Document 9NO4YnXeDbaMZKkoKZ0b06dR

i 0. I Crackles In the Early Detection of Asbestosis12 MM, i V, * sizer vim rAYMOND L. H. MURPHY, JR, EDWARD A. GAENSLER, STEPHEN K. HOLFORD, one ELIZABETH A. DEL BONO, and GARY EPLER Am ation Pi 6. )ecovior. Lon- StCTTJ uan. :am- >sioQ tical Jrnia B, ernal a. N Risk eef- su~ enor yses. Introduction fylinimal interstitial pulmonary fibro sis caused by the inhalation of asbestos can be difficult to detect (1, 2). Sympto matology is subjective and pulmonary function studies are often nonspecific. Roentgenographic manifestations of garly interstitial disease are often diffi cult to distinguish from normal sha dows (1-3), and observer variability in interpretation of radiographs is also a problem. Discontinuous adventitious lung sounds (crackles, rales) have been recognized as prominent features of pulmonary asbestosis (4-6) and are thought to be an early finding (7-9). Although chest auscultation is regarded as highly subjective, recently developed technology has brought the promise of objective quantification of lung sounds. One such method, called time-expanded waveform (TEW) analysis, consists of tape recordings of sounds for analysis by a computer. Amplitude is then plotted versus time, with a time axis of 800 mm/s or more, allowing visualiza tion of minute details of pulmonary sounds (10). Because TEW analysis provides documentation of the pres ence of crackles, it can be used to train technicians in auscultation (11). Moni toring of exposed workers by ausculta tion is attractive because it is noninvasive and inexpensive. The purpose of this investigation was to quantify the relationship between the prevalence of crackles assessed by a technician, whose performance had been objective ly validated, to other criteria for asbes tosis in exposed workers. Methods The 386 workers included in this study were exposed to asbestos during the manufacture of paper and insulation materials and in a shipyard. Lung sounds were recorded at 4 Preselected basilar sites during breathing with the mouth open while the worker was a sitting position. Workers were asked to breathe slightly more deeply than the usual tidal breathing as in standard clinical prac tice- These sites were the right and left bases tn the midscapsular line, the right midda- SUMMARY We studied 386 workers exposed to asbestos to assess the value of chest ausculta tion by a trained technician In detecting asbestosis as defined by previously reported clinical, physiologic, and roentgenologic criteria. The presence and degree of crackles were assessed at preselected basilar lung sites by a technician whose performance was validated by comparison with computer-generated time-expanded waveforms of tape recordings of lung sounds. Asbestosis was present in only 2.8% of the total population, but it was present In 8.6% of those with over 25 yr or more of employment. The technician correctiy identified ail the workers In whom the diagnosis was most certain, that Is, those with all criteria positive. The overall true positive rate was 55%. The majority (94.8%) of those with no abnormal criteria were correctly classified. Auscultation by an objectively validated technician can be a useful noninvasfve method for screening industrial populations exposed to asbestos. AM REV RESPtR 01S 19S4; 129:375-379 vicular line, and the left anterior axillary line. The number of crackles was estimated at each auscultation site using a scale rang ing from 0 (none) to 3.0 (indicating many), and a weighted crackle score (WCS) was then calculated. Scores from the right and left bases posteriorly were doubled and added to the scores from the anterior sites. The result ing WCS ranged from 0 to 18. Lung sounds were recorded at the time of auscultation, and selected tape recordings were analyzed for frequency content and TEW analysis with and without 800-Hz filtered sound. These were compared with the independent observations of a trained technician. For this assessment, worker identification was masked by use of code numbers. Comparison of the WCS was made by 2 observers who interpreted the TEW ana lyses while simultaneously listening to the sounds. The technician's results were also compared with the pulmonary physician's listening during the surveys. The Kappa statistic was calculated according to the method of Fleiss (12). Kappa is a numerical method of separating agreement from chance association, and it varies from 0 (no agreement) to 1.0 (complete agreement). For interpretation of the intermediate values, we employed the criteria of Landis and Koch (13): a Kappa value of 0.81 to 1.0 indicates almost perfect agreement, 0.61 to 0.80 in dicates substantial agreement, 0.41 to 0.60 indicates moderate agreement, and 0.4 and below indicates slight to poor agreement. Other data were collected using a ques tionnaire on respiratory history and occu pational exposure, a physical examination including independent observations on lung sounds by 2 or more chest physicians, chest roentgenographic readings according to the ILO scheme, and detailed pulmonary func tion testing (14). Asbestosis in these exposed workers was considered to be present when 3 or 4 of the following criteria were present: (/) radiograph showing irregular opacifica tion of a profusion of 1/2 or more, (2) single-breath diffusing capacity less than 807e predicted, (3) vital capacity less than 8097o predicted, and (4) crackles at 2 or more basilar sites. Results Asbestosis by our criteria was present in only 2.397b of the workers and did not occur in workers with less than 11 yr of exposure. In the 93 workers with over 25 yr of employment, were considered to have asbestosis. The pre valence of abnormal findings in each of these criteria is shown in table 1. In general, the prevalence of these criteria increased both with exposure and with age, as is expected in populations wherein age and duration of exposure are closely correlated (figure 1). Pleural abnormalities also showed a similar re lationship to duration of exposure and age (figure 2). The relationship between the tech- (Received in original form May 26, 1982 and in revisedform October 18, 1983)* * ' From the Tufts University School of Medi cine, the Boston University School of Medicine; and the Pulmonary Service, Faulkner Hospital, Boston, Massachusetts. * Requests for reprints should be addressed to Raymond L. H. Murphy, Jr., M.D.. Director, Pul monary Service, Faulkner Hospital, 1153 Centre Street, Boston MA 02130. 375 376 MURPHY, GAENSlER. HOLFORO. DEL BONO. AND TABLE 1 SELECTED RESULTS IN 386 ASBESTOS-EXPOSED WORKERS Number Percent VC < 80% pied Dl < 80% pred Radiographic UICC 1/2 or more 1/1 or more Bilateral rales on routine auscultation by MD Weighted Crackle Score 3 or more 2 or more Asbestosis Criteria Positive: 0 1 2 3 4 47 61 14 29 24 31 51 289 63 23 7 4 12.2 15.8 3.6 7.5 6.2 8.0 13.2 74.9 16.3 6.0 1.8 1.0 Oetimtion ol abbreviations: VC s vital capacity; Ol = diffusing capacity; MO s physician. nician's auscultatory findings and the number of asbestos criteria found in individual workers is presented in table 2. The technician correctly identified all the workers in whom the diagnosis was most certain, that is, those with 4 of 4 positive criteria. The majority of those (94.8^0) with no abnormal criteria were also correctly identified. If the crackles heard by physicians were omitted from the assessment to avoid the possibility of circular reasoning, the results were similar. The technician identified 807o of those with all 3 of the remaining cri teria. The observations of the tech nician while recording at the industrial site correlated well with interpretations of waveforms (table 3). The epidemio logic diagnosis of asbestosis related poorly to symptomatic and roentgeno logic evidence of obstructive disease (table 4). The weighted crackle score (WCS) was associated with pack-years of cigarette smoking (table 5) but re lated poorly to the diagnosis of bron chitis with obstruction, as defined by the combination of a history of pro ductive cough plus a coexistent forced expiratory volume in one second of 707o. The interaction of smoking his tory, years of exposure, and crackle index was statistically evaluated using logistics regression, using both a trichotomous and dichotomous response variable (15, 16). Using the trichotomous response variable, age appeared to be the most significant effect, and the smoking-asbestos interaction was nonsignificant. The effect of asbestos exposure was present only when com paring a WCS of 3 + with a WCS of 0, PLEURAL ABNORMALITIES - EXPOSURE AGE " Fig. 2. Pleural thickening and/or caicificati shows an increase in prevalence with incrAaSin exposure and age. as expected in a Poooiatle^ with exposure to asbestos. 0(1 suggesting that there was no asbestosrelated differences between a WCS or 1 or 2 and a WCS of 0. The effect of smoking was more difficult to inter, pret. The smoking effect was present only when comparing a WCS of 1 or i with a WCS of 0. The results suggest the WCS of 1 or 2 may be the result of smoking, but a WCS of 3 + is associ ated with asbestosis. A dichotomous model was then examined. W'hen comparing a WCS of 0, 1, or 2 with a WCS of 3 +, an as bestos effect was seen (p = 0.06) but no smoking effect was found. Age was shown to be highly significant (p = 0.0099) when there is no smoking, asbestos interaction (p = .85). In comparing a WCS of 0 versus 1 + , a smoking effect is seen (p = 0.024), no asbestos effect is found (p = 0.85), the age effect is again highly significant (p =* 0.00026), and no smoking-asbestos interaction is found (p = 0.65). The results of the dichotomous model con firm those of the trichotomous model RALES (M.o.) % POSITIVE *r u si <e ii si n m m n OL %PRED.4 80 ] _/n'A'/ : - `ASBESTOSIS* /" y__ V.C.S.*3 .: EXPOSURE AGE Fig. 1. Relationship of the weighted crackle score and of abnormalities consistent with pulmonary asbesto sis and our diagnosis of asbestosis to duration of exposure and age in 386 asbestos-exposed workers. In general, an increasing prevalence is seen with increasing exposure for each of the criteria and for asbesto sis. The weighted crackle score (WCS 3+) also shows a simitar ex posure and age relationship. For definition of abbreviations, see table 1. Oiscussion Prior to discussing the possibility of our false positives reflecting slight forms of asbestosis we will briefly review the history of crackles in asbestosis. Fifty years ago Wood and Gloyne (17) pointed out that crackles were a common feature of pulmonary asbestosis. Since that time, reports have varied as to the importance and speci ficity of auscultation findings. Hunter (18) said crackles occurred "sometimes" and Wyen (19) said they were "gen erally" present. Wyers believed that the adventitious sounds tend to be "evanes cent," and that the dry crackling sounds could disappear altogether. Smither (9) believed that the crackles of asbestosis were characteristic in their sound and distribution, present first at the bases in the midaxillary lines, and tending to spread to the posterior bases. As the disease advanced, the ' pl* I IN eARLY oetection of asbestosis Nation basing Nation astos. >or l ct of interesent or 2 Sgest lit of soci- thea -S of i asJt no was P* :ing- .uO , the it (p istos The :on* del. of :ght efly in and kies ary ave ed* iter es' en e es* n er. of etf. or he \ ^ckles were distributed upwards from [he base at ^e scaPuIar level. The belief that crackles are an early jlgn is also consistent with many pub lished reports. Crackles have been re garded as a feature of pulmonary asbestosis for over 50 yr, and even in ^ly reports were considered to be resent with minimal disease (17). finical features of their nature and distribution on the chest were described in some detail by Smither (9). Mitchell jnd coworkers (20) found crackles more closely related to duration of exposure to heat-resistant and friction composites than was vital capacity, and they recommended that chest ausculta tion be included as a biologic monitor 0f the work environment. Reported rates vary, but about half of the per sons considered to have asbestosis on clinical grounds are reported to have crackles (21-23). These represent selected cases. In order to understand the strength of an association between an exposure and a health effect, it is im portant to know the status of all per sons exposed. If only patients are studied, the effects may appear more striking than if asymptomatic persons are also included. Population-based studies, therefore, allow more precise examination of the frequency of crackles in asbestos-exposed persons. These show prevalences ranging from about 10 to 20%. Such prevalences depend on a variety of factors including the method of auscultation, the severity and duration of exposure, the age of population, the prevalence of the diseases causing crackles, etc Neverthe less, the association between crackles and asbestos exposure is clearly estab lished by: (/) high frequency in diag nosed cases, (2) common occurrence in exposed populations, and (3) increased frequency with increased duration of exposure (24, 25). Furthermore, theprevalence in control populations is low (24-27). The correct detection of those in whom the epidemiologic diagnosis of asbestosis was most certain suggests that a trained technician can screen a group of exposed workers to estimate die prevalence of asbestosis. Despite the high true positive rate there are errors in the method. In our study, 5 of the 7 workers with 3 of 4 positive cri teria were missed (false negative). Al though this may represent technician error, it is known that not all those with asbestosis have crackles. Indeed, Epler td coworkers (7) reported crackles in TABLE 2 RELATIONSHIP OF AUSCULTATION BY A TECHNICIAN TO NUMBER OF ASBESTOS CRITERIA Number of Positive Criteria per Worker of 4 Possible Asbestosis Criteria Weighted Crackle Score 0 1 2 3+ Total 0 1 2 3 4 Total 225 34 15 15 289 50 5 3 5 63 13 3 2 5 23 3202 7 0004 4 291 44 20 31 386 377 TABLE 3 TECHNICIAN VALIDATION: COMPARISON OF TECHNICIAN'S WEIGHTED CRACKLE SCORE MADE AT THE INDUSTRIAL SITE TO WAVEFORM INTERPRETATIONS OF THE SAME LUNG SOUNDS* Waveform Interpretations Weighted Crackle ScoreT 3 or More Less than 3 Total Positive for crackles Negative for crackles Total 6 0 6 2 14 16 ` 8 14 22 * Kappa = 0.80; 20/22 = 90.9%. t On-site tech observations. TABLE 4 RELATIONSHIP OF ROENTGENOGRAPHtC EVIOENCE OF EMPHYSEMA AND SYMPTOMS OF CHRONIC BRONCHITIS TO ASBESTOSIS AND WEIGHTED CRACKLE SCORE Asbestosis* Weighted Crackle Score Yes No Total Yes NO Total Emphysemat Yes 0 5 5 1 4 5 No 11 343 354 28 326 354 Total 11 348 359 29 330 359 Chronic bronchitis* Yes 3 56 59 8 51 59 No 8 319 327 23 304 327 Total 11 375 386 31 355 386 ' Asbestosis defined as 3 or 4 criteria positive, t Emphysema defined by roentgenographic criteria. * Chronic bronchitis defined by American Thoracic Society criteria. TABLE 5 RELATIONSHIP OF SMOKING HISTORY TO WEIGHTEO CRACKLE SCORE* Smoking History (pack-years) Weighted Crackle Score 0 1,2 3 + Total 0-20 21-40 41-60 61 + Total 169 (153.4)t 52 (57.54) 32 (33.93) 17 (25.08) 270 27 (38.64) 20 (14.49) 9 (8-55) 12 (6.32) 68 12 05.91) 6 (5.97) 5 (3.52) 5 (2.60) 26 208 78 46 34 366 *X> * 19.35. p< 0.005. t Numbers in parentheses refer to the expected value for that cat). 378 MURPHY. GAEN$!.R. HOLFORD. DEL BONO. AND gplEj| #only 60% in their pathologically proved cases of usual interstitial pneumonia. The magnitude of the false negative rate for each asbestos criteria was assessed by comparing the number of positives by the test with those workers positive by each of the remaining tests. The diffusing capacity was abnormal in 4 of the 5 workers positive for each of the remaining tests. Corresponding rates were 4/7 for vital capacity, 4/6 for radiograph, 4/5 for physician-crackles, and 4/5 for technician crackles. The false negative rate of crackles for the identification of asbestosis was, there fore, similar to the false negative rate for each of the other criteria, when con sidered separately. In any case, a false negative rate would have to be accepted for auscultation, as is true for any of the other available screening methods. Some of the false negative results may have been due to the method we employed. Mitchell and coworkers (8) pointed out that inhalation to total lung capacity can abolish or decrease the number of fine rales heard in a sub sequent inspiration. Conversely, Shirai and associates (26) have shown very high prevalences by breathholding volumes prior to deep breathing. Pre sumably this maneuver accentuates collapse of dependent lung regions, which reexpand on the subsequent in spiration. This is consistent with the theory that the opening of collapsed airways is the mechanism of crackle generation (28). We have also found that crackles are more common after this maneuver, particularly if the initial inspiration is performed slowly. Un fortunately, observer variability and artifacts appear to be somewhat greater, and more careful investigations will be required before this becomes an accept able method. The "false positives," that is, those with a WCS of 3 + but otherwise not meeting our criteria for asbestosis, are of particular interest. Possible explana tions for this include: (/) technician error, (2) crackles caused by conditions unrelated to exposure, or (3) crackles indicating lesseT degrees of disease than do the other methods employed. Technician Error Such error is likely small in circum stances such as this study wherein the echnician is highly trained and focuses jn a specific task. We evaluated the accuracy of the technician in assessing the presence, degree, and quality of crackles, using objective methods of Pig. 3. A single crackle from time-expanded wave form analysis. On the basis of 2 waveform features, Holford (30) devised a classifier for separating fine from coarse crackles. The first measurement (the initial deflection width or IOW) is the time in milli seconds of the first deflection of an identifiable crackle above or below the baseline. The twocycle duration is the time in milliseconds for 2 S-shaped waves or cycles to occur. A discriminant curve on the classifier then separates fine from coarse crackles based upon both IDW and 2CO measurements tor each crackle. detecting crackles by tape recordings and time-expanded waveform analysis. Indeed, the association between the observations of the technician and the waveform interpretation was over 90%; the Kappa statistic likewise was high, indicating that this was due to agree ment rather than to mere association, as it was in a more extensive study of observer variation in chest auscultation (29). The accuracy of the technician in this study in classifying crackles as fine or coarse compared with the computer classifier devised by Holford (30) was 84.4% (Kappa, 0.68). Crackle measure ments made on TEW analyses are illustrated in figure 3. Other Causes of Crackles Numerous conditions other than in terstitial fibrosis, such as chronic bronchitis, cause crackles. Our tech nician was carefully trained by listening to tape recordings of crackles from patients with asbestosis to facilitate recognition of these sounds. Indeed, the prevalence of crackles did not cor relate significantly with the symptoms of chronic bronchitis, pulmonary func tion showing obstruction, or with roentgenographic evidence of emphy sema. There was, however, a significant correlation between WCS 3 + and pack-years of smoking. The degree to which crackles reflect the industrial exposure rather than the effects of cigarette smoking is difficult to de termine in these workers. Weiss (31) has suggested that it is important, but not all studies confirm this opinion (32). The study of Rossiter and Berry (21) is interesting in this regard. These inves tigators showed an interaction between smoking and asbestos exposure in the production of crepitations. In an asbes tos textile factory, those who did not smoke or smoked less than 5 cigarette* per day were less likely to develop cren? tations or be certified for asbestosj than those who smoked more, although the amount of asbestos exposure was similar. Clarification of this p0im awaits more thorough investigation 0f the relationship of smoking to intersti tial fibrosis and more careful delinea tion of the differentiating character! istics of crackles in bronchitis and asbestosis. As cigarette smoking and resulting chronic obstructive Inndiseases are common in industrial workers, this is an important considera tion. The effect of removing persons with a ratio of forced expiratory volume in one second to vital capacity of less than 70% on the age and dura tion of exposure plots is shown in fig. ure 1. These tend to strengthen the re lationships. The fine crackles of interstitial fi. brosis are similar in character to those caused by congestive heart failure, The latter diagnosis can usually be readily distinguished on clinical grounds. Iq this regard, it is important to mention that our criteria were designed for epidemioiogic screening and in no way were intended to obviate the need for careful clinical evaluation of suspected positives. Crackles as an Early Diagnostic Sign of Asbestosis Whether our false positives reflected lesser degrees of asbestosis than other tests is unknown. However, the average duration of exposure of the workers with high crackle scores and no other criteria was longer than that of the en tire group (19.6 and 13.8 yr, respec tively). Most investigators consider crackles to be an early finding (7, 26, 33). Not all investigators have concurred. In the study of 386 ship repair workers by Selikoff and associates (34), radiologic changes were considered to be a far more sensitive index than were crackles. Some of this variation in opinion is likely related to the criteria employed. We have discussed the rationale for our choice previously (10, 35, 36). Selikoff and associates apparently considered s small opacity of 0/1 or greater or an1 pleural change to be evidence of asbestosis. Indeed, in an asbestos worker in the course of developing asbestosis, it is likely that microscopic changes of alveolar thickening peribronchial fibrosis precede any J0* entgenologic abnormality. It is - EARLY DETECTION OF ASBESTOSIS 379 * "AS Point m of erstj. ineaicier. and and lung stria] ierasons ttory achy iurafig* e re. 1 fi. hose The tdily . In tion epi. cted iher age kers ;her enw- kies Mol the by jgic far les. i is ed. our off dt any j bi^\c that those who develop severe I >:?L0$is have roentgenographic pro ton from UICC Srade 0/0 t0 3/4 i* datively orderly fashion. Our re * anCe to regard the lesser UICC %es of small opacifications as prima He evidence of an effect of asbestos is Jvcd the reliability of such evi^ce. First, the early diagnosis of in- stitial Fibrosis has long been rerded as one of the most subjective in Sology (2, 3) (other causes for ro t* cn0iogicaIly similar shadows in?gde soft tissues, inadequate inspire * etc.). Second, it is likely that a Variety of other illnesses and exposures 'reduce similar findings (previous pul- nonary infection, other silicates, drugs, digestive heart failure, etc.). Third, in 'r studies, lesser grades of opacifica tions in asbestos workers could not be Languished reliably from shadows seen on roentgenograms of the control sub jects (10. 36). Decreasing the degree of disease required to classify an indiriduai as a positive case in epidemio logic studies frequently leads to an ufltase in sensitivity. Unfortunately, this also often decreases the specificity or accuracy of the observation (13). fle relative "earliness" of pulmonary function abnormalities, roentgenopaphic findings, and auscultation are ill subject to this phenomenon. The re bant issue is which test or set of tests provides reliable monitoring of workers to assure, as far as practicable, their afety. In this regard, a significant advantage of auscultation is its noninwsiveness, as worker acceptance must also be considered. Roentgenograms offered by employers are frequently refused by workers because of the fear of exposure to radiation or personal considerations. The results of these studies indicate that auscultation, by a properly trained ind objectively validated technician, can be a useful method for screening industrial populations with exposure to ubestos. Although true positives may be underestimated, the false negative w is low and the method is inexpen- With the use of modern recording ud analysis techniques, chest ausculta tion can be verified and permanent words can be made. Further studies of & sensitivity and specificity are war ded in view of the noninvasive natore of auscultation. Acknowledgment The writers wish to thank Drs. J. Giassroth, J. 0`Brien, M. L. Petusevsky, L. Sicilian, C. P. Wall, and D. Woodford who participated in the industrial surveys, and Drs. R. J. McCunney and B. Raff for their participa tion in observer variability studies. We also thank R. C. Bloom who performed the logis tics regression analysis, J. G. Berera for tech nical assistance, and L. M. Starr for secre tarial help in the manuscript preparation. References 1. Parker WR. Occupational lung disorders. London: Butterworth & Co., Ltd, 1974:305. 2. Mackenzie FAF. The radiological investiga tion of the early manifestations of exposure to asbestos dust. Proc R Soc Med 1971; 64:34. 3. Fletcher DE, Edge JR. The early radiological changes in pulmonary and pleural asbestosis. Clinical Radiol 1970; 21:355. 4. Heard BE, Williams R. The pathology of as bestosis with reference to lung function. Thorax 1961; 16:264. 5. Mitchell CA. Asbestos and health today. Mod Med Aust 1976; 14-6. 6. Wedler HW. Klinik der lungen asbestose, Arbeit und Oesundheit. In: Manineck G, ed, Sozialmedizinische Schriftenreihe aus dem Gebiete des Reichsaxbeitsministeriums. Vol 34. Leipzig: Georg Thieme, 1939. 7. Epler GR, Gaensler EA, Carrington CB. Crackles (rales) in the interstitial pulmonary diseases. Chest 197S; 73:333-9. 8. Mitchell CA, Chamey M, Schoenberg JB. Early lung disease in asbestos-product workers. Lung 1978; 154:261-72. 9. Smither WJ. Secular changes in asbestosis in an asbestos factory. Biological effects of asbesto sis. Ann NY Acad Sci 1965; 132:166-81. 10. Murphy RLH, Holford SK, Knowler WC. Visual lung sound characterization by timeexpanded wave-form analysis. N Engl J Med 1977; 296:968-71. 11. Workum P, Del Bono E, Shirai S, et at. Observer variability in breath sound interpreta tion. Presented at the Fourth International Con ference on Lung Sounds, Chicago, Illinois, Sept. 1980. 12. Fleiss 3L. Statistical methods for rates and proportions. New York: John Wiley and Sons, 1973. 13. Landis RJ, Koch CG. The measurement of observer agreement for categorical data. Bio metrics 1977; 33:159. 14. Epler GR, Saber FA. Gaensler EA. Determi nation of severe impairment (disability in in terstitial lung disease). Am Rev Respir Dis 1980; 121:647-59. 15. Pagano M, Trichler D. An algorithm for logistic regression with polychotomous response. Technical Report. Dept, of Biostatistics and Epi demiology. Sidney Faber Institute, Boston, MA. 16. Cox DR. Analysis of binary data. London: Chapman & Hall, 1970. 17. Wood WB, Cloyne SR. Pulmonary asbestos. Lancet 1930: 1:447. 18. Hunter D. The diseases of occupations. London: English University Press, 1955:877-8. 19. Wycrs H. Asbestosis. Postgrad Med J 1949; 25:631-8. 20. Mitchell C, Charney M, Schoenberg J. A survey of pulmonary disease in the friction materials industry. Am Rev Respir Dis 1974; 109: 728. 21. Rossiter CE, Berry G. The interaction of asbestos exposure and smoking on respiratory health. Bull Eur Physiopathol Respir 1978;J4: 197-204. 22. Huuskonen MS. Clinical features, mortality and survival of patients with asbestosis. Scand J Work Environ Health 1978; 4:265-74. 23. Segarra F, Baselga MM. Lopes IP, Perez NJ. Asbestosis in a Barcelona fibrocement factory. Environ Res 1980; 23:292-300. 24. Dreesen WC, Dailavale JM, Edwards TI, et al. A study of asbestos in the asbestos textile industry. Public Health Bulletin No. 241. Wash ington, D.C.: Government Priming Office, 1938. 25. Murphy RLH, Ferris BG, Burgess WA, Worcester J, Gaensler EA. Effects of low concen trations of asbestos. Clinical, environmental, radiologic and epidemiologic observations in shipyard pipe coverers and controls, N Engl J Med 1974; 285:1271-8 26. Shirai F, Kudoh S, Shubuya A, Sada K, Mikami R. Crackles in asbestos workers. Auscul tation and lung sound analysis. Br J Dis Chest 1981; 75:386-96. 27. Kleinfeld M, Massitte J, Kooyman O, et al. Effects of asbestos dust inhalation of lung func tion. Arch Environ Health 1966; 12:741. 28. Forgacs P. Applied cardiopulmonary physi ology. The functional basis of pulmonary sounds. Chest 1978; 73:399-405. 29. Davidson F, Murphy RLH. Gravity depen dence of crackles. In: Nair S, ed, Computers in critical care and pulmonary medicine. New York: Plenum Press, 1983:279-86. 30. Holford SK, Discontinuous adventitious lung sounds: measurement, classification and modeling. Sc.D. Thesis. Massachusetts Institute of Technology, 1982. 31. Weiss W. Cigarette smoking and diffuse pulmonary fibrosis. Am Rev Respir Dis 1969; 99: 67-72. 32. Chew PK, Chia M, Chew SF. Asbestos workers in Singapore Arch Environ Health 1973; 26:290-3. 33. Elmes PC. Health parameters other than mesothelioma. Proceedings of Asbestos Sym posium, Johannesburg, S. A., Oct 3-7, 1977. Randburg: National Institute for Metallurgy, 1978:9-16. 34. Selikoff IJ, Nicholson WJ, Lilis R. Radio logical evidence of asbestos disease among ship repair workers. Am J Ind Med 1980; 1:9-22. 35. Murphy RLH, Gaensler EA, Ferris BG. Diagnosis of "asbestosis." Observations from a longitudinal survey of shipyard pipe coverers. Am J Med 1978; 65:488-98. 36. Murphy RLH. Asbestosis in pipe coverers engaged in new ship construction (thesis). Har vard School of Public Health, Boston, MA, April, 1968. did n> Iso