Document Z8ymzk1ERYN64E18455YN8nEd

____ 316 Occup Environ .Wed 2000;57:3 16-324 Occupational exposures and pancreatic cancer: a meta-analysis I Ameli Ojajirvi, Tim0 J Partanen, Anders Ahlbom, Paolo Boffetta, Timo Hakulinen, Nadia Jourenkova, T h o P Kauppinen, Manolis Kogevinas, Miquel Porta, H a m U Vainio, Elisabete Weiderpass, Catharina H Wesseling Department of Epidemiology and Biostatistics, Finnish Institute of OccupationalHealth, TopeIiuksenk JlA, 00250 HekinLi, Finland I A Ojajini TJ Pananen T P Kauppinen Institute of EnvlrO~Clllpl Medicine, KaroIinska krstitutet, Stockholm, Sweden A Ahlborn C H Wesseling Institute ofMedical Epidemiology, Kprolinska Inatitutet, Stockholm, Sweden E Weiderpass Unit of Environmental Cancer Epidemiology, Internadoad Agency for Research on Cancer, Lyon. France P Boffcna Unit of Cbemoprrvention, International Agency for Research on Cancer, Lyon, France H L'V w i o Finnish Cancer Registry, Helsinki, Finland T Hakulinen Insdtut National de La SantC et de la Reserche Medicale, Villejuif, France N Jourenkova lnstitut Municipal d'hvestigacio .Medica, Barcelona, Spain .M Kogevinas .\I Porta Universidad S a c i o n d . Heredia, Costa Rica C H Wesseling Cnrreopondencc to Dr h n c l i O i a i s m aoia uoccuphcalth n k c p i e d 30 Dcrcmher I QoQ Abstract interactions between genes and the en&- Objectivedonsolidation of epidemio- ronment. logical data on pancreatic cancer and (Occup Envimn ,Med 2000;57:316-324) worksite exposures. Methods-Publications during 1969-98 Keywords: pancreatic cancer; occupational exposure; meta-analysis were surveyed. Studies without verified exposures were excluded. Meta-analyses were conducted on data fkom 92 studies Some 180 000 pancreatic cancers are regis- covering 161 populations, with results for tered annually in the world. It is highly and 23 agents or groups of agents. With a rapidly fatal and represents the fifth lea- standard format, five epidemiologists ex- cause of deaths from cancer in indusmalised tracted risk estimates and variables of the counmes and is 50%-100% more common in structure and quality of each study. The men than in women. It is not consistentlyasso- extracted data were cenwally checked. ciated with socioeconomic status within na- Random meta-models were applied. tional populations, although there is a tendency Results-Based on 20 populations, expo- toward higher age adjusted risk in richer than sure to chlorinated hydrocarbon (CHC) poorer countries.'-' Incidence has risen in solvents and related compounds was asso- industrialised counmes since the 1960s and ciated with a rneta-risk ratio (MRR)of 1.4 subsequently levelled off in several (95% confidence interval (95./0 CI) 1.0 to populations.' 1.8). Nickel and nickel compounds were considered in four populations (1.9; 1.2 to 3.2). Excesses were found also for chromium and chromium compounds (1.4; 0.9 to 2.3), polycyclic aromatic hydrocarbons (PAHs) (1.5; 0.9 to 2.9, organochlorine The causes of pancreatic cancer are mostly unknown.Tobacco smoking is the single established common cause. The proportion of cases enributable to smoking has been estimated at 5%-50%, depending on the population.' Epi- demiology of pancreatic cancer has suffered from bias due to high misclassification rate, insecticides (1.5; 0.6 to 3.7), silica dust notably prominent when case definition has (1.4; 0.9 to 2.0), and aliphatic and alicyclic been based only on death certificates. This has hydrocarbon solvents (1.3; 0.8 to 2.8). Evi- resulted in inconsistencies in the results on the dence on pancreatic carcinogenicity was aetiological role of environmental and occupa- weak or non-positive for the following tional dererminants of pancreatic cancer.)d agents: acrylonitrile (1.1; 0.0 to 6.2); We identified published epidemiological arsenic (1.0; 0.6 to 1.5); asbestos (1.1;.0.9 studies about pancreatic cancer 'dnd job titles, to 1.5); diesel engine exhaust (1.0; 0.9 to industrial branches, and occuparionai expo- 1.3); electromagnetic fields (1.1; 0.8 to sures, and conducted a meta-analysis of the 1.4); formaldehyde (0.8; 0.5 to 1.0); flour role of 33 chemical or physical agents presezt dust (1.1; 0.3 to 3.2); cadmium and in the working environment that affect the aeti- cadmium compounds (0.7; 0.4 to 1.4); ology of pancreatic cancer. gasoline (1.0; 0.8 to 1.2); herbicides (1.0; 0.8 to 1.3); iron and iron compounds (1.3; 0.7 to 2.5); lead and lead compounds (1.1; 0.S to 1.5); man-made vitreous fibres (1.0; 0.6 to 1.6); oil mist (0.9; 0.8 to 1.0); and wood dust (1.1; 0.9 to 2.5). The occupational aetiological fraction of pancreatic cancer was estimated at 12%. In a subpopulation exposed to CHC solvents and related compounds, it was 29%; to chro- Materials and methods The literature search covered the Medline, Toxline, and Cancerlit databases for the period 1969 to May 1998, with the following search conditions: (1) (occupational OR agriculture) AXD neoplasms AND morbidity (2) (occupational OR agriculture) .WD mium and chromium compounds, 23%; to neoplasms .L"D mortality NOT morbidity nickel and nickel compounds, 47%; to ( 3 ) (occupational O R agriculture) .WD insecticides, 33%; and to P.SHs, 33%. neoplasms AND incidence NOT mortality Concitcsion-Occupational exposures NOT morbidity may increase risk of pancreatic cancer. (4)(pancreatic OR digestive) .WE occupa- High quality studies are called for on tional interactions between occupatioaal, envi- (5) (pancreatic OR digestive) .WD case ronmental, and lifestyle factors as well as AND (control OR referent) The search was accompanied by a scan ofthe The most unbiased estimate if there is a lists of reference of the identified studies. In 811, choice and choose: 1902 studies were identified. A total of 365 0 Estimates adjusted for at least known risk studies remained after exclusion of studies that factors for pancreatic cancer (age, sex, did not report on pancreatic cancer; that did tobacco smoking), if there was a choice not represent the most recent update; that 0 Social class adjusted risk ratios over those reponed i n s d c i e n t data for the mea-analysis; unadjusted for social class that did not report data for any job or occupa- Risk estimate nearest to 20 y latency tional agent; that did not report original results period, if there is a choice. (reviews); that reported on part of a larger population reported elsewhere; and that reponed on job categories or agent categories too broad or outside our list of job titles and agents. The agents were based on the FINJEM job exposure matrix.' The list of job titles covered 150 entries in the Finnish social status categories 3, 4, and 5. Data for categories 1 and 2 represented the highest social categories and were excluded because the relevant occupational chemical and physical exposures were minimal or nonexistent. The chemical and physical agents considered were the following The extracted data were then centrally checked for consistency (AO, TP) and finalfy entered into a database and checked for correcmess. We recovered missing 95% CIS with Byar's approximadon'@' in cohort studies, and with variance of log odds ratio in case-control studies. Simple random effects models1" were ap- plied in estimating the meta-risk ratios (MRR). Fixed effects models were used only on occasion for Comparison with results from random effects models. Standard errors of log of 23 agents or groups of agents: aliphatic and alicyclic hydrocarbon solvents; aromatic hydro- Tab& I Charactenintion of the 92 agmr specific studies carbon solvents (excluding aromatic amines); arsenic; asbestos; cadmium and cadmium compounds; chlorinated hydrocarbon (CHC) Srudy ppe: solvents and related compounds (excluding Adminitrntiw (linkage of administrative 23 organochlorine insecticides); chromium and chromium compounds; diesel engine exhaust; electromagnetic fields, flour dust, formalde- records or PYRPCMRXIOR SNdies) Indusmal cohon Industry based (nested) cax-control srudy Popularion or hospital based case-conml aa 7 43 hyde, fungicides, gasoline, herbicides, insecti- study cides, iron and iron compounds; lead and lead cases: Exocrine pancreanc cancers only 32 compounds; man made vitreous fibres; nickel All pancreatic cancers and nickel compounds; oil mist (including Unspecified Diagnosis of cases: machining fluid and cutting fluid); polycyclic Histological I27 2 47 aromatic hydrocarbons (PAH);silica dust; and wood dust. The studies were divided into (a) agent spe- Other (clinical: radiology, nccmpsy, ctc) Morraliry fila Mixed UnlmOWn 2 96 9 7 cific studies with direct risk estimates for one or several of the 23 agents, or for job titles with verified exposure(s) to the agent(s), and ( b )job Ascertainment of cases: Mortality film Cancer registry Ales Hospiral records 98 40 21 specificstudieswithout risk estimates for any of .Mixed Unspecified the selected agents but instead for one or more sex: I I of the 150job categories without verified expo- Men sure(s) to the agent(s). This report is based on Women Both men and women or unspecified the data set from agent specific studies only (92 Risk measure: I I2 8 41 studies presendng data for 161 different exposed populations).ILw Standardised data extraction forms (avail- S M R (SmndPrdLcd mortality ratio) SIR (scandudired Incidence ntio) PMR (propomonal mOKality nuo) .MOR imortaliry d d s r a m ) 68 17 18 1 able from the corresponding author) covered characteristics of the study (publication year, country, study w e , case definition, source of cases, reference population, selection of control subiects, follow up period, loss to follow up, HR (hazard mtiol OR (odds raao) RR :nsk ratio). Source of exposure data: Industrial hvgtene measuremcnrs Job exposure matrix Expert assessment 3 50 4 4 15 25 response rates, sources of exposure or job data, time reference for exposure or job; risk measure, cohort admission, lag periods, Job urlest Othe* Mmed Unknown 57 19 37 4 exposure-response, job title coding applied), risk estimates, latency periods, and numbers of exposed cases. Five epidemiologists (AO, TP, NJ, EW, Cw) read the reports and extracted the necessary Time reference of exposure: Last or around diagnosis Earlier cross section Lifetime lon~cudmal Less than lifeome longitudinal Other Unknown 17 9 47 80 1, data. using- .predefined rules. The main Drinci- Total I61 ples of extraction were to extract: 'Ratio of risk, or cumulatr\e incidences In exposed and Relevant, unbiased estimates of relative unexpnsed iohnrts nsk :Job tttlr5 uich \pecihcall\ i e n h e d C X ~ ~ S U I X ~ SiJo the rclesant Measures Of risks associated rgentr I) atdl $Emplmmenr Jur~tim. biol~gicai rnunltonng. ernplover specific exposures and job titles rcctwd. registn 01 ihemicdb. relf reportmg 318 Qajiirvi, Partanen,Ahlbom, et al risk ratios (RRs), when not given, were cancers, irrespective of type. As most studies recovered as the log of the ratio of upper and considered mortality, the diagnosis was in most lower 95% CIS divided by 3.92.h test of the populations obtained from the death record. heterogeneity was performed as a ;C test with Less than 5% of the populations were women. degrees of keedom equal to one less than the Exposures were assessed in 57 populations number of populations. The public domain through job titles; in 25 through expert assess- software package computer programs for epide- ments; in 15 through job exposure mamces miologic analysis (PEPI; http:lhvww.usd- (JElMs); and in 60 through other, mixed, or inc.com/pepi.htlm), version 3.0, and the statis- unexplained methods. Industrial hygiene tical software package Szuzu release 5 were measurements were explicitly applied in four used. lo' populations only. Exposure assessment was Population aetiological fractions were esti- longitudinal for 127 populations. mated as PEF=p-p/MRR, and aeaological Reference populations in cohort studies repfractions among the exposed groups as resented predominantly national or other large EEF=l-l/MRR for each agent where excesses were observed. The proportion of pancreatic cancer cases who were exposed @), was calculated as Eexposed casevZall cases, summing over studies that provided the necessary numbers. The PEFs were based on MRRs from population based and hospital based case-control studies only, whereas the EEFs were based on MRRs from all except propor- tional studies. Aetiological b c t i o n s were not calculated for agents with no indication of increased risk. The data were organised and analysed by populations rather than studies. This was because most studies considered more than one subpopulation defined by exposure. A total of 161 populations were covered in agent specific studies (table 1). Most (85) populations were from North America, closely followed by populations (57 studies). Fewer studies (23) used local populations as the reference; even fewer (seven) used an internal reference. Follow up for case ascertainment began usually during 1940-79 (99 studies) and spanned in most studies a period of 2 1 0 years (98 studies). Losses to follow u p in agent specific cohort studies were minor or moderate (6% in 44 studies but unknown in 51 studies) but were more marked in job branch cohort studies. Agent specific cohorts were rather evenly dismbuted benveen entry cohorts, cross sectional cohorts, and mixed cohorts. Case-conwl studies used variable periods of case ascertainment. Most agent specific casecontrol studies (44) used cancer or population controls. Response rates were S80% for the cases, and S 90% in the controls. Sixteen studies did not report response rates. Western Europe (63). There were fcw populations from Central and Eastern Europe, Oceania, and none from Middle and South America, Asia, or iifrica. The annual number of studies has been rising considerably during 1 969-8 (two studies during 1969-79; 58 studies during 1980-9, and 101 studies during 1990-8). Industrial cohort (88 populations) and case- control (50 populations) studies were the most c o m m o n agent specific studies (table 1). The cases represented predominantly all pancreatic Results The aggregated results for the occupational agents are shown in table 2. Random effects models without covariates were used. Propor- tional studies (four) representing 18 populations, were excluded from most analyses because of poor quality.T h e study by Magnani t t SI,'' although reported as a case-control study, used job and branch data as well as diagnoses of pancreatic cancer &om death cenificates only. We therefore treated it as a propor- tional mortaliry rate (PMR) stud^ Gable 2 Populunons (n), mrrarrsk esnnrutrs (.MPRsr) (95'6 CIS),ranges of point rstimutes and p raitirs for hrwogeneze bv aaqxanonai ~ g e n r r(pmporrionai srudier zxciuded:simple randonr rffecrs mo&h uith no cmmrrurrs) Significant excesses were found for nickel and nickel compounds (MRR 1.9; 1.1 to 3.9; four populations) and CHC solvents and Ranpr of potnr related compounds (1.1;95% CI 1.O to 1.8;20 .Agmr n .\IRR YSph CI zsnmae p I'Ylue populations). Non-significant excesses over .Aliphatic 3nd ~licyclichydrocarbon iolvents 3 I . 3 .Aromaochydrocarbon solvents 13 1.0 0.8 to 2.0 I .&I .6 0.8to 1.3 0.5-2.9 0.09 0.2 .MRR>1.3 were found for P.4Hs; organochlorine insecticides; silica dust; and chromium ,Arsenic Asbestos Cadmium and cadmium compounds Chionnated hydrocarbon solvents and 4 10 24 1.1 2 0.7 0.610 1.6 0.9-1.4 0.9 to 1.4 0.5-3.6 0.4 to 1.4 0 . 7 4 . 8 0.9 and chromium compounds, 0.004 0.9 Table 3 shows the MRRs after stratification by sex and diagnostic quality. Only seven stud- related cornpounds Chromium and chromium compounds D i e d engine exhaust Elecuomagnetic fields Flour dust Formaidehvde Fungicides Gasoline Herbicides Insecticides Iron and iron compounds Lead and lead compounds .Man made wtrcnus fibres Nickel and nrckel compounds 011mist Polycyclic ammatic hvdrxarhons P.4Hs) Silica dust wmd .lust 20 1.4 9 1.4 7 1.0 5 1.1 1 1.1 5 11.3 2 1.3 I 1.0 10 1.1) 3 1.5 1 1.3 4 1.1 i I0 4 1.0 h 0.0 I 1.5 3 1.4 I 1.1 1.010 1.8 0.34.9 0.9 to 2.3 0.6-20 0.9 to 1.2 0.5-1.4 0.8 to 1.4 0.6-2.4 0.3 to 3.2 0.5 IO 1.0 0.5-1 0 O.4ro3.8 1.3-1.4 1).8rcr 1 3 0.8 to 1.3 0.610 3.7 (1.7 to 2 . 5 0.9roI.5 0.0 to 1.6 1 2 to 3.2 (18 Io 1.0 (7.9 to 1.5 0.2-1 I O.b-5.9 0.8-21.0 1.0-1.4 0.34.8 1.2-3 0 0.3-1 I) 1.3-3.0 0.9103.0 1.1-2.0 0 . R t 0 1.) 0.7-1.i 0.05 0.2 0.2 0.04 0.3 0.9 0.5 03 0.1 1 .0 0.03 0.9 0.5 0.8 0.2 0.3 ies presented results for women. For CHC solvents and related compounds, the MRR for women (1.8; 95% CI 0.7 to 5.8; three populations) was higher than for men (1.3;0.9 to 1.9; 14 populations). The MMRR remained essentially unchanged irrespective of whether histological verification of diagnoses of pancreatic cancer was done or not. For chromium and chromium compounds and for P.\Hs, the excess risks disappeared in studies that had histological diagnoses, whereas for nickel and nickel compounds, the .MRR was increased in populations with and without histological veri- Scation of the cases. Tab& 3 Populations (n), nrerarisk esnmates (MRRF) (95% CIS)by occupatiovd agenu, sex, and qualip of diagnosis: (propordonal srudies are exclude& simple random effects mo&k unth caianates) Agenr Sex .Mm n MRR 95hGI mmzn n .MRR 9SO.oCI Unspecified or borh n MRR 95%CI Hirrologicd diagnosl Ye5 n MRR 950r GI NO n MRR 9S%GI Aliphanc and alicyclic 2 1.3 0.8 102.0 hydrocarbon sulvents Aromatic hydrocarbon 6 0.7 0.6 to 1.0 7 1.3 0.910 1.7 4 1.0 0.6 to 1.8 9 1.0 0.7 to 1.2 solvents Anenic 3 0.9 0.6 to 1.6 2 1.1 0.5 to 2.7 2 0.9 0.5 IO 1.6 Asbestos 18 1.3 1 . 0 ~ 01.5 3 0.8 0.4 to 1.7 3 0.6 0.4t00.8 3 0.7 0.4 to 1.2 21 1.2 1.0 to 1.5 Chlonnatedhydrocarbon I4 1.3 0.9101.9 3 1.8 0.7104.6 3 1.6 0.8to2.9 4 1.2 0.7to2.1 16 1.4 1.0to2.1 solvents and related compounds Chromium and 7 1.8 0.9 to 3.6 2 0.9 0.5 to 1.8 3 1.0 0.7 to 1.6 6 2.3 0.9 to 5.8 chromium compounds Diesel engine exhaust 5 1.1 0.9 to 1.3 2 0 . 9 0.6100.9 3 1 . 1 0.9101.2 4 1 . 0 0.9co1.1 Electromagnetic fields 6 1 . 1 0.9 to 1.3 6 1 . 1 0.910 1.3 Formaldehyde 3 0.8 0.5 to 1.3 2 0 . 6 0.3to1.1 2 0 . 5 0.3to0.9 3 0 . 9 0.7101.3 Gasoline 2 1.0 0.8 to 1.3 2 1.0 0.8 to 1.4 2 0.9 0.610 1.3 Herbicldw 8 1.2 0.8 to 2.0 2 0.9 0.7 to 1.2 9 1.1 0.810 1.5 Insecricides 2 0.7 0.4 to 1.5 2 3.7 0.3ro43.3 3 1.2 0.3 to 4.3 Lead and lead 3 1.1 0.8 to 1.6 2 1.1 0.8 to 1.6 2 1.0 0.5 to 2.2 compounds .Man made n m o u fibres 2 1.8 0.8 to 3.8 3 0.9 0.7 to 1.2 3 0.8 0.5 co 1.2 2 1.2 0.8 to 1.6 Nickel and nicM 3 2.0 1.2 to 3.5 2 2.0 1.2to 3.2 2 1.6 0.4t06.9 compounds oil mirt 4 0.9 0.8 to 1.0 2 0.5 0.3 to 1 . 1 6 0.9 0.1to 1.2 Polycyclic aromaat 3 1.8 0.8 to 3.4 2 1.4 0.8 to 2.3 2 3.0 0.7 to 13.2 hydrocarbons (PAHs) . Siliudwt 2 1.1 0.8 to 1.5 2 1.S 0.8to 2.6 Wood dust 3 1.1 0.7 to 1.7 3 1.2 0.9 to 1.6 Table 4 presents results by study type. For CHC solvents and related compounds, cohort studies with internal reference and case-control studies yielded an MRR of 1.4 (95% CI 0.8 to 2.4; four populations). In studies that used general populations as the reference (SMWSIR studies), MRR was 1.3 (0.9 to 2.0; 16 populaaons). For the lower quality proportional studies, it was 1.1 (0.9 to 1.3; four populations). For nickel and nickel compounds the MRR was highest for case-control studies (2.0; 95% CI 1.2 to 3.2; two populations). For chromium and chromium compounds it was highest in SMWSIR studies (2.3; 0.9 to 5.8; six populations), as for PAHs it was (3.0; 0.7 to 13.2; two populations). For asbestos, the 20 S W S I R populations yielded a significant MRR of 1.2 (95% CI 1.O to 1 . 9 , whereas the four case-control populations resulted in an MRR of 0.7 (0.5 to 1.0). Two cohort studies with an internal refer- ence yielded a significant MRFt of 1.4 (95% CI 1.3 to 1.9) for diesel engine exhaust, which was, however, not confirmed in case-control and SMWSIR studies, the overall MRFt remaining at 1.o (95% CI 0.9 to 1.2). For insecticides, the overall MRR was 1.5 (95% CI 0.6to 3.7). For the two case-control studies, borh based on cytological diagnoses, it was 3.7 (95% CI 0.3 to 43.3), which was not confirmed in the one occupational cohort mortality study. AU insecticide results Tabk J h p d a n o n s (n), mrtarisk rams (MRRs) (95% CIS),by study gpe (propornonal s r u d i a excluded simp& random effecrr mo&ls uith no covanutes) Apl Car-connoisncdus and <ohon studies uwh i n m d reference n .MRR 95O0 CI Aliphatic and alicyclic hydrocarbon solvents Aromatic hvdrocarbon solvents Arsenic Asbestos Cadmium and cadmium compounds Chlunnated hydrocarbon solvents and related compounds Chromium and chromium compounds D i e l engine exhaust Electromagneac fields Flour dust Formaldehyde FunQcides Gasoline Herbiodcs Insecricides Iron and iron compounds Lead and l e d compounds .Man made virreous nbres Nickel and nickel cnmpounds O i l mist Polvcvclic ~rornatich' ydrocarbon5 PAHs: Silica dust Wood Just 2 4 3 4 1 4 3 5 I 1 I 2 I 2 I 3 4 1 3 2 2 4 1.3 0.9 I.:. 0.7 0.8 I ..I 1.0 I.! 1.1 0.5 1.3 1.0 0.9 1.7 1.3 11 I) 9 2.0 0.3 I4 1.5 I2 0.8 to 2.0 0 . 5 to 1.6 0.5 to 2.6 0.5 to 1.0 0.2 to 2.9 0.8 to 2.4 0 . 7 to 1.6 0.9 to 1.4 0.3 to 3.2 0.3 to 1.6 0.3 to 3.8 0.8to 1.4 0 . i to 1.8 0 . 3 to 43.3 0.; to 2 . 5 5 8 t o 1.6 0.4 to 1.2 1.2 IC :2 0.6 (0! 3 d.8 to 2 3 10.8to 2 7 0.9to 1 b SMRISIR rturiu: n .WRR OS% CI 9 1.0 I 0.9 20 1.2 I 0.; 16 1.3 6 2.3 2 0.9 5 1.1 3 0.9 2 0.9 9 1.0 I 0.8 1 I .0 I ii 2 i.0 3 0.Q I 3.0 I 1.2 0.8 to 1.3 0.4to 1.5 1.0 to 1 . 5 0.3 to 1.4 0.9 to 2.0 0.9 to 5.8 0.8 to 1.1 0.8 to 1.4 0.7 to 1.3 0.6 to 1.3 0.8 to 1.3 0.3 to 1.7 0 . i to 2.1 0 . R to 1.5 0.4 to 0.9 0 7 to 1.0 0.7 to 13 2 0.6 to 2 I '! t 1 1 i ! iI 1 : i I ! I L 1 i i t i i I i 320 Qajarvi, Pananen, Ahlbom, et al Table 5 PopuIariOns (n), population a~KiOkOgIcdfrc2CllOiotrr(PEFs), aeriologiculfiuctions dmong exposed (EEF), (9S%CIs) Popurorion ~ndogrcolfmcwn(PEF) Roporrwn n dxpsed MRR 95?6CI PEF 95% CI Aerio&cdfiacrwn nmong cxposed (EEF) n .MRR 95%CI EEF 95?6CI Aliphatic and alicyclic hydrocarbon solvents 2 Chlorinated hydrocarbon solvents and related compounds 3 Chromium and chromium compounds 3 Fungiades I Insecticides Iron and iron compounds I Lead and lead compounds 2 Nickel and nickel compounds 2 Polycyclic aromatic hydrocarbons (PAHr) 2 Silica dust 2 Wood dust 4 0.094 0.061 0.034 0.0034 0.067 0.11 0.063 0.029 0.063 0.062 1.3 0.6to 2.0 0.022 0.00 to 0.047 2 1.3 1.2 0.7 to 1.9 0.010 0.00 to 0.029 20 1.4 1.1 0.7 IO 1.6 0.003 0.00 KO 0.013 9 1.4 1 . 1 0.3 to 7.2 0.001 0.00 to 0.0029 2 1.3 3 1.5 1.3 0.7 to 2.5 0.016 0.00 to 0.040 1 1.3 1.1 0.8 IO 1.6 0.010 0.00 to 0.041 4 1.1 2.0 1.2 to 3.2 0.032 0.011 to 0.043 4 1.9 1.4 0.6 to 2.3 0.008 0.00 to 0.016 4 1.5 1.5 0.8 to 2.7 0.021 0.00 to 0.040 3 1.4 1.2 0.9 to 1.6 0.01 0.00 to 0.023 5 1.2 0.8ro2.0 0.23 0.00 to0.50 1.0 to 1.8 0.9 to 2.3 0.4 to 3.8 0.6 to 3.7 0.7 to 2.5 0.8 to 1.5 1.2 to 3.2 0.9 to 2.4 0.9 to 2.0 0.9 IO 1.6 0.29 0.29 0.23 0.33 0.23 0.09 0.47 0.33 0.29 0.17 0.00 to 0.44 0.00 to 0.57 0.00 to 0.74 0.00 to &73 0.00 to0.60 0.00 to 0.33 0.17 to 0.69 0.00 to 0.58 0.00 to 0.50 0.00 to 0.38 Only agents wnrh both .MR& >1.O considered. Negative lower confidence bounds of PEFs and EEFs forced to zero. MRR=meta risk ndo, as rsumated from population based and hospital based ase-conuol studies for PFF, and all but proportional studies for EEE ' concerned organochlorine compounds. There were 12 further studies that considered unspecified pesticides. This group of agents was considered to be too heterogeneous in exposures and was dropped h m the analysis. Given the low proportions of the populations that were exposed, the PEFs remained low, from 0.1 %-3%(table 5). Assuming independence between exposures, summing up of the PEFs resulted in an overall aetiological h c d o n of 12% for workplace exposures. The agent specific EEFs ranged from 9% to 17%, with wide 95% CIS. In a subpopulation exposed to nickel and nickel compounds, the EEF was 4796 (95% CI 17 to 69%); to PAHS, 33% (0 to 58%);to insecticides,33% (0 to 73%); to CHC solvents and related compounds, 29% (0 to 44Y0); to chromium and chromium compounds, 2996 (0 to 5794); and to aliphatic and alicyclic solvents, 23% (0 to 50%). Discussion EVIDENCE The excess risk found for CHC solvents and related compounds was based on 10 populations. Heterogeneity of RRs was nearly significant 3nd may be explained by differences in the qualitv and exposure level of the agents. Various compounds with variable carcinogenic potential were mentioned as worker exposures: trichloroethylene, ;enachloroethylene, 1,1,1trichloroethane, methylene chloride, vinyl chloride, ethylene chlorohydrine, ethylene dichloride, bis(chloromethyi)ether, and polychlorinated biphenyls. Intensities and long term doses were characrerised in most of the studies either poorly or not at all. The risk for nickel and nickel compounds was most evident in population based casecontrol studies. For chromium snd chromium compounds, the .MRR was non-significantly increased in all studies, but was not in excess in population based case-control studies. For P.\Hs, a non-significant increase was present in all studies, in population-based case-control studies, and in the two S L W S I R studies. These findings could have occurred by chance. 'The excess of silica dust reached significance in one" of three studies. This same Finnish population-based case-control study found a significant excess for aliphatic and dicyclic hydrcw3rbon solvents. This finding was qgre- gated with the finding of no excess for alkanes (C5-C,,)in another population based study from MontrealY7t' he result being an MRR of 1.3 (95% CI 0.8 to 2.0). Two case-control studies," both based on cytological diagnoses, one S h 4 R study," and one PMR study" considered exposure to organochlorine insecticides. The insecticidal agents listed as exposures were lYl-bis(4- chlorophenyl)-2,2,2-trichloroethane (DDT), chlordane, heptachlor, endrin, aldrin, dieldrin, bulan, chlorfenethol, chloropropylate, dicofol, ethylan, methoxychlor, and tetrachlorodiphenylethane (TDE). Excluding the PMR study, the aggregated M R R was 1.5 (95% CI 0.6 to 3.7). In case-connol studies it was 3.7 (0.3 to 43) based on the random effects model, and 1.7 (0.9 to 3.2) based on the fixed effects model (heterogeneity ~ ~ 0 . 0 9T)h.e highest RR was obtained for exposure to the DDT family (DDT, ethylan, DDD; OR 21.0; 95% CI 2.6 to 966; five exposed cases) in a case-controi study nested in a chemical manufacturing cohort." Potential confounders included nitrophenol derivatives, clays, lV,iV-dimethylformamide, dispersing agents, octane, and carbon tetrachloride. The other case-conuol study" was populauon based (.Michigan, US), with self reported exposures. Based ob 31 exposed cases, it yielded an OR of 1.5 (95O6 CI 0.8 to 2.?) for organochlorine insecticides. Assuming an etfect, the difference between the IWO point estimates might be due to qualitative and quantitative differences in exposures between manufacturing and agricultural application. There is a possibility of effect modification of environmental or occupational determinants by lifestyles (tobacco, alcohol, coffee) or dietary factors. These interactions were not considered in the studies. Also, genetic factors may interact with environmental or occupational exposures. Cc7UBISARILITi AND H E T E R O G E S E I n Epidemiological meta-analyses have imperfect combinability of results associated with differ- ent study types, methods, populanons, exposure circumstances, and diagnostic specificities. We calculated .LIRRs including and excluding the pooresr quality studies (proportional studies). We also calculated separate .MRRs for cohon studies with internal controls. I \ ! k i i case-control studies, SMR/SIR studies, and the few proportional studies. Differences 'in results from different study types were not consistent. Several populations were poorly character- ised. There were even studies that did not spec@ whether the cohort consisted of men, women, or both. We analysed the data for known male and female populations separately, and found data for women to be associated with a slightly higher MRR than data for men for CHC solvents and related compounds. There was in all likelihood substantial heterogeneity across populations in the quality and intensity of exposure categories, in the intake route (respiratory, dermal, or other) of exposure, time aspects of exposure (period, latency, duration, quality, and intensity), and applied scales of exposure, as well as in the quality of diagnosis of pancreatic cancer. rI -1.5 ,: 0.2 0.4 8 I,I,I.i 0.6 0.8 1.0 1.2 1.4 1.6 SE (In (RR)) Funnel plot for chlorinated hydrocarbon solvenu. Natural log of nsk ratio (In ( R R j j plonrd against imprecision (SE of in ( R R j j . The unit of thc.p b r iC study popularwn. Qualitative and quantitative differences in exposures have already been exemplified in connection with CHC solvents and related which histological vedcation was applied, compounds and organochlorine insecticides. compared with no histological verification. For I Based on a rough statistical test, significant CHC solvents and related compounds, the two heterogeneity of risk was found for asbestos, MRRs were practically identical. For PAHs elemmagnetic fields, C H C solvents and and chromium, the MRRs were higher in related compounds, and man made vitreous populations with histological diagnoses. fibres. Some studies did not document the exposure aspects at all, and no study provided PUBUCATIOK BVSS a 1 1 1 documentation. Expert assessments, Publication bias is not likely in this study, as which represent an imperfect yet acceptable very few small studies expressly considered the method of exposure assessment, were used in occupational determinants of pancreatic can- 25 populauons. Industrial hygiene rneasure- cer. Non-positive occupational findings from ments that represent a certain degree of objec- small studies therefore were not likely to tivity were used as the prime source of remain unpublished. Also, a funnel plot for exposure data in only four populations. Expo- CHC solvents and related compounds, for sure assessment based on job titles (57 popula- which an excess risk was found (figure),did not tions)are of lower quality, unless the exposures identify a concentration of small studies (high are highly homogeneous within job titles. SE of In (RR)) at high RRS. Publication bias However, some of the populations represented therefore may be claimed to be minimal or rather homogeneous single title cohorts. Job non-existent. Negative results based on reason- exposure mamces assess exposures better if the ably large numbers are unlikely to remain matrix is specific for branch and job title, even unpublished. for company and period. The JEMsof variable A counterargument may be raised about degree of specificity were applied in 15 popula- cohort studies with multiple end points. Some tions. Most were relatively unspecific and of these studies deleted results based on small thereby induced exposure misclassification. numbers, occasionally for pancreatic cancer. Misclassification, however, was likely to be This omission may have minor influence on the non-differential, resulting in underestimation metaresults. Some case-control studies may of the MRR. Multiple sources of exposure data have omitted results for rare exposures, with were applied in 37 populations. Exposure data similar minor effect on the metaresults. were longitudinal in 127 populations and lifelong in 47 populations. The longitudinality of exposures was thus well covered. Msclassification rates for pancreatic cancer are marked, particularly when the diagnosis is based on death certificate only. This adds to the bias in the mcta estimates towards the null value. Garabrandt et u f zshowed what the magnitude of the impact may be in a single study. SEIECTION OF STUDIES We used major databases and lists ofreferences of the studies for the identification of studies. Studies not found in major databases are probably of lower quality. For the same reason and because of the cost we did not try to identify unpublished studies. They compared, in a case-control study of pancreatic cancer, ORs for the DDT family tzxmcnoN between cases representing death certificates Extractor bias was minimised by the formal and cases representing cytolog~alverification. extraction procedure between the extractors, For death certificate cases the ORs ranged and the central checking of the extraction. The kom 0.8 to 2.6; for cytologicallyverified cases, procedure was also intended to guarantee the from 15.4 to infinity. In our rneta-analysis, extraction of the relevant risk estimates in agent specific MRRs were higher for nine studies that offered several alternative risk esti- agents but lower for IO agents in populations in mates. 24 Brown DP. Mom!ity of workers emploved at organochlo- 57 k t T,Aquavetla J, Lynch C, et al. Cancer incidence rine pesticide manuhccuring plants: ah updare. ScandJ among alachlor manufacruring workers. A n f Ind M e d I@E&mmn Health l992;18:155-61. 1996;30:300-6. 25 Broun DP, Dement JM,Okun A. Morralify panems among 58 Levine RJ, Andjelkovich DA, Shaw LK. T h e mortality of female and male chrysotile asbestos textile workers. f OCCUP ,%fed 1994;36:882-a. Ontario undertakers and a rniw of formaldehyde- related mortality studies. 7 Occup .$fed 1984;26:740-6. 26 Cammarano G,Crosignani P, Berrino H, er a/. Additional 59 L y n p E. A follow-up study of cancer incidence among follow up of cancer mortality among workers in a thenno- workers in manufacture of phenoxy herbicldes in Den- clmaic power plant. ScandJ Wbrk Emiron Health 1986;12: mark. BrJ Cancer 1985;52:259-70. 631-2. 60 Lynge E,Andmen A, Nilsson 2, er af. Risk of cancer and 27 Checkoway H, Heyer NJ,Seixas N S , cr d.Dose-response exposure to gasoline vapors. A m f &puiznrwlI997;145:449- associadons of silica with non-malignant respiratory 58. disease and lung cancer mortality in the diatomaceous earth industry. AmJEpidemiol 1997;145:680-8. 28 C h i n e LJ, Ference LD..MortaliU among PVC-fabricaang 61 .Mack TM, Peters JM, Yu MC, er a f . Pancreas cancer is unrelated to the workplace KI Lor Angeles. A m J lnd Med 1985;7:253-66. employees. Em'mn Htalth Pmpccr 1981;41:13743. 29 Claude j,Freneel-Beyme R Mortality of workers in a Ger- man rock-wool factory: a second look with extended follow up. Scandf Mrh Em.mn Health l986;12:53-60. 30 Cocco P, Hua F, Bofkna P,TI d Mortality of Italian lead smelter workers, Scand J W r k Emimn Healrh I997;23:15- 23. 31 Cocco P, Blair A, Congia P,er al. Long term health effects of the occupational exposure to D D T A preliminav report. Ann A' Y Acad Sn' 1997;837:246-56. 32 Coggon D, Pannen B, 'Sinter PD, et ol. .Mortality of workers exposed to 2 methyl4 chlorophenoxyacetic acid. Siand J Wrk Em.mn Health 1986;12:448-54. 33 Decoufle P. Further analpis of cancer mortality panerns among workers exposed to cumng oil musts. J Sad Cancer INL 1978;61:1025-30. 34 Decouflc P, Blamer WA, Blair A. Mortality among chemical workers exposed to benzene and other agents. Emim R u 1983;30:16-25. 35 Dement JM, Brown DP, Okun A. Follow up srudy of chrysoak asbestos tunile workers: cohon mortalin. and use-conuol anolysn.A m J Ind ;Wed 1994;26:431-+7. 36 Elinder CG, Kiellsu6m T, Hogstedt C, zr d. Cancer rportality of cadmium worken. Brf Ind .Wed 1985;42:65I - 3. 37 Enterline PE, Hvrley J, Henderson V.Asbestos and cancer: a cohort followed up to death. BrJlnd .Wed l987;44:396- 401. 38 Enterline PE, Day R,,Mush GIM. Cancers related to exposure to arsenic at a copper smelter. Occup Emiron .Urd 62 Magnani C, Coggon D, Osmond C, R al. Occupaoon and five cancers: a case-control study using deadr certificates. BrJ Ind Med 1987;44:769-76. 63 Magnani C,Ten-acini B, Ivaldi C, et al. Tumor mortality and death from other causes in asbestos cement workers at the Casale Montfernto plant. .Urd Lav 1996;87:133-46. 64 McDonald JC, Liddell FD, Dukesne A, et al. The 1891-1920 birth cohort of Quebec chrysotile miners and millers: mortality 1976-88. Br J lnd M e d 1993;501.073- 81. 65 Mikoczy 2, Schiia A, Stromberg U,et a / . Cancer incidence and specific occupational exposures m the Swedish leather tanning i n d u s e : a cohort based case-control study. &cup Emvan .%fed 1996;53:463-i. 66 .Milham SJ.Mortality in workers exposed to elecrromagneoc fields. E m m n Health Perspccr 1985;62:297-300. 67 .Milham SJ.Increased monality in amateur radio operators due to lymphaoc and hematopoietic molignanrtes. Am f EpldOnWl1988;127:50-1. 68 .Moulin JJ,Wild P,.Mur JM, 61 d. Risk of lung, larynx, phar- YN[, ana buccal cawty cancers among carbon electrode manufacruring workers. ScandJ Ly6nc E m m Health 1989; 15:30-7. 69 Nalramura R A momlity s n d y of vinyl chlonde workers in Japan. Sang)o lka Daigaku Zasshi 1983;20:49-57. 7 0 Ohlson CG, Klaesson B, Hogstcdt C. Mortality among asbestos-expased workers in a &ad wokshop. ScandJ Mrh Em'mn Health 1984;10283-91. 7 1 On .MG,Olson RA,Cook RR,cr al. Cohort mortality s ~ d y 1995;52:28-32. of chemical workers with porennal exposure to the higher 39 F m c h i n i I, Magnvu F, Murd A. .Mortality u p e n e n c c chlomatcd di0Xms.J O C C UM~d 1987;29421-9. among chromcpla- workers. Inroll lindings. Stand J 72 k r k RM, Mirer FE. A survey of mortality at two Envimn He& 1983;9247-52. aufomoovc engvle manufacturing plants. .4m J fnd Med 40 Frenuel-Beyme R, Thins AM, Wieland R S u n e y of mor- 199630664-73. d i t y among employee engaged in the manufacture of s ~ - 73 Ramlow JM, Spndacene N W ,Hoag SR, rt af. .Morrility in a rcne and polystyrene at the BASF Ludwgshafen works. cohort of pentachlorophenol m a n u i a c m workers, ScandJ W d E m n m Health 1978;4231-9. 1940-89. Am J lnd ,Wed 199630:180-94. 4 I Fryrek JP,G a r a h r DH, Harlow SD,:I ,al. A casc<onuol 74 Sathrakumar N,DelzeU E,Cole P. Mortality among workers srudy of self reponed aqosum to pesmder and pancreas at w o rrivtne herbicide maoufacruring p h a . A m f I d cancer ln s o u t h w t e m Michigan. Inrf C a m 1997;72:62- M d 1996;29143-5 1. 7. 75 Sathiakumar N, Deltcll E, Hoviaga M,er af. Mortality from 42 G a n b n a t DH, Held J, Homa D. Rcsponre ( D D T and pan- C m O C cancer). 7 Noll Cancrrlnn 1993;85:328-9. cancer and other causes of deab in synthetic rubbet workm.Occup E m m n M i d 1998;55:230-5. 43 Gardner MJ,Pannett B, Winter PD, et al. A cohon study of 76 Seidman H. Selikoff IJ, Gem S K .Mortality expenencc of workers elposed to formaldehyde VI the British chemcal amorire asbestos factory workers: dose-response relation- i n d u s q an update. E r J Ind &fed 1993;50827-34. ships 5-40 years after onset of short-term work exposure. 14 Gibbs GW, .hxl j, Soden K A cohort mortality studv of Am J lnd Med I986;10479-j14. cellulose macetate-fibre workers exposed to methylene i 7 Selikoff IJ, Seidmnn H. Cancer of the panncar among chlonde. f Occup E m n m Med 1996;38:693-7. asbestos insulation workers. C Y K 1~981;47:1.169-73. 45 Greenland S,Salvan A, Wcgman DH, et al. A casc-control 78 Sheffet A, Thind I, ,Miller A M , ct d Cancer mortality in a study of cancer mortality at a transformer-assembly pigment plant utilizing lead and zinc chromates. .+rh E m - faality. lnt Arch &up Envimn Health 1994;66:49-54. m He& 1982;37:4452. 46 Gustanson P, Reutemall C. Mortality and incidence af 79 Siemawcki 1. Risk fouors for cancer in rhe wrkdacc. Boca cancer among Swedish gas worken. BrJ Ind 'Wed I990;47: Raton,'FL. k R C , f991 ' 169-74. 80 Simonato L, L'Abbe K.4, Andmen A, et d.A collabonove 47 H m J. O h JH.Formaldehyde and cancer morbidity among male employes in Denmark. Cancer Couses Cmrrol study of cancer lncidencc and mortality among wnyl chlo- nde workers Scand 7 ubr* Emmn He& 199I , I 7.159- 1995;6:354-60. 69 48 Hearne R,Pifer jU', Grox F. Absence of adversc mortality 9 1 Sinks T,Steele G,Smith AB, ad.Mortality among workers effects m workers exposed to methylene chlonde: an exposed to polvchlonnated biphenvls. A m 7 Eplriemrol update. J Ocsup .Mzd I990;32:23440. 1902136.389-98. 49 Hooiveld M, Heedenk DJJ, Kogevlnas .M. er d Second 92 Smulewch VB, Fedotova N, Filatova VS. Increasing follow-up of a Dutch cohort occupationally exposed :o evidence of the rise of cancer in workers exposed to vinyl phenoxy herbiades, chlorophenols, and contaminants. .4m chlonde. B r J lnd .Med 1988;45:93-7. 7 EpJemwi I 99a;147.89I -90I. 83 Spirtas R, Stewart PA, Lee IS,rt ol. Retrospecove cohort 50 Howe GR Fraser D, Lindsay J, rr d.Cancer mortality mortality study of workers at an aircnft maintenance f a d - i1965-77)in relauon to diesel fume and coal exposure in 3 icy. I. Epidemioloacal results. E r f fnd .Mid 1991;48:5I5- cohort of retired railway workers J Sat1 Concn Insr 30. 1983;701015-9. 84 Stayner LT, EUiOK IBlade L R d.A reuospecwe cohort 51 Kano K, H o n k a w .M, Ununormya T, +I d.Lung cancer mortality s t u d y of workers exposed to formaldehyde in the mortalicy among a cohort of male chromate pigment work- garment indusuy. A m J lnd Med 1988;13:667-81. ers ln Japan. lnr 3EpuLmwl 1993:22:16-22. 85 Sun J, Shabata E, Nisanaga N. et d.A cohort mortality 52 Kaupplnen T,Partanen T.Degerth R cr d.Pancreaoc can- study of construction workers. .-lm 7 Ind M e d 1997;32:35- cer and occupational exposures. Epdemwl I995;6:498- 41. 502. 96 Swaen GAMv.an Vliet C, Slangen JJ,el d.C a n c a morralify 53 Kogcwnas .M, Fern, G. Andersm A. a d.Cancer mortality among licensed herbicide applicators. S c ~ n d JMrk E m m n in a hutoncal cohort study of workers exposed :o styrene. Heolth 1992;18:2014. Siondf ITnrb Emirnn Health 19Q4:?0251-n1 87 Tolbert PE. Eisen E.4, Poduer Lj, et d..Mortality rtudics of 54 Kugmnas ,M. Becher H. Benn T, et d Cancer mortalin. in machmmg-nuid exposure m the automobile industry. 11. workers exposed to phenoxy herbicides. shlorophenolc. ksb associated wth rpmfic flrud npn. Scow' 7 W r h m d dioxins. .in expanded and updated iniernitional Eminin Hedth 100?:18:35I-b0 iohon itud\ .-lm7Epdonrd IQo7:l45~1na1-75. 98 Tollestrup K. Daling JR..Ulard J. .Mortalin. in a cohort of 55 L n e s SF, Rothman KJ, Drever S.4. zr 41 .Unrtalin. updare tmhard workers exposed :o lead arsenate peticide $pray. oiiclluln5e hbcr prnducrion worker< S d n d 7 WhrR Entvrou A r ~ hEmwon H d r h 1W5.50.221-9 H e d i h 1093;19.12b-Y. $0 Tomenson JA. Bonner S.M. Heline CG. if A. Mortality of in Lnpnrd S.Andersen A. Ravnestad J. Incidence of Cancer workers exposed methvlene chlonde cmploved JI a planr Jmong ferrochromium m d ferrmilicon workers m ex- prnducing cellulose :nacetaic tilm base Oc,-up Ernwon .Urd tended observation period. Br.7 1n.i .Mrd I oOO:4i 14-0 i Qo:,sd 4-0-0 4 i 324 ! I I ... Qajarvi, Parranen,A h , et al 90 Tsai SP,Waddell LC,Gilatrap E L P d..Mortality among mamtcnance employees potentially exposed to asbestos in 3 refinery and petrochemical plant. A m 3Ind .Urd 1996;29 89-98. 91 Tynn T, h d e r s e n A, Langmark F. Incidence of caocer in Nonvegtan w o r k m potentially exposed to rlecmmagncuc fields. A m 7 Epldnniol 199?;136:81-8. 92 Tyner T, Reitan JB,Andmen A. Incidence of cancer among workers in Nonwegian hydroclenric power companies. ScandJ l&&Envirrm Health 1994;203394. 93 Van Den Eedcn SK, Friedman GD. Exposure to engine exhaust and risk of subsequent cancer.3 Occup Mad 1993; 35307-1 1. 94 Wilczynska U,S7.cxenia D a b m h N,Szymcuk W.Mor- tality from malignant neoplasms in men occupationally exposed to asbestos dust. .Urd R 1996;4743743. 95 Woitowiu HJ,L q c HJ,Beierl L R d..Mormlity rates in the Federal Republic of Germany following previous occu- pational exposure to -os dust. Im Arch Occyp E& Health 1986;57:161-71. 96 Wong 0.An industry wide mortality study of chemical workers occupadonaUy exposed to benzene. I. General mults. B r 3 Ind Med 1987;44:365-81. 97 Wong 0,Whorton .w),Folian DE, cr d.An industry-wide epidemiologic study of vinyl chloride w o r k s , 1942-82. Am 7 Ind M a d 1991;20317-34. 98 Wong 0,Trent LS,Whmon iw).An updated cohort mortality study of workers expored to styrene in the reinforced plastics and composites industry. Occup Emimn h i i d 1994; 511386-96. 99 Yassi A, Tate R, Fish D. Cancer mortality in workers employed at a transformamanufacturing plant. .4m 3 I d Mad 1994J53425-37. 100 Breolow NE,Day NE.S&,tkal merhodc in c a w meatrh. Vol II. lh&I& a d andyszs of w h n studits. Lyon: Inter- national Agency for Research on Cancer, 1987. 101 DerSimonian R, Laird N. .Meta-analysir in c l i n i d mals. Connd CIin T d 198k7177-88. 102 Sharp S, Steme J. Meta-analysis S w T2chni.d BuUuin Reprints 1997;7:1004. . - Occupational and Environmental Medicine - http://www.occenvmed.com Visitors to the world wide web can now access Occupational and Environmental Medicine either through the BMJ Publishing Group's home page (http:ilwww.bmjpg.com) or directly by using its individual URL (http://www.occenvmed.com)T.here they will find the following: Current contents list for the journal Contents lists of previous issues 0 Members of the editorial board 0 Subscribers' information 0 Instructions for authors 0 Details of reprint services. A hotlink gives access to: BMJ Publishing Group home page British ,Medical Association website 0 Online books catalogue 0 BiMJ Publishing Group books. The web site is at a preliminary stage and there are plans to develop it into a more sophisticated site. Suggestions from visitors about features they would like to see are welcomed. They can be left via the opening page of the BMJ Publishing Group site or, ~lternativelyv, ia the journal page, through "about this site".