Document 4423oDo3JwVrw0LJr2JY6pvOQ

484 THE LANCET, MARCH 4, 1978 Occupational Heai THE HYGIENE STANDARD FOR CHR ASBESTOS* 'TILE Julian Peto D.H.S.S. Cancer Epidemiology and Clinical Trials Vr 9 Kcble Road, Oxford Summary Previous studies, including the ana; ^ on which the current 2 fibres/cm hygiene standard is based, may have underestimated the risk of morbidity or mortality following exposure to low levels of asbestos dust. Accurate dose-response data at levels below 2 fibres/cm1 are unlikely to be available for the foreseeable future, and the biologically plausible assumption that excess cancer mortality is approximately proportional to dust level should be provisionally accepted. It may be reasonable, however, to postulate a safe threshold for mortality from asbestosis. If excess morta lity from asbestos-related disease is proportional to dust level for each cause, approximately 10% of male asbestos workers might, under certain assumptions, eventually die of asbestos-induced disease after 50 years' exposure at 2 fibres/cm3. Peritoneal mesothelioma is usually due to crocidolite (blue asbestos) or other amphiboles, but exposure to chrysotile (white asbestos) alone may lead to a substantial risk of pleural mesothelioma. These predic tions are based on rather small numbers in a single fac tory, and further studies in other working environments are required. Fibre counts based on optical microscopy are likely to be less relevant than total counts by elec tron microscopy, and excess mortality is virtually con fined to men first exposed more than 20 years ago, when little or no accurate data on dust levels were collected. THE PRESENT STANDARD The current hygiene standard of 2 fibres/cm3 for chrysotile asbestos dust was based on a study conducted (in 1966 by the British Occupational Hygiene Society (B.O.H.S.).1 Currently employed asbestos textile workers who had been exposed for at least 10 years were examined, and the prevalence! of crepitations (basal rales) was related to cumulative dust exposure , (fibre/cm3 years). A log-normal dose-response Curve fit ted the data adequately and led to the prediction that the cumulative dose corresponding to a 1% lifelong risk i of developing crepitations was approximately 100 fibre/cm3 years, or 2 fibres/cm3 for 50 years (fig. 1).J Most men with cumulative exposures of less than 100 fibres/cm3 years had been employed for between 10 and ^20 years at average dust levels ofless than 10 fibres/cm3, 1 and the absence of crepitations in this group was the basis for the prediction that the effect of 50 years' expo *' sure at 2 fibres/cm3 would be minimal. , . .. ,, This prediction would be reasonable If crepitations 'Based iVn'u'rcport presented to the Advisory TjSmmil tee 'on Asbestos Ulculth and Safety Executive) in latndon on June 28, IV77. ............. fTbe prevalence of a sign is the proportion or percentage of workers with the ign at a particular time. This must be distinguished from incidence, which is., the rate of appearance of new cases in previously unaffected men, and h usually expressed as a proportion or percentage per annum. '' ` (an early sign of the effect of accumulated asbestos expo sure) rarely appear or progress after exposure has ceased. However, other assumptions consistent with these data could lead to a predicted risk an order of mag nitude higher than 1% following 50 years' exposure at 2 fibres/cm3. For example, suppose: (1) that pulmonary damage caused by inhaled asbestos dust is progressive, either because fibres remain in the lung or because the fibrosis they engender progresses after they have been eliminated; (2) that crepitations rarely appear less than 10 years after first exposure; (3) that men are usually removed from employment when they develop crepita tions; and (4) that incidence (rate of appearance of new cases) is approximately proportional to cumulative expo sure after an initial lag of 10-15 years. The B.O.H.S. study comprised all men who had started work since 1933, had completed 10 years' service, and were still employed in 1966. The proportion of men with crepi tations rose from zero at exposures below 100 fibre/cm3 years to over 16% at 400 fibre/cm3 years. Under assumptions (1) to (4) the lack of cases in the lowerexposure groups might merely reflect a delay between first exposure and the appearance of crepitations, while the prevalence at 400 fibre/cm3 years would, if men were usually removed from employment within, say, 4 years of developing crepitations, suggest that the annual risk of developing crepitations after a cumulative expo sure of about 400 fibre/cm3 years might be 4% per annum, /This figure depends on the average time from first detection of crepitations to early retirement, which has arbitrarily been taken to be 4 years. If employment normally continued for 8 years after the detection of cre pitations the predicted incidence corresponding to 400 fibre/cm3 years would be roughly halved.) If incidence is proportional to cumulative dose the incidence at 2 fibres/cm3 might thus rise from zero during the first 10-15 years to about 1% per annum after 50 years, when the cumulative exposure would be 100 fibre/cm3 years. The risk of developing crepitations by retirement after 50 years' exposure at 2 fibres/cm3 would then be about 20%, and some further cases would appear after retirement. 'This figure is not presented as a useful pre diction of the likely risk. It is intended only to illustrate the importance of the assumptions implicit in such extrapolation, particularly when prediction of the long term effects of low exposure is based on initial response to relatively hiEh exposure. MORBIDITY AS A BASIS FOR A HYGIENE STANDARD- The B.O.H.S. analysis is significant because it'is the basis of the current standard. However, the risk of de veloping crepitations may not be an appropriate basis for a hygiene standard. The sign itself is virtually asymptomatic and may not progress to severe breathless ness or entail a substantially increased risk of bronchial carcinoma or mesothelioma. In the absence of direct observation of the prognosis in affected workers a stan dard based on such early signs may thus be unreasona bly restrictive. Moreover, if data on severe disability or death are sufficient to evaluate the significance of the sign they' will also be sufficient to assess the serious risk directly, so there may be little advantage in considering the prevalence of such a sign in framing a hygiene stan dard. This is not to say that such early signs are not the lancet, march 4,1978 485 worth studying. They contribute to un unde: anding of the natural historv of the disease process, r increased mortality in a small proportion of workei vith crepi tations can be detected and estimated bet. the overall excess approaches statistical significance. It should also be pointed out that a 1' elong risk of developing such a mild condition may h, stringent a criterion. THRESHOLD AND LINEAR MODELS FOR C,V ! R INDUCTION Since the pragmatic implications of a non-linear doseftsponse model such as that in fig. 1 and the assumption of a threshold (absolutely safe level) arc similar 1 shall apply the term "threshold" to both. The distinction is of scientific importance, but in the absence of biological understanding epidemiological data cannot establish whether a safe exposure level exists. Crump et al. argued on theoretical grounds that legislators should assume linear dose-response below the range of observation, even when the data suggest a quadratic model, particu larly for carcinogens,3 but for asbestos this is an aca demic point. The apparent threshold for crepitations could be an artefact for the reasons cited above, while .. 100 ZOG _.,300 : 400 500 .j|- Cumulative dose (years xfibre/cc) Pl|. 1--Doie-response relationship for prevalence of basai riles in workers in a chryaotiie asbestos factory (data from Berry ). j Log-normal curve. Numbers ofcases in italics. for lung cancer and mesothelioma there is neither theor etical nor statistical evidence to suggest that the relation between dose and response is anything other than linear. PREVIOUS STUDIES RELATING CUMULATIVE EXPOSURE AND LUNG-CANCER MORTALITY The suggestion that the eventual proportional excess (relative risk minus one) of bronchial carcinoma is approximately proportional to cumulative asbestos expo sure4 is supported by published work. Enterline's data-* suggest such a linear relationship, but he rather arbi trarily superimposed a normal distribution (fig. 2) and concluded: "Had larger numbers been available it might have been possible to identify more precisely a t.l.v. (threshold limit value) for respiratory cancer in this type of population. The available data suggest, however, that large increments in respiratory cancer appear at a cumu lative asbestos-dust exposure of somewhere between 100 and 200 mpef-years." Similar data were reported by McDonald et al.6 These authors inferred that "excess mortality was virtually confined to men with exposure equivalent to at least 400 mpcf-ycars" and concluded that "considering al! facets of disease--death, roentgcnographic changes, pulmonary function changes, and respiratory symptoms--the l^b risk is reached bv men in our third dust exposure category (100-200 mpefyearsi". However, a straight line appears to fit their data well (fig. 3), and the corresponding proportional excess of respiratory cancer at 150 mpef-years--about 30*( --would alone constitute a lifelong risk greater than 1, ignoring morbidity and asbestosis mortality. A cohort briefly exposed to amositc at the beginning of the 1939-45 war has been followed up by Seidman et al.1 This study is particularly valuable for testing models of the likely long-term effects of asbestos expo sure, since the workers' cumulative doses (fibre/cm3 years or mpef-years) have remained constant over more than 30 years, and the recorded duration of exposure, which was less than 2 years for most of the cohort, is probably closely proportional to cumulative dose. The relative risk for lung cancer rose for about 20 years and 0 . \ ' 200 - 400 600 800 - ,1000 Cumulative dose (mpef-years) - ' Wf. 2--Relative risk (observed/expected) for respiratory cancer i in retired asbestos workers (data from Entcrlineet pi.'). *-----------------Original cumulative normal curve. Numbers of Keiths in italics. . ., . - ' ' - (1004 1200-1 '(4004 / lobulor categories . ..0 300 600 Cumulative dose (mpef-years) Fig. 3--Age-ad|usted respiratory-cancer mortality in chryaotiie mine and mill workers (data from McDonald et al.`). The highest category (800 or more mpef-years) is plotted at 1000 mpef-years. Other categories are plotted at their mid-range. Numbers ofdeaths in ii,:" .- 486 then remained nearly constant. The otion that proportional excess risk is proportional mulativc dose may therefore be a useful approxinta: - retired workers or groups such as the asbestos tex orkers described below whose exposure during recen s has been low compared with their earlier exposu. ta more sophisticated model is required to describe i, w initial rise. If the duration of brief exposure can be as a measure of total dose the results of this st appear to suggest that for respiratory cancer lower li may be proportionately more dangerous, and Knox .. ferred that the cancer risk is proportionately higher at lower dust levels or shorter exposures. However, neither these data nor the studies he reviewed are described in enough detail to confirm such a pessimistic conclusion. Such apparent non-linearity could be due to selective removal of workers who develop respiratory symptoms, saturation at very high exposures, underestimation of expected numbers, or random error in estimates of dose or duration ofexposure. A HYGIENE STANDARD BASED ON FIBRES/CM5 The studies described in the preceding section support the suggestion that dose-response is likely to be approxi mately linear for bronchial carcinoma, and the first two studies illustrate the common error of inferring that there is a safe threshold because mortality in the lowest- exposure groups is not significantly increased. They may be of little quantitative relevance in framing a standard based on fibres/cm1, however. The correspondence between fibres and particles is very variable,1' and the estimates of risk shown in figs. 2 and 3 differ by a factor of more than two, perhaps reflecting differences of par ticle size and type in different environments. Pleural mesothelioma should be analysed separately. Excess mortality is likely to be proportional to the amount of dust that penetrates the bronchus for bronchial car cinoma or reaches the pleura for pleural mesothelioma, but the relationship with airborne concentration is not known. Different types and sizes of fibre are likely to be distributed and eliminated differently. It may be reason able to relate excess mortality to estimated particle or fibre counts in a particular environment, but different working conditions may not affect the risk uniformly for each disease. ` THE LANCET, MARCH 4,197 `t, . 01 - Y- , ------------ -------------- ---- 1---------- IQ30 192.0 1959 1960 1970 Fig. 4--Average dust levels, weighted by the number of men each level, in an asbestos textile factory, 1936-72. ; I>utu from Peto el al.ln). MORTALITY IN ASBESTOS TEXTILE WORKERS The 679 men in the asbestos textile factory in which the 11.0.U S. study was conducted who entered sche duled areas after Jan. 1, 1933, and had worked for at least 10 years bv Dec. 31, 1972, have been followed up to the end of 1974.10 The B.O.H.S. cohort is thus in cluded, together with men who retired or died before June 30, 1%6, and those who completed 10 years' ser vice between June 30, 1966, and Dec. 31, 1972. Deaths from bronchial carcinoma, pleural mesothelioma, other respiratory diseases, and other causes are compared with expected numbers based on national rates in table I. This shows a substantial excess due to cancer and respir atory disease beyond 25 years after first exposure (35 observed; 15-74 expected). Mortality due to other causes has been normal, except for the "healthy worker" effect between 10 and 15 years after first exposure (14 obser ved; 21-63 expected). These men were still employed at the start of the period of observation and were therefore unlikely to be chronically sick during this first period. DUST EXPOSURE -- Estimated average dust levels over the period of the study are summarised in fig. 4.10 The average (weighted by the number of men at each level) was approximately 13 fibres/cm' up to 1*950, falling to about 5 fibres/cm5 by 1955. Individual exposure histories have noi yet been compiled for the whole group, but for diseases for which excess risk is approximately TABLE 1---MORTALITY IN 679 ASBESTOS TEXTILE WORKERS AFTER 10 OR MORE YEARS- EXPOSURE Years . since lirst ^7 - exposure Manyears Mesothelioma -------- ------------------ Rate per (lbs annum Bronchial carcinoma --- Ohs Exp Other respiratory disease -- Other causes -- Ohs 1-xp Ohs I-xp h - All causes - Ot>s I-xp At Excess Attributable respiratory excess deaths* mortality iObs-Expi (Hi (B/A) HI - . = 2727 15 ' 2612 . 2(1- / 1964 .-'25 - V"'-'; USX ' 30 ;?)- 630 ;U189, ^c-Tmal v , 793IO 0 0 1 1 2 ./,, 1, ;:r5 0*0000 0-0000 0-0005 00008 0-0032 ,0 0053. ,2 7 5. : r - r5 0-0005 - 29 3-07 4-25 4 50 3-57' 2-31 ` 0 69 , - ,2 6 56 10 v5 3-54 4-95 . 5-56 , 5-07 325 ; 0x5 18-39 31 - 23-22 " 14 - 27 35 23 16 6 121 21-63 28-65 3003 25-21 I5.qQ -*,4.50 126 01 19 28-24 40 37-85 47 40 09 42 -: 33-85 28 21-55 10 604 1X6 16'62 -1-61 3-80 1-94 10-36 6 44 2 46 23-39 0 00 -0 to oor 0-31 0-30 041 0 14 ` Ot observed.Exp-expected. * Combined excess (observed minus expected) for mesothelioma, bronchial carcinoma, and respiratory disease. the LANCET, MARCH 4,1978 4N~ Proportional to dust level these average U s can be related to overall excess mortality, as 100 men c,| d to 1 libre/cm3 rod 100 at 3 fibres/cmJ would show sub i it excess morta lity similar to that of 200 men exposed to res/cm' over a comparable period. Such aggregation is do: mplicitly in any Hudy. Measured levels vary considerable .er time in each *TOa, so even a group nominally expose : > 2 libres/cm3 will . b*ve experienced a wide range of dust U I he excess risk < appears to be largely confined to the pen eyond 25 years *fter first exposure (table U, when the exc. mortality from . respiratory disease, including bronchial viiioma and - roesthelioma, exceeded the expected mortalitv : ;> all causes *n 31% (19-26/61-44). The average exposure a h caused (his excess can he roughly summarised bv total s., 'vment during the 30 years following first exposure, fable it, which u based on the 158 men followed up for at least 30 years, sug gests that this exposure consisted on average of about 11-6 Jeros before 1954 at an average dust level of about 13 >res/cm3 followed by 9-4 years at an average of about 5 fibres/cm3--a cumulative dose close to 200 fibre/cm3 years, y Dust levels in fibres/cm3 were estimated in each area from onbrane-filter measurements at fixed sampling points from 1961 onwards. Thermal-precipitator measurements in par- bdeVft3 taken between 1951 and 1960 were converted to fibres/cm3 using the observed ratio of parallel measurements obtained by the two methods in 1960 and 1961. Routine , measurements were not made before 1951. Estimated levels in 4 TABLE II--EXPOSURE (TOTAL SERVICE IN SCHEDULED AREAS) IN f *MLE ASBESTOS TEXTILE WORKERS IN THE 30 YEARS FOLLOWING $ FIRST EXPOSURE* Year of first exposure 1933-34 1935-39 1940-44 Total distribution of Average ex posure (yr) No. total exposure (yr) of men 10- 15- 20- 25- up to 1955 _ from 1954 (approx. (approx. 13 fibres/cm3) 5 fibres/cm3) 6 17 0 67 17 104 25 19 19 37 48 23 19 31 27 14 1 120 10 5 7-7 93 100 158 24 18 25 33 116 94 * * Men followed up for less than 30 years arc omitted. <4% " :...... ,,- 4 fibres/cm3 in each area of the factory are therefore based on (direct observation after 1961 and conversion from particles/ft3 to fibres/cm3 between 1951 and 1960. Estimates before 1951 - re based on the assumption that levels in each area fell from I) 1-5 times the 1951 values between 1933 and 1950. It is diffi4cult to assess the error in this factor. Excess mortality docs not 3 *Ppear to have been much higher in men first employed - between 1933 and 1950 than among those first employed after / 1950,10 and in view of the technical continuity in many areas between 1933 and 1950 it was considered likely that dust levels (did not fall substantially over this period. However, this cru?dal assumption is open to dispute and cannot be proved until .further foilow-up has provided a more accurate estimate of ,,excess mortality in more recent employees. Exposure was pre dominantly to chrysotile, although a small proportion of the : fibre processed was crocidolite at various times after 1933. |.| BRONCHIAL CARCINOMA . ?' Mortality due to lung cancers other than pleural mesothesnoma in this cohort has been approximately twice the national jtverage beyond 25 years after first exposure (14 observed; s6-57 expected). There is no evidence that the relative risk injereases beyond 25 years, which perhaps reflects the relatively negligible increase in cumulative dose in recent years. Dust Ilevels in this asbestos textile factory have been much reduced since 1055. I hi- interpretation is supported by an earlier study, l're-1933 workers, who were exposed to very high levels before 1933 and comparatively low levels subsequently, showed a similar lack of progression beyond 20 years for both bronchial carcinoma and asbestosis mortality.11 If the eventual proportional increase in mortality due to bronchial carcinoma is approximately linearly related to cumulative exposure and the risk is doubled at 200 fibre/cm3 years, a level of 1 fibre/cm' might, after 50 years' exposure, increase mortality due to this cause by 25% after retirement, corresponding to a lifelong risk| ol bronchial carcinoma attributable to asbestos of about 3% in smokers.4 The risk to non-smokers is probably very small. Asbestos and cigarette smoking appear to enhance each other's effects in causing bronchial carcinoma,13 but mortality due to this cause is so low in asbestos workers who have never smoked that the risk due to asbestos cannot be estimated with any confidence from published work. A MORE GENERAL LINEAR MODEL FOR BRONCHIAL CARCINOMA This analysis is based on the assumption that the relative risk for bronchial carcinoma is increased in proportion to cumulative dose. As this is not precisely true a less restrictive assumption may be preferable. Bronchial carcinoma accounts for an approximately constant proportion of male deaths in England and Wales (12% at age 50-54, falling to 10% at age 70-74), and various studies suggest that the relative risk for this cause does not fall with the passage of time after exposure to asbestos has been substantially reduced or ceased. The observed excess mortality due to bronchial carcinoma as a pro portion of total expected mortality beyond 25 years (7-43/61-44, or 12%) is therefore a reasonable estimate of the lifelong attributable risk in this cohort. It would follow, under any model which predicts that the increase in risk is approxi mately proportional to dust level, that a similar period of expo sure at a reduced dust level would produce a proportionately ' lower risk. Thus a group that had on average been exposed to 1-3 fibres/cm3 rather than 13 fibres/cm3 for about 12 years fol lowed by 0-5 fibres/cm3 rather than 5 fibres/cm3 for about 9 years would probably have a lifelong attributable excess risk due to bronchial carcinoma of about l-2'7. The conclusion that about 20 years' exposure at about 1 fibre/cm3 is likely to cause bronchial carcinoma in roughly 1 man in 100 thus fol lows from any linear model, but a more specific assumption is required to predict the elfect of 50 years' exposure. PLEURAL MESOTHELIOMA As only 5 cases of pleural mesothelioma have occurred in this 'group, any model-fitting or extrapolation must be very speculative. These 5 cases are tabulated by man-years of obser vation in tabic t, which suggests that incidence rises sharply with increasing time since first exposure. A formal residence time model with linear dose response for cancer incidence is given by the equation: , Incidence at time T after first exposure a ,T-t)k c(t).dt ... . where c(t) is the dust level at time t. Tlie value k=2 fits the incidence rates in table I well and corresponds closely to Newhouse and Berry's finding that incidence rose as (t--9)3 after heavy mixed exposure to chrysotile and crocidolite had ceased.1314 Applying the observed rates in table l, the predicted ^Diseases fur which the cicest risk persists beyond retirement cannot be sum marised by relative risk or cumulative prevalence at a pantcular age. I he appnv priaic statistic is the proper!am who will eventually die of the disease as a result oftheir exposure. I his is described throughout as the "hfdung nvk'\ . _ 488 TUF LANCET, MARCH 4,1978 prevalence after 50 years' exposure at 1 fibre/cm*1 correspond ing to this model is 1 -6'< . The corresponding lifelong attributable risk to a man exposed to 1 fibre'em' from age 15 lobs js approximately 2',. It must be emphasised that this estimate is necessarily utirch able. It is based on only 5 cases, and 2 of these were in men employed in an area where raw asbestos was handled who should perhaps be excluded from the calculation, almost halv ing the predicted risk. BRIEF OR NON-OCCl'I'ATIONAI. FXPOSt'KF Malignant diseases such as mesothelioma or bronchial car cinoma for which the incidence rises as the third or higher power of time since first exposure to a carcinogen cannot be understood without such formal analysis, for example, there had up to the end of 1975 been 10 cases of pleural mesothe lioma among approximately 20 000 men first employed in this factory after 1933. To express this as a risk ol 0 05', is mean ingless. Idle lifelong risk to a man employed continuously since 1933 may exceed 10',. On the other hand, the suggestion that brief exposure is extremely dangerous is equally misleading. Preliminary results of a study including the entire past and present workforce support the suggested model, and only 1 case is known to have occurred in approximately 15 000 men exposed for less than 2 years since 1933. Isolated case-reports with a history of minimal exposure must be viewed against some hundreds of thousands of men and women with slight occupational exposure and several millions with casual nonoccupational exposure. A recent study confirming that the risk of mesothelioma is dose-related suggests that only very high dust levels produce an appreciable risk after brief exposure.15 MINIMAL CKOCIDOI.ITE EXfOSIRI IN CIIKVXOt li t WORKERS It has been argued that pleural mesothelioma is due to croci- dolile even when it occurs in workers w hose exposure has been largely to chrysotile with minimal exposure to other fibres. Evidence from three sources appears to contradict this inter pretation: (1) 9 cases1 have occured in Canadian chrysotile miners and millers with no exposure to crocidolite;1'' (2) the in cidence in our study has been comparable to that in Newhouse and Berry's" cohort of workers with heavy crocidolite expo sure. Their data show a clear dose-response for both severity and duration of exposure. Although average exposures Were longer in our cohort, total exposure to crocidolite was so much lower that the risk of mesothelioma should have been far less than that observed in the factory we studied if crocidolite were the sole cause of mesothelioma; (3) workers exposed to crocido lite show a comparable incidence of peritoneal and pleural mesothelioma. For example, 22 of the 45 mesotheliomas reported by Newhouse and Merry" were peritoneal in origin. None of the 34 that have occurred in workers in the factory we studied (including women and pre-1933 employees! or the 9 cases in Canadian miners and millers were peritoneal. An obvious explanation for this difference is that chrysotile causes only pleural mesothelioma, whereas crocidolite causes both pleural and peritoneal tumours. ' ". ASBESTOSIS . . The lifelong risk of death from asbeslosis following exposure to 1 or 2 libres/cm3 cannot be predicted with any confidence. There are no grounds for assuming linear dose-response for such a generalised progressive disease, and although a qualita tive dose-response has been demonstrated at very high expo sure levels" there mav well be a safe or virtually safe thresh- ' ^Assuming that k 2 and that incidence 30 years after first exposure following 11-6 yean at 13 fibres/cra* and 9-4 yean at 5 libres/cm1 liable it! is 3 per 1000 per annum - table I, col. 31. I hie model can he iuetilied biologically, lent for prac- 1 tiea! purposes it is irrelevant whether incidence rises roughly as t\ as this model predicts, of as (t-911. These factors are closely proportional for values ol I y between 20 and 90 years, and betuw 20 yean the incidence is negligible. - , ' _ old. I hero were 1!" deaths 9.17 expected: due W iu'ii malignant respiratory disease mcr 25 wars alter firs* exposure m our study, including ' attributed to asbestosis. If it is assumed that incidence is proportional to cumulative dost, the corresponding life long attributable risk alter 50 yean exposure at 1 fibre cm' would exceed l',.4 but this model is presented for lack of another rather than because il is partial- ^ tarty plausible biologically. Further data on early pulmonary signs may indicate w hether a threshold exists, but even if the i incidence of crepitations is linearly related to dust level the prognosis in affected workers may be related to the severity preceding exposure. , DISCISSION ; There have been too few deaths in our study to esti mate the risks or evaluate the suggested models adequa- : telv for each asbestos-related disease, but the overall excess is substantial. There were 80 deaths (61-44 expected) more than 25 years after first exposure, in cluding 14 y6- 57 expected) due to bronchial carcinoma, 17,9-17 expected) due to non-malignant respiratory dis ease, and 4 due to pleural mesothelioma. The excess lor asbestos-related disease thus exceeds the overall expected number by about 30',. The implications of these figures are perhaps obscured by the necessarily complex statisucal models presented for each separate cause. If dose-res ponse were linear for all causes (i.e.,- the risk of death due to asbestos exposure were proportional to dust-level lor any given period of exposure) it would follow that about 20 years' exposure at about 1 fibre/cm3 would in crease total mortality by about B'y 25 years alter first exposure. The trend in the right-hand column of table . i suggests that this proportional excess is likely to persist with the further passage of time, so such exposure might be expected to lead to fatal asbestos-induced disease in about Vi of men. The assumption of dose-linearity is dubious for asbestosis mortality, but both epidemiology ; cal findings and contemporary knowledge of carcino genesis suggest that excess mortality from bronchial car cinoma and pleural mesothelioma is likely to be , proportional to dust level. -. , These predictions might be grossly inaccurate. Indi- . vidttal cases have not been investigated in detail, and ; previous employment in dusty working environments ; may have contributed to the excess mortality in this j cohort. Employment histories are now being re-exa- j mined, but the estimates of pre-war asbestos exposure are not reliable, and the assumption that there is no safe ' threshold cannot he confirmed without long-term pro- ^ speetive observation of workers exposed to low levels. The recent I.A.R.C. monograph on asbestos,17 which ' provides a comprehensive reference source and summary, of recent- work, concludes that "at present, it is not pos..sible to assess whether there is a level of exposure in humans below which an increased risk of cancer would not occur." There are thus three alternatives; v ; 1. To conclude that the economic consequences of more re strictive legislation are not justified unless a further 15 or more . years' observation of workers exposed to low levels of pure ! chrysotile shows the current standard to be dangerous. 2. To base a decision on these analyses. Ignoring sampling - error, our results suggest that about S-1QT of men exposed for ; 50 years to chrysolite asbestos concentrations of 1-Jte fibres/cm3 are likely to die of asbestos-induced disease. PIauv ___ _________________ ______________ ----i"- * A further mesothelioma occurred in a woman in this population who had prv . viously been exposed to crocidolite. ., \, ... the lancet, march 4,1978 48d ble alternative models might reduce or increase thi factor of perhaps 2 (2^-5'f or 10-20','), hut the , which to base such refinements will not be available foreseeable future. e by on the 3. To accept that linear dose-response models are appro Pnate, at least for cancer, but to defer a decision until our est, Rates are more precise and analogous studies have been con ducted in other environments. It may not always be possible 10 assign meaningful individual exposures, but average expo*ures of the sort shown in fig. 4 could be estimated in manu factories. The incidence of bronchial carcinoma and mesothefwma must be studied in environments with different fibre sizes before such predictions can be generalised. It should be emphasised that such studies will not afford much information R the effect of very low doses. Their only value is to give inde pendent estimates ol the slope of the dose-response curve in Other industrial settings. The epidemiological evidence that pleural mesothe lioma can be caused by chrysotile alone is supported by the observation that chrysotile is quickly eliminated from the lung parenchyma but remains in the pleura, while amphiboles are present in very much lower con centrations in the pleura than in the lung.18 This may be because amphiboles are more penetrant and less liable to dissolution, and having left the lung are more widely distributed elsewhere but less likely to remain in the pleura. The reported excesses of peritoneal mesothe lioma and other non-respiratory cancers in crocidolite workers,14 neither of which occurred in our cohort,10 appear to support this interpretation. The persistence of some chrysotile in the pleura but very much less in the lung perhaps reflects differential dissolution in these sties, Electron-microscope studies of the peritoneum and other organs are urgently required to confirm or refute these speculations. Total fibre counts of both air and post-mortem tissue samples, in conjunction with morta lity data, would greatly increase the value of compara tive studies ofother working environments. I should like to thank Turner and Newall and T.R.A. Industrial Products for their collaboration in the collection of these data, and various colleagues, particularly Sir Richard Doll, for criticism of ear lier drafts. The views expressed in this paper are mine alone, however. REFERENCES 1. British Occupational Hygiene Society Ann. occup. //vg. 1968, It, 47. 2. Berry, G. in Biological Effects of Asbestos; p. 145. International Agency for Rcsearchon Cancer, Lyon, 1973. - 3. Crump, K. S., Hoel, It. G., Langley, C. H., I*eto, R. Cancer Res. 1976, 36, 2973. 4. Peto, J. in Environmental Health: Quantitative Methods (edited by A. Vt'hil- iemore). Society for Industrial and Applied Mathematics, Philadelphia, 1977. U.K. distributors: tlcydcn and Son, Ltd., Spectrum House, Aider- ton Crescent, London NW4 3XX. - 5;Knterline, P., dc Coufle, 1*., Henderson, V'. Hr. J inj Med. 1973, 30, 162. 6. McDonald, ]. C,, Becklake, M. R,, Gibbs, G. W., MclXtnald, A l),, Rossiter, C. V.. Archs cnt'ir. tilth, 1974,28,61. 7. Setdman, H., I.ilis, R., Selikoff, I. J. Short-term asbestos exposure and - delayed cancer risk. Presented at 3rd International Symposium on Detec - non and Prevention ofCancer. New York, 1976. .; , 8. Knox, E. G. Br.J. inj. Med. 1973,30, 54. ' `".'T. 9. Gibbs, G. VI'., latchance, M. Archs envir. Hlth, 1974,28,69. ' 10. Peto, J., Dull, R., Howard, S., Kinlen, L. J., I-cwinsohn, 11. C. Br.J. ipd. Med. 1977,34, 169. .. ' 11. Knox, J. E., Holmes, S., Doll, R., Hill, I, Yi.fbid. 1968,25, 293. ' ' ' 12. Doll, R.J1R. statist. Sue. A, 1971,134,133. . . T vi - 13. Newhouse, M. L., Berry, G. Br.J. ind. med. 1976,33, 147. ' : . 14. Newhouse, M. L. in Biological Effects of Asbestos; p. 203. International - Agency for Research on Cancer, Lyon, 1973. .... . ' 15. Whitwell.F., Scott, J.,Grimshaw, M. TAorox, 1977,32,377. . 16. McDonald, J. C. Personal communication. ........... -. .. 17. International Agency for Research on Cancer. Asbestos. I.A.R.C. Mono- i : graphs on the Evaluation of Carcinogenic Risk of Chemicals to Man, vol. 14. Lyon, 1977. . .. ., . T -18. Le Bouffant, L., Bruyfre, S., Martin, }. C, Tichoux, G., Normand, C. Rev, . fr.Malad.resp.\^(s,49svp^i2t\2\. ' Questionable Routines CRITICAL EVALUATION OF ADENOIDECTOMY J. Hibbert P. M. Stell Department of Dto-Rhino-Laryngology, University of Liverpool anil Liar, S'ose and Throat Infirmary, Liverpool 1.7 7DF ` Summary In two matched groups of thirty-two children, one which had tonsillectomv alone and the other which had tonsillectomy plus adenoidectomy, the symptoms generally attributed to adenoidal hypertrophy were equally common in both groups before operation and improved with equal fre quency after operation whether or not the adenoids were removed. INTRODUCTION Temporary improvement in symptoms after adenolonsillectomy may be due to an operation rather than the operation;1 the effects of adenotonsillectomy should thus be compared with those of another operation, to exclude the placebo effect. W'e have compared the im provement in the symptoms of adenoidal hypertrophy in two groups of patients, only one of which had an adenoidectomy. PATIENTS and methods Thirty-two children listed for adenoidectomy and tonsillec tomy on the waiting-list of the Liverpool F.ar, Nose, and Throat Hospital were matched for age and sex with thirty-two children listed for tonsillectomy alone (table i). The parents were interviewed and asked about the symp-\ toms attributed to adenoidal disease (nasal obstruction, snor ing, rhinorrhtea, speech problems, cough, and headache' and the children were examined for corresponding signs--mouth breathing and abnormalities on anterior rhinoscopy.3 At the time of examination the examiner did not know what oper ation the children were to have. ' The children were then operated on and reassessed six weeks later, again without the examiner knowing what operation they had had. The number of positive scores for each symptom and sign in each group was analysed by a y1 test, with a 2x8 contin gency table. The proportion of children who improved in each group (x) was calculated and a modified arc sin transformation done on the proportion (v-arcsin (x/n +1) ^ + arcsin (xtl/(ni 1)* t. The transformed proportions were then com pared by a t test for paired data. ' RESULTS - In the two groups, there was no significant difference in the number of children who'showed signs and symp toms of adenoidal hypertrophy before operation TABLE I---- PATIENTS ,. -7 Sex -M ...... " "~ * . ... F - V - ~ " Tonsillectomy No. Age range 16 5 yr 5 mo. to 9 vr 6 mo. 16 -3 yr 6 mo. . to 10 yr 8 mo. Tonsillectomy and adenoidectomy No. Age range 16 J yr 6 mo. 1 u> 9 vr 6 mo. 16 3 yr 9 mo. to - 10 yr 5 mo.