Document Ozp6n46VRKMaqyqZ01o2LoNY1

(WIIO^MlsTa 23. 444-4Mi PLAINTIFFS EXHIBIT now-33? The Interpretation of Spirometric Measurements in Epidemiologic Surveys Albert Miller1 and John C. Thornton Pulmonary Laboratory. Pulmonary Ditision. Department of \teJkine: Environmental Sciences Laboratory: and Department of Biostatisties. Mount Sinai Scb.sol of Medicine, City L'nri ersiry of Sen York. Sen York. Sen York 10029 Received March 5, 1910 Spirometry is a fundamental tool in pulmonary epidemiology. Published standards provide specific information on instrumentation, techniques of testing, calculation of results, and physiologic mechanisms. They do not provide guidelines on how to interpret the test results once obtained, e g., on such questions as which predicted values to use. what value is abnormal, how many people in the general population are likely to be abnormal? It is w-ith these questions that this paper is concerned. Standard normal values in wide use are not ideal but should be utilized for aD subgroups ofthe population investigated, as well as for the _ comparison population. Conventional definitions of abnormality lack statistical rationale and result in many false-positive and some false-negative results. This is especially true for fixed ratios such as forced expiratory volume (-1 sec I forced vital capacity <0.70. A statistical definition of abnormality would pl.ee 95'r of the normal population above the lower limit (lower 95li confidence limit). Such a definition may be more applicable to oven than to early disease. The spirometric values of well-defined diseased groups must be characterized before it is possible to state that abnormality (a value below the lower limit of normal) means disrate. The distribution of va'ues (including the mean and standard devia tion) should be used to assess epidemiologic risk factors. Prevalence rates separate abnor mal from normal subjects. Such data may be used to identic those requiring preventive intervention, medical care, or compensation. Pathogenic mcchani-ms can be more easily isolated when abnormal subjects are separated from the majority of normal. Frequencies of spirometric abnormality in standard or normal populations serve as a reference to compare rates of similar abnormality in a population under investigation. Contmts. I. Recording and manuring. A. Performance artifacts: The need for recording flow-volume or spirometric curves. B. "Best effort." II. Sarmul values. A. What is a normal population? B. Which normal values to use? C. The influence of race. D. The influence of age. E. Other factors influencing spirometric results. F. Normal values for instantaneous maximum flow rates (V*.,). III. Comparison populations. /V. Lon er limits of normal din's). A. Percentage of predicted. B. Ratios to FVC. C. The 9517 confidence lower limit. D. Limitations of a lower limit of normal based on a "normal" population: Docs "not likely to be normal" mean abnormal? V'. Hon- to compare groups: The distribution of observed values. VI. Prevalence rates for spirometric abnormality. A. Why look at preva lence rates? B. Use of multiple measurements to define abnormality. C. Prevalence rates in general populations. Several excellent guides to the use of spirometry have been published in the last 5 yean. These include: (i) "The Assessment of Ventilatory Capacity," a State ment of the Committees on Environmental Health and Respiratory Physiology of the American College of Chest Physicians (1) in 1975: (ii) "Clinical Pulmonary To whom reprint requests should be e'nt at: Pulmonary Laboratory, Anncnberg 7*-?6, Mount Sinai Hospital. 1 Gustave L. Levy Place. New York. N.Y. 10079. OtM- V.V*I Ml0n0444-:;$07.000 e.f. t*v vc-.Vr-c r-f- te. \T Ml f.* " IHtN. 4X4 zssnoois ti DOU 06793 SPIROMETRIC MEASUREMENTS IN EPIDEMIOLOGIC SURVEYS 445 Function Testing: A Manual of Uniform Laboratory Procedures," edited by R. E. Konner and A. H. Morris (2) and published by the Intermountain Thoracic Soci ety Standardization Task Force in 1975; (iii) "Chronic Obstructive Pulmonary Disease," a handbook of the American Lung Association (3) in 1977; (iv) an American Thoracic Society Statement based on the Snowbird Workshop on Stan dardization of Spirometry (4) held in 1977; and (v) a definitive supplement to the American Review of Respiratory Disease, the Epidemiology Standardization Project (5) published in December 1978, for which Benjamin G. Ferris served as principal investigator. These are excellent references for definitions and physiologic rationale of the various tests, standards for calculation and reproduci bility of results, calibration of equipment, instrument response, and other engi neering details. These matters might be called the "hardware" of spirometry. Another aspect of spirometry needs fuller discussion for the benefit of many who use it as an epidemiologic (and clinical) tool. This might be called the "software": What do the results mean? Are they "normal" or "abnormal" and what do these terms mean? It is with this aspect of spirometry that this paper w ill be concerned. I. RECORDING ANO MEASURING A. Performance Artifacts: The Seed for Recording Flow-Volume or Spirometric Curi es Curves should be recorded whenever spiromerry is performed. Erroneous re sults brought about by improper or incomplete performance are often detectable only from the curve. This is especially true of early termination of expiration (Figs. 1 and 2), since this can occur when most of the forced vital capacity (FVC) has already been expired and its value may be read as "normal." Early termina-'v lion increases mean and instantaneous flow rates at middle and low lung volumes / derived from the maximum expiratory flow-volume (MEFV) curve, as well as the ratio of the forced expiratory volume -1 sec (FEV,) to FVC. Many investigators believe that subject effort is best monitored by the MEFV^ FVC 20 L TFS sTTS'iJimrei Fio. I. The last two ofmany spirograms (water spirometer) compered with the MEFV curve (inset) obtained at the same session, showing significantly greater FVC anj FEV, when effort was monitored with the MEFV curve. (Displays arc ATPS for spirographic and BTPS for MEFV curves.) DOW 06794 5STOOD 552G 446 MILLER AND THORNTON Pic. 2. Multiple MEFV curvet made during one tettion. The FVC nd FEV, in efTorti I and II are virtually identical but were not accepted because of evidence of early cetiaiion of expiration on the MEFV cunt, at confirmed by the next three curvet. EfTont IV and V again ho virtually identical value! for FVC. which are J-s'S greater than thotc initially obtained. No attempt * at made to record full impiratory loop*. curve (6-8). Figure 1 shows the last two of many spirograms performed by a patient with airway obstruction in 197]; they were reproducible and believed to represent maximum effort. The inset is the MEFV curve performed during the same session (one of the first such curves recorded in our clinical laboratory). N'oie the significantly greater FVC and FEV,. We have reviewed spirographic tracings obtained in our laboratory before 1970, when MEFV curves were intrtK^ duced, and found many which reveal early termination. In subjects with norm^* values, the flow rates and flow* volume ratios calculated from these tracings were somewhat high. In some patients with clinical airflow- obstruction, the values were those of restrictive rather than obstructive impairment, while in others, the degree of obstruction was consistently underestimated. Because of performance artifacts and the w ide range of "normal" values, ap parently normal results do not mean optimum, reproducible values. Should less than maximal values be used, the calculated regression equations--which may then be used as controls, as baselines for longitudinal studies, or to assess risk fitters for various groups--will be in error. The errors introduced by an incomplete or poorly performed effort point out thi^ importance of the technician (or physician) performing the test. His instructions,/ encouragement to maximaleffort, and eje for detecting departures from accept- bOL) 06795 SPIROMETRIC MEASUREMENTS IN EPIDEMIOLOGIC SURVEYS 447 ahle procedure are paramount. No computer-programmed monitoring system can easily replace these qualities. His experience will cause him to obtain additional efforts when existing ones seem (o be reproducible and maximal (Fig. 2). and allow him to avoid wasting time and energy in getting repeated inadequate curves from the subject who simply cannot perform. B. "Best Effort" There is agreement that for FVC and FEV, the largest values be used, even if\ these do not come from the same effort. For calculation of forced expiratory flow1' (Ftt-h-:. rfcf;3.u. and instantaneous flow rates, a "best effort" or "best curve" must be selected. This has been simply and workably defined (4) as the curve with the largest sum of FVC and FEV,. In our experience with many thousands of suPj^Jls on epidemiologic surveyj,_lhg--best curve" asjJeflned above usually contains the largest individual values iz27c) for FVC. and_FEV1. Successive MEFV curves are overlain at high gain and are virtually identical for most subjects (Fig. 3). II. NORMAL VALUES A.What Is a S'onnal Population? With Hutchinson's publication "On Capacity of the Lungs and on Respiratory Function with a View of Establishing a Precise, and Easy Method of Detecting Disease by Spirometer" in 1846, it was recognized that spirometric measurements iff can be interpreted only be comparing them to value/ derived from a "normal" h J population. A normal population is one from which abnormal subjects are ,, 1-J excluded. This must be distinguished from a general population, which should\ reflect the distribution of impairment, symptoms, specific disease states, smoking / habits, and occupational or environmental exposures in a community. - a The normal values for North Americans in wide use until recently (9, 10) were to derived from a narrow range of subjects, generally hospital personnel or patients .he with nonpulmonary illness. In the last decade, a broader range of subjects has >> been studied (11-19), and individuals have been excluded because of: hie (1) an objectively obtained history of respiratory (sometimes including car ro diac) disease and symptoms (11, 13-13, 17-19); ta) (2) abnormal physical findings (13, 18); :re (3) an abnormal chest roentgenogram (13, 18); ire (4) a history of smoking (13. 13, 17, 18); ree (3) likely exposure to pulmonary hazards (II, 13). It may be seen that most series lack physical or roentgenographic examination. 3P- Many series have relied on volunteers; the proportion of those asked who actually tit participated has generally not been stated. The ethnic composition may not be iay specified, or may be limited to a specific group (11,13). Several recent series have tk included smokers (13, 19) "because they are found in an average population" (13). thus confusing a normal with a general population. Even the definition of the "smoker" has varied. One investigator (18) excluded smokers but defined a sub -ns. ject as a smoker only if he had smoked more than a pack of cigarettes a day for 3 .`pi- years. ronnnrre DOW 06796 I STUUU5320 448 MILLER AND THORNTON Fic. 3 (A-O- MEFV cur'*' performed in triplicate by three consecutive subjects in a field survey, showing the consistency in configuration end results routinely obtained with motivated. weD-coached persons. All efforts were recorded. The deflections of the pen are at 0.3. 1.0. and 3.0 sec. What comprises a positive history has varied. It is often defined only as chronic bronchitis. At the oihcr viewpoint, an editorial (10) advocated that a history or electrocardiographic c' Mu nee of coronary artery disease should e.\clude subjects from a normal population, since the FVC and FEV, are lower (21). This approach was followed in a survey of 26S4 Israeli men aged 45 years and older (22). Results reported for FVC and FEV, have either been the largest of three satis* DOW 06797 SPIROMETRIC MEASUREMENTS IN EPIDEMIOLOGIC SURVEYS 449 factor>* efforts or the mean best three of five. The Snowbird Workshop (4) con-^ eluded that "there is no need to obtain more than three acceptable tests." Both approaches involve at least three efforts, yet in a frequently cited series (II),. spirometry was performed "at least twice" on each subject. The results of spirometry have been used to define normal subjects for spirometric prediction equations. Normal values for the MEFV curve (including VC) were reported for subjects whose spirometry was norm,:! by "existing" stan dards (18). In another study, persons with disease were excluded from the com munity sample from which the FVC and FEV, were derived (14). "Disease" con sisted of symptoms and/or a spirometric abnormality, FEV,/FVC <0.6. B. Which Normal Values to Use? For most users of spirometry the question really is: Which published normal ' values to use? The references cited in the beginning of this paper (1,2.4,5) make no recommendation on this important matter. Moms' data (11) are probably the most widely used: they have been cited in the American Lung Association handbook (3) / and programmed by many manufacturers for their automated spirometers. Sub jects, who came from a nonsmoking population living in an unpolluted rural val ley, did not have histories or symptoms of respiratory disease. Problems with these data include: '*(1) Epidemiologic Applicability to urban populations. (2) Methodologic (a) Physical and roentgenographic examinations were not made; heart disease was not on the questionnaire. (b) No more than two efforts were obtained from an unknown number of subjects. (c) The Kory method was used for calculation of FEV,. This results in a value which may be up to 179 cc smaller (23) than that obtained by the cunently recommended method, extrapolation. The resulting differ ence in FEV^FVC x 100 may be 4.5 for an FVC of 4 liters. The Kory method was used in several other series of predicted values (10, 12, 13). Surveys of several large representative populations have provided regression equations for spirometric measurements. In the Tecumseh, Michigan, survey, 6365 subjects >10 years of age (5101 of whom were >16 years of age) were tested (24). Separate regressions were developed for FVC, FEV,, FEV,/FVC, V*,^. FEFu.-j. FEFfct-u, and peak flow (PF) for male and female subjects without cough, phlegm, or dyspnea in the age groups 10-15, 16-19, and 20-74 years. A wedge spirometer was used. Only two tracings were obtained and the best effort was defined as the one with the larger FVC. Schoenberg et al. (25) recorded MEFV curves in 3046 asymptomatic individuals, all of whom were nonsmokers except for some black males. A computerized pneumotachygraph was used. The two studies (out of five) with the largest FEV, were averaged, provided values agreed within l(fi. Subjects resided in rural Lebanon, Connecticut, urban Ansonia, Connecticut, and semirural Winnsboro, South Carolina. Between 54 and r-STOOD 5529 OU 06798 450 MILLER AND THORNTON 755? of all residents >7 years of age were tested. Of the total, there were -455 white and 114 black females >20 years of age and 194 white and 120 black males >18 years of age. Since simple linear regressions overpredicted at each end of the> age range and underpredicted in the middle range, more complex models (utilizing ( weight as a variable as well) were developed. The lower 955? confidence limits,; thus obtained were closer to predicted values. 1 ! j i C. The Influence of Race It is now well accepted that blacks residing both in Africa and in the New World's. ^ j(26-30), East Indians (31, 32), yellow-skinned peoples (28, 33, 34), and Pacific Islanders (35, 36) have smaller regressions for lung volumes against height than / whites. This knowledge has not consistently been applied in the clinical interpre- j tation of pulmonary function tests or in the epidemiologic screening of popula tions. For blacks, most studies have shown the difference to be in the range of 10-155? and a 135? scaling factor has been proposed to adjust values measured in blacks to those predicted for whites (37). A similar factor (12.55?) "converted" the FEV, of East Indians (31). -- D. The Influence of Age It is well know n that FVC. FEV,, and flow rates in adults decline w ith age. The FVC increases up to age 24 and then remains stable to age 35 (25). Simple lineal's regressions against age, by using a single negative term for age over the range/ 20-70 years, overpredict values in young adults. The difference is not large nr absolute terms (100- 200 cc for FVC in white males, height 175 cm, age 20). Nevertheless, Schoenberg (25) suggests that there might be important conse quences in determining occupational risk. When 18- to 20-year-old workers siarj jobs, preemployment tests may be considered below normal. If the worker is not\ refused employment, significant function loss shown on subsequent tests may be/ overlooked because the rate of decline appears no greater than expected foir all) adults. It is less well recognized that FEV,/FVC varies w ith age. Most cross-sectional \ surveys have shown a decline w ith age and then a slight increase in the elderly (11, ' 13, 24, 38). Schoenberg (25) demonstrated this increase at age 55 in both sexes. A recent longitudinal study of 259 elderly subjects conducted over a 5-year span (39) disclosed a greater absolute and relative decline in FVC than in FEV, with a consequent increase in FEV,/FVC. These changes were greater in men than women, and in those 70 - 90 years of age than in those 62-69. The 42 men >70 years of age lost a yearly average of 125 cc FVC (45? of baseline) and 70 cc FEV, (3.15? of baseline). Since reduction in FEV,/FVC is considered by many authors as a basic impairment which reflects symptomatic status, interpreters of spirometry must recognize that a single value for this ratio cannot be used as the\ lower limit of normal for all age groups. However, review of the literature show*' that this practice continues. Recently, mean transit time (or effective time) of forced expiration has been proposed as the single most sensitive test for obstruction, in both centra) and peripheral airw ays (40-43). This measurement can be applied over the full range r-S-T0 0 0 544-0 DOW 06799 SPIROMETRIC MEASUREMENTS IN EPIDEMIOLOGIC SURVEYS 451 of growth and shows no difference between the sexes (41). However, age remains'^ an important factor; mean transit time increases linearly with age (41, 45). E. Other Factors Influencing Spirometric Results VC. PFR, FEV,, and airway conductance display circadian rhythm (44 - 46). Variations are small and probably negligible in interpreting individual satues but introduce a small systematic error in comparing mean values of different groups or of the same group at various intervals. The time of year may influence spirometric results, perhaps through differences in levels of pollutants or incidence of respiratory infections. In addition, FEV (47) and airway resistance (48) vary with the hours of daylight, the former with an amplitude of 100 cc. Altitude may affect spirometric results. Schoenberg et at. (25) in re view ing theirS own and other workers' findings, concluded that FVC in young adults increases ca. 100 cc for every 300 m above sea level, up to an altitude of 4000 nt. Smoking immediately preceding spirometric testing is another variable which should be controlled. The acute effects of smoking even one cigarette include decreased flow rates (49). "n^ The variability of spirometric results is greater w hen they are abnormal. When ^ measurements were made at two and three weekly intervals in patients with stable irreversible chronic obstructive airway disease, the coefficients of variation were .v large (11.1% for FVC and 14.8% for FEV,) (50/. In the opinion of the inves tigators, this variation must be considered w hen assessing the effects of treatment. In an earlier study, Fletcher (51) found that 5% of apparently "normal" men-, showed a difference in FEV, >400 cc between two determinations at a 10- to' 12-day interval. Variation was greater when values were abnormal. Acute, self-limited, noninfluenza! viral upper respiratory- infection (simple URI) has not generally been found to decrease spirometric measurements in adults (52,. 53). However, obstruction in the small airways has been demonstrated in spon- / taneous and induced URI bv the following tecKn7auesrnreouenrt"de'BgnitgTrrrBr compliance-(53, 54), increased closing volume in smokers (55). decreased flows at low lung volumes on breathing helium-oxygen (55),.increasediotal respiratory resistance by forced oscillometry'(56/7*and greater sensitivity to parasym- - pathomimetic challenge (56). Changes persist up to 8 weeks.(VI. 16) Therefore, the decrease in Dows at low lung volumes*(FEF,,.ri, FEF;,.. FEF;1, etc.) de scribed in small airway obstruction would not be unexpected during an URI. In longitudinal studies on children under.7 years of age (57), FVC. FEV,, and PFR as well as FEFa.a and FEF* were noted to decrease during coincidental URls. Ef fects may be greater in young children because small airway conductance is lower in early childhood than in later years (58). F. Normal Values for Instantaneous Maximum Flow Rates (V^,,) Like FEF,,_,, and FEF;J., these flows are inherently more variable than FVC and FEV,. A pneumotachygraph or low-resistance spirometer, with expired vol ume measured at the mouth, is generally used in the field to obtain these flows and generate MEFV curves. Values for instantaneous flow rates using expired volume have been published by Chemiack and Raber based on 899 subjects (15) and Bass ' ST000553I DOW 06800 zcccnnoiS-i .452 MILLJ AND THORNTON based on 247 (18). In the latter paper, normal was partly defined by spirometric results, criteria for which were not indicated. In two other investigations, volume was measured in the body plethy smograph. Black etal. used a composite of two curves showing the higher flow (19) and more recently, Knudson et al. (17) reported flows from a composite MEFV curve representing the subject's best performance at each decile of volume. Use of such an "outer envelope" curve to derive optimal flows has been reported to reduce variability (59). Values may differ from those derived from a single "best curve." Whether for this reason or for reasons relating to selection of subjects, the flows in Knudson's series are higher than those in others. Although the body plethysmograph and composite curves were used by these investigators in the field, this approach is less applicable to surveys in which the time available for testing a large number of subjects is limited. Stanescu et al. (60) have recently w ritten that "concerning variability within an individual, values for Vmw measured in relation to the expired volume are as good as (and more simple to measure than) Vmu recorded with a body plethysmograph." Values for ^ have been standardized for volume to achieve greater compand bility between subjects of different size. Although TLC is preferable, it is not generally available in the field. As early as 1967, Lapp and Hyatt (61) published values for FEF^FVC and FEF;*FVC; means in 34 normals were 0.98 and 0.45, respectively. We have found similar values in normal subjects (62). Knudson eial. (17) showed little change in the FEFM'FVC and FEF;yFVC with age, although Chemiack and Raber (15) found that FEF- had a large negative correlation with age. III. COMPARISON POPULATIONS A comparison population should be equivalent to the surveyed population in alN respects except exposure. Such a group is difficult to find and, if found, to mod* vate to undergo testing. In the absence of a comparison population, it would seem advisable to compare the results in all subgroups of the surveyed population with the normal values in w ide use in order to uncover impairment common to all in the study population. This has not always been the practice. Results for subjects in different exposure categories have been analyzed only relative to each other. Some investigators have used as a comparison population those in the group at risk w ho have had the least burden and/or duration of exposure or who are asymptomatic (63). Since values in this group are set at lOOCc, any impairment resulting from their level of exposure would be inapparent. Flows at different lung volumes were related to smoking and bronchitis in coal miners (64). Mean values were compared for smoking and nonsmoking, bronchitic and nonbronchitic groups matched by age and height. Compared to predicted values, the results for FEF; seem low in allN the groups, including the "comparison" set of nonsmokers w ithout bronchitis. / IV. LOWER LIMITS OF NORMAL (lln's) In the use of spirometry, the question "Who is abnormal?" becomes "What value should be considered abnormal?" Many surveys have not been concerned with this question and, therefore, cannot be used to obtain the prevalence of DOW 06801 SPIROMETRIC MEASUREMENTS IN EPIDEM10L0CIC SURVEYS 453 spirometric impairment, even when prevalence rates of various symptoms (in cluding "chronic bronchitis") are reported. The investigators have shown mean values and regression equations of spirometric measurements for the groups at various risk. These permit them to find statistically significant differences related /` to risk factors, but do not provide any information on prevalence of abnormality. A. Percentage of Predicted Until recently, the conventional criteria for abnormality in clinical use have been carried over to epidemiologic screening. Although less than 8CKT of the predicted value has generally been considered abnormal, this guideline has not / been uniformly followed. Particular confusion centers on a value of "800 of predicted," which most laboratories interpret as "within normal limits." However, the handbook "Chronic Obstructive Pulmonary Disease" of the American Lung Association (3) shows 800 as abnormal for FVC and FEV, (normal is greater than 800) and 750 of predicted as abnormal for FEFu.;t and FEF-,_tt (after Moms (65); normal is greater than 750). On the other hand. McFadden et at. (66), in assessing various tests for small airway disease, defined an abnormal FEFu_;i as less than 800 of predicted (800 was normal), although they used Morris' predicted values. B. Ratios to FVC 1. FEVJFVC. Although 0.75 had been widely used as the lln, recent practice^ tends to 0.70. Since FEV,/FVC falls with increasing age and height (13, 25, 38)/ use of a constant value for all subjects is invalid. Using Moms' means and stan dard deviations (38) and mean - 1.64 SD (see below) as the lln, the cutoff value for a man 69 in. in height ranges from 0.69 at age 20 years to 0.59 at age 60. (These valuef are systematically low because of the Kory method used to calculate FE V,;^ see above.) The result of using a constant value for FEV,/FVC is that prevalence } of impairment appears to increase as a population ages. This may falsely be attributed to duration of occupational exposure, total burden of occupational exposure, or pack years of cigarette smoking, since all of these correlate with age. There is evidence that certain athletes (67) and others whose work activities develop the pectoral muscles (including the accessory muscles of respiration (68,x 69)) have a relatively greater than normal FVC compared to FEV,, with a resultingly lower FEV/FVC. This might be true of many industrial workers. 2. FEFu-iJFVC. This ratio has been used to classify patients as to restrictive / or obstructive impairment (70); a lln of 0.65 has been proposed (2, 71). Since ` FEF-j.-, - FVC/2 midexpiratory time (MET), the ratio to FVC V6MET. It would be simpler to use the MET as a flow-related measurement which is inde pendent of volume with an upper limit of normal of 6.77. C. The 95% Confidence Lower Limit A rational statistical basis for the conventional "80*3 of predicted" does not exist. For the FVC and FEV, of some investigators (11), S0% of predicted has a\ confidence approximating 95% at the mean age and height of the normal popula- ' r.ooi s V UCSG001S i 454 MILLER ANO THORNTON tion they measured. This is not true for different ages and heights. The matter has been lucidly discussed by Sobol (72): "Applying the 80 percent rule to those functions defined by a regression equation adds another dimension to the error ... percent of predicted will deviate from the regression line less for small values than it will for large values. Therefore, small predicted values, such as occur in the short and aged, w ill result in a higher incidence of abnormal findings. The use of a fixed percentage can be more or less stringent than the statistical limits depending upon the magnitude of the variance around the regression, but it will always result in a disproportion between the number of abnormal findings among small versus large values." Thus, use of 807? of predicted as the lln, especially for flow rates, will lead to the identification as abnormal of many subjects whose values are w ithin 957? confi*^ dence limits, particularly when they are shorter or older. On the other hand, the/ values for FVC and FEV, of taller or younger subjects may erroneously be con sidered "normal" because they are >807! predicted when in fact they are belo'wK the 957? confidence limits. This is illustrated by Fig. 4, which shows the relation^ ship between the regression line for predicted FVC, the line for 807? of predicted, and the line for the 957? confidence limit. A useful approach to statistically meaningful lln's is to consider as "normal" 957? of the normal population whose measurements arc above a specified value. For such a one-sided test, the lln - mean (in litcr.sec, etc.) - 1.64 SEE (standard error of the estimate (72, 73)). The lln may also be expressed as a percentage of predicted, i.e.: Below what percentage predicted can we consider, with 957? con fidence, a subject's value to be abnormal? As show n for FVC in Fig. 5 for a man 70 in. in height, this ranges from 817? at age 20 to 767? at age 70. We have drawn up simple tables showing the lower 957? confidence limit cutoff values based on Morris' data for percentage predicted FVC, FEVlf and FEF.U and for FEV,/ FVC for individuals of different ages and heights (74). Another way to express statistically valid lln's while retaining the percentage Fic. 4. The relationship between the regrevsi.'n line for predicted FVC at a heijhi of 70 in. slightly modified from Morn? (IIU. the line for ftr~ of predicted. nj the line for 95C! confidence. The lust was so minimal!) cursed that it is dr<sn as a straight line. Note that M" of predicted is increasingly above the 93*7 confidence limn from age 30 on. For the decade 70-30 years of age. on the other hand (an age range of many industrial workers). the 93'f confidence limit is above the tine for KKJ of predicted. 063 DOU SPIRO METRIC MEASUREMENTS IN EPIOEMIOLOCIC SURVEYS *I I i SI9U* l r airi i9 ! aZi l. SCrV*3*i*gam 9*>i*i 455 . --------- 2 0 ~ 1 1 ` ' 2 4 6 8 i0 >2 i Fic. J. The distribution of values for two (roups with an identical prevalence (J'f) below the lover Kmit of normal, yet with distributions (anj means) which show important differences. Reproduced, with permission, from A. Miller.J. C. Thornton. H. Smith. Jr., and J. F. Morris (1980). Amer. J. InJ. hied. t, No. I. published by Liss. New York. predicted and 80?? rules of thumb has recently been proposed by Sobol (75). This involves "normalizing" all subjects by determining the value of the mean when mean - 2 SD (1.64 would be preferable for a one-sided distribution) represents 80C? of the mean. Using this procedure, 1 SD below the mean would be 90C? of predicted. The statistical lln for flow rates such as the FEFa.T* may be as low as 20<"r ofv predicted in older, shorter subjects. The greater variance of such tests, however, ) is more or less offset by their greater decrements w ith disease (77). Values this low / are frequently encountered in clinical practice. While use of such a lln for patients with oven disease may present no difficulty, it may be less applicable to early disease. Thus, w hile McFadden (66) joined Morris (76) in concluding that percent age of predicted flow rates is highly sensitive in detecting small airw ay obstruction, Marcq and Minette (78) using predicted -1.64 SEE found that these tests did not improve the ability of conventional indices (FEV, and FEV,/FVC) to detect abnormality in smokers. "Statistical" definitions of the lln (based on the SEE) have been inconsistent. Most investigators use 1.64 SEE below the mean; some use 2.00 SEE or 1.00 SEE and others use half the mean (79). The National Institute of Occupational Safety and Health Laboratory in Cincinnati defines "abnormal" as the value aboveS which 99S5 of their normal population fall (mean -2.33 SEE) and classifies thosv^ in between the 1st and 3th percentiles as "borderline." D. Limitations of a Lower Limit of Normal Based on a "Normal" Population: Does "Not Likely to Be Normal" Mean "Abnormal?" The values used to define the lln were obtained from studies of normal popula tions. The limits were selected to provide 9We of the subjects with a classification of "normal" and 3?e with a classification of "abnormal." 1: is important to realize', that some normal subjects will be classified as "abnormal." \ r ST0005535 W 06804 436 MILLER AND THORNTON Studying large numbers of normal subjects does not define an abnormal (in the sense ofdiseased) group. It allows us to classify an individual as abnormal only in v the sense of not belonging in the normal group. Such a classification does noy state the probability that a subject is in a diseased group (e.g.. patients with obstructive airway disease, or patients with interstitial lung disease). To do so requires defining and analyzing the distribution of values for the diseased group, and comparing it w ith the normal population. As an example of this approach, we applied conventional and statistical lln's to 100 randomly selected patients re* ferred to our clinical pulmonary laboratory with the diagnosis of "COPD," "chronic bronchitis,** and/or "emphysema" made by a Pulmonary Division physician. (We were aware that some patients are referred to the laboratory w ith a tentative clinical diagnosis which may be disproven, and that airflow obstruction need not be present in all patients with chronic bronchitis.) Using conventional criteria, we found 899* abnormal for FEV, percentage predicted (<S0), 909c for FEVj/FVC (<0.70) and 959* for FEF;j.t percentage predicted (<75). Using 1.64 SEE below the predicted mean as the lln, 859c were abnormal for FEV,, 739* for FEVj/FVC. and 609c for FEF:j.:l. This groups of patients with clinically defined disease .was limited and is re* ferred to for purposes of illustration. Tashkin ei al. (80) defined an abnormal group to provide an'estimate of test sensitivity. Two respiratory specialists made a diagnosis "based on their own professional judgment and experience, as to the presence or absence of chronic obstructive respiratory disease, taking into con* sideration the results of all the lung function tests, responses to the respiratory questionnaire and reported findings on physical examination." This "clinical'' definition is based to an unknow n extent on the results of the tests under consid* cration. V. HOW TO COMPARE GROUPS: THE DISTRIBUTION OF OBSERVED VALUES The distribution function of a random variable describes the probability that the random variable assumes a value in any given interval. Therefore, the probability that a random variable satisfies any specified condition can be calculated fiom a knowledge of the distribution function. An estimate of the distribution function can be obtained from the collection of individual observations which result from a random sampling of the population. For a Gaussian distribution, know ledge of the mean and variance (standard deviation) allows calculation of probabilities. If tw o or more groups are to be tested for differences in their pulmonary funai tion related to an environmental exposure, the comparison should be based on^N their distribution of observed values, not their prevalence of abnormality. Infor*/ mation is lost when each subject's response is replaced by a classification or normal or abnormal. For example, the percentage of individuals in each of two groups with values below the lln may be identical, yet the two groups are truly different. This is seen in Fig. 5. In group A, the majority of observations are only slightly above the lln while in group B, the majority of observations arc far above . the lln. Although the prevalence of abnormality is identical, the two groups have very different distributions. On the other hand, there may be a considerable dif* DOW 06805 ! ST00Q553G SPIROMETRIC MEASUREMENTS IN EPIDEMIOLOGIC SURVEYS 457 fcrence in prevalence (of 255r, as shown in Fig. 6) with no meaningful difference in distributions. The latter situation is often seen in epidemiologic studies. Two groups (Croup A thought to be at risk because of its exposure to a to xic agent and \ Group B. similar in other respects but lacking this exposure) are compared to a/ third group (C). a standard population. Both test groups (A and B) have meaning* fully different distributions from the standard group (C) even though differences between A and B are minimal. In both situations illustrated above, looking only at prevalence rates can beN misleading. Therefore, to establish that a certain group is different (in this discus*./ sion, different in its spirometric measurements) and that this difference is related to a quantifiable variable (e.g., environmental exposure), one should analyze the distribution of observations. Small differences in the distributions can point to a significant effect of the exposure. When the distributions of large groups are compared, even tests with greater inherent variability whose clinical interpreta tion remains unclear (such as V'aMX) may provide useful information. That differ ences are "statistically significant" does not mean that they are "clinically im portant." VI. PREVALENCE RATES FOR SPIROMETRIC ABNORMALITY A. Why Look at Prevalence Pates? 1. Prevalence rates are based on the identification of "abnormlaall'" individuals^^, Identifying who is abnormal individualizes the problem on ther one hand and y demonstrates its social impact on the other. The uses of prevalenice data depend / on the definition of abnormality. Identifying early lung disease allows intervention at a more remediable stage, either by the individual to stop smoking or by society to ameliorate occupational exposure. It is generally based on tests which showgreater variation in normal individuals. Overt disease is generally manifest by symptoms and other clinical findings. However, dyspnea is a very subjective complaint and airway obstruction does not correlate well w ith the symptoms of cough or expectoration (81, 82), or with radiographic changes. Demonstrating (and i Ftc. i. The distribution of value* for two (roups with a practical difference in distribution of values (or mcansl. difference in prevalence, yet with no I DOW 06806 458 MILLER AND THORNTON quantitating) disease identifies subjects who may require clinical care or disability compensation, and allows society to allocate medical or financial resources. 2. Once abnormality is identified and quantitated, trends can be followed, and correlations made to changes in work practices, atmospheric pollution, smoking habit, etc. 3. Several large, carefully done studies have demonstrated the prognostic value of abnormal spirometric findings. Those at greater risk of progressive disease and death can be identified. Petty et el. (83) studied residents of a small Colorado city 7 years after an initial survey, and found that those who had had an FEV,/FVC <0.59 had a greater decline in FVC and FEV, and a threefold greater mortal* ity. Those with a normal FEV, initially remained normal. Bates (84) followed 216 men with chronic bronchitis over a 10-year interval, and noted a significant de cline in FEV, and FEF-.-, in 1012 (whom he termed as having "malignant bron chitis"). He suggested a fall in FEF-*-: *0.6 liters'sec in 1 year with no im provement the following year as an indicator of progressive disease. Fletcher's monumental studies of patients with chronic bronchitis (81, 82) have demon-\ strated that cough and sputum production do not point to a poor prognosis, which \ is indicated by a greater decline in FEV, with age. Cessation of smoking not onljr decreases mucus hypersecretion but returns the slope of decline in FEV, to normal. (iv) Most diseases, especially those of environmental origin, are multifactorial. Reporting findings as prevalence, by identifying the affected individuals, may uncover other casual mechanisms, e.g., smoking, other exposures, ethnicity, genetic influences, etc. (See Fletcher on this matter, below.) B. Use of Multiple Measurements to Define Abnormality Since no test will detect 10012 of abnormals, and we are often interested in finding early disease, for w hich no single simple measurement is available, tnany authors have used multiple tests to define abnormality. When this approach is used, the results are very different when a subject is classified as abnormal be cause any one test is abnormal or because more than one test is abnormal. This is illustrated in Table 1. Each of three tests classifies a normal individual as "abnor mal" with a probability of 0.05. If an individual is to be considered abnormal wheh> one or more of the three tests are positive (i.e., one, two, or three tests are positive), the probability of classif)ing a normal subject as abnormal rises to, 0.1426 (if the tests are independent: somewhat less if they are not). An investigator who wishes to define abnormality with 9517 confidence by using any one of three tests may do so by setting each test at the 0.017 rather than 0.05 level. Cl c_r. PiomiiiuTY or \ TABLE I Positive L'sivc T**et KoirtvotM Tests to DsitCT Ainoimsiitv Niinhrr of jvviiiv* iis 1 tof 31 : tof )> (All) J ,' Probability 0.1354 0.0071 00001 DOW 06807 SPtROMETRIC MEASUREMENTS IN EPtOEMIOLOCIC SCRVEVS 459 On the other hand, if tm> or more positive results are requireJ to classify a subject as abnormal, the corresponding probability is 0.0072, i.e.. the probability of a false positive is much less than when one test is used. Requiring all three to be positive reduces the probability to 0.0001. Many authors have defined their abnormal group by abnormality of one or more of a large number of tests, e.g.. FEV./FVC, FEF:,.n. and closing volume (S3). In the past w e defined air flow- obstruction as a decrease in any one of these three tests. FEFu--,. V'IMX.J/FVC, or FEV,/FVC (86). Whatever limits of normal are used for the individual tests, such a definition will increase the preva lence of impairment, perhaps obscuring differences between groups. The artifact introduced by defining abnormality on the basis of one or more abnormal tests has been discussed. Sobol (72) concluded that: "One final source" of trouble caused by multiple tests is that the likelihood that one of them will be abnormal in a group of normal subjects increases as the number of tests used increases." Knudson et at. (87) examined this problem, using mean - 1.64 SEE as their definition of abnormality. When impairment was defined as abnormality of any one of three tests (FEV,, FEV/FVC. and FEF;.:,). 10.8% of the normal subjects, from whose values "normal" was defined, were now abnormal. This is not far from the 14.3Cc expected from Table 1, especially when one considers that the three measurements are not independent. C. Prevalence Fates in General Populations Prevalence rates of impaired spirometry in the general population are not read ily available. Investigations have been characterized by inconsistencies in test procedures, source of predicted values, and definitions of abnormality. In addi tion, many investigators have presented mean values without indicating what percentage of the population was abnormal. As Fletcher has pointed out (88), "Most epidemiologists have avoided this problem (where ... the lower limits of normal are to be set) by simply recording differences in mean test values between groups of subjects. One objection to this custom, in relation to causative factors such as smoking, is that only a minority of smokers appear to be sufficiently susceptible to smoking to develop significant airflow obstruction. Thus, with aging, the distribution of FEV levels or of severity of emphysema at autops/ becomes skewed as a majority of subjects remain normal, but there develops a "tail" of clinically significant abnormalities. The mean value of a group or popu lation with such a distribution is not the best index for comparing the effects of smoking or any other agent: for the effect on the susceptible minority tends to be overw helmed by the unaffected majority." Even though many papers present prevalence of respiratory symptoms or of chronic bronchitis in various populations, few have reported prevalence of spirometric impairment. Bouhuys (89) compared Spanish hemp workers with a control population of farm and marble workers for whom he found FEV, values <80*7 of predicted in 12% of those 20-69 years of age. 13.6% of those 40-69 years of age. and 31% of those 50-69 years of age. The survey of Chilliwack (90), an unpolluted town in Canada, found 12.6% of men to have severe obstructive lung disease: about 70% of these had an FEV,/FVC *0.60. The FEV^FVC was be- GfSSOOOlSj DOW 06808  460 MILLER AND THORNTON low 0.75 in 26.4'7 of 18.403 British male civil serv ants 40-64 years of age (911. The prevalence of this `abnormality" increased from 19.5% in the age group 40-49 years lo 28.8% at ages 50-59. and 40.8% at ages 60-64. Kuperman and Riker (71) noted that 21 of IJ9 subjects (1895-) in a community survey who were "normal by clinical criteria" had FEV,/FVC values below 0.75. Bower (79) reported that FEFjj.., was "less than half" the predicted value in 19% of U.S. male bank employees 40 years of age. Spirometry was performed on a representative sample comprising one-fifth of the adult population of a small Colorado city (92). Of the 281 men tested in the age range 20-69,37 (13%) had an FEV,/FVC <0.59. Of those 49-69 years of age. the prevalence of this impairment was 17% and of those 60- 69, 32%. Only 2% of the 328 women aged 20-69 had this abnormality. In Tucson. Knud`on and associates (87) found that 24.4% of asyptomatic male smokers and 35.3% of symptomatic males had an abnormal FEF;,. Corresponding rates of abnormality for the entire survey population can be estimated, since the authors* criteria for abnormal were the fifth percentile values for nonsmoking asymptomatic subjects, w ho comprised ca. 25% of the total. In a stratified sample of 206 residents of unpolluted Manitoba, Nelson and Cherniack (93) defined ab normality in a similar way: more than 2 SD beyond the mean value for the asymptomatic nontmokers. As in Knudson's investigation, 13-18% of male smokers had an abnormal FEV,, FVC. If one is interested in prevalence of spirometric impairment as it is convention ally defined in a "normal" population, the population from which the normal values in general use are derived (11) is an obvious group to analyze. As seen in Table 2, when conventional tin's are applied to the values for FVC and FEV,, 5.5 and 5.9%. respectively, of all men have abnormal values (74). These frequencies are in agreement with a statistical definition of abnormality at the 0.05 level. However, 10.2% of subjects *50 years of age have a decreased FEV,. Indeed, a decreased FEV, (or FVC) is here defined as <79% of predicted,' as is the corrimon practice. If the American Lung Association (3) criteria (<80% of predicted) are used, the prevalence of abnormality would be greater. Prevalence rates for airflow obstruction are much higher. Using FEV,/FVC as the test and 0.70 as the lln, 18.1% of all the subjects are abnormal. At age 50 and above, 37% are abnormal. Using FEF:>.:j as the test and 75% of predicted as the lln, 17.8% of all subjects are "abnormal": for those *40 years of age, 20.6% are abnormal. Again, use of the lln recommended by the American Lung Association (76% of predicted) would result in a slightly higher prevalence of abnormality, as show n in Table 2. The prevalence of airflow "abnormality" in this selected nonsmoking, healthy population is compared in Table 3 with rates reported in populations which in cluded smokers and those with symptoms and overt disease. The prevalence in the "normal" population is less than in most but not all of the comparison popula tions. T1 GO <_n GH ' Sonwl defined at any value which odd roend off at the fourth decimal to StH of predicted or greater, i.e., 79.5000 uas formal. 79.4999 was decreased. DOW 06809 TAIII.K I Pbcvai.cmcc o r Abnobmai.* Smbomi.ibv in Mobbii' Nonsmokino Min (74) nssooois j SPIROMETRIC MEASUREMENTS IN EPIDEMIOLOGIC SURVEYS 461 *s i<s8 p* r iet rsC <it*. pi pi R M^ fl m ssssss: ?is;s=S # ^* S* *r** ^ *-- ^ g g* 2 asc ^s p<i ip-*.. V e> "W .3 ? ? ? ? E j 5 >5 u: c. N b- . '^ t5 CSC 24 ^S2^S .&3 Uu *** *S$ i* ;!?! va *s si it 'Zw JZ *w -3 < r +. _ *at Stf '3ISISa .S b< C .*, J1 02 DOW 06810 462 MILLER AND THORNTON I 1 I2 |ll 2e Ss 3 Ii X-- sS *5 - i8 Ivsriv *5 I 3 E* a 3 3 TAULF. 1 r n ( V ifH C I Ol S riM H E lH U ' A n H n n M A I.ilV IN V a NKIUS COMrAHIMIN IN iru l AI ION* S| Jt1 ZwV A-- t NX* -9 MM tO - a~ * 5S ?X C *>"t 9I 9^I 0rIi ??as = 53 *AA i I * p* -- 1*o> SVto . *sf X . * A* SSt 5^3 9 :AA A A A 93 2 l 'll E 5 = *A 2A c~ zf ie JI M9wt&e > U > S. >HI 2H ISm i c/r c-n. C_"!.: rv) DOU 06811 SPIROMETRIC MEASUREMENTS IN EPIOEMIOLOCIC SURVEYS 463 table 4 Mess Vallss o> Snao'iiTmi Tests in thi Whiii Mm Porno ion or Micmigsn. 197*tax S'ioni'C Hisronx) Nontmokers (n - IJ4) Es-smokers (a - 157) Current smokers (n IS61 FVC percentage prejictej (II) FEV, percentage predicted FEV,. FVC FEF:. n percentage predicted Age lyearsl Height (in.) 99." 105.7* O.IO: (0.756197.5* 4:.os 69. IS 97.51 101 64 0.770 IS. 16 41.80 61.15 94 74 96 15 0.775 S0.4S 51.47 69.56 * VfJin of FEV/FVC in th Morris series (II) at ih mean age and height of the nontmoken in Michigan. Additional data on the prevalence of abnormal spirometry (as conventionally defined) in the general population would be useful. Together with Dr. Irving J. Selikoflf and his group at the Mount Sinai School of Medicine Environmental Sciences Laboratory, we have recently completed a clinical, radiographic and spirometric survey of a cross-sectional population of a large industrial state (Michigan) (94). Subjects were representative of the general population of the state for residence (urban versus rural), occupation and age. Spirometric data were obtained on 983 white adults (497 men) with know n smoking histories. Table 4 shows the distribution of spirometric values. Values for the 154 male nonsmokers w ere similar to those of Morris. Expressed as percentage of Morris* predicted values, the range was from 97.58" (for FEF-j.;j) to 105.7SCc (for FEV,). Values for FVC agreed most closely with Morris* values. The higher values for FEV, were expected since results were obtained by extrapolation whereas Moms used the Kory method which leads to lower values. The sljghtly lower values for FEF.j_rj may relate to more complete expirations achieved when efTon was mon itored with the MEFV curve. Frequencies of abnormal spirometric tests using conventional criteria (Table 5) were similar to those in Morris' population. (The considerations for FEV, and FEFit.n in the preceding paragraph apply to the prevalence rates as well.) A recent study of Vermont male dairy farmers and nonmineral industrial workers tables PuvAitNCi or SniOMtitir Anoshlitiis* in tmi Whits Mali Population or Michigan. I97gitx Smoking Hisroax) Cigarette smoking sutus NcscrFormer Current Number examined 154 157 116 Abnormal FVC 9 (6%l [6%]* 15 *10*7* 2S(I5%) Abnormal FEV, 211'*) [O'*]* 17(11%) 24 115%) Abnormal FEFa.B 24(16%) [18%J* 60 (58%) 10 (45%) * Conventional criteria: see text. * Prevalence of "abnormality" using similar criteria in the Oregon Respiratory Association Sur vey III). DOW 0681 464 MILLER AND THORNTON (95) showed prevalence rales of abnormal FVC (also defined as <80% of Morris' predicted) remarkably similar to those in Table 5. Among nonsmokers. 6% (farm ers) and 7% (industrial workers) were abnormal; among former smokers. 14 and 12%, and among current smokers, 13 and 10%, respectively. These studies give support to regarding the frequencies of spirometric abnormality show n in Table 5 as reference rates for a white population in the United States. When abnormality is conventionally defined, rates should exceed these before a group is thought to have an increased prevalence of spirometric impairment. ACKNOWLEDGMENT W would t!Lc ie acknowledge the support we received from th Jack Martin Fund and from tht NIEHS Cenicr No. ES009:t. REFERENCES 1. Committee Recommendations (1975). The arrestment of ventilatory capacity. Statement of the Committees on Environmental Health and Respiratory Physiology, American Cotleae of Chest Physicians. Chtti 67, 93 - 97. 2. Konncr. R. E.. and Morris. A. H. tEds.) < 1973). "Clinical Pulmonary Function Tcsung. A Manual of Uniform Laboratory Procedures for the Intermountain Area." Inlermountain Thoracic Soci ety Pulmonary Function Standardization Task Force. Salt Lake Gty. 3. American Lung Association (1977). "Chronic Obstructive Pulmonary Disease." 5th ed. 4. Gardner. R. M. (1979). ATS Statement--Snowbird workshop on standardization of spirometry. Amer. Rtv. Http. Dit. 119, 131-131. 5. Ferris. B. G. (Principal tnvesiigator) (1971). Epidemiology standardization project, Amtr. Her. Retp. Dit. Ill, Pan 3. 35-11. 6. Lord. G. P.. Gazioglu. K.. and Kaltreider. N. (1969). The mavimum expiratory flow-volume in the evaluation of patients with lung disease. A comparative study u ith standard pulmonary-function tests. Amer. J. Med. 46, 72 - 79. 7. Peters. J. M.. Mead. J.. and Van Gansc. W. F. (1969). A simple flow , volume device for measuring ventilatory function in the field: Results of workers exposed to low levels of toluene diisocya nate. Amer. Rev. Retp. Dit. 99, 617-622. I. Hyatt. R. E.. and Black. L. F. (1973). The flow.volume curve. A current perspective. Amer. Rev. Retp. Dit. 107, 191-197. 9. Baldwin. E. F.. Cournand. A., and Richards. D. W. (I9zg). Pulmonary insufficiency. 1. Phy siological classification, clinical methods of analysis, standard values in normal subjects. MeJitine 27, 243-271. 10. Kory. R. C.. Callahan, R.. Boren. H. G.. and Syner. J. C. (1961). The Veterans Administration--Army cooperative study of pulmonary function. I. Ginical spirometry in nor mal men. Amer. J. Med. 30, 243-251. 11. Morris. J. F . Koski. A., and Johnson. L. C. (1971). Spirometric standards for healthy nonsmok ing adults. Amer. Rev. Rryp. Dit. 103, 57-67. 12. Dickman. M. L.. Schmidt. C. D.. Gardner. R. M.. Marshall. H. Wt., Day. W. C., and IVaraer. H. R. (1969). On-line computerized spirometry in 758 normal adults. Amer. Rev. Retp. Dit. 100. 780 - 790. 13. Schmidt. C. D.. Dickman. M. L.. Gardner. R. M.. and Brough, F. K. (1973). Spirometric stan dards for healthy elderly men and women: 532 subjects, ages 55 through 94 vears. Amer. Rev. R, yp. Dit. 109.933-939. 14. Ferris. B. G. Jr.. Anderson. D. O.. and Zickmantel. R. (1965). Prediction values for screening tests of pulmonary function. .4-;rf. Rev R, yp. Dit. 91. 252-261. 15. Chemiack. R. M.. and Raber, M. B. (1972) Normal standards for ventilatory function using an automated wedge spirometer. Amer. Rev. Retp. Dit. 106. 31-46. 14. Dischcr. D. P.. and Palmer. A. H. (1972). Development of a new motivational spiromctemiionale for hardware and software. J. Oecup. Med. 14, 679- 615. DOW 06813 STUD 05 5 0 t 71- nJ x 5 iy to the the -CM tual OCK *>- ter. (he i 111 nng :ya- . 1. it>. jni vtok- r.tt. Du LinllT. iw ulc SPIROMETRIC measurements is epidemiologic SURVEYS 465 17. Knud<on. R. J., Slatin. R. C., L*N'it*. M. D.. and Burrow*. B (I97oi. The rruvinul expiratory flow-volume curve. Normal standards. variability an<i effects of age. ,Wr. Rev. /tap. Pi\. 113,317-600. II. Ban. H. (I973i. The flow volume loop: Normal standards andabnormJitiev in chronic obstructive pulmonary disease. Chest 63. 171-176. 19- Black. L. F.. Offord. K.. and Hyail. R. E. (1974). Variability in the maximal expiratory flow volume curve in asymptomatic smokers and nonvmokers. Amer. Rev. Resp. Dis. 110, 282-292. 30. ScMevinjtr. Z. (1975). Nomograms for pulmonary ventilatory- function: shoulJ coronary heart disease patients be included? (Editorial). Chnt 67. 379-310. 21. Higgins. 1. T. T.. Higgins. M. W.. Lochshin. M. D.. and Canale. N. (1969). Coronary disease in mining communities in Marion County, West Virginia. J. Chroo. Dis. 22, 163-179. 22. Schlesinger. Z-. Goidbourt. V.. hfedali*. J. H.. Oron. D.. Neufeld. H. N.. and Riss, E. (1973). Pulmonary ventilatory function values for healthy men aged 43 years and older. Chest 63. 320-324. 23. Smith. A. A., and Caensler. F. A. (1973). Timing of forced expiratory volume in one second. Amer. Rev. Resp- Dit. 112. 112-133. 24. Higgins. M. W.. and Keller. J. B. (1973). Seven measures of ventilatory lung function. Population values and a comparison of their study to discriminate between persons with and without chronic respiratory symptoms and disease. Tecumseh. Michigan. Amer. Rev. Resp. Dit. 101, 231-272 23. Schoenberg. 3. B.. Beck. C. J.. and Bcuhuys. A. (1971). Growth and decay of pulmonary function in healthy blacks and whites. Retp. Physiol. 33. 367-393. 26. Abramow itr. S.. Leincr. G. C.. Lewis, W. A., and Small. M. J. (1965). Vital capacity in the negro. Amer. Rrv. Resp Dis. 92, 247 - 292. 27. Lapp. K. L.. Amandus. H. E.. Hall. R.. and Morgan. W. K. C. (1974). Lung volumes and flow rates in black and white subjects. Thorex 29, 113-113. 23. Seiner. C. C., Sicgclaub, A. B.. Friedman. G. D.. and Collen. M. F. (1974). Differences in pul monary htnetion related to smoking habits and race. Amer. Rer. Resp. Dit. 110. 398 - 601. 29. Mustafa. K. Y. (1977). Spirometric lung function tests in normal men of African ethnic origin. Amer. Rev. Resp. Dis. 116, 209-213. 30. Lanes*. R. R.. Keller. M. D.. Foley. M. F.. and Underwood. E. H. (197*). Differences in pulmo nary function tests among Whites. Blacks, and American Indians in a textile company. J. Occup. Med. 20. 39-44. 31. Corey, P. N.. Ashley. M. J.. and Chan-Yeung. M. B. (1979). Racial Differences in hmg (Unction: Search for proportional relationships. J. Orrup lled. 21. 393-398. 32. Coles. J. E.. and Malhotra. M. S. (1965). Differences in lung function between Indians and Euro peans. J. Physiol. (London) 177, 17P-IIP. 33. DaCosta. J. L. (1971). Pulmonary lUnction studies in healthy Chinese adults in Singapore. Amer. Rev. Resp. Dis. 104. 128-131. 34. Ching. B.. and Horsfall. P. A. L. (1977). Lung volumes in normal Cantonese subjects. Preliminarystudies. Thorax 32, 332-335. 35. Brown. P., Sadow jlcy. D . and Gajdusek. D. C. (1978). Ventilatory lung fanction measurements in Pacific Island Micronesians. Amer. J. Epidemiol. 108, 259-265. 36. Glass. W. I. (1962). Ventilatory (Unction differences between Polynesian and European rope workers. .V.Z. Med. J. 61. 433-444. 37. Rossiter. C E.. and W*t0. H. (1974). Ethnic differences in lung function. EviJence for propor tional differences, lot. J. Epidemiol. 3, 55-61. 38. Morris, i. F.. Temple. W. P.. and Koski. A. (1973). Normal values for the ratio of one-second forced expiratory volume to forced vital capacity. Amer. Rev. Resp. Dit. 109. 1000-1003. 39. Milne, i. $. (1971). Longitudinal respiratory studies in older people. Thorax 33. 347 - 534. 40. Pride. N. B. (19791. Analysis of forced expiration--a return to the recording spirometer? (Edito rial). Thorax 34. 144-149. 41. Gottis. T.. Kontos. J.. and Gardikat. C. (1979). Effective time of the forced expiratory spirogram in health and airways obstruction. Thorax 34, 187-193. 42. Liang. A.. Macfic. A. E.. Karris. E. A., and Whitlock. R. M. L. (1979). Transit-time analysis of SV5SG001S . DOW 06814 . I * ' ` ( " - - -....................... 466 MILLER AND THORNTON the forced expiratory spirogram during clinical remissions in juvenile asthma. Th,*ut N, 194-199. 43. Tockman. M. Menkes. H.. Cohen. B.. Fermott. S.. Benjamin.}.. Ball. C.. Jr.. and Tonscia.J. (1976). Comparison of pulmonary function in male smokers and nonsmoken. Amer. An. Hrsp. Dit. 114, 711-723. 44. Kerr. H. D. (1973). Oiumal variation of respiratory function independent of air quality. Experi ence with an environmentally controlled exposure chamber for human subjects. Arch. Em iron. Health 26. 144--152. 45. Reinbcrg, A., and Cervxis. f. (1972). Circadian rhythms in respiratory functions, with special reference to human chronophysiology and ehronopharmacolx'sy. Bull. Phytioputhol. Hetp. 8, 663 - 675. 46. McDermott. M. (1966). Diurnal and weekly cyclical changes in airways resistance. J. Fhyiiol. (London) 186, 909-929. 47. McKerrow. C. B.. and Rossitcr. C. E. (1961). An annual cycle in the ventilatory capacity of men with pneumoconiosis and of normal subjects. Thorat 23. 340-549. 4*. Spodnik. M. J. Jr.. Cushman. C. D.. Kerr. D. H . Blide. R. W.. and Spicer. W. Jr. (1966). Effects of environment on respiratory function. Weekly studies in young male adults. Arch. Em iron. Health 13, 243- 254. 49. Hamosh. 9., and DaSilva. A. M. T. (1977). The effects on expiratory Dow rates of smoking three cigarettes in rapid succession. Chau 72. 6)0-613. 50. Mungall. 1. 9. F.. and Hainswonh. R. <19*9). Assessment of respiratory function in patients with -- chronic obstructive airway disease. Thotai 34,254 - 251. 51. Fletcher. C. M. (1964). The problem of observer variation in medical diagnosis w ith special refer* ence to chest diseases. Methods Inform. Med. 3, 91-103. 52. Cate, T. R.. Roberts, J. S.. Russ, M. A., and Fierce, J. A. (1973). Effects of common colds on pulmonary hinction. Amer. Rev. Hesp. Dir. 10S, 658-664. 53. Ficken. J. J., Niewoehner, D. .. and Chester. E. H. (1972). Frolonged effects of viral infections of the upper respiratory tracts upon small airways. Amer. J. Med. 52, 738 - 746. 54. Blair. H. T.. Greenberg. S. B.. Stevens. 9. M.. Bilunos. 9. A., and Couch. R. B. (1976). Effects of rhinovirus infection on pulmonary function ofhealthy human volunteers. Amer. Her. Hasp, Dit. 114,95-102. 55. Fridy. W. W. Jr.. Ingram. R. H. Jr.. Hicrholzer, J. C.. and Coleman. M. T. (1974). Airway tone* lion during mild viral respiratory illness: The effect of rhinovirus infection in cigarette smokers. Ann. fniern. .Mod. 80. 150-ISS. 56. Hall, W. J.. Had, C. B .and Speers. D. M. (1978). Respiratory syncytial virus infection in adults: Clinical, virologic and serial pulmonary function studies. Ann. Intern. Ited. 88. 203-205. 57. Collier. A. M.. Fimmel. R. L.. Hassclblad. V.. Clyde. W. A.. Jr.. Knelson. J. H.. and Brooks, J. G. (1978). Spirometric changes in normal children with upper respiratory infections. Amer. Her. Hrtp. Dit. 117, 47-53. 58. Hogg. J. C., Williams. J.. Richardson. J. 8.. Macklem, 9. T.. and Thurlheck. W. M. (I9'0). Age as a factor in the distribution of lower-airway conductance and in the pathologic anatomy of obstructive lung disease. .V. Enfl. J. Med. 282. 1283-1287. 59. Feslin, R.. Bohdana. A.. Hannhart. B.. and Jardin. 9. (1979). Comparison of various methods for reading maximal expiratory flow-volume curves. Amer. Her. Hop. Dit. 119. 271-277. 60. Stanescu. D.. Veritcr. C, VanLeemputten. R.. and Bras<eur. L. (1979). Constancy of effort and variability of maximal expiratory flow rates. Chett 76, 59 - 63. 61. Lapp. N. L., and Hyatt. R. E. (1967). Some factors affecting the relationship of maximal expira* lory flow to lung volume in health and disease. Dit. Chen 59. 475-481. 62. Miller. A.. Tcirstein. A. S.. Jackler. I.. Chuang. M.. and Siltzbach. L. E. (1974).'Airway function in chronic pulmonary sarcoidosis with fibrosis. Amer. Her. Hop Dit. 109, 179-189. 63. Weill, H.. Ziskind. M. M.. W.i^enspack. D . and Rossiter. C. E. U975). Lung function consc* tiucnces of dust exposure in j*hetos cement manufacturing plants. An h. Eniiron. Hi ahh 50, S8-97. 64. Hankinson. J. L.. Reger. R.. and Morgan. W. K.C. (1977). Maximal expiratory flows in coal miners. Amer. Her. Hetp. Dit. 116.175 --180. ( J ! ! ! I | -im s fio o is : Dow 06815 .. JU. fir %p. ^CvUl nW. fmea .Ani wmrtll *eC*rJ% %* ftlfW* 'U'fW .full*. mrr. Age ut tanJ pir^ ;uxi ifK* 4 SO. not t SPIROMETRIC MEASUREMENTS IS EPIDEMIOLOGIC SURVEYS 467 (5. Morris. J. F. (19741. Spirometry in the evaluation of pulmonary function. Hi W. J. ,\ltJ. 1'5, 110-11*. 44. MeFadden. E. R. Jr., Kiker, R., Holmes. B. and OeGrooi. W. J. (1974). Small aira> Ji-tjvt: An assessment of the tests of peripheral airway Amotion. Amer. ]. Med. 57, 171 - 142. 47. Shapiro, W\. and Patterson, J. F., Jr. (1943). Effects of smoking and athletic conditioning on ventilatory mechanics including observations on the reliability of the forced e spirogram Amer. Rev. Retp. Dit. EJ, 191-199. 41. Cotes. J. E. Lung Amction. (I97J). "Assessment and Application in Medicine." p. ?<4. Blackwell. Oxford. 49. Cresbic. W. A., Clarke, M. B.. Cos. R. A. F.. Mclvcr, N. K. I.. Anderson. I. K.. Evans. H. A.. Liddlc. C. C.. Cowan. J. L., Brookings. C. H.. and Watson. O. 0. (19771. Phy tical character* istics and ventilaiory function of 404 commercial divers working in the North Sea. frit. J. lad. Mid. 34, 19-25. 70. Wilson. K., Miller, W',, Blair. T., and Stevens, P. (1974). "Flow-Volume Relationships in Obstructive and Restrictive Lung Disease." American College of Chest Physicians. Atlanta. October 34. 71. Kuperman. A. S.. and Riker. J. B. (1973). The predicted normal maximal midexpirutory flow rate. Amer. Rev. Kelp. D.t. 107. 331-334. 72. Sobol. B. J. (1974). The early detection of airway obstruction: Another perspective. Amer. J. Med. 40, 419-434. 73. Knudson. R. J.. and Lebowitz, M. D. (1977). Comparison of flow-volume and closing volume variables in a random population. Amer. Rev. Rnp. Dii. 116. 1039- 1045. 74. Miller. A., Thornton, J. C.. Smith. H.. Jr., and Morris. J. F. Spiromctric "abnormality" in a normal male reference population: Further analysis of the 1971 Oregon survey. Amer. J. tnd. Med., in press. 75. Sobol. B. J.. and Sobol. P. O. (1979). Percent of predicted pt the limit of normal in pulmonary function testing: a statistically valid approach (Editorial). Thorax 34, 1-3. 74. Morris. J. F.. Koski. A., and Brecse. J. D. (1973). Normal values and evaluation of forced expi ratory flow. Amer. Rev. Rnp. Dit. Ill, 753-763. 77. Sobol. B. J.. Parks. S. S.. and Emergil. C. (1973). Relative value of various spiromctric tests in the early detection ofchronic obstructive pulmonary disease. Amer. Rev. Retp. Pit. 107,753 - 763. 71. Marcq, M.. and Mincttc. A. (1976). Lung function changes in smokers with normal conventional spirometry. Amer. Rev. Retp. Dit. 114, 733-734. 79. Bower, C. (1941). Respiratory symptoms and ventilatory Amction in 172 adults employed in a bank. Amer. Rev. Retp. Dit. 43, 644 - 639. 10. Tashkin. D. P.. Detels. R.. Coulson. A. H.. Rokaw. S. N.. and Sayre. J. W. (1979). The UCLA population studies of chronic obstructive respiratory disease. II. Determination of reliability and estimation of sensitivity and specificity. Environ. Ret. 20. 403-424. 41. Fletcher. C.. Peto. R.. and Tinker. C. (1976). "The Natural History of Chronic Bronchitis and Emphysema: An Eight Year Study of Early Chronic Obstructive Lung Disease in Working Men in London." Oxford Univ. Press. New York. 43. Fletcher. C.. and Peto. R. (1977). The natural history of chronic air flow obstruction. Brit. Med. J. 1,1443-I4J4. 43. Petty, T. L. Pierson. D- J.. Dick. N. P-. Hudson. L. D.. and Walker. S- H. (1976). Follow-up evaluation of a prevalence study for chronic bronchitis and chronic airway obstruction. Amer. Rev. Retp. Dit. 114. 441-490. 44. Bates. D. V. (19731. The fate of the chronic bronchitic: A. report of the ten-year follow -up in the Canadian Department of Veteran's Affairs Coordinated study of chronic bronchitis. Amer. Rev. Retp. Dit. 104. 1043-1945. 45. Braun. S. R,, DePtce. G. A.. Tsiatis. A.. Horrath, E.. Dickie, H. A., and Rankin, J. (1979). Farmer's lung disease: long-term clinical and physiologic outcome. Amer. Rev. Retp. Dit. 119, 145-191. 46. Miller. A.. Teirsiein. A. S.. Chuang. M.. Selikoff. I. J.. and Warshaw, R. (1973). Changes in pulmonary Amction in workers exposed to vinyl chloride and polyvinyl chloride. Ann. .V, Y. Aeod. Sri. 246. 43-52. t M w n n is 1 DOW 06816 <568 MILLER AND THORNTON 17. Knudson. R. J.. Burrows. B.. and K'N'wita. M. D. (1979). The maximal evpintory fl.'wvr'-.-me tune: In u<< in the detection of ventilatory abnormalities in a population study. ,<;rr Rn Resp. liit. 114. *71-179. *1. Fletcher. C. M. {IVTgi. Terminology in chronic obstructive lung disease. J. CriJemint. C.-;m. Hruhh it >9. Bouhuys. A.. Barbero. A.. Schilling. S. E.. and Van de Wocstijnc. K. P. (1969). Chronic respira tory disease in hemp workers. Amrr. J. A/rd. 46. 526-337. 90. Anderson. 0. U.. Ferris. B. G. Jr.. and Zickmjntel. R. (1963). The Chilliwack respiratory survey. 1963: Pan III. The prevalence of respiratory disease in a rural Canadian town. Canud. A/rd. Aisoc.J. 92, 1007-1016. 91. Reid. D. 0.. Brett. C. Z.. Hamilton. P. J. S.. Jarren. R. S.. Keen. H.. and Rose. C. (19*4). Cardiorespiratory diseases and diabetes among middle-aged male civil servants: A study of screening and intervention. Lower t. 469 -473. 91 Mueller. R. E.. Keblc. D. L... Plummer. J.. and Walker. S. H. (1971). The prev alence of chronic bronchitis, chronic airway obstruction and respiratory symptoms in a Colorado City. Amer. Rev. Retp. Dit. 103. 209 - 221 93. Manfreda. J.. Nelson. N.. and Chemiack. R. M. (19*1). Prev alence of respiratory abnormalities in a rural and an urban community. Amtr. Rev. Reap. Dis. 117, 213-220. 94. SelikofT. I. J.. and Anderson. H. A. (1979). "A Survey of the General Population of Michigan for the Health Effects of PBB Exposure." Submitted to the Michigan Department of Public Health. September 30. 93. Babbott. F. L.. Gump. D. W.. Sylwester. D. L.. MacPherson. B. V., and Holly. R. C. (I9W). Respiratory sv mptoms and lung function in a sample of Vermont dairymen and industrial workers. Amrt. J. Rub. Hi-ulth 70. 241-243. | j . ( ; ' BTC'S" 00 0IS i I III I t DOW 06817