Document aJzgLpoXLd3756zVRyQENr32M

FILE NAME: RT Vanderbilt (RTV) DATE: 1988 DOC#: RTV 124 DOCUM ENT DESCRIPTION: Journal Article - Analysis of Excess Lung Cancer Risk in Short-Term Employees ACompeyrriicgahnt &J o19u&rSnbaylTohfe EJopihdnesmHioolpkoignys. University School of Hygiene and Public Health All rights reserved PriVntoe.d1in2?U, N.So..A6. ANALYSIS OF EXCESS LUNG CANCER RISK rN SHORT-TERM EMPLOYEES STEVEN H. LAMM.' MARSHAL S. LEVINE.*12*4'1 JENNIFER A. STARR,u AND SAN'DRA L. T1REY" Lamm, S. H. (Consultants in Epidemiology and Occupational Health, Inc., Washington, DC 20007), M. S. Levine, J. A. Starr, and S, L- Tirey. Analysis of excess lung cancer risk in short-term employees. Am J Epidemiol 1988;127:12029. An excess of lung cancer found in a cohort of 741 New York State tremolifc talc workers observed from 1947 through 1978 has been shown paradoxically to be concentrated in short-term workers. Review of past work histories suggests that the excess of lung cancer in these short-term workers may be accounted for by prior exposures rather than by exposures at the employment under investi gation. This finding has significant implications in view of the developing practice of including short-term workers in occupational cohort studies in contrast to the more traditional practice of excluding short-term workers. The traditional practice was based on the assumption that the inclusion of short-term workers with little exposure, and thus little risk, might dilute an otherwise apparent association between mortality and exposure. This study suggests that in certain instances the inclusion of short-term workers may magnify rather than dilute the estimation of risk, reflecting the presence of confounding variables. lung diseases; Jung neoplasms; occupational diseases; talc New York tremolitic talc workers have been the subject of numerous epidemiologic studies over the past 50 years. Eariy reports (1-4) focused on morbidity and mortality from pneumoconiosis and related pulmo nary conditions. Excess cancer mortality was reported in 1967 and 1974 by Kleinfeld eal. (5,6) in aproportional mortality study of 220 New York State talc miners and finaRlefcoerimvedAfuogrupsutb2l8ic, a1t9io87n. December 19,1986, and in cupAabtiborneavliatSioafnest:yNaiOndSHH,eNalathti;onSaMl iRn,stisttuantedafordrizOecd mortality ratio. He1aCltho,nIsnucl.t,an24ts28inWEispciodnemsinioAlovgeynuaen,dN\OVc.cWupaasthioinngal ton, DC 20007. (Reprint requests to Dr. Steven H. La2mCmu.r)rently rt National Aeronautics and Spac--e Ad min1Cisutrrarteinotnly, WataWshihnagrttoonn,SDcCho. ol of Finance, Univer sity of Pennsylvania, Philadelphia, PA. 4Currently at Chemical Manufacturers Associa tion, Washington, DC. millers. Among 91 deaths, the only excess cancer mortality was for lung and pleural cancer. This excess was limited to those workers who bad been hired prior to 1945 (at which time wet drilling industrial hy giene controls were introduced) and who had died prior to 1960 at age 60 years or greater. In 1973, Kleinfeld et al. (7) examined 39 workers from an upper New York State talc plant which had begun operations in 1947 and had used only the modern wet drilling techniques. No malignancies, and only one case compatible with pneumoconiosis, were observed among these workers whose only known exposure to commercial talc was at the plant studied. The National Institute for Occupational Safety and Health (NlQSH) subsequently conducted indus trial hygiene studies at this talc plant aswell as undertaking cross-sectional morbid ity and historical mortality studies of its ANALYSIS OF LUNG CANCER IX SHORT-TERM EMPLOYEES 1203 workers iS-10). From a cohort mortality study of 39S white males initially hired between January 1,1947, and December 31, 1959, and followed for their vital status as of June 30. 1975. NIOSH reported finding a statistically significant excess mortality from lung cancer and from nonmalignant respiratory diseases. .. ' An independent cohort mortality study of employees at this same talc plant was performed by Siille and Tabershaw (11). Their cohort study of the 655 white males employed between January 1, 1948, and December 31.1977, assessed vital status as ostfaDtiestciecmalblyers3ig1n, i1f9ic7a8n.tTehleeyvarteiopnosrteodf nluonng- and respiratory cancer mortality in the to tal cohort and showed a significantly in creased risk of respiratory cancers only for employees with any history f prior em ployment. The authors concluded that this increase was probably due to exposures that had occurred prior to employment at this talc plant. Due to the conflicting findings with re spect to excess mortality from lung cancer in this population of talc workers, a re analysis of the data gathered by Stille and Tabershaw and by NIOSH has been under taken. M aterials and m ethods Study population The study population consisted of the cohort of 741 men and women who had ever worked at this talc plant from 1947, when the plant opened and began operations, through December 31,1977. All of the em ployees in the study population were white. During this period, 705 men and 36 women had been hired. Mortality follow-up of the' 36 women was conducted but not included in the statistical analysis because of the small numbers. A total of 425 men worked at the plant for one year or more; 280 woforokbesderfvoartiloenssinthtahne osntuedyyeaerx.teTnhdeedpefrrioomd the date of hire to either the date of death, the date of loss to follow-up, or December 31,1978, whichever date came first. The data utilized for this analysis were originally collected by Tabershaw Occupa tional Medicine Associates as described previously (111. These data were abstracted from employment applications compieted at the time of application for hire and from job cards and employment files maintained by the company. Death certificates were obtained by Tabershaw Occupational Med icine Associates and by NIOSH from state health departments. The cause ofdeath was coded by nosologists using the Interna tional Classification of Diseases, Eighth Re vision. ' For this reanalysis, additional data on missing dates of birth for 16 workers were obtained from local hospital records. For the remaining 17 whose birth dates were missing from the records, an estimated birth date was developed by G. Block of the National Cancer Institute, using the method by Block et al. (12) based on the year of issuance of the social security num ber. Prior employment history Each prior employment experience indi cated on the initial job application at this talc plant was categorized on the basis of its likelihood of association with an in creased risk of lung cancer. This categori zation was developed by a senior industrial hygienist without knowledge of the individ ual's vital status or cause of death. The basis for classification was' as follows: "Lung cancer risk" jobs included metal mining (primarily lead/zint mines) (13,14), steel-making (15), foundry (16), roofing (17), welding (18), construction ("19), and paper mill (20). Jobs such as retail sales and restaurant work were clearly classifia ble as jobs with "no prior risk." Other em ployment categories, such as carpentry, electrical work, transportation, military, farming, and engineering, which are not clearly either at risk or free of risk were for .this analysis also classified as "no prior risk" employment. Such a classification would tend to bias against finding a differ ence between the two risk classifications. 1204 LAMM ET AL. The previous analysis by Stilie and Tabershaw (U) stratified only on the presence or absence of a prior work history on the job application, while our analysis stratifies on the basis of specific employment informa tion contained in the prior work history. Individual workers were then classified on the basis of the categorization of their prior employments as ever having had "prior risk" (353, or 50 per cent), as having "no prior risk" (214, or 30 per cent), or as "unclassifiable" (138, or 20 per cent) be cause their job application gave no indica tion of their prior work history. Statistical methods Observed-versus-expected deaths for all causes and for specific causes of deaths were developed and compared, using Na tional Center for Health Statistics causespecific death rates for US white males and using the analytic program developed by Monson"(21) to calculate the expected numbers. Comparison between the ob served and the expected number of deaths was expressed as a standardized mortality ratio {SMR), where SMR equals 100 times the observed deaths divided by the expected number of deaths. Statistical significance of the difference of a standardized mortality ratio from 100 is presented as a two-tailed p value based on a Poisson distribution. Results Mortality for the entire male cohort {ta ble 1) was significantly elevated {SMR = 141) as were the risks of death from all cancers {SMR = 165), from the subclassi fications of respiratorycancer (SMR ~ 246) and of lung cancer (SMR = 240), and from all nonneoplastic respiratory disease (SMR = 236). There was no significant excess mortality from nonrespiratorycancers. The recorded causes of the 10 nonneoplastic respiratory disease deaths were pneumonia (foyr), pneumoconiosis (two), emphysema {two) and other lung diseases (two). In ad dition, an electrician at the talc plant died from mesothelioma 15 years after being hired. He had previously worked 20 years in various employments as a miner, a millet, or in construction. Mortality for the 36 females revealed no deaths from lung cancer or other diseases of particular con cern. A similar analysis (table 1) was per formed for the 425 men 160 per cent of the cohort) who had at least one year's duration of talc plant employment (experienced workers). The significantly increased mor tality from lung cancer noted previously in the total group was no longer evident. The significant excess risk for all nonneoplastic respiratory diseases (SMR = 278) was seen to be restricted to deaths from noninfectious, nonneoplastic respiratorv diseases (SMR = 370). When mortality of the total cohort, of those employed for at least one year (i.e., one year or more), and of those.employed for less than one year are compared for all causes, for all respiratory cancers, and for lung cancer liable 2), it becomes apparent that the excess mortality from lung cancer in this cohort is concentrated in those em ployees who were employed for less than one year. Furthermore, the excess mortality from noninfectious, nonneoplastic respira tory' disease is found only in those who were employed for one year or more. When prior exposure history is added to the analysis {table 3), it can be seen that lung cancer mortality risk also appears to be related to prior occupational exposures. Thus, lung cancer risk is found to be greater in those who had less than one year of employment in the talc plant, and seems to predominate in those who _had potential lung cancer risk exposure prior to their talc employment. In contrast to the findings for lung cancer mortality, mortality from noninfectious, nonneoplastic respiratory disease (table 4) is only demonstrable in those workers with one year or more of employment and, un like lung cancer mortality, is not concert* trrisakte.d in employees with prion exposure The 12 cases of lung cancer observed in ANALYSIS OF LUNG CANCER IN SHORT-TERM EMPLOYEES Otatfnn'f m tv/wvfni niunfvrs of dvnth*t and xlandrirdhrci murtaffty ntliav (SM its) (W17- IV7H) ttmmifi male ufistatv. New York lulu filant |H I l I H I p S i !! IT ***! 3 Ii p s 3S333SS233g S 555 z- -- -- m m u U n fg* Es|ss3SiHf!!S*li Z ~ SSi u %n |35333S353Si 35 sss *rtg=W <S-t50-C4 S * cp as c- J ? i *Um ii v I '" !M lll "Wj| rr c 1206 LAMM ET AL. Table 2 All cause, respiratory cancer, lung cancer and nanin/ectious. nonneoplastic respiratory disease mortality risks Less than one year OEbxspeercvteedd/ * SMRt All causes Respiratory cancer Lung cancer Noninfectious, nonneoplastic respiratory disease 55/33.8 163*_ 6/2.0 300* 6/1,9 317* 0/1.0 *p < 0.05, two-tailed test, t SMR, standardized mortality ratio. Duration ofemployment Oneyear or more ^OEbxspeercvteedd/ SMR 63/49.S 126' 7/3.3 213 6/3.1 193 ....... 6/1.6 ... 370* ' H ill il ' Total cohort OEbxspeercvteedd/ SMR 118/83.6 141* 13/5.3 246* 12/5.0 240' 6/2.6 .... U S & l T able 3 Lung cancer mortality risks (1347-197$! gy duration of employment and prior exposure history among male upstate New York talcplant workers ^ u Wclaosrskifhiciasttoiorny Less than one year OEbxspeercvteedd/ SMR+ Duration of employment One year or more OEbxspeercvteedd/ SMH TotL cohort OEbxspeercvteedd/ SMR Prior risk 3/0.9 333 6/2JCT 305' No prior risk 1/0.4 2S8 0/0.5 Undassifiable 2/0.6 313 0/0.7 Total cohort 6/1.9 317* 6/3.1 193 *p < 0.05, two-tai led test. +SMR, standardized mortality ratio, t Prior employment in job with lung cancer risk (see text). No prior work history on job application (see text). 9/2.9 36* 1/0.9 ' 117 2/1.3 154 12/5.0 2i0* - the total cohort are reviewed in table 5. Duration of employment of these workers varied from several days to over 17 years, with a mean of 4.7 years (median of 20 months). Ages at time of hire varied from 24 to 57 years, with a mean of 34.8 years (median of 32.5 years). D iscussion Inclusion of short-term workers in the analysts of occupational cohort studies is generally felt to dilute the measure of as sociation of mortality with the exposure of interest because of minimal exposure du ration. To compensate for this perceived difficulty, short-term employees are fre quently omitted from the analysis. We have presented am-example of mor-" tality from lung cancer occurring in an occupational cohort of talc workers which demonstrates an increased mortality con fined precisely to those who are tradition ally excluded from analysis, the employees with short-term exposure (less than one year). Risk from occupational exposures is generally expected to increase with dura tion of employment. Here the risk does not appear to do so. The paradoxical nature of this finding is emphasized by the contrast ing fact that mortality from noninfectious, nonneoplastic respiratory disease in this cohort does follow the anticipated pattern of increased incidence with increased du ration of employment. When "the lung cancer mortality is ana lyzed by duration'Of employment (table 2) and by prior exposure risk categories (table 3), it appears that the lung cancer mortality ANALYSIS OF LUNG CANCER IN SHORT-TERM EMPLOYEES 1207 Table 4 Naninfcctioaa, normeopiostic respiratory disease mortality risks (1947-1STB) by duration of employment and by prior exposure history among male ups,,htalte New^York tcjlcplant workers Duration of employment Wfaolrskifhiciasttoiorny Lass than one year Observed/ One year or more Observed/ SMR* Total cohort Observad/ SMR Expert ed Expened .Expired Prior risk): 0/0.4 3/0.9 319 3/1.3 -224 No prior risk 0/0.1 3/0.3 1200* -3/0.5 666* Unclassifiabfei 0/0.5 0/0.4 0/0.9 154 Total cohort 0/1.0 6/1,6 -_ ' Y..^.H 370* ,6/2.6 227. ` p < 0.05 two-tailed test, t SMR, standardized mortality ratio, t Prior employment in Job with lung cancer risk (sec text). 5 No prior work history on job application (see text). Table 5 Characteristics of lung cancer deaths <1947-19781 among'male upstate New York tak plan--t w:--or1kerr&s , ----- Cnaos.e eDmuprlaotyiomneonft Dahtiereof Yaegaersatof hye, emPrpilooryrmjseknt Date ol , i 8,0 days 1948 57 2 19.0 days 1949 43 3 2.0 months 1948 31 4 2.5 months 1948 26 5 3.0 months 1948 24 6 11.0 months 1948 34 7 2.5 years 1950 30 3 2.8 years 1948 39 9 3.8 years 1949 27 10 12.0 years 1953 37 11 16.8 years 1956 45 12 17.5 years 1952 25 t Also 5 months employment at another talc mine, t Also 12 years employment at another talc mine. $ No prior work history on job application (see text). linclassifiablel Unclassifiabie None ___M eal mining _ . Metal mining Metal mining* Metal mining Metal mining Metal mining Paper miDJ Metal mining construction Paper mill 1970 1970 . 1971 1971 1970 1973 1974 1964 . 1961 . 1975 1973 .i 76 is significantly increased for short-term employees (SMR = 317) but not for longer term employees (SMR = 193), and is sig nificantly increased for those with prior risk employment (SMR = 316) but not for those whose prior work history indicates no such employment (SMR = 117). Several factors might explain these find~ ings, including significant exposures occur ring prior to (or subsequent to) employ ment at this talc plant, greater exposures for those with the short period of employ ment at this talc plant, differences in smok ing habits, or perhaps some other factors that distinguish the short-term employees. Analysis of prior work histories as dis cussed above demonstrates that employ ment in industrial jobs associated with an increased lung cancer risk prior to employ ment at this talc plant may be significantly related to death from lung cancer. Little information on possible exposures after leaving this employment is available for evaluation. It is possible that unusual working-con ditions existed during the time of employ ment of the short-term workers, as opposed to that of employees who remained longer at the talc plant, but support of this possi bility Is not evident. It might be that work 1208 LAMM ers who stayed for a shorter time began with dustier jobs, but the distribution of initial job assignments was no different for those who left within one year than for those who remained employed for at least one year. The lung cancer mortality was found primarily among miners (.nine of 12 cases) rather than millers, although mea sured exposures were greater for the mill ers. The risk of lung cancer did not appear strongly related to the date of hire. Five of the six deaths from lung cancer in short term employees occurred in workers who had worked three or fewer months, with two having had, respectively, only eight and 19days employment at this talc plant (table 5). The presence of unusual working con ditions during these short employment time periods thus appears to be an unlikely explanation of our findings. Variations in other potentially confound ing variables in short-term workers, such as cigarette smoking habits, are also possi ble. Limited anecdotal information on these cases indicates that those who were known to be smokers were considered to be heavy smokers. Additional data collection might assess whether the smoking habit for these workers was greater than for com parison groups and whether the smoking habit of short-term workers differed from those of longer term workers. Other sociobehavioral or environmental factors that affect subsequent cause and age ofmortality may distinguish short-term employees from long-term employees. The "healthy worker effect" is usually applied to the distinction between the ever em ployed and the never employed or the gen eral population (22). There may also be a "short-time employment effect" or a "very short time employment effect"on mortality patterns of workers. Gilbert (23) found in hi.s Washington State study of nuclear -plant employees that the greatest risks for cancer and noncancer' mortality" were among those employed less than two years. We found half the lung cancer deaths in our study of talc workers to be in those employed less than one year and one-third ET AL. in those employed less than three months. Shindell et al. (24) found no differences in mortality risk by duration of employment in their study of chemical workers; how ever, their study was restricted to workers employed for three months or more. Fur ther studies should be done to determine the differential mortality pattern of short term or very short-term employees. The adverse health, behavior, or work-related practices of persons who remain at a spe cific company for only a short period of time may differ from the practices of those who remain for a longer period of time or those who were not hired. In this paper, our method has been to determine the vital status of all persons ever employed and then to analyse and assess the results by duration of employment. Mortality from noninfectious, nonneoplastic respiratory disease was used as a surrogate measure of fibrogenic lung dis ease in this population. If the talc exposure experienced by workers at this plant is be lieved to be both fibrogenic and carcino genic, one would expect a similar exposurerelated mortality pattern for both lung can cer deaths and noninfectious, nonneoplastic respiratory disease. However, unlike lung cancer mortality, the mortality risk for noninfectious, nonneoplastic respira tory disease (table 4) is concentrated in those employees who worked more than one year (SMR = 370) and in those whose prior work history did not indicate prior risk (SMR = 666). Analysis of the mortality risk by duration of employment reveals that the risk from lung cancer in this study decreased with duration of employment while the risk from noninfectious, nonneoplastic respiratory disease increased with duration of employ ment. This suggests that the sources of these risks are different. C onclusions The increase in lung cancer mortality observed in a cohort of talc workers has Jjeen. shown to be concentrated in short term employees. This increased lung cancer ANALYSIS OF LVNG CANCER risk in short-term workers is most likely due to risk acquired elsewhere, such as prior employments, or to differences in smoking experience or other behavioral character istics. The inclusion of short-term employ ees in this occupational epidemiologic study has affected the apparent relation observed between exposure and outcome in a mag nifying rather than a diluting manner. The analysis in this study suggests that the lung cancer risk demonstrated in this cohort may reflect risks acquired before this, em ployment rather than during this employ ment. The assessment of risk by duration of employment can reveal unusual patterns that may indicate confounding variables. References 1. Dreossen WC. Effects of certain silicate dusts on 2. tPhoerr|uonFg\sY. J. PlnadttoHnygJR19, 3H3:o1b.5b.6s6A-7A3.. Pneumocon i1o9s4is2:4in7:5t0h7e-24ta.ic industry. Am J Roentogenol 3. Siegal \V. Smith AR, Greenburg L. The dust haz oalrodgiicnaltrefimndoilnitges tainlc tmalciniwngo,rkienrcsl.udAimng JroReonetgnetan- 4. Kgelneionlfe1l9d43M:4.9M:1e1s-s2i9te. J.Tabershaw 1R. Talc pneu moconiosis. AMA Arch lnd Health 1955:12:66-72. 5. KtalleiitnyfaemldoMng, tMalcesmsiitneeJr.s Kanodovmmilalners0 .ineNt aelw. MYoorrk 6. sKtalerien.fAelrdchME. nMviersosniteHJe,alZthak1i9M67H:1.4:M66o3r-t7a.lity ex perience among talc workers: a follow-up study. J 7. KOlceciunpfeMldeMd .19M74e:s1s6it:e34J5,-5L9i.nger AM. A study of workers exposed to asbestiform mineral in com mercial talc manufacture. Environ Res 1973; 6:132-i3. 8. DIne:mOecnctuJpMat,iZonuamlwexalpdoesRurDe.tEontvailrconcmonetnatianlinstgudays. bestos. NIOSH Technical Report February 1980:5-10. (DHEW Publication no. 80-115). 9. Gamble JF, Fellner W. DeMeo MJ. Cross sec IN SHORT-TERM EMPLOYEES 1209 tional morbidity study. In: Occupational exposure to talc containing asbestos. NIOSH Technical Report. February 1980:19-27. (DHEW Publica 10. Btioronwntot .D80P-,11W51a.goner JK. Retrospective cohort study of mortality. In: Occupational exposure to tale containing asbestos. NIOSH Technical Re pnoo.rt8.0F-1e1b5r)u.ary 1980:29-33. (DHEW Publication 11. Stille WT, Tabershaw IR. The mortality experi ence of upstate New York talc workers. J Occup 12. BMleodck1G98,2M:2a4t:a4n8o0w-4-s.ki GM, Seltser RS. A method for estimating year of birth using social security number. Am J Epidemiol 1983;113:377-95. 13. Axeison O, Rehn M. Lung cancer in miners. Lan cet "1971:2:706-7. 14. CWoLst.eRlloomJ. MWoLr.taMlietyrcohfamnteJtaAl,meidnse.rHs.eIanl:thWiasgsuneesr related to metal and nonmetallic mining. Boston: 15.. LBluotytderwJWor,thL, o1n9g8-3t:e2r2m7-4m0.ortality study of steel workers. V. Respiratory cancer in coke plant , workers. J Occup Med 1971:13^3-68. - 1-6.i Pfoaulmnderry\wVoGr,keSrcso: ua WrevDie.wL.uAnmg clanndceHrvign AfessrroocuJs . _ 1981:42^329-40. 17. HInahmamlatoionnd oEfCb.enSzeplyikreonffe Ia.nJ.d Lcaanwctehreirn PmLa.n.eAt naln. NY Acad Sci 1976:271:116-24. 16- Stern RM. Assessment of risk of lung cancer for 19. Hwealrdreinrsg.toAnrcJhME.nBvliorot n\VHJ,eHalothov1e9r8R3.N38, :e1t4a8l-.5L5.ung cancer in coastal Georgia: a death certificate " analysis of occupation. JNC1 1978;60:295-8. 20. International Agency tor Research on Cancer. iARC Monograph on the Pulp and Paper Industry 21. M19o8n0s;o25n:1R91R-.2.Analysis of relative survival and proportional mortality. Camput Bio Med Res 22. M19c7M4;i7c:h3a2e5l-3A2J.. Standardized mortality ratios and tthhee s"uhrefaaclteh.yJ wOoccrkuepr Mefefdec1t"9:76sc;1r8a:t1c6h5in-8g. beneath 23. Gilbert ES. Some confounding factors in the study oEfpimdeomrtiaollit1y98a2n;d11o6c:1c7u7p-a8t8io. nal exposures. Am J 24. .SwbolnrkdeerlleSf,feWcte"i--sbfearcgt RoFr, GariteiffaerctE?EJ. TOhecc"uhpeaMltheyd 1978;20:807-11.