mendella LA, Manfreda J, Warren CPW, An-thonisen NR. Steroid response in stable chronic ob-structive pulmonary disease Ann Intern Med 1982;%:17-21.185. Guyatt GH, Townsend M, Nogradi S, Pugs-ley SO, Keller JL, Newhouse MT. Acute responseto bronchodilator, an imperfect guide for bron-chodilator therapy in chronic airflow limitation.Arch Intern Med 1988; 148:1949-52.186. Lorber DB, Kaltenborn W, Burrows B. Re-sponses to isoproterenol in a general populationsample. Am Rev Respir Dis 1978; 118:855-61.187. Dales RE, Spitzer WO, Tousignant P,

12l8 AMERICAN THORACIC SOCIETY

Schechter M, Suissa S. Clinical interpretation ofairway response to a bronchodilator: epidemiologicconsiderations. Am Rev Resir Dis 1988: 138:317-20.188. Watanabe S, Rentetti AD Jr, Begin R, Big-ler AH. Airway responsiveness to a bronchodila-tor aerosol. Am Rev Respir Dis 1974; 109:530-7.189. Sourk RL, Nugent KM. Bronchodilator test-ing: confidence intervals derived from placebo in-halations. Am Rev Respir Dis 1983; 128:153-7.190. Anthonisen NR, Wright EC, and the IPPBTrial Group. Bronchodilator response in chronicobstructive pulmonary disease Am Rev Respir Dis1986; 133:814-g.191. Tweeddale PM, Alexander F, McHardy GJR.Short term variability in FEV, and bronchodilatorresponsiveness in patients with obstructive ventila-tory defects. Thorax 1987; 42:487-90.192. Ehasson 0, Degraff AC Jr. The use of criteriafor reversibility and obstruction to define patientgroups for bronchodilator trials: influence of clin-ical diagnosis, spirometric and anthropometricvalues. Am Rev Respir Dis 1985; 132858-64.193. Brand PLP, Quanjer PhH, Postma DS, etuf. A comparison of different ways to express bon-chodilator response Part 2. Am Rev Respir Dis1990; 14l:A20.194. Olsen CR, Hale FC. A method for interpret-ing acute response to bronchodilators from thespirograms. Am Rev Respir Dis 1%8; 98301-2.195. Boggs PB, Bhat KD, Vekovius WA, Debo MS.The clinical significance of volume adjusted maxi-mal mid expiratory flow (Ho-volume FEFu_,&in assessing airway responsiveness to inhaled bron-chodilator in asthmatics. Ann Allergy 1982; 48:139-42.1%. Berger R, Smith D. Acute postbronchodila-tor changes in pulmonary function parameters inpatients with chronic airways obstruction. Chest1988; 93:541-6.197. Committee report. Criteria for the assessmentof reversibility in airways obstruction: report of thecommittee on emphysema, American College ofChest Physicians. Chest 1974; 65:552-3.198. Vedal S, Crap0 RO. False positive rates ofmultiple pulmonary function teats in healthy sub-jects. BullEur Physiopathol Respir 1983; 19263-6.199. Knudson RJ, Burrows B, Lebowita MD. Themaximal cxpiratory flow-volume curve: its use inthe detection of ventilatory abnormalities in a popu-lation study. Am Rev Respir Dis 1976; 114871-g.200. Ash KO, Clark SJ, Sandberg LB, Hunter E,Woodward SC. The influences of sample distribu-tion and age on reference intervals for adult males.Am J Clin Path01 1983; 79574-81.201. Herrera L. The precision of percentiles in es-tablishing normal limits in medicine J Lab ClinMed 1958; 5234-42.202. Shulz EK. Willard KL Rich SS. et UL Im-proved reference-interval estimation. Clin Chem1985; 31:1974-8.203. Dybkaer R. lntemational Federation of Clin-ical Chemistry (IFCC)l),2) The theory of referencevalues Part 6. Presentation of observed values relat-ed to reference values. J Clin Chem Clin Biochem1982; 20841-5.204. Elveback LR, Taylor WF. Statistical methodsof estimating percentiles. Ann NY Acad Sci 1969;161:538-48.205. Linnet K. Two-stage transformation systemsfor normalization of reference distributions evalu-ated. Clin Chem 1987; 33:381-6.206. Solberg HE. T’he theory of reference values.Part 5. Statistical treatment of collected referencevalues. Determination of reference limits IFCC Sci-entific Committee proposed recommendations. JClin Chem Clin Biochem 1983; 21:749-60.207. Gardner MJ, Altman DG, eds. Statistics with

confidence Br Med J 1989; l-140.208. Becklake MR. Kalica AR. NHLBI workshopsummary. Scientific issues in the assessment of re-spiratory impairment. Am Rev Respir Dis 1988;137:1505-10.209. Evaluation of impairment/disability second-ary to respiratory disorders. Am Rev Respir Dis1986; 133:1205-g.210. Kanner RE, Renzetti AD, Stanish WM. Bark-man HW, Klauber MR. Predictors of survival insubjects with chronic airflow limitation. Am J Med1983; 74249-55.211. Traver GA, Cline MG, Burrows B. Predictorsof mortality in chronic obstructive pulmonary dis-ease. Am Rev Respir Dis 1979; 119:895-902.212. Anthonisen NR, Wright EC, Hodgkin JE.Prognosis in chronic obstructive pulmonary dis-ease. Am Rev Respir Dis 1986; 133:14-20.213. Kannel WB, Lew EA. Vital capacity as apredictor of cardiovascular disease: the Ftamin-gham study. Am Heart J 1983; 105:311-5.214. Kannel WB, Lew EA, Hubert HB, CastelliWP. The value of measuring vital capacity for prog-nostic purposes. ‘Bans Assoc Life Insur Med DirAm 1981; 6466-83.215. Tockman MS, Comstock GW. Respiratoryrisk factors and mortality: longitudinal studies inWashington county, Maryland. Am Rev Respir Dis1989; 14O(Suppl:56-63).216. Foxman B, Higgins ITT, Oh MS. The effectsof occupation and smoking on respiratory diseasemortality. Am Rev Respir Dii 1986; 134x549-52.217. Annesi I, Kauffmann F. Is respiratory mu-cus hypersecredon really an innocent disorder? AmRev Respir Dis 1986; 134:688-93.218. Ottmeyer CE, Costello J. Morgan WKC,Swecker S, Peterson M. The mortality of Ap-palachian coal miners 1963 to 1971. Arch EnvironHealth 1974; 29:67-72.219. Peto R, Speizer FE, Cochrane AL, et al. Therelevance in adults of air-flow obstruction, but notof mucous hypersecretion, to mortality from chroniclung disease: results from 20 years of prospectiveobservation. Am Rev Respir Dis 1983; 128:491-500.220. Tockman MS, Anthonisen NR, Wright EC,Donithan MG. Airways obstruction and the riskof lung cancer. Ann Intern Med 1987; 106:512-8.221. Kuhn LH, Ockene J, Meilahn E, SvendsenKH. The relation of forced expitatory volume inone second (FEV,) to lung cancer mortality in themultlule risk factor Intervention trial IMRFIT). AmJ Epihemiol 1990; 132235-74. .222. Lebowita MD, Quackenboss J, Camilli AE,Bronnlmann D, Holberg CJ, Boyer B. The epide-miological importance of intraindividual changesin objective pulmonary responses. Eur J Epidemi-01 1987; 3:390-8.223. Cochrane GM, Prieto F, Clark TJH. In-trasubject variability of maximal expiratory flowvolume curve. Thorax 1977, 32171-6.224. Morris JF, Koski A, Johnson LC. Spiromet-ric standards for healthy nonsmoking adults. AmRev Respir Dis 1971; 103:57-67.225. Chemiack RM, Raber MB. Normal standardsfor ventilatoty function using an automated wedgespirometer. Am Rev Respir Dis 1972; 106r38-44.226. Rota J, Sanchis J, Augusti-Vidal A, et aLSpirometric reference values from a Meditetrane-an nooulation. Bull Eur PhvsioDathol Resnir 1986:22-217-24.

_ _

227. Joharmsen ZM, Erasmus LD. Clinicalspirometry in normal Bantu. Am Rev Respir Dis1968; 97:585-97.228. Miller GJ, Ashcroft MT, Swan AV, BeadnelIHMSG. Ethnic variation in forced expiratory vol-ume and forced vital capacity of African and Indi-an adults in Guyana. Am Rev Respir Dis 1970;102:979-81.

229. Rossiter CE, Weill H. Ethnic differences inlung function: evidence for proportional differ-ences. Int J Epidemiol 1974; 3:55-61.230. Lap NL, Amandus HE, Hall R, MorganWKC. Lung volumes and flow rates in black andwhite subjects. Thorax 1974; 29185-8.231. Cookson JB, Blake GTW, Faranisi C. Nor-mal values for ventilatory function in RhodesianAfricans. Br J Dis Chest 1976; 70:107-11.232. Patrick JM, Femi-Pearse D. Reference valuesfor FEV, and FVC in Nigerian men and women:a graphical summary. Med J Nigeria 1976; 6380-S.233. Billiet L, Baisier W, Naedts JP. Effet de lataille, et du sexe de l’age sur la capacite de diffu-sion pulmonaire de l’adulte normal. J Physiol (Paris)963; 55:199-200.234. Teculescu DB, Stanescu DC. Lung diffusingcapacity. Normal values in male smokers and non-smokers using the breath-holding technique StandJ Respir Dis 1970; 51:137-49.235. Van Ganse WF, Ferris BG Jr, Cotes JE. Cig-arette smoking and pulmonary diffusing capacity(transfer factor). Am Rev Respir Dis 1972; 105:30-41.236. Frans A, Stanescu DC, Veriter C, ClerbauxT, Brasseur L. Smoking and pulmonary diffusingcapacity. Scand J Respir Dis 1975; 56165-83.237. Marcq M, Minette A. Lung function changesin smokers with normal conventional spirometry.Am Rev Respir Dis 1976; 114723-38.238. Salorlnne Y. Single-breath pulmonary diffus-ing capacity: reference values and application inconnective tissue diseases and in various lung dis-eases. Stand J Respir Dis 1976; 96(Suppl: l-86).239. Crap0 RO, Morris AH. Standardized singlebreath normal values for carbon monoxide diffus-ing capacity. Am Rev Respir Dis 1981; 123:185-g.240. Paoletti P, Viegi G. Pistelli G, et al. Refer-ence equations for the single-breath diffusing ca-pacity. A cross-sectional analysis and effect of bodysize and age. Am Rev Respir Dis 1985; 132:806-13.241. Knudson RJ, Kaltenbom WT, Knudson DE,Burrows B. The single-breath carbon monoxidediffusing capacity. Reference equations derivedfrom a healthy nonsmoking population and effectsof hematoait. Am Rev Respir Dis 1987; 135:805-11.242. Rota J. Rodriguea-Roisin R, Cobo E, Bur-gos F, Perez J, Clausen JL. Single-breath carbonmonoxide diffusing capacity prediction equationsfrom a Mediterranean population. Am Rev RespirDis 1990; 141:1026-32.243. Hall AM, Heywood C, Cotes JE. Lung func-tion in healthy British women. Thorax 1979;34:359-65.244. Black LF, Offord K, Hyatt RE. Variabilityin the maximal expiratory flow volume curve inasyrnptomatic smokers and in nonsmokers. Am RevRespir Dis 1974; 110:282-92.245. Crap0 RO, Morris AH, Clayton PD, NixonCR. Lung volumes in healthy nonsmoking adults.Bull Eur Physiopathol Respir 1982;. 18:419-25.246. Grimby G, Soderholm B. Spirometric studiesin normal subjects. III. Static lung volumes andmaximum voluntary ventilation in adults with anote on physical fitness. Acta Med Stand 1%3;173:199-206.247. Reed TE. Caucasian genes in AmericanNegroes. Science 1969; 165:762-8.248. Ghio AJ. Craoo RO. Elliott CG. Referenceequations used to predict pulmonary function atinstitutions with respiratory disease training pro-grams in the United States and Canada: a survey.Chest 1990; 97:400-3.249. Gaensler EA, Smith AA. Attachment for au-tomated single breath diffusing capacity measure-ment. Chest 1973; 63:136-45.