Document nkO74yepRk0nRJ693b53NNZNm

S T l) U!) !>i I !) Th# Us# of Pulmonary Function Testing and Questionnaires as Epidemiologic Tools in th# Study of Occupational Lung Disaasa* Brum A BothUtks, US. tad Jamas A Merchant, US, Dr. fjf. Screening has been defined u "thn presumptive idea* ^ tifieatiao of unrecognized dittos* or defect by tbe application of tests, examinations. or other procedures which eeu be applied rapidly to sort out apparently well persons who probably have a diteaia from thoee who probably do nob"1 2C3riteria for tests useful for screen ing or other epidemiologic studies differ somewhat from those applicable to laboratory investigations. Tbe equip ment and procedure must be relatively simple to prevent malfunction and delays far a field setting. To obtain high rates of participation, tests most net only be safe, but also relatively free of discomfort for subjects. Test results should not be influenced by subject cooperation or observer bias, although this is never completely true for any test. Spirometry and standardized questionnaires meet these criteria quite well and have found wide use as epidemiologic tools in studying occupational lung disease. Two other attributes of screening tests are in*portant: reproducibility or precision and validity, or the ability to correctly identify persons with the disease or cosdittos being sought These will be discussed in de tail for spirometry and respiratory questionnaires. REPnoDoaBUJTT Table 1--HigkBghlt af "grsodsrdisalfso af Spiromatrj* (Statmnont by tbe American Thoracic Sociclj&) Equipment (minimum requirements) 7 titer volume Accumulate volume for 10 second* Volume accuracy 3% of reading or 50 sal, whichever i* greater Flow range 0-12 L/ace Resistance <1-5 cmHiO/L/aec at 12 L/aee flow Flow accuracy 551* of reading or 0.2 L/see, whichever ia greater T*t Performance and Measurement Minimum of 3 acceptable maneuvers, (no leaks, obstruction of mouthpiece, cough, Valsalva, or early termination) Start point defined by back extrapolation or equivalent method Back extrapolated volume <10% of FVC or 100 ml, whichever is greater Best 2 of 3 acceptable curve* vary for FVC by no more than '' 5% of leading or 100 ml, whichever is greater Bert FVC and FEVi used regardless of whether they occur on same curve attended a National Institute for Occupational Safety and Health (NIOSH) approved training course in pulmonary function testing (Table 2) to *-- pro ficient in a standardized technique for administering pulmonary function tests. When multiple lasts on an individual are to bo compared, biologic variation can bo minimized by maintaining conditions during testing as nearly identical as possible (eg, same time of day). This is also important if resubs from two groups of subjects are to be compared. Even when spirometry is Tabic 2--Highlight! af NIOSH Approved Pulmonary' Function Training Caarcm Tbe reproducibility af spirometric test results is determined by two factors: (1) measurement error, caused by equipment limitations and the observer who carries out the test and measures or calculates the . 1. Course Director: MJD., sum or health professional with advanced degree in pulmonary physiology or related field with training and experience with equipment and pro cedures. results, and (2) true biologic variation. Measurement errors can he reduced by setting accuracy requirements for the equipment standardizing procedures and cal culations, and requiring frequent calibration. This has recently been done for spirometry by the American Thoracic Society and die Department of Labor. Highlights of the minimum equipment requirements and recommendations for test performance and calcu lations are given in Table L Thirteen of 18 currently available spirometers met tbe recommended accuracy requirements for measurement of tbe forced expiratory volume in tbe first second (FEV,) when tested on a laboratory simulator.4 5Spirometric tests performed to fulfill medical surveillance requirements of an Occupa tional Safety and Health Administration (OSHA) standard, such as that for occupational exposure to cotton dust, now must be carried out by a technician who has 'National Institute for Occupational Safety and Health, Appalachian Laboratory for Occupational Safety and Health, Morgantown, West Virginia. Rrprint rrcusstc Dr. Merchant, 944 Chaitnut Bids* Baud, Uorgontoum, West Virginia <26505 2. Minimum of 1 instructor and spirometer system ineluding calibrating syringe per 4 students. 3. Minimum of 16 hours of instruetion with at leaat:/ a. 4 hours lecture and/or A-V material / h. S hours small group practical instruction c. 2 hours (valuation and testing of each student's skills ia spirometry testing 4. Content to include: a. Basie physiology b. Instrumentation requirements l-t.-li-g calibration a. Performance of testing d. Measurement of tracings and calculations a. Discussion of data quality including routers of error, corrective action, reproducibility f. Supervised use of equipment 5. Application for approval to: Director, Division of Training nod Manpower Develop ment, NIOSH Robert A Taft Laboratories 467S Columbia Parkway Cincinnati, OH 45226 114S AMERICAN COLLEGE OF CHEST PHYSICIANS CHEST, 79: 4, APRIL, 1981 SUPPLEMENT ocr.cnoois i TabU 3 tfiw Intrmtubjml Cm^o'm / Yurioiimm latercet Day*1 FVC (%) za FEV, m <%) 3.1 (%) u lOweekst i e II 13 1 year! Smokers 4 10 19 Non-smokers 3 0 16 IMnwi 3 fRifmetl ^Utinw 7 performed properly oa accurate equipment under op timal condition*, the result* very oa repeated testing. Table 3 contains the coefficient of variation (standard deviation divided by mean value and expressed as a percentage) for the forced vital capacity (FVC), forced expiratory volume in the first second (FEV,), and the maximum expiratory flow rate after 50 percent and 75 percent of the FVC has been exhaled (V--- and V--respectively) oa repeat measurements to the same individual Clearly the FVC and FEV, are \ the most reproducible. No significant difference to the y reproducibility of these measurements between smokers and nonsmoleers was present Reproducibility is especially important when the change in a measurement over a short period of time is of interest, eg, the change to pulmonary function over a workshift Cochrane and co-workers1 *h3av4e stated that to short-term situations where the subject is used as his own control, changes of 8 percent of FEV, are' significant, but at least a 19 percent change to V___.. is needed "to provide a similar degree of confidence." Reproducibility must also be considered in longitudinal studies. Figure 1 shows two theoretical patterns far change to FEV, with age to three individuals. The variability depicted in the bottom panel results from both errors to estimation of the true FEV, for each individual at the two ages and true differences in the rate of decline of FEV, between individuals. If the follow-up period is relatively short, estimation enure are of major importance.* Note that a cross-sectional study can provide an accurate estimate of the mean rate of change to FEV, with age to either situation, but cannot provide an accurate estimate of the variation between individuals or within an individual over time. 'The latter makes prediction of long-term changes from short periods of observation difficult Figura 2 shows the FEV, of an asymptomatic smoker without other ex posure to respiratory hazards, repeatedly measured to our laboratory over four years. Diving die first year, the decline was over 700 ml, yet without change to smoking habits, the overall rate of decline for the four yean was approximately 80 ml per year. The inaccuracy of using-' the decline in the first year to predict long-term results is obvious. Responses to questionnaires may be used to define s condition (Monday chest tightness for byssinosis), to provide information on the presence of factors coo-' founding the relationship between pulmonary function and the exposure being studied (eg. smoking and previous exposures), or to estimate total cumulative occupational exposure to an agent when quantitative measurements for an individual are not available. In- terviewers must be weD trained in a standardized method of administering respiratory questionnaires, if reproducible ana vatm to be obtained. Table "4 lists some features of interviewer training. The reporting of smolong and occupational history, although showing some variability, is generally quite Table 7rciasif fiiiim'iwri for (be Vto of Rmpiratory Quodtionnoirot to top figure. 1. Questionnaire ted instructions should be studied to detail and difficulties discussed with an experienced interviewer or investigator 3. Interviewer should apply the questionnaire to 10 or mors subjects who hsre some respiratory symptom* (eg. pul monary clinic patients) 3. Practice toterriewa should be witnessed by an experienced colleague or preferably tape-recorded for correction end clarification later 4. The importance of following exactly the printed wording of the questionnaire must be stressed 3. Acceptable explanatory statements should be included to the instruction manual and strictly followed when needed CHEST, 79: 4. APRIL, 1981 SUPPLEMENT INTERNATIONAL CONFERENCE ON BYSSINOSS 11SS DOW 06819 4J*r 30 YEAR 0L0 MALE SMOKER CM* mange to smoking taMti) FV, tu a a* i ;< ;<r, non sj 3M I 2 TIME (YEARS) reproducible,* as shown m Table 5. However, avenge cigarette consumption is reported to be less reproduci-) (ble than current consumption. This may result in un-y reliable estimates of "pack-years" in older subjects.1' The reported prevalence of symptoms may be quite variable. Table 6 shows that the symptoms of cough and phlegm increased significantly over a one-year period during which the workers were exposed to an educational program concerning the risks of asbestos. Schilling and oo-workers11 compared the prevalence of bysstnosis in 183 male cotton workers as recorded by two observers. Table 7 shows that complete agree- TabU 5 -- Reliability of Smoking end Occupational Riiiariat Mean Reported Survey 1 Survey 2* R Smoking (pack-years) 23.3 24.9 .81 Yean employed as pipe coverr 9.0 10.3 .95 Yean asbestos exposure 10.8 11.3 .8* Adapted from reference 9 "Surrey 2 performed 1 year after survey 1 R --correlation coefficient Tabla 6--RadaWiry of Qmottionnairo Response* Symptom Prevalence (%) On Clinical Questionnaire Surrey 1 Survey 2 Agreement Dyspnea L.7 6.7 .93 Cough 33.2 51.7 .65* Phlegm 38.7 56.7 .63" Wheex* 8.4 13.4 J3 History of pneumonia 25.0 20.0 J2 Sinus trouble 26.7 35.0 JO Adapted from reference 9 *P<05 FNcoim 2. Actual measured FEV, in 30I year-old male smoker with m change in 4 rmolriwf habits or exposures over four yearn. X indicates measured value. ment was obtained in only 76 percent of but agreement on the presence or absence of disease was achieved for 93 percent of the workers. These studies\ point out that symptom prevalence is variable on J repeat measurement even when recorded by trained ' interviewers using standardized questionnaires. Table 7--ProooUnco of ByuinotU: Coos, Crodings By Two Observers in 183 Molo i|s*Md '/ Workon Observer A Normal Grade I Grade II 79 67 17 Normal 78 a I Grade I 70 Grade II 35 72 6 1 6 47 14 0 17 20 Total 183 Total 183 Complete reproducibility in 72+47+30--139 (76%) Agreement oa presence or sbteaee of byminosis in 139+14+17-170 (93%) Normal -- do evidence of byseinoeis / Grads I -- chest tightness on Mondays only ` Grads II - chest tightness on Mondays and other working days (reference 11) Table 8--Predictive Vmluo of Positive Tost % Sensitivity 95 95 95 95 60 % Specificity Prevnlencn 95 50/100 95 5/100 95 1/100 95 1/1.000 98 1/1,000 Predictive Value <%> 95 50 16 3 3 nss AMERICAN COLLEGE OF CHEST PHYSICIANS CHEST. 79: 4, APRIL, 1981 SUPPLEMENT DOW 06820 Ttem 3. Mean direct maximum voluntary ventilation ef cotton workers and healthy controls (ref 13). Vaudttt The usefulness of a screening test lies in its ability to distinguish those who probably have the disease in question from those who probably do not The prob ability that a person with a positive soecainf test result actually has the disease sought is called the predictive value of a positive test11 It is important to recognize that this probability depends not only; on the validity of the test as measured by its sensitivity | Jand specificity, but also on the prevalence of the dis- ease in the screened population (Table 8). Thus, if a disease is rare, many false positives will occur, even with highly specific test, before a true poritivo is found. To determine the validity of a screening test, sons* other measure must be used to define true positives and true negatives. The validity of die grading of bvssinosis by questionnaire responses has been evalu ated by comparing pulmonary function test results for workers classified in the different grades.1*'14 Figure 3 shows that for groups of workers, pulmonary func tion is progressively reduced below normal as byssinosis grade increases. Workers with grade 2 byssinosis by questionnaire response not only have lower hateline values for FEV, than grade 0 or grade 1/2 workers, but also show persistent decrements in FEV, over the workihift throughout the work week (Fig 4). Thus, the grading of byssinosis by symptoms reported^ on a standardized questionnaire appears to separate l groups of workers with objective differences in pal* 1 monaiy function. Such validation against an external/ standard should be done whenever possible. In large surveys the use of a separate measure to define true positives is usually not possible for spirom etry, since no other pulmonary function tests are done. Rather, an individual's results are compared with a predicted normal value derived from a pulmonary function survey of a group of presumably healthy pes sons. Factors which affect the predicted value include theft population from which the reference group was chosear -the criteria for normality used to select individuals for inclusion, and the methodology for testing and calcu lating results. The discussion which follows draws heav ily on tha work of Miller and Thomton." Przsictd Normal Values for Spirometry The predicted normal values derived from a refer ence group depend upon the distribution within that group of age, race and ethnic group, socioeconomic class, occupation, urban-rural residence, and other fee- FEVyj DURING FIVE OCT OUST EXPOSURE OF 2S CARDERS ZSSS0001Sj CHEST, 79: 4. APRIL, 1981 SUPPLEMENT FYcorb i Mean FEV, of orders with dif fering grades of byedaosis during five days of dust Mposutu (ref 14). INTERNATIONAL CONFERENCE OH BYSSINOSIS 117$ DOW 06821 Table > Paiwawy Fmmetien / Blech And White Cmel Mlnen Matched Far <4|, Height, And Smelting /Hilary Mean White Black Difference Ago (yea) 34.7 34.9 - Height (ia) 69.9 70.0 - FEV, a> FVC a) 4.18 331 13.9% 5.20 4.43 14.4% TLC (l) 7.10 6.01 13.3% re/eresea 16 ton. These characteristics must be considered wheu evaluating the applicability of the predicted values from e given reference group to the current stndy group. Several studies have documented a difference for FVC and FEV, between healthy white and blade men of the same age end standing height Table 9 shows results from the study of Lapp and co-workers1* who found the FEV, of black subjects to be append* matdy 14 percent lower than white subjects. Although simple adjustment of predicted values obtained from studies of white reference groups for use in blacks is not completely satisfactory, use cl such a scaling factor is preferable to ignoring the difference until adequate prediction formulae for blacks are available. Schoen berg and co-workers1' have produced regression equa tions for spirometric values for black subjects, but the number of adult men was relatively small (120), and the equations contain the In (weight) and (age)' as variables, thus making computations somewhat cum bersome. Rossiter and co-workers1* suggest reducing the predicted values of FEV, and FVC from white reference groups by 13.2 percent when they are applied to blacks. Since the FVC and FEV, appear to be re duced approximately proportionately, no adjustment for the FEVj/FVC ratio is necessary. Lanese and co-workers1' found similar differences between white and black textile workers. The common assumption that the use of age in the prediction formulae derived from e reference group completely controls for any differences in age distribu tion from the study group is not totally accurate. This is discussed under considerations of the "normal range" given below. The other factors mentioned are more difficult to evaluate and have usually been ignored in deriving normal values. Criteria used for selection of "healthy" individuals for inclusion in the reference group in various studies have varied considerably. While most modem studies exdude subjects with respiratory symptoms, past his tory of severe or chronic respiratory or cardiac disease, and smokers, only recently have the effects of occupa tional exposure been considered. Very few studiea have required chest radiographs or physical examinations, for inclusion. Although theoretically one might wish to j obtain a reference group as totally free from respiratory u system impairment as possible, the more selective the/) Jcriteria for inclusion, the more lately that any compari- son of this group with an unscreened group from the Table 10--tnfluinee of ftumber / Kummf, Performed on The Ueaeured FEF, .i - . ------ ,-- Bt FEV, Maneuver 1 29% 2 19% 3 17% 4 18% * 19% Beet FEVi on maneuvers 4 or 3 exceeded beet on maneuvers 1-3 by: 3% ie 8% of subjects 10% in 2% of subjects Adapted from reference 20 general population will show a statistically "significant" prevalence of abnormalities in the unscreened group. The practical significance of such e difference may be difficult to determine. Methodologic factors also influence predicted values. Equipment and test procedures have already been dis cussed. The optimum number of maneuvers performed has been debated citing a "learning effect** for the first few maneuvers followed by a "fatigue effect." Table 10 is adapted from the work of Xnudson and associates** who found that the largest FEVv occurred in the first three trials in 65 percent of over 3,000 sub jects tested in a genera] population survey. When the largest FEV, occurred on the 4th or 5th trial, it ex ceeded the largest from the first three maneuvers by more than 5 percent in only 8 percent of subjects end by more than 10 percent in only 2 percent of sub jects. Tager and co-workers11 found the largest FEV, in the 4th or 5th trial in 31 percent of subjects and found the average single maximal FEV, to be 105 ml larger than the mean of the three largest values. These differences ere not necessarily negligible when large groups are studied and small absolute differences may be statist:caDy significant Nevertheless, the recommen dation of the Nationel Heart, Lung, and Blood Insti tute Epidemiology Standardization Project" is that\ "three acceptable"--maneuvers be obtained. Whatever \ practice is followed, it is essential that the same pro- I cedure be used for the study end the reference os/ "control" groups. It is also important to obtain an accurate measure ment of the maximum FVC Falsely low values for FVC will produce artificially high FEV,/FVC ratios'! and very significantly increased flow rates near the end ] of expiration. The flow rate measured at e given per-*' centage of .the falsely low FVC is actually occurring at a higher lung volume than the flow rite measured at the seme percentage of the true FVC Hanldnson and Petersen" have shown that at least 10 seconds of v expiratory time must be recorded if 94 percent of sub jects with true FEV,/FVC ratio less than .70 are to i be correctly classified (Table 11). The predicted values./ 118$ AMERICAN COLLEGE Of CHEST PHTSICIANS CHEST. 79: 4, APRIL, 1981 SUPPLEMENT r s is rin m s ; DOU 06822 : STOOD5554 Table 11--Effoti of Spiromotor Cat-*/ Time On Obaorood FF,/FFC fbtie Cut-off Time % of 204 Subjects Correctly Claarified With (vee) FEV,/FVC <.70 4 47 C 67 7 SI S 86 9 93 10 .' 94 11 97 12 97 reference 23 % from three recent surveys for FVC, FEVlf end FEV,/ FVC for a 50-year-old man who is'70' tall are given in Table 12. Morris and co-workers***** used the Kory technique*' for measuring FEV,. This method has been shown to result m values which are an average of 179 ml lower th the bade extrapolation technique." The FVC is unaffected by this method of measurement and agrees with that of Petersen and co-workers** who used a rolling seal spirometer and a method equivalent to back extrapolation to measure the FVC and FEV,. Knudson and associates** used a pneumotachograph to measure flow and then integrated to obtain volume. Their value for FEV, agrees with that of Petersen et al, but tbe FVC value is lower, resulting in a higher predicted FEV,/FVC ratio. If the FVC is falsely re duced, predicted values for flow rates near the end of expiration will also be too high. The interpretation of these flow rates is made difficult by this sensitivity to Table 12 -ProAietod KJus For Male Apod SO Pear* and 70* TmU rvc a> FEV, (L) FEV,/FVC (7.) Peterten et *1" 4.88 3.70 76* Morris et al**-** 4.87 3.48 73 Knudton et sla 4.64 3.70 81 'Obtained bv talcing tbe ratio of predicted FEV, to predicted FVC Tabic 13--Freactance o/ "AboormoT Spiromairy in A flaeldr Population errors in measurement of the FVC. Because healthy individuals of the same sex, race, age, and height exhibit considerable variation in rpiremetric results, classifying a subject as "abnormal- de pends upon some arbitrarily defined lower limit of normal. The most common clinical practice is to the lower limit uf normal at some percentage of theN mean predicted value, often 80 percent, and to label all values below this as "abnormal.' This practice is not( completely satisfactory. As discussed previously, prediction equations derived from a given reference group are never completely applicable to any other group. Also, because a given percent predicted deviates less from die mean value for small mean values than for large ones, more abnormalities "rill always be\ found among older and short subjects when this method J for defining "abnormal* is used. Miller and co-worktx**y have calculated the prevalence of "abnormal" spiremetric values among the participants in the survey of ' healthy subjects by Morris.** Table 13 shows tbe remits for men where "abnormal" is defined as a value 79 percent of mean predicted for the FVC and FEV, and 75 percent of mean predicted for the FEF,*_,1%. The prevalence of "abnormal" values in this healthy group increases with age so that 12 percent of men over 60 yean would have an 'abnormally low' FEV, and 29 percent an "abnormal" FEF,,_,,% by this definition. Use of a statistical definition of "abnormal" is one .. means to avoid the problem just discussed. By setting^ the lower limit of normal at the mean predicted value \ Jless some multiple of the standard error of the esti- mate (SEE) of the regression line, the number of healthy, persons arbitrarily called abnormal is prede- termined. A lower limit of normal of the mean value --1.64 SEE will place 5 percent of the group below normal; if two SEE's are used, only 2.5 percent of the group will be below the limit This method may be more or less stringent than using 80 percent of mean predicted value depending on the SEE. Figure 5 shows a comparison of the mean --1.64 SEE and COMPARISON OF MEAN -1.64 SEE AN0 80% OF MEAN PREDICTED AS LOWER LIMIT OF Melee Age (yrs) 39 40-49 40-49 *60 FVC (%) 4.4 U 7J 7.1 FEV,<%) 4.2 6.8 9.1 11.9 FEFtt-i, (%) 17.2 17.9 19.7 28.4 Calculations from data of Morris'* with abnormal FVC or FEV, defined as lees than nr equal to of mean predicted and FEFu.)t75% of mean predicted.1* Ficoax 5. Comparison of mean --1.64 SEE and 80 percent of mean for defining louver limit of normal for a theoretical regression line of FEV, and age (ref 29). CHEST. 79: 4, APRIL, 1981 SUPPLEMENT INTERNATIONAL CONFERENCE ON 3YSSIN0S1S 119S DOW 06823 80 percent of the mean for a theoretical iegression equation relating ago and FEV,. At younger ages die statistical definition is more stringent, while the oppo site holds true far the older ages. Although use of a statistical definition of the norma] range alleviates the problem of increasing prevalence of "abnormality" fopt there with smaller predicted values, it does assume that) the SEE is the same far any point along the regression fine. This may not be true, especially if the reference group has proportionately fewer members at the ex tremes of age. Also, some variables do not exhibit a Gaussian (normal) distribution about the mean vahieA If the distribution is skewed or the SEE large, the 1 mean value less 2 SEE may include zero, thus invalidat-' mg this definition of the normal range." When the reference group is large enough, the most useful method far individual comparisons may be to determine the peremtage of observations in the refer ence group above given values of the variable in ques tion (percentile ranks). Results for an individual can . then be classified by the percentile tank In which they would fall in the reference group. Since the probability* of being truly "abnormal" increases as the percentile! rank decreases, the physician does nor need to arbitrar-/ ily interpret one level as the "lower limit of normal." Comparison or Pulmonary Function Between Two Groups Population studies may be undertaken to determine the prevalence of disease, to elucidate the natural his tory of a disease, or to establish an association of dis ease with a given agent In occupational lung disease epidemiology, the objective is often to determine a dose-effect relationship between an occupational ex posure and reduction in pulmonary function. Proper interpretation requires not only defining "abnormal" for DISTRIBUTIONS WITH DEFERENT PERCENTAGES BELOW LOWER LMTT Of NORMAL umrmmt 10CR LIMIT of nommsc tcroMSwroe individuals, but also detection and interpretation of any differences between the study group is a whole and a proper!y chosen reference ("control") group. Wha comparing the pulmonary function of the torn gruupsX the basis should be the distribution of observed values \ --not the percentages of observations classified as "ab- T normal." Miller and Thornton11 point out two/ groups may have significantly different percentages of observations below any arbitrary lower limit of normal, yet not differ in the mean value of observations (Fig 6). Contrariwise, two groups may have the same per centage observations below a lower limit of normal bet markedly different distributions (Fig 7). The most powerful techniques for comparing study and reference groups are the analysis of covariance and\ matching procedures. Matching requires a large pool ) of subjects in the reference group and the knowledge / of which factors are important Analysis of covariance allows direct estimation of the difference between groups while adjusting for the effect of confounding variables. Certain assumptions (eg, linearity of re sponse) are made in the analysis, and care must be taken that true differences between the groups are not obscured by limitations of the technique. Statistical consultation is necessary when attempting to use either of these techniques. Before beginning a comparison of this type, the in vestigator must also select what magnitude of differ ence in function he considers important to -detect. This determines the sample size he must use. The overall significance of a difference in pulmonary function be- - tween two groups depends on the basic question the study is attempting to answer and should not neces sarily be equated with either "statistical significance" or "clinical significance." In some situations, small""' absolute differences between groups may have import tut implications while in others statistically rigaificanf"^ differences may not <--3 DISTRIBUTIONS WITH PERCENTAGES BELOW UMlT OP NORMAL SIMILAR LOWER ~ `--' 01 MtiraMr <-n VALUE Of PARAMETER Ficnaa ft Distributions with markedly different per centages of observations below arbitrary lower limit of normal but oot significant difference in mean value (adapted from ref IS). 120S AMERICAN COLLEGE OF CHEST PHYSICIANS Ftcvmx 7. Distributions with similar percentages el ob servations below arbitrary lower limit of normal but markedly different meen values (adapted from ref IS). CHEST, 79: 4, APRIL, 1981 SUPPLEMENT DOW 06824 Th* prescne* of smokan and ex-smaken in a study group presents a difficult problem. Although the effect of smoking ou pulmonary function can be quantitated to acme Client for large groups,11 the impact on an individual is so variable that no useful predicted values for pulmonary function for smokers are available. Flet cher and co-workers11 state that "if the whole of the population were exposed to (say) 15 cigarettes/day for 30 years, the resultant FEV losses would have t skew distribution with many quite smell lose*, but a laiT of clinically significant loses." Thus, simply match-\ mg two groups for percentage of smokers and ax- } smokers or even pack-years of smoking does not oeose- ) sarily control for the effect of smoking on pulmonary function in producing differences between the groups. Summary Pulmonary function testing and questionnaires are valuable tools in epidemiologic studies of occupational lung disease. Accurate equipment and standardised methodology are vital to obtain reproducible responses. Far spirometry, the FVC and FEV, show the least in trasubject variability and on questionnaires; oocupational and smoking history are more reproducible than symptoms. The limitations of any method used to de fine a lower limit of normal should be kept in mmd\ and, whenever possible, groups should be compared by V. use of the distribution of observations in the two \ groups--not just the prevalence of "abnormal" findings: J BimBOS 1 Commission on Chronic IDoess: Chronic illness in the United States, Vol 1, Cambridge: Harvard Univenity Press, 1957 2 Standardization of Spirometry--Statement of the Ameri can Thoracic Society. Am Rev Respir Dis 1979; 119: 831-39 3a 29 CFR 1910.1043 Appendix D, Occupational exposure to cotton dust. Final Mandatary Safety and Health Standards. h 20 CFR 718.103 Appendix B, Standards for determin ing coal miners' total disability or deads due to pneu moconiosis. 4 Gardner R, Haakioson J, West B. Evaluating commer cially available spirometers. Am Rev Respir Dis 1980; 121:73-82 5 Cochrane C, Prieto F, Clark T. ZntrMnbfeet variability of maximal expiratory fiow volume curve. Thorax 1977; 32:171-78 8McCarthy D, Craig O, Cherniak R. Intzaindividual variability fe maximal expiratory fiow-vehme and closing volume in asymptomatic mbjacts. Am Rev Respir Dis 1975; 112:407-12 7 Tattemll S, Benson M, Hunter D, Mansell A, Pride N, Fletcher C The use of tests of peripheral king function for predicting future disability from aitfiow obstruction in middle-aged smokers. Am Rev Respir Dis 1978; 118:1035-50 8 Berry C. longitudinal observations, their usefulness and limitations with special refereaos to die forced expira tory volume. Bull de Physio-Path Respir 1974; 10:643-55 9 Samct ], Spcizer F, Cacosier E. Questionnaire reliability and validity in asbestos exposed workers. Bull de ' Physio-Path Respir 1978; 14:17748 10 Eckert HL, Petersen MR. Reger RB, Hahon N. Reli ability of smoking histories is standard questionnaire interviews (abstract). Annual Meeting, Profaanonal Association of tho VS. Public Health Service, San Francisco, CA: April 34, 1977 11 Schilling R. Hughes J, Dingwall-Fordyce L Disagreemeat between observers in an epidemiological study of respiratory disease Br Med J 1955; 1:8548 12 Vecchio TJ. Predictive value of a single diagnostic lest in unteketed populations. N Engl J Med 1988; 274: 1171-73 13 Schilling R, Hughes J, DisgwaD-Fordyea I, Cilaon J. An epidemiological study of byxsiaosis among Lanca shire ooRon workers Br J ladustr Med 1955; 12:217-27 14 Merchant J, Halpria C, Hudson A, Kilbum K. McKen zie W, Bermanzolin P, Hurst D, Hamilton J, Getmiao V. Evaluation before sad after exposure--the pattern of physiological response to cotton dust Ana NY Arad Set 1974; 221:38-43 15 Miller A, Thornton J. The interpretation of ipirometric measurements in epidemiologic surveys: Standards and Incoosistenciei. Presented at the Annual Meeting; ACC? Subcommittee on Criteria for Occupational Lung Dis tant, Washington, D.C, 1978 18 Lapp NL, Amandus H, Hall R, Morgan WKC Lung volusas and flow ratas in black and whita subjects. Thorax 1974; 29.-18548 17 Schoenberg J, Beck C, Bouhuys A. Growth of decay of pulmonary function in healthy Masks and whites. Respir Physiol 1978; 3:387-93 18 Raoitar C, Weill H. Ethnic difference* in hing funo* tien: Evidence for proportional differences. Int J Epi demiol 1974; 3:5541 19 Lines* RR, Keller MD, Foley MF. Underwood EH.' Differences in pulmonary function tests among whites, blacks, and American Indians io a textile company, j Oecup Med 1978; 20:3944 20 Knudson R. Statin R, Lebowitz M, Burrows B. Tho marimal expiratory flow-volume curve normal rtandardi of variability and effects of age. Am Rev Respir Dis 1976; 113:587400 21 Tiger I, Spcizer F, Rower B, Prang G. A comparison between the three largest and throe kit of five farced expiratory maneuvers in a population study. Am Rev Respir Dis 1976; 114:120143 22 Epidemiology Standardization Project (Ferris BC, Principal Investigator). Am Rev Remit Dis 1978; 118: Part 2 23 Hanlrinson J, Petersen M. Data analysis for spirometty' instrumentation rtasdarda (abstiact). Am Rev Respir Dis 1977; 115:Supplcment, 118 24 Morris J, Kositi A, Johnson L. Spirometric standards for healthy non-smoking adults. Am Rev Respir Dis 1971; 103:57-87 25 Morris J, Tcmpk W, Kosld A. Normal values for the ratio of one-second forced expiratory volume to forced vital capacity. Am Rev Respir Dis 1973; 108:100043 If Kory R, Callahan R, Boren H, Syner J. The Veterans Administration--Army cooperative study of pulmonary function. L Clinical rpuomctiy in normal men. Am J Med 1981; 30:24348 27 Smith A. Caenskr E. Timing of forced expiratory vol ume in one second. Am Rev Respir Dis 1975; 112:88285 28 Peterses M, Amandus H, Reger R, Lapp N, Morgan CHEST, 79: 4, APRIL, 1981 SUPPLEMENT INTERNATIONAL CONFERENCE ON BYSSINOSIS 121$ : ST00Q5 556 DOW 06825 WIC Ventilatory capacity in normal coal miaoi pro-' dicrioa formulae for FEV, and FVC I Occnp Mad 1873; 15:099402 29 Sobol B. The aarlr detection of airways obstruction: another perspective. An J Mad 1970; 60:919-24 30 Millar A. Thornton J, Smith H, Morris J. Prevalence of "bnormal" spirometry ia a normal reference mala population--* tocoosidanboa of tbo 1971 Oregon wmy. Am J Indus* Med (ia pram) 31 Burro*! B, Knudaoa R, Cliaa M, Lebowitx M. Quan titative relationships between dgaratta aad ventilatory function. Am Rav Raspir Dia 1977; 115: 195-2b5 32 Flatchar C ?ato R, Tinker C Spaiaar F. Tha natural history of chronic bronchibs aad emphysema Odord; Oafotd University Press: 1976, 129 Worker Monitoring in Byssinosis* Harold 1L Imbue, MJ>, SeJD.f n late 1970, we began our program of medical sur I veillance of workers exposed to raw cotton dust This program followed initial research efforts with Duke University and the North Carolina State Department of Health, which gave insights into the cause and in cidence of byssinosis. Accordingly, see embarked on a program that included worker monitoring and environ mental control We describe the following three aspects of worker monitoring: procedures, some of the results achieved, and some of the problems involved. PnocEstnos The procedures used ia our program are testing, initial follow-up, referral, and notifiesbon of result*. Testing Our testing program, begun ia 1970, consists of a forced expiregram, namely, the one-second forced expiratory vol ume (FEVj), the forced vital capacity (FVC), and the British Research Council (MRC) Respiratory Questionnaire Modified for Byssinosis. At first, this was a team effort con sisting of aunts aad myself fram die corporate head quarters ia additioa to the plant aurse who had beea trained ia spirometry and administrabon of the question naire. Later the program was carried out on a yearly basis by the plant nurse, with results monitored by the corporate headquarters. Pulmonary function testing is done on a waterless spirom eter on the first day of the work week, before the worker goes to his assigned place of work. The worker returns between 49 to 0 hours into the work shift, and the spirometry is repeated. Pulmonary function testing before and after the worker has been exposed to cotton dust is compared to determine if there has been a significant dacrease. Currently, the Occupational Safety and Haalth Administrabon (OSHA) defines a 5 to 10 percent decrease as significant The graphs are read by the plant muse and recorded on a form that also contains the remits of the questionnaire. Theta results are then processed by the computer, which records each employee's FEV,, FVC percentage of predicted, and any decrement of FEV, during the work shift Also; the computer records the frequency with which the worker complains of Monday tightness under the Schilling Classifiesbon System. Under Schilling's system, the frequency of these complaints places * From Burlington Industries, Inc Creensboro, North Caro lina. tDirector of Health and Safety. 122S AMERICAN COLLEGE OF CHEST PHYSICIANS a worker in a byssinosis grade of I, 1, or 2 based on symptoms only. Initial Foflow-up If the nurse suspects unsatisfactory results owing to technical problems, such as unsatisfactory effort, or any other reason for obtaining a poor pulmonary function test result, she will repeat the test at another data before send ing the results ia on a computer form. Once the rails an obtained for each individual, than is a follow-up for cer tain categorias of workers. This follow-up is made by either a physician or a nurse who has had special training and wide experience ia the program aad who comes from the corporate headquarters. The following categories of workers are interviewed: (1) those with an FEV, of las than 00 percent of predicted units they have been previously interviewed aad an ia a stable condition; (2) those with a shift decrement of their FEV, of S percent or aton; (3) those who have bysrinosis symptoms whether or not then is any impairment; (4) those whose predicted FEY, has decnased by non than S percent baoe the baseline test the first test performed. During this interview, eddiboaal questions an asked about symptoms, other diseases, smoking habits, and med ical trcaonent Another pulmonary function test may bo administered on the spot if then is any quesbon about1 the technical quality of die test results that indicated problem. Maoy of tbo workers wbo an interviewed nri assigned a surveillance schedule requiring them to undOS go retesbng oooe every six months. Those who have beS interviewed by the corporate staff have tha results cfa fiadinp discussed with them, including abnormally leyi pulmonary funebon or reaction to cotton dust, either ia tfen form of a decrement of pulmonary function or the Monday-] tightness. If the employee appears to be a reactor, be be advised to wear a respintor or to transfer and will be placed under spedal rurveiDanee. Smoking habits are also discussed, and the worker is advised that smoking can contribute to the problem. Asftnal We refer to independent specialists, at company expense and on company tune, those wbo have an FEV, below 60 percent of predicted, wbo have as accelerated deteriora tion in pulmonary function, or who have other problems or complaints not readily explained. This evaluation is done by physicians who have spedal expertise in chest mcdicioe. This evaluation must include as a minimum a complete history, physical examination, ECG. SMA 12 or SMA 23, urinalysis, posteroaatcrior aad lateral chest x-ray film, spirometry, usually arterial oxygen and carbon dioxide ten dons, and any other study that the specialist believes to be important. Since it has been shown that emphysema is not re lated to cotton dust exposure, the specialist in tha future wiD CHEST. 79: 4. APRIL, 1981 SUPPLEMENT DOU 06826 Ronnai 1 Morgagni JB. The Mt tod cusses of diseases investigated by anatomy. 1788 Translated by B. Akzandcr. London: Millar. 1769; 1:381 Hafaer Publishing. 4 Van Coetsem. De la pocumonk produitt par la poutsiiie do coton. Aimaks da k Medicine Balga at Ebmngtre (Brussels) 1838; 3:3 3 Creenhow H. Tba pathology of flax dressers long. Trans action! of tha Pathological Society of London. 1869; 20: 48-59 4 Schilling 1C Ubar dk Schadlichea ElawiHtnnftn das Baumwellstaubes anf dk Abnungsorgsna. Dent Arch f Klin Mad. ,1925, 146-183. Abstracted in Public Health Bulletin, Washington No. 297, 1947 tf | 5 1 *"-<< HRM. The relation of organic dust to pneumo coniosis. J lod Hygiene 1923; 7:1 6 Shaw Dura J, Sheehan HL. Report on tea autopsies an cotton-mill workers. Appendix 111, Report of tbe Depart mental Committee oa Dust in Card Rooms in the Cotton Industry. London: HMSO 1932, 68-70 7 Ruttner JR, Spycber MA, Engekr ML. Pulmonary fibroais induced by cotton fibre inhalation. Pathol at Microbiol 1968;32:1-14 8 Edwards C, Macartney J, Rook# C. Ward F. Tbe pathology of the lung in bytsiaotica Thorax 1975; 30:812-823 9 Takiaawm T, Thurlbeck WM. Muscle and mucous gland ske in tbe major bronchi of patients with chronic bronch itis, asthma and asthmatic bronchitis. Am Rev Rcspir Dk 1971; 104:331-36 Factors Influencing the Interpretation as volume-time or flow-volume plots of tbe maximal forced expiratory vital capacity maneuver. The measurement! in of FEV| Declines Across the Working Shift* / cluded the 7-scc forced vital capacity (FVC). tbe forcod expiratory volume in J end 1 second (FEV, and FEV,), and f- forced expired flow between 25 and 75 percent of the forced vital capacity (FEF25-73S). The beginning of Henry Gtindmeyer Iff, DSngj John Diem, fU),Janet Hufher. PhJJj Robert N. Jonee, MJ>, F.C.C.P,- end Bene WeiU, M-D,, F.C.CJ. time for calculating tbe FEV, and FEV, was obtained by tbe method of backward extrapolation, and all measuremants were corrected to tbe body temperature and premia Blu nted with water taper (BTPS). Subjects were tested in Guidelines for the evaluation and management of workers exposed to cotton dust have been defined tbe standing position, nose dips seen used, and at least four masimal espintory maneuvers wen performed at aach testing session. by the Department of Labor in die recently adopted A Mined interviewer administered a British MSC quae- \ cotton dust standard. The succcess of this standard is Jtionnave modified for detection of bytstnosis. Responses to dependent on the precision with which an acute or die questionnain wen analysed for tbe prevalence of bye- chronic change in lung function can be detected at the sinosis by tbe Schilling definition and for chronic bronchitis, workplace. Although strict requirements have been atopy, dyspnea, and groups of upper and lower respiratory adopted for test instruments and procedures, the in herent variability of measurements of lung function still pose a significant problem in distinguishing re actors from nonreactors. According to the standard, possible acute reactors are defined as those exhibiting a fall in Monday FEV, in excess of 5 percent or 200 ml, whichever is less. Since for some persons the withinsubject SD for FEV, is in excess of 200 ml, individual beet symptoms. An analysis of variability within nibjects for FEV, was performed using tbe four Urges! preshift and postshift vol umes obtained on 192 subject! tested at tbe four miHf in 1977. For each individual, withiiwuhject SD was computed as the square root of tbe average preshift end postshift FEV, variances. This w-ithin-subject SD was analyzed in relation to symptoms, smoke status, and work shift. The ' SD of tbe preihift-postshift FEV, difference for each in variability can cause the cutoff for reacton to be ex dividual was computed by multiplying tbe within-subject ceeded, yielding both false-positive and false-negative SD by tbe square mot of (1/N, + 1/N,), where N, and results. To quantify this effect, the individual and group variability of data collected at four cottonseed crushing mills was examined relative to symptoms, smoldng status, and work shift N. represent the number of preshift and postshift msarur*mestx. Acute changes in lung function in relation to shift were analyzed for all mills and far all visits. Only subjects with acceptable over-the-shift spirometry values were included Matehals akb Mmoos in analyses. Acceptable spirometry required visually ac ceptable curves, wish tbe added stipulation that the two Poor rnltnnind cradling mills in the southern United States were visited in 1975, 1977, and 1978. Lung function measurements were performed before work and after tbe largest FVC values differed by less than 3 percent. Ac ceptable over-tbe-sbift spirometry was obtained at least once for 185 lister area workers at three vfsite. subjects had been away horn the miO for at least 38 hoots and after at kast five hours at the workplace. Spiromebic RKSOZ.TS tests were conducted on a dry, rolling-seal spirometer with the output displayed on a Hewlett-Packard XYY recorder The individual withia-subfect SD for the 192 subjects studied in 1977 ranged from .020 to .428 L, with 74 From tbe Tulane University School of Medicine, New Orleans. This work was supported by U.S. Public Health .Service grant HL-15092 of tba National Heart, Lung, and Blood Iostitute. Reprint requeue: Dr. CUndmeyer, Depertment of Pulmo nary Diseases, 1700 fordido Street, New Orleane 70112 percent less than .100 L. The avenge within-subject variability was .102 L. SD across the shift ringed from .014 to .302 L, with .072 L as a representative value. Figure 1 displays the average FEV, within-subject variability between groups answering 'yes'* or `no* to CHEST. 79: 4, APRIL, 1381 SUPPLEMENT INTERNATIONAL CONFERENCE ON BTSSINOSIS 71S I UU'J 'j ' $ jr j u' DOW 06827 2<h 0< JO- ><0 14 HOOI 9 LRS URS CB DYSP SYMPTOMS .15 > .10 BYSS Fteoan 1. Aei| FEV, withia-eubfect variability btwaaa group* iniwsring "yes" or *ao* to .lower respiratory bact symptom* (LBS), upper respiratory tract symptoms (UBS), chronic bronchiti* (CB), dyspnea (DYSP), tad byariaoris (BYSS). Neabar of subjacm ioclndad ia aach category si* listed at top of tacb Hr EV .05 Ficoas 2. Average FEV, witbia-nbjaet variability displayed for current (CURB), cxsrnoken (EX), aad aevor smokers (XEV), aad for workers oa lb* morning (MORN), evening (EV), and night (NT) shifts. Number of subjects included ia each category are listed at top of each bar. ^ " Smoker Shift lower or upper respiratory tract symptom complexes, chronic bronchitis, dyspnea, and byssiansi*. The vari ability of groups answering "yes" to brtmrhiHt and byssmosis (.141 L and .192 L, respectively) was sig nificantly different (P = .001) from the nonbronchitic subjects and the subjects without symptoms of bys. sinosis (.099 L and .102 L, respectively). No statistical difference in variability was exhibited between groups answering "yes" or "no" to lower or upper respiratory tract symptom cosnplrrea or dyspnea. Figure 2 displays the average within-subject vadability fe FEV, as a function of smoking status and drift Far smoldng, then was no significant difference in vsriability between current, exsmokers, and nevarsmoleers. In addition, the morning, evening, and night-' shifts yielded statistically similar repeatability. Changes over the working shift were related to time of day for all three visits and for all four mills, even though preshift measurements were not significantly _ related to shift For FVC, FEV,, and FEF2S-75S, there was no significant difference between morning and-- night shifts. However, there was always a siprificant difference for all of these parameters between evening shift and the morning and night shifts. Table 1 shows the mean change for all visits collapsed for all mill. The significant difference in FEV,, FVC and FEF2575* between shifts persisted even though there was no significant difference between shifts in smoking ** " symptoms, exposure (all were linter area workers), race, age, yeais employed in mill, and wade arm. The ability with which one can detect a meaningful change in FEV, Is related to drift and is an inversa Table 1--Afaan Change In 5*im<ria Faina* /ar Ml ritiu la Ml Mills Shift Morning Evening Night Mean Time of Prmbift Spirometry 74)0am 2:30 n 11:00 m Mean Change Over All Yiaiu (Linter Area Workers) N FVC FEV, FEFM-75-c 72 -M2 -4131 84 -.on -.135 --.14S -.463 40 -.031 -.030 -.0!>S 72S AMERICAN COLLEGE OF CHEST PHYSICIANS CHEST, 79: 4, APRIL, 1981 SUPPLEMENT t: * DOW 06828 function of within-rubjeet vuriabiBty. This variability it related to symptoms of bysrinosis and chronic bron chitis. The current practice of comparing individual decline in FEV, across the shift to a fixed value. eg, > 200 ml, in order to classify reactors, is misleading, . since the across-shift SD of some subjects is in excess of 250 mL Conversely, very repeatable subjects (SD < 50 ml) should be classified as reactors if their acrossshift decline k in excess of 100 mL In addition, the excessive variability of subjects with symptoms of bys sinosis is a possible explanation of the lack of correlation between acute decline and symptoms. The drift effect implies that an adverse change in FEV, observed across the day shift or night shift would be significantly lower thansn equally meaningful change observed in the evening. Therefore, a more accurate classification of reactors should taka into account symptoms, shift; and an assessment of individual variability. SusotAxr An analysis of variability within individuals was conducted for FEV,, obtained hom subjects employed in the cottonseed industry. Individual SDs ranged from .020 to .428 1. The within-subjeet SD of the preshiftpostshift difference ranged from .014 to .302 I* When within-subject variability of FEV, was analyzed in rela tion to symptoms, smoking history, and shift, the only statistical difference occurred in the 14 bronchitic sub jects (.141 L) relative to the 179 without bronchitis (.099 L), and the three subjects with bytsinosis (.192 L) relative to those without symptoms (.102 L). Changes over the working shift were significantly dif ferent for evening relative to morning and night shifts, even though there was no significant difference in smok ing status, symptoms, exposure, race; age, yean em ployed in the mill, and work area. In addition, baseline measurements were not significantly related to shift Therefore, the effect df individual variability, symp toms, and shift should be considered if an accurate classification of reacton based on change in FEV, across the shift is to be obtained ACKNOWLEDGMENT: The authors would Uka to aeknowledge the programming and technical assistance of Daniel Seale, B-i, in the preparation of this manusnipt. The Effect of Mediator Modifying Drugs in Cotton Bract-Induced Bronchospasm* . Neil Schechttr, M.D^ F.C.C-F,-t Steven Brown, BS^ Eugenia Zuikm, M-D,- Gerald /. Beck, FAD-.-? Marion Buck, FAD,- Barbara Kolock, B-S,- end Arend Bouhvys, 1U>, FAD.fi Byssinosis is a chronic occupational lung disease as sociated with die inhalation of cotton and other textile dusts.1-1 In its early phases acute reversible symptoms, such as wheezing chest tightness, and shortness of breath, accompany reversible changes in pulmonary function.1 The therapy for this disease re mains incompletely defined However, considerable evi dence from both in oitro and in eioo studies indicates that the mechanism underlying acute airway obstiuction in this df*---- is related to the nooaatigeaic re lease of histamine and possibly of other mediators by the action of an airway constricting agent hi cotton bracts.*-* To test the therapeutic implications of these findings, we studied the effect of mediator modifying drugs on cotton bract-induced bronchospasm in healthy subjects. Two classes of agents were studied a oomone, dimdium cromoglycate, end two antihistamines, an H,blocking agent, chlorpheniramine, and rimetidine, an H,-blocking agent From do Department of Internal Medicine, Tale Univer sity School of Medicine, New Haven. tAssociata Professor of Internal Medicine. :Assistant Professor of Biostatistics. {Deceased June 15. 1979 (Professor of Internal Medicine). Reprint teoueati: Dr. Schachter, Tele Unwenity School of Mediant, 333 Cedar Street, New Haotn 06S11 Decreases In lung function owing to cotton dust can be reproduced in the laboratory in healthy subjects, never before exposed to textile dust, after they inhale an aerosolized aqueous extract of cotton bract Most healthy subjects will show some decrease in pulmonary' function when exposed to these extracts, but previous studies have shown that the degree of induced bronchospasm to a standardized dose will vary with the in dividual* These differences permit investigators to classify healthy subjects as responders and nonre sponders. The ability of mediator modifying agents to protect against cotton bract extract-induced broncho spasm was studied in healthy responders. Subjects and Methods A total of 21 haahhy subjects whose ages ranged from 18 to 34 years were selected from a larger group o 31 subjects on the basis of their increased reactivity to cotton bract extract. Nineteen subjects participated In our study with cromolyn sodium, and tea partiripated in our study with antihistamines Sh individuals bom the group nnobad, and 15 wen then noamokets. None of the subjects related a history of asthma or other atopic dlmeees. Pulmonary function data showed normal lung function for the group as a whole (Table 1). The study consisted of three protocol days for dm disodium cromoglycate study and four protocol days for dm antihistamine study. On the first day of each study, the subject's response to cotton bract extract waa determined. Baseline pulmonary function was obtained Immediately before dm challenge with cotton bract and subsequently was measured 30, 60, 90, 120, and 150 minutes after the exposure. This permitted us to group our subject* into respoodeft and nonretponders. The protocol was followed exactly oa each subsequent study day, with the exception that the drug or placebo was administered before the chal- CHEST. 79: 4, APRIL, 1981 SUPPLEMENT INTERNATIONAL CONFERENCE ON BYSSINOSIS 73S cn <-n cn o G3 DOW 06829