Occupational and Environmental Medicine 1994;51:541-547

Cancer mortality among workers in the meatdepartment of supermarkets

Eric S Johnson

Environmental andMolecularEpidemiology Section,Epidemiology Branch,National Institute ofEnvironmental HealthSciences, P 0 Box12233, ResearchTriangle Park, NorthCarolina 27709, USAE S JohnsonCorrespondence to:Dr E S Johnson,Department of Biostatisticsand Epidemiology, Schoolof Public Health andTropical Medicine, TulaneUniversity Medical CenterSL18, 1430 Tulane Avenue,New Orleans, LA 70112,USA.

Accepted 8 April 1994

AbstractObjectives-The aim was to study therisk of dying from cancer among workersin the meat department of supermarketspotentially exposed to oncogenic retro-viruses and fumes during the wrapping

and labelling ofmeat.Methods-Cancer mortality for theperiod 1949 to 1989 was compared in a

previously studied cohort of 10 841 mem-bers of a local meatcutters' union inBaltimore, Maryland who worked in themeat department of supermarkets, afteran extended follow up of nine years(1981-9). Person-years and deaths were

apportioned in five-year intervals by sex,

age, and calendar year, and standardisedmortality ratio (SMR) and proportionalmortality ratio (PMR) analyses were con-

ducted. The United States general popu-

lation was used as the comparison group.

Analyses of SMR and PMR were alsoconducted for a control group of workersfrom the same union who worked exclu-sively in non-meat companies.Results and discussion-Among women,

an SMR of 1-6 (95% confidence interval(95% CI) 1.1-2.2) and a PMR of 1-5 (95%CI 1-0-2-0) for lung cancer were found.For men, the SMR for cancer of the buc-cal cavity and pharynx was 1-8 (95% CI1-0-3-0), and for colon cancer it was 1-5(95% CI 1.1-2.1). The respective PMRswere 1-9 (95% CI 1-1-3.1) and 1-5 (95% CI1-1-2-1). Whereas the role of non-occu-

pational factors needs to be taken intoaccount before occupational factors can

be implicated in the occurrence of theexcess of cancer of the buccal cavity andpharynx, and colon cancer in men, thereis reason to suspect that occupationalfactors may be responsible for the lungcancer excess in women. Thus exposuresthat occur predominantly in women,such as exposure to fumes during wrap-ping and labelling, should be investigatedas to their role in this excess.

(Occup Environ Med 1994;51:541-547)

Excess risk of certain cancers, particularlylung cancer, has consistently been reportedamong workers in the meat industry. 1-12

Occupational exposures with carcinogenicpotential included (1) the oncogenic virusesof cattle and chickens'317: (2) fumes emittedwhen heat is applied through a hot wire or rod

to cut the plastic film used to wrap meat, andwhen heat is applied to a plate or wire ele-ments to activate the adhesive in price labels.These fumes contain hydrogen chloride, plas-ticisers phthalates and adipates, phthalicanhydride, and small amounts of benzene andother hydrocarbons'8-21; (3) nitrosaminesformed during the curing of meat22; (4) theantioxidants butylated hydroxytoluene (BHT)and butylated hydroxyanisole (BHA) used aspreservatives23 24; and (5) polynuclear aro-matic hydrocarbons (PAHs) formed duringthe smoking of meat.25We previously studied mortality for the

period 1949-80 in a cohort of 28 900 mem-bers of a local meatcutters' union inBaltimore, Maryland.lA The union drew itsmembership from the meat department ofsupermarkets, abattoirs, meat packing plants,and chicken slaughtering plants, and fromoutside the meat industry (non-meat) such asthose working in the soft drinks and soupmanufacturing, oyster shucking industries etc.The population comprised subjects who werealready members of this union as from 1 July1949, and subjects who joined subsequentlyuntil 1979. Membership in the union wascompulsory within 30 days of the start ofemployment.

Certain cancers including lung cancer werein excess. In men, the lung cancer excess wasrestricted to those working in abattoirs andmeatpacking plants. In women, the lung can-cer risk was much higher than that in men,and was more generalised, occurring not onlyin those working in abattoirs and meatpackingplants, but also in those working in chickenslaughtering plants, and in the meat depart-ment of supermarkets.2 We therefore con-ducted a nested case-control study of lungcancer within the cohort.5 Fairly strong evi-dence was provided from this study linkingexcess lung cancer in men with putative expo-sure to oncogenic viruses. The study was toosmall, however, to adequately investigate theother occupational risk factors such as curingand smoking of meat, fumes during wrapping,etc. Also, the risk in women was not investi-gated specifically because of small numbers.A study of workers in the meat department

of supermarkets would potentially offer aunique opportunity for investigating specifi-cally the role of exposure to fumes duringwrapping and labelling in the occurrence ofexcess cancer in the meat industry. The rea-son, is that processing activities such as smok-ing and curing of meat are typical of meatpacking plants, and do not normally occur in

541 on 28 M

ay 2018 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.8.541 on 1 August 1994. D

ownloaded from

52ohnson

supermarkets. Hence exposure to nitros-amines, PAH, BHT, and BHA would beabsent in supermarkets, except possibly fornegligible amounts of residues in the meatitself. On the other hand, useful informationon the role of the remaining two candidateexposures (oncogenic viruses and fumes dur-ing wrapping and labelling), may be reliablyobtained, if cancer risk is studied separately inmen and women in the meat department ofsupermarkets, for the following reasons: Inthe meat department of supermarkets, there isa difference in the tasks men and women do.The department ordinarily consists of threesections: (a) a raw meat section where meatreceived from abattoirs or meat packing plantsis cut up into smaller customised parts,weighted, wrapped, and labelled; (b) a deli-catessen section where partially or fullycooked meat is cut, weighted, wrapped, andlabelled. Some cooking such as barbecuing ofchickens in closed electrically heated vesselsmay be carried out in situ; (c) a seafood sec-tion where weighing, wrapping and labellingare also done. In the delicatessen and seafoodsections men and women do exactly the samejobs. In the raw meat section, however, menand women work side by side, and handle thesame meat, but their jobs differ. Women arealmost exclusively engaged in wrapping andlabelling of meat and spend virtually all theirtime in this activity, whereas men are primarilyengaged in meat cutting, and will only occa-sionally wrap meat when there is a functionalshortage of staff.

Because of these sex differences associatedwith meat cutting and meat wrapping activi-ties, excess cancer seen in women, but appre-ciably less or absent in men, would likely beassociated with meat wrapping and labellingactivities, whereas similar excess risk in menand women or a slightly higher risk in men,would likely be associated with exposure tothe other candidate-namely, oncogenicviruses. (The intensity of exposure to onco-genic viruses is expected to be either the samefor men and women, or slightly higher inmen; men and women handle exactly thesame meat, except that those men who aremeat cutters are more prone to cuts. Hencefor them, there is an increased likelihood forviruses to penetrate the skin.

Before 1975, exposure to fumes duringmeat wrapping and labelling was a problem insupermarket workers in the United States. Itcaused acute noxious effects on the respira-tory mucosa and generally referred to asmeatwrappers' asthma.26 Different plasticshave been used to wrap meat, starting withcellophane in the early 1 950s, rubberhydrochloride, polyethylene, polystyrene,polyvinylidene chloride, and polyvinyl chlo-ride (PVC), which has been in use since1963.2728 Benzene is a thermal decompositionproduct of virtually all of the different plasticsused to wrap meat.2930 Smoke produced byburning PVC is believed to be mainly due tothe presence of aromatic pyrolytic decomposi-tion products, especially benzene." The majorcomponents resulting from the pyrolysis of

PVC or polyethylene resin include hydrogenchloride, benzene, toluene, and PAHs such asnaphthalene.'83' Constituents of fumes fromPVC film pyrolysis, include, hydrogen, chlo-ride, phthalates, phthalic anhydride, adipates,benzene, PAHs such as naphthalenes, andother hydrocarbons.'8 21 31 32 Thus these fumescontain low concentrations of at least threecompounds that are known carcinogens: ben-zene, which causes leukaemia and possiblylymphoma in humans,3334 and is a multicar-cinogen in animals, causing a variety oftumours including those of the haemopoi-etic/lymphatic systems and lungs35; phtha-lates, which cause hepatocellular carcinomasand possibly myelomonocytic leukaemia inrodents'6; and PAHs, such as benz(a)pyrene,which induced lung tumour and leukaemia inanimals,'7 and are constituents of tobaccosmoke, hence contributing to lung cancer inhumans. Exposure to coal tar pitches and coaltar that contain PAHs results in the develop-ment of lung cancer, leukaemia, and othercancers in humans."'8

Benzene concentrations in area air sam-pling of up to eight hours were 0-02-0-04parts per million (ppm; 0-06 to 0-14 mg/m3),and personal air concentrations of as much as0 03 to 0-16 ppm (0-1-0-5 mg/m3n) have beenrecorded in connection with use of a labellingmachine.'9 Similar values when extrapolatedover an eight hour shift were obtained in asimulated experiment cutting PVC plasticfilms.32 In a biomarker study of meat wrappersand meat cutters in supermarkets we detectedconcentrations of muconic acid (a metaboliteof benzene) in the urine of a few workers,compatible with exposure to 1 to 2 ppmbenzene (unpublished data). These concen-trations are important, as they approach levelsof exposure at which benzene causes cancer inhumans.404' Although subject to some statisti-cal uncertainty, current risk estimates indicatethat exposure to as low as 1 to 5 ppm benzenefor 40 years results in a threefold risk ofleukaemia in humans,40 and to 1 ppm for 15years results in a twofold risk of tumours ofthe haematopoietic/lymphatic system.4'

Phthalates have been reported in thesefumes at concentrations of up to 3-7 mg/m3.39Although specific PAHs such as naphthaleneshave been found in these fumes, actual valuesare not available.We report here on cancer mortality after an

additional nine years of follow up (1981-9) inthe subset of the Baltimore cohort, consistingof 10 841 men and women who ever workedin the meat department of supermarkets, andthe comparison group of 6265 men and womenin the same union who worked exclusively innon-meat companies. The total number ofdeaths in supermarket workers for the first fol-low up (1949-80) was 865. In this extendedfollow up a further 766 deaths were recorded.This paper reports on standardised mortalityratio (SMR) and proportional mortality ratio(PMR) analyses of the total of 1631 deathsthat had occurred among supermarket work-ers and 1202 deaths in non-meat workers, dur-ing the combined follow up period of 1949-89.

542 on 28 M

ay 2018 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.8.541 on 1 August 1994. D

ownloaded from

Cancer mortality among workers in the meat department ofsupermarkets

MethodsThe original cohort has been described indetail before.' As we had conducted a com-prehensive follow up in the first study for theperiod 1949-80,' 2 for this extended follow uponly the National Death Index and SocialSecurity Administration Mortality File wereused. Also, we manually searched for match-ing death certificates at the State Departmentof Vital Records, from all identifying informa-tion we had on our records, for the rest of thecohort who were not identified as dead bythese two methods (including those withmissing date of birth or social security num-ber). After this, persons not identified as deadfrom these death file searches were assumedto be alive. Based on the methods of follow upwe had employed, death ascertainment wasregarded as virtually complete.

Although the records for this union werecomplete to the extent that the record of any-one who had ever been a member could beretrieved, the race and sex of members werenot routinely recorded and date of birth wasmissing for a few members. Thus althoughthese items were available for all dead subjectsfrom their death certificates, they were miss-ing for persons not known to have died. Wecarried out both SMR and PMR analyses, asthis deficiency was a factor in the SMR butnot in the PMR analysis. We carried out thefollowing procedures for the SMR analysis:For persons not known to have died, (1) werandomly assigned a race to each subjectbased on the racial distribution of dead sub-jects with death certificates.' This distributionwas found to be virtually the same among asample of current members of the union; (2)sex was determined based on first name; (3)date of birth was artifically assigned to 672subjects in supermarkets and 668 subjects innon-meat companies without this informa-tion, as described previously.'

Cause specific SMRs and PMRs were cal-culated with the O/E system computer pro-gram developed jointly by the NationalCancer Institute and IMS Inc as a modifica-tion of the Monson program. The referencepopulation was the United States general pop-ulation.42

In the PMR analysis, the expected numberof deaths for a given cause was based on theproportion of all deaths due to that cause inthe United States population multiplied bythe total number of all deaths in the studygroup, by age, calendar time, race, and sex.The PMR is given as the ratio of the observednumber of deaths divided by the expectednumber. For significance testing, the Mantel-Haenszel X2 test with one degree of freedomwas used; the variance was determined by thePoisson approximation (the mean of theexpected deaths). Ninety five per cent confi-dence intervals (95% CIs) were based onMietinnen's test based formula.42 In the SMRanalysis, all subjects without death certificates(which includes all subjects with missingdemographic information) were assumed tobe alive. Expected deaths for a given causewere calculated by partitioning person-years

at risk within the study group into age, calen-dar time, sex, and race categories and multi-plying by the respective rates for that cause inthe United States reference population.Enumeration of person-years began as from 1July 1949 for subjects who were already in theunion before this date, or from the date of firstpayment of dues for subjects who joined-sub-sequently, until 31 December 1989, or untildeath, whichever was earlier.

In the enumeration of person-years in theSMR analysis, date of first payment of uniondues was used as a surrogate of date of firstexposure or employment, because, as men-tioned, union membership was compulsorywithin 30 days of date of employment. Thiswas appropriate only for subjects who firststarted work in a company after it joined theunion. This measure is not valid for subjectswho had worked in companies for a variablenumber of years before their companiesjoined the union. Such workers would haveaccumulated in some cases significant dura-tions of exposure before coming under obser-vation. Such a group has been referred to as a"prevalent" or "cross sectional" cohort. It hasbeen shown that bias may result in risk esti-mation when a prevalent cohort constitutespart of the study population.4344 To examinethis in the SMR analyses, we further selecteda subset of 7740 supermarket workers whowere first employed at least one year aftertheir companies had joined the union-thusall of these workers would have come underobservation from their date of first employ-ment or exposure, and constitute an "inci-dent" cohort.43We prefer to present the full results using

the SMR analyses, even though they are basedon incomplete demographic data, for the fol-lowing reasons: (1) The SMR does not havethe inherent disadvantage of a PMR, which isthat the investigator does not know whethercause specific PMR results have underesti-mated or overestimated the true risks, becauseof the dependency of these PMRs on the over-all mortality in the study population com-pared with that in the comparison group.4546(2) As the racial distribution of the union hasnot changed with time, it was reasoned thatthere should be negligible error, in randomlyassigning race as we did, as the proportions ofperson-years assigned to each race for subjectsnot known to have died would closely resemblethe true ones. (3) The SMR analysis allowedmore readily for more detailed analyses suchas examination by year of termination ofemployment, year of first employment, inci-dent cohort etc. (4) The SMR analyses per-mitted the determination of the direction ofbias in the PMR results, which was based oncomplete data.

ResultsTable 1 gives the race and sex distributions ofall deaths in the supermarket workers for eachperiod of follow up.

Table 2 gives the cause specific SMRresults. For men in the meat department of

543

on 28 May 2018 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.51.8.541 on 1 A

ugust 1994. Dow

nloaded from

4Johnson

Table 1 Distribution of deaths by race and sex for each follow up period

Racelsexgroup

White men(6343)* [2081]tNon-white men(563)* [1822]tMen(6906)* [3903]tWhite women(3749)* [1167]tNon-white women(186)* [1195]tWomen(3935)* [2362]tAll race and sexgroups(10 841) [6265]

Previous follow up

Supermarketsl Non-meatgrocery stores companies

676

57

733

177

10

187

920

236

216

452

136

145

281

733

Present study

Supermarkets Non-meatgrocery stores companies

1112

92

1204

414

13

427

1631

3951

340t

735:

2311:

236t

4671

1202t

*Estimated number of subjects in supermarkets (since race was estimated)tEsumated number of subjects in non-meat companies (since race was estimated)tIncludes subjects who worked in both meat and non-meat companies.

supermarkets, statistically significant excessesof cancer of the buccal cavity and pharynxand colon cancer were found. For women,only lung cancer was significantly in excess.(In supermarket women, the results reflect

Table 2 Standardised mortality ratios (SMRs) for subjects who ever worked in the meatdepartment ofsupermarketslgrocery stores, andfor those who worked exclusively in non-meat companies

Cause of death(ICD)

All causes

All cancers(140-209)

Buccallpharynx(140-149)

Stomach(151)

Colon(153)

Pancreas(157)

Lung(162)

Breast(174)

All uterus(180-182)

Other genital(183-184)

All genital(180-184)

Prostate(185)

Lymphoma reticulosarcoma(200)

Hodgkin's disease(201)

Myeloma(203)

Leukaemia(204-207)

Other lymphatic(202, 203, 208)

All haematopoeitic/lymphatic(200-209)

Supermarket

Men Women

0.9 1-0[0-9-0-99]* [0-9-1*1](1204) (427)1-0 1-0[0-9-1-2] [0-9-1-2](290) (153)1-8 1-5[1.0-3.0]* [0*3-4-4](15) (3)1-0 0-9[0-5-1 8] [022-27](12) (3)1-5 06[1.1-2.1]* [0-3-1-2](36) (8)1-0 0-9[0-6-1-7] [0 3-2 0](14) (5)11 16[0-9-1-3] [1-1-2 2]*(102) (40)- 07E = 0-4 [0-5-1-1]

(26)NA 1 3

[0 7-2-3](14)

NA 1.1[0 6-2 0](12)

NA 1-2[0-8-1-8](26)

1-2 NA[0-7-1-8](19)0 7[0-1-2-0](3)0-6[0- 1-2-2](2)1-3[0 4-3 0](5)0*5[0-2-1-2](6)0-7[0-2-1-4](6)0-6[0.3-0.95]*(17)

1*1[0-1-3-9](2)0 9[0 0-5-0](1)1 -6[0 3-4-8](3)1-3[0-5-2-8](6)1-6[0 6-3.3](7)1-4[0 8-22](17)

Non-meatt

Men Women

09 08[0.9-0-99]* [0-7-0.9](710) (455)1-0 0-9[0-9-1-2] [0-7-1-0](160) (126)1 7 05[0 8-3-2] [0-0-2 6](10) (1)09 2-0[0-4-1 9] [0 9-3 6](7) (10)0 9 1-3[0-4-1-6] [0 8-2 0](10) (18)1.1 04[0-5-2-1] [0- 1-1-2](8) (3)13 12[1-0-1-6]* [0-8-1-8](69) (25)- 0*7E = 0 2 [0 4- 11]

(21)NA 0 9

[0-5-1-5](14)

NA 1.0[0-5-1-9]9

NA 0 9[06-1 4](23)

0 7 NA[0-3-1 -4](8)

E= 1 9

0-7[0 0-3 7](1)

E = 2-5

0-8[0.2-1-9](4)0-2[0-0-1*1](1)0-4[0 2-0-9]*(6)

E = 1 5

E = 0-8

1-1[0-2-3-3](3)

E = 4*1

0 7[0-1-1*9](3)04[0-1-0-9]*(4)

mostly the experience of white women, as

there were only two cancer deaths in non-white women.)When the cause specific SMRs given in

table 2 for male and female supermarketworkers were compared with the correspond-ing cause specific PMR results (not pre-

sented), they were all found to be virtuallyidentical. When these were compared withcause specific results from the incident cohortanalysis, again virtually identicial results wereachieved, even when the groups were brokendown by race and sex. There were two excep-tions: the SMR for cancer of the uterus inwhite women in supermarkets of 1-5 (95% CI0-8-2 5) was similar to the correspondingPMR of 1*5 (95% CI 0O8-2-6), but differentfrom the corresponding SMR of 2-3 (95% CI1.2-4.0) in the incident cohort. This may beof no relevance as the SMR in the controlgroup of non-meat white women was also 1-5(95% CI 0-6-3-1). In non-white supermarketmen, although the SMR of 2-5 (95% CI0-7-6-4) and that of 2-2 (95% CI 0-0-12-3) inthe incident cohort for stomach cancer were

similar, the corresponding PMR was 3-7(95% CI 1.0-9-4) (table 3). It is likely thatthis PMR is artifically inflated, resulting fromthe lower all causes mortality in non-whitesupermarket workers (SMR = 0-6) comparedwith the United States general population.

Table 3 compares statistically significantresults of the two SMR and the PMR analysesin this study, with those obtained in the previ-ous follow up, by race and sex. To rule outany bias in the SMR results due to artificialassignment of race, further analyses (worstcase scenario) were carried out in which allsupermarket workers were classified as whitesby sex, and then as non-whites by sex, usingthe appropriate United States reference deathrates each time. In Table 3 it is seen that forcancer of the buccal cavity and pharynx inmen, for which there are pronounced racialdifferences in the United States rates, theSMR results obtained for supermarket work-ers were similarly divergent. On the otherhand, for colon cancer in men and lung can-

cer in women, for which the correspondingrates in the United States population were

similar in both races, the results were essen-

tially unchanged. These findings suggest thatthe possibility of any bias resulting from our

artifically assigning race in the main SMRanalyses can be ruled out for the findings ofexcess colon cancer in male and excess lungcancer in female supermarket workers.

Table 4 gives the lung cancer SMR resultsby calendar year of termination of employ-ment (exposure), year of first employment,and calendar year of death, in the incidentcohort. Examination of risk by year of termi-nation and by year of first employment showsthat for men, no significant excess risk wasfound for those who worked in supermarketsafter 1954. By contrast, despite the muchsmaller number of deaths in women, therewas a statistically significant excess of lungcancer in women who stopped work as earlyas 1955-9 and as late as 1970-4. Similarly,

* p < 005(X( test).tExcludes all subjects who ever worked in the meat industry.tFigures in parentheses are numbers of observed deaths; figures in brackets [ ] are 95% CIs;E = number of expected deaths; NA = not applicable.

544

on 28 May 2018 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.51.8.541 on 1 A

ugust 1994. Dow

nloaded from

Cancer mortality among workers in the meat department ofsupermarkets

Table 3 Summary of statisticaly significant results for supermarket workers compared with the previous follow up, andthe incident cohort, and classifying al subjects as white or non-white by sex

SMR

Present study

All AllSupermarket PMR subjects subjectsworkers classified classifiedRacelsex Cancer Previous Present Previous Full incident as white asgroup site follow up study follow up cohort cohort by sex non-white

White Buccal NA 1.9 1-8 1 9 2-4 1-9 0 9men cavity and [1-1-3-3]* [0.7-3.6] [1-1-3-2]* [0.9-5-2] [1-1-3 2]* [0-5-1-5]

pharynx n=14 n=7 n=14 n=6 n=15 n=15Colon NA 1-6 1-8 1-6 14 1-5 1-5

[1*1-2-2]* [1 0-2-8]* [1*1-2-2]* [0-7-2-5] [1*1-2 1]* [1*1-2-1]n=35 n= 18 n = 35 n= 10 n= 36 n = 36

White Lung 2-2' 1-5 19 1-7 1-7 1-6 1-7women NA [1.1-2.1]* [1.0-3.1]* [1-2-2-3]* [1-0-2 6]* [1-2-2-2]* [1-2-2-4]*

n = 40 n= 14 n = 40 n = 20 n = 40 n = 40

*Statistically significant at the 95% confidence level. Figures in brackets [ ] are 95% CIs; NA = not available.

there seems to be an increased risk for womenfirst employed for each interval between 1950and 1974. Examination of risk by calendaryear of death shows that in both sexes, risk oflung cancer began to fall after 1974.We refer briefly to results for tumours of

the haematopoietic and lymphatic systemsbecause we were especially interested in thesetumours, as they are the sites commonlyinvolved in virally induced tumours in cattleand chickens, and in benzene induced cancer.

Although there was a definite deficit oftumours of the haematopoietic and lymphaticsystems (ICD 8th revision 200-209) in super-market men which was also seen in the con-

trol group, women in supermarkets seem toshow an excess (SMR = 1 4), even whencompared with women in the control group(SMR = 0A4; 95% CI 0-1-0.9)). This is moreapparent, considering that even though thetotal number of all deaths among men of 1204was almost three times the total number ofdeaths in women of 427, the total number ofdeaths from tumours of the haematopoieticand lymphatic systems in both sexes was 17each. There were four lymphomas in mencompared with five in women, and fivetumours of the erythroid and myeloid series inmen compared with eight in women.

Although these data are limited, they seem toshow an excess of these tumours in womencompared with men.

DiscussionThe consistency between the PMR and SMRresults (including those of the incidentcohort) indicates that the results overall prob-ably approximate closely to the true risks, andany bias is likely to be minimal.

Previous studies have reported an excess

risk of cancer of the buccal cavity and pharynxamong butchers, meat cutters, and workers inmeat packing plants.'47 The excess in super-market men in this study, however, may be ofno relevance, as a similar risk was also seen inthe control group of male non-meat workers,and thus may simply reflect a high back-ground rate of this disease in the study area.

The general population of Baltimore city fromwhich most of the study population is derived,is known to have a significantly higher rate ofcancer of the buccal cavity and pharynx thanthe United States general population.48

Similarly, the excess risk of colon cancer inbutchers and meat cutters, and in foodprocessors has been reported before.'249 Thecause of the excess of colon cancer in male

Table 4 Lung cancer SMRs for incident cohort, by year of termination of employment, by year offirst employment, andby calendaryear of death

White men White women

SMR for SMR for SMRfor SMRforsubjects who subjects SMR for subjects who subjects SMRforterminated who were first subjects terminated who were first subjects

Period employment employed who died employment employed who died

1950-4 3-2 2-1 - 11 2-4[1-0-7-4]* [0 9-4-2] E = 0 0 [0-0-6-3] [0-8-5-6] E = 0.0

1955-9 07 07 - 2-8 1-6 24-5[0 2-1-4] [0 3-1-2] E = 0 2 [1-2-5.5]* [0 7-3 2] [0-140]

1960-4 0-6 1-3 - 05 1 1[0-2-1-4] [0-6-2-4] E = 0-7 [00-2-8] [0-2-3-3] E = 0-2

1965-9 1-6 1.0 1-8 2-1 2-8 2-5[0-7-3-1] [0 4-2-0] [0-4-5.2] [0-4-6 0] [1 0-5 7]t [0-14]

1970-4 1-2 - 1-6 3-0 1-6 6-7[0-4-2-7] E = 2-4 [0 6-3-5] [1-24-62]* [0-2-5-5] [2-5-13.1]*

1975-9 07 - 11 10 - 09[0-3-1 5] E = 0 6 [0-4-2-3] [0 3-2 5] E = 0 4 [0-1-3-3]

1980-4 - - 09 - 08[04-1-8] [0 2-2 3]

1985-9 - - 07 - 1-7[03-1-3] [08-3 1]

*Statistically significant at the 95% confidence level.tApproaching statistical significance. Figures in brackets [] are 95% CIs; E = expected number of deaths.

545 on 28 M

ay 2018 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.8.541 on 1 August 1994. D

ownloaded from

6Johnson

supermarket workers in this study is notknown. Viral exposure is a possible candidatefor investigation as no excess was found inwomen. It has been shown experimentallythat the presence of wounds (which are morelikely occur in male meat cutters than infemale wrappers) facilitates the developmentof cancer induced by oncogenic retrovirusesof chickens.50 On the other hand, substantialdata identifying dietary factors, particularlyanimal fat and meat consumption, as risk fac-tors for colon cancer exist in the literature. Inone of these studies, it was noted that no vari-able could reduce the correlation with meatconsumption to less than 0 70.5' In view ofthe fact that one study has reported increasedconsumption of meat among butchers com-pared with the general population,49 the roleof non-occupational factors (particularly diet)needs to be taken into account in futureattempts to investigate the disease in this pop-ulation.

Excess risk for lung cancer has consistentlybeen reported among men in the meat indus-try.l 3-12 Evidence indicates that exposure tooncogenic viruses may be associated with thisexcess,5 1' although exposure to PAHs andnitrosamines in connection with smoking andcuring of meat could also be relevant.4952 Wehave been the only group to report on lungcancer risk in women, in our original reporton the entire Baltimore cohort. We found ahigh risk of the disease in women who workedin meat packing plants, chicken slaughteringplants, and the meat department of supermar-kets.2 In this report, we also noted excess riskin women in supermarkets, but not in the menwho work side by side with them. We there-fore suspect that exposure to fumes duringwrapping and labelling may be associated withincreased risk of this disease in women, for thefollowing reasons: (1) This is the only majortask that women perform in the meat depart-ment of supermarkets; (2) in the meat depart-ment of supermarkets in the United States,meat wrapping and labelling are predomi-nantly female activities-men usually do notwrap or label; when they do, it has been esti-mated that they spend no more than 15% oftheir time on this activity (DH Wegman,unpublished data); (3) as well as supermar-kets, this is the only exposure common towomen in abattoirs, meatpacking plants, andchicken slaughtering plants, the other areas inwhich this excess risk has been found; (4) the"hot wire" machine used for wrapping hasbeen in use since the early 1950s when plasticfilms were first used to wrap meat. (Not allsupermarkets in the study had this machine.)A significant reduction in exposure to wrap-ping and labelling fumes occurred in super-markets in the United States after 1975, withthe substitution of the "cold rod" in place ofthe "hot wire" in the wrapping machine. OurSMR results by year of termination, year offirst exposure, and year of death showing thatexcess lung cancer risk has been present insupermarket women from the 1950s till theend of 1974 seem consistent with this change.It is noted that this fall in risk is also mani-

fested by the fall in SMR for these women inour previous study2 from 1 9 to 1-7 in this fol-low up; (5) wrapping and labelling fumes con-tain benzene, PAHs, and phthalates that areassociated with the occurrence of lung cancerand leukaemia in humans or animals. Thusour finding of excess lung cancer and possiblyleukaemia and lymphoma in these women, isalso consistent.

For these reasons, it is appropriate to sus-pect that exposure to fumes during wrappingand labelling may be related to the excess lungcancer seen in female but not in male super-market workers. These results should beregarded as preliminary, however, as theyneed to be confirmed in other populations.Employment history was not updated for thisfollow up, and we did not control for tobaccosmoking. Because of these caveats, cautionshould be exercised in the interpretation ofthe findings. Studies that will take intoaccount non-occupational confounding fac-tors like tobacco smoking are needed to fur-ther investigate this apparent associationbetween exposure to fumes during wrappingand labelling, and occurrence of lung cancerin women.

Although use of the "cold rod" has defi-nitely resulted in substantially reducing expo-sure to fumes, the hot wire machine is stillbeing used in some supermarkets and in otherareas of the meat industry such as chickenprocessing plants and in other industries(personal observation). Also, incorrect use ofthe cold rod machine by operating it abovethe recommended temperature, a practiceknown to occur among some supermarketworkers, may still be associated with notice-able exposure to fumes; field measurementshave shown that about 10% of the time a"cool rod" was as hot as a "hot wire" (DHWegman, unpublished data). It has beenshown that operating the cool rod cutter attemperatures of 210'C or more results inemission that exceed those obtained with apoor practice hot wire technique.5' Thus thepotential for exposure to these fumes,although significantly reduced, has not beencompletely eliminated.

I thank Glen Heartwell, and Charles Knott of Survey ResearchAssociates, and Sukon Kanchanaraksa of the Johns HopkinsUniversity, for their valuable assistance in the conduct of thisstudy.

1 Johnson ES, Fischman HR, Matanoski GM, Diamond E.Cancer mortality among white males in the meat indus-try. 7 Occup Med 1986;28:23-32.

2 Johnson ES, Fischman HR, Matanoski GM, Diamond E.Occurrence of cancer in women in the meat industry. BrJf IndMed 1986;43:597-604.

3 Johnson ES. Mortality among nonwhite men in the meatindustry. J Occup Med 1989;31:270-2.

4 Johnson ES, Fischman HR. Cancer mortality amongbutchers and slaughterhouse workers. Lancet 1982;i:913-4.

5 Johnson ES. Nested case-control study of lung cancer inthe meat industry. J Nad Cancer Inst 1991;83:1337-9.

6 Fox AJ, Lynge E, Malker H. Lung cancer in butchers.Lancer 1982;i: 165-6.

7 Griffith W. Lung cancer in butchers. Lancet 1982;i:399.8 Lynge E, Andersen 0, Kristensen TS. Lung cancer in

Danish butchers. Lancet 1983;i:527-8.9 Doerken H, Rehpenning W. Lung cancer in butchers.

Lancer 1982;i:561.10 Reif JS, Pearce NE, Fraser J. Cancer risks among New

Zealand meat workers. Scand J Work Environ Health1989;15:24-9.

546 on 28 M

ay 2018 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.8.541 on 1 August 1994. D

ownloaded from

Cancer mortality among workers in the meat department ofsupermarkets

11 Caggon D, Pannett B, Pippard EC, et al. Lung cancer inthe meat industry. BrJInd Med 1989;46:188-91.

12 Siemiatycki J, ed. Risk factors for cancer in the workplace.Boca Raton, Baton Rouge, CRC Press, 1991:225-6.

13 Avian Leukosis. GF De Boer ed. Boston: MartinusNijhoff, 1987.

14 Witter RL. Reticuloendotheliosis virus. In: Hofstad MS,Barnes HJ, Reid WM, Yoder HW Jr, eds. Diseases ofpoultry. Ames, Iowa: Iowa State University Press,1984:406-17.

15 Payne LN, ed. Marek's Disease. Scientific basis and methodof control. Boston: Martinus Nijhoff Publishing, 1985.

16 Burny A, Mammerickx M, eds. Enzootic bovine leukosis andbovine leukemia virus. Boston: Martinus NijhoffPublishing. 1987.

17 Campo MS. Papillomas and cancer in cattle. Cancer Surv1987;6:39-54.

18 O'Mara MM. High temperature pyrolysis of poly(vinylchloride): gaschromatographic-mass spectrometricanalysis of the pyrolysis products from PVC resin andplastisols. Journal ofPolymer Science 1970;8:1887-99.

19 Takahashi T, Serizawa M. Pyrolysis of polyvinyl chloride.Plastics Industry News 1972;12:17-21.

20 Polakoff PL, Lapp L, Reger R. Polyvinyl choride pyrolysisproducts. Arch Environ Health 1975;30:269-71.

21 Vandervort R, Brooks SM. Polyvinyl chloride film thermaldecomposition products as an occupational illness. I.Environmental exposures and toxicology. J Occup Med1977;19:188-91.

22 Sen NP, Miles WF, Donaldson B, et al. Formation ofnitrosamines in meat curing mixture. Nature 1973;245:104-5.

23 Williams GM, Maeura Y, Weisburger JH. Simultaneousinhibition of liver carcinogenicity and enhancement ofbladder carcinogenicity of N-2-fluorenyl-acetamide bybutylated hydroxytoluene. Cancer Lett 1983;19:55-60.

24 Ito N, Fukushima S, Hagiwara A, Shibata M, Ogiso T.Carcinogenicity of butylated hyroxyanisole in F344 rats.J Natl Cancer Inst 1983;70:343-52.

25 Nordholm L, Espensen I, Jensen HS, Holst E. Polycycicaromatic hydrocarbons in smokehouses. Scand Jf WorkEnviron Health 1986;12:614-8.

26 Sokol WN, Aelony Y, Beall GN. Meat-wrapper's asthma. Anew syndrome? JAAMA 1973;226:639-41.

27 Vandervort R, Brooks SM. Health hazard evaluation deter-mination. December 1975. Cincinnati: US Departmentof Health, Education, and Welfare, National Institutefor Occupational Safety and Health, Center for DiseaseControl, 1975. (Report No 74-24,92,95-246.)

28 Encyclopedia of Polymer Science and Technology.Plastics, resins, rubber, fibers. New York: IntersciencePublishers, vol 9, 1968.

29 Madorsky SL. Thermal degradation of organic polymers.Robert E. Krieger Publishing Company, Huntington,New York: 1975:27-167.

30 Michal J, Mitera J, Tardon S. Toxicity of thermal degrada-tion products of polyethylene and polypropylene. Fireand Materials 1976;1: 160-8.

31 Quinn EJ, Ahlstrom DH, Liebman SA. Effect of structureof chlorinated vinyl polymers on smoke evolution.Presented at the American Chemical Society Meeting,1973. Polymer Preprints 1973;14:1022-7.

32 Cook W. Industrial hygiene evaluation of thermal degrada-tion products from PVC film in meat-wrapping opera-tions. Am Ind HygAssocJ 1980;41:508-12.

33 International Agency for Research on Cancer. Monographson the evaluation of carcinogenic risks of humans. Overallevaluations of carcinogenicity: an updating of IARC mono-graphs Volumes I to 42. Supplement 7. Lyon, France:IARC, 1987.

34 Adamson R, Seiber SM. Chemically-induced leukaemia inhumans. Environ Health Perspect 1981;39:93-103.

35 Maltoni C, Ciliberti A, Cotti G, Conti B, Belpoggi F.

Benzene, an experimental multipotential carcinogen:Results of the long-term bioassays performed at theBologna Institute of Oncology. Environ Health Perspect1989;82:109-24.

36 Kluwe WM, McConnell EE, Huff JE, Haseman JK,Douglas JF, Hartwell WV. Carcinogenicity testing ofphthalate esters and related compounds by the NationalToxicology Program and the National Cancer Institute.Environ Health Perspect 1982;45:129-33.

37 International Agency for Research on Cancer. Monographson the evaluation of carcinogenic risk of the chemical to man.Certain polycyclic aromatic hydrocarbons and heterocycliccompounds Volume 3. Lyon, France: IARC 1973.

38 International Agency for Research on Cancer. Monographson the evaluation of carcinogenic risk of chemicals tohumans. Polynuclear aromatic compounds, Part 4, bitumens,coal-tars, and derived products, shale-oils and soots. Volume35. Lyon, France: IARC 1985.

39 Frederick L, Wilcox T, Moseley C. Health hazard evaluationdetermination. Cincinnati: US Department of Health andHuman Services, National Institute for OccupationalSafety and Health, Center for Disease Control, PublicHealth Service 1981 (Report No HE 79-011-1011.)

40 Rinsky RA, Smith AB, Hornung R, Fillson TG, Young RJ,Okun AH, Landrigan PJ. Benzene and leukemia: Anepidemiologic risk assessment. N Engl J Med 1987;316:1044-50.

41 Wong 0. An industry-wide mortality study of chemicalworkers occupationally exposed to benzene. II. Doseresponse analyses. BrJ Ind Med 1987;44:382-95.

42 O/E System. Observed versus expected events. User guide-Version 3-7, January 1990. Epidemiology and biostatisticsprogram, National Cancer Institute and IMS Inc.

43 Johnson ES, Matanoski GM. SMR estimations in "preva-lent" cohorts and "incident" cohorts. Med Lav 1987;78:263-71.

44 Weiss W. Heterogeneity in historical cohort studies: asource of bias in assessing lung cancer risk. J Occup Med1983;25:290-4.

45 Wong 0, Decoufle P. Methodological issues involving thestandardized mortality ratio and proportionate mortalityratio in occupational studies. J Occup Med 1982;24:299-304.

46 Johnson ES. PMR and relative risk, (letter). Br J Ind Med1986;43:214-5.

47 Milham Jr S. Occupational mortality in Washington State1950-1971. Vol 1. Cincinnati: US Department of HealthEducation and Welfare, Center for Disease Control,National Institute for Occupational Safety and Health.1976. (Contract No CDC-99-74-26.)

48 Gittelsohn A. Mortality Surveillance Project (1979-1981).Comparison of mortality in the 700 geographic areas in theUS, with population greater than 10,000 for the period1968-1978. Report to the US National Center forHealth Statistics. (Unpublished.)

49 Guberan E, Usel M, Raymond L, Fioretta G. Mortalityand incidence of cancer among a cohort of self employedbutchers from Geneva and their wives. Br Y Ind Med1993;50: 1008-16.

50 Sieweke MH, Thompson NL, Sporn MB, et al. Mediationof wound-related Rous sarcoma virus tumorigenesis byTGF-beta. Science 1990;248:1656-60.

51 Amstrong B, Doll R. Environmental factors and cancerincidence and mortality in different countries, with specialreference to dietary practices. Int J Cancer 1975;15:617-31.

52 Johnson ES, Dalmas D, Noss J, Matanoski GM. Cancermortality among abattoir and meatpacking plant work-ers: an update. Am J Ind Med 1994 (in press).

53 Smith TJ, Cafarella ll, Chelton C, Crowley S. Evaluation ofemissions from simulated commercial meatwrappingoperations using PVC wrap. Am Ind Hyg Assoc J1983;44: 176-83.

547

on 28 May 2018 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.51.8.541 on 1 A

ugust 1994. Dow

nloaded from