Document pBpB6Q2GZbBK63VXyDvJDa61X

FILE NAME: Doubt Science (DBTS) DATE: 2005 DOC#: DBTS013 DOCUMENT DESCRIPTION: Egilman Journal Article from IJOEH - Abuse of Epidemiology: Auto Manufacturers Manufacture a Defense to Asbestos Liability Abuse of Epidemiology: Automobile Manufacturers Manufacture a Defense to Asbestos Liability DAVID S. EGILMAN, MD, MPH, MARION A. BILUNGS, M SC Masbuecshtoosf ethxepo"sduerbeatfero"maboauuttotmheobreilleatibornaskheipwboertkwaenend athsbaet shtaovse-ibndeeuncefduncdaendcebry hcaosrpboeraetniofnusewleidthbbyillsitoundsieast satsabkeestoins-litnoerdt lbirtiagkaetiomna. nTufhaectuaruetrhsorhsaveexpcloorrreuphtoewd mfuenddiecdalblyitetrhaetsuerecotomepsacnaipees ltioabeilnitayb, laenathlyezmingtostculdaiiems mtheastowthoerlkiomwiat.h Tahsebyesrteovseablrahkoew litnhiengcsomnpevaneriescahuasvees rceaduesfei-neefdfecsct ireenltaitfiiocncsrhiitpesriaanfdormthaenidpeutleartmedinsactiieonntifoicf dBautta tthoe gaibvesetnhcee iomfpervesidsieonnceofisannoatbesveindceencoef eofffetchte. aabsbseesntcoes; ocforarnupteifofnec;t.ocKceuypawtioordnsa:l mheeasoltthh;elibormakae; mechanics. INT J O C C U P ENVIRON HEALTH 2005; 11:3 6 0 - 3 7 1 In recent years, thousands of automobile and "shade-tree" (amateur) mechanics have sought compensation for asbestos-related disease from mthaant uthfaecsetubrerarskeosf aasrbee4st0o-s7-0li%nedasbbreasktoess., Dtheespbirtaektehemfaanct ufacturers have vigorously denied that they contributed in any way to a single case of asbestos-caused disease.1 The brake manufacturers have dismissed the fact that there is no known threshold of exposure associated with mesothelioma (a rare cancer of the pleura or peri toneum) , and that in some cases the only documented worker exposures are those that have occurred as a result of exposures to asbestos from brakes,2 Rather than accept the well-established scientific facts, the industry has funded scientists and lawyers to develop arguments and methods for defending against lawsuits brought by workers and their household members who htoavaesbdeesvtoesloipnedbrmakeesso.1th3"e6liToomsauapspoarrtetshuelitroafrgexupmoesnutrse, the automobile manufacturers have hired consultants Received from the Department of Community Health, Brown UniAvedrdsirteys,sPcroovrriedsepnocne,dRenhcoedeanIdslarnedp.rint requests to David Egilman, MD, 8 North Main Street, Suite 404, Attleboro, MA 02703, U.S.A.; telephone: (508) 226-5091; e-mail: <degilman@egilman.com>. to reanalyze previously published hygiene and epi demSoiomloegimc satnuudfieasc.t4u-7rers of asbestos-lined brakes-- ihnacvleudisnpgonGsMor,eFdordfo, uDrainmoletarb-Clehryesxlaemr, palneds Boefndsiuxc--h research since the late 1990s. In 2001, Otto Wong, an escpiiednecmeisolocogmistpawnoyr,kinfigrefdor taheCafliirfostrnisaa-lbvaoseodfhethaeltihr defense.6Although Wong's paper is couched as a cri tique of regulation of exposures to asbestos-lined brakes and EPA risk assessments, Bendix (an automo bile brake manufacturer) first presented it as an expert report to defend against a lawsuit by a worker who had been exposed to asbestos brake linings and had contracted mesothelioma.8Three years later other manufacturers of asbestos-lined brakes (GM, Ford, and Chrysler) sponsored Goodman et al. to write a second brake-related lung cancer meta-analysis com bined with a duplication of Wong's mesothelioma analysis.5 The automobile companies then sponsored Hsioesnseolfeatpaal.pteorwbryitSepairtthaisrdofptahpeeNr wIHhitchhawt ahsaadnalerxetaedny baneaelnysiisn.c4luFdineadllyin, tGheooidnmduasntreyt saul.bsainddizeWdoanng'sinmcoemta plete analysis of historical exposures to asbestos in braTkheewseorcko.3mpanies have spent millions of dollars to generate these epidemiologic studies in order to refute claims of causation and thereby avoid com pensation payments to victims and their families. Accordingly, all four papers demonstrate the use of two practices associated with the industry's involve ment in scientific research: the redefinition of scien rtiefliacticornistehriipas faonrd tthhee mdeatneirpmuilnaatitoionnofosfciceanutisfeic-edffaetac.t In this paper we critique these methods, drawing eabxoavmep. leTshferomfirstht esfeicrtsitotnhrdeeelionfetahteesfotuhrepapproecrsescsiteodf rceaduesafitnioitni,onfoollfoswceiednbtiyfica cmriotreeriainf-odredpethtedrmiscinusastiioonn ooff specific methods of scientific manipulation utilized Ains thaewshtuodleie,swbey Ganoaoldyzmeantheet avla.,luWe oonfg,thainsd tHypeesseol.f rbeestewaerecnh einxproegsuarredsttooiatssbcelasitmossdaubroinugt tbhreakreelawtoiornksahnipd induction of mesothelioma. 360 REDEFINITION OF SCIENTIFIC CRITERIA FOR DETERMINATION OF CAUSE-EFFECT RELATIONSHIPS This method, first developed by the tobacco industry, rinevdoelfviensetwthoeptayrptse.9aFnidrsta,minoduunsttroyfla"wpyroerosfa'nredqsuciireendtisttos luengfaalvlyoreasbtaleblcislahssceasuosafteiovnid.9enTcheisthisatdoarneeubnyfadveorroagbalteintgo industries'litigation and regulatory positions and insist iinngdutshtaryt eivsidmeonscteimthpatordtoaenst noorteesxsiesnttioarl itsofaevsotarabblilsehtoa cause-effect relationship in court. For example, when a suspect carcinogen is found to cause cancer in human epidemiologic studies but not in animals, companies argue that animal studies are required to prove causa tion.9On the other hand when animal studies are posi tive and human epidemiology is incomplete or nega btievfeo, rceomthpeagnoievseranrgmueentthcaatnhruemgualnateevtihdeenscuebisstarnecqeuiarnedd befFoorer ewxoarmkeprlse,antodboacthceorscocmanpbaneiecsomarpgeunesathteadt.9human epidemiologic studies cannot establish causation but instead only establish some "non-definitive" association between exposure to tobacco smoke and cancer.10They have argued that epidemiologic studies cannot distin guish a genetic link between the propensity to smoke and the propensity to contract cancer.10This critique of human epidemiologic studies is true, but irrelevant since there is no biologic basis for such speculation. On the other hand, when early researchers failed to induce cancer in animals that were exposed to tobacco smoke, tobacco companies insisted that only animal inhalation studies could "prove" causation.10After the now-famous beagle studies, these companies emphasized the impor tance of mechanistic understanding as a causal require ment.10They then argued that unless one could explain exactly how tobacco smoke caused cancer they had no "scientific proof" of causation.10 Some chemical manu fmacetnutrsewrshheanvestmudaidese ocfomthpeairrapbrloedeupcitsstehmaovleogfoicllaolwaerdgua wsimheinlarrefsaecatrcphaetrtserfna.ileIdn tothiendeuacrleyc1a9n7c0esr,infoarneimxaamlspfloel, lowing benzene exposure, industry expert witnesses setsatatebdlisthhatthathteasesumbsistasnincge awnaismaahl udmataanwcearrecirneoqgueirne.d11'1t2o The second element of this method is industry's insistence on direct causal evidence related to specific forms of exposure in individual cohorts of workers. Human epidemiology is often lacking for most chemi cal-exposure cancer effects. Most chemical companies, therefore, argue that these missing "animal or other studies," or human studies in which the exposed work etursalairme peoxspsoisbeildityto),thareesruebqsutainrecde itno qeusteasbtliiosnh opnrloyo(faovfira caraguusees-etfhfeacttarneilmataiol nasnhdipi.nInvittrhoesltautdtieers caarsee,reinndduesrterdy invalid as evidence of causation.9 Since cancer effects have long latent periods and workers are generally sub ject to potential confounding exposures, these studies are rare and easily criticized.9 When clear epidemio logic evidence of causation exists, as is the case of the relationship between asbestos and mesothelioma, com pinajnuireesdawnodrkthereicranexepveerntsfihleavaecalarigmuefdortchoamt bpeefnosraetioann utwalosetphiadtefminiodloagsitcatsitsutidciaelslyinsigsinmifiilcaarlnyt edxopuobsleidnginodfivtihde risk must be published in peer-reviewed journals.13 Some Texas courts have accepted this argument and even appear to require the publication of human epi demiologic studies for subgroups of particular indus tries with statistically significant rate ratios above two before a worker can even file a lawsuit.14 Greenland calls the use of this criteria a "methodologic error that hasTbheecsoemsetrainsgoecniatl plergoabllermeq."u1i8rements can be con trasted with the normal physician practice of assigning causation based on an analysis of all available evidence with no specific rate ratio or statistical requirements.16 In general, practicing physicians attribute lung cancer chaisutsoartyioonf tsomtookbiancgcoansymcoigkainregttiefst,hreergeaiisdleevsisdeonfcberaonfda, for often unspecified time frames prior to disease diag wnohseist.h9erPraapcutibcliinsghepdh,ypseiceira-nresvwieowueldd pnaopterchsheocwk etdo thseaet smoking ten cigarettes per day caused a statistically sig nificant doubling of the risk of contracting lung cancer prior to making the determination that a patient's smoking had caused or contributed to his or her cancer. Physicians also take for granted that all brands smoked contributed to disease causation even in the absence of studies evaluating individual brands. The asbestos-product manufacturers and their experts have constructed an epistemological straw man, asserting that epidemiologic evidence of causation is required in each and every worker cohort that has a bdeofcourme eevnitdeednceexcoefsscaeuxspatoisounrecantobae phraersmefnutel dsuinbsctoanucret Companies argue that, to prove causation, both the type joufstascboemstpoasraabnlde tehxeponsauturer)e mofusetxbpeosculerearl(ysadmoceujmobe,nnteodt itnheandeevpeildoepmmieonlotgoifccsatnucdeyr.a1s3Ehavveinngthae ctoabuasaclcoefcfeocmt poan nies have not gone so far as to argue that an expert ccaanncneort sausbsteyrpt eth(aet.gV.,irbgrionniachSoligmesniccaucsaencaers)peicnifwicolmuneng absent an epidemiologic study of smokers of that par ticular cigarette showing a statistically significant twofold increased risk in that particular cohort of female smokers for that particular subtype of lung cancer. Yet the asbestos companies arid their experts alirsghuceatuhsaattisoenpaforartethceohsaomrtestduidseieassearfeorneeaecdhedfitboeresttyapbe and each product composed of that fiber despite the VOL 11/N O 4, OCT/DEC 2005 w w w .ijoeh.com Abuse of Epidemiology 361 fact that almost all workers are exposed to more than oneAsfibaerrestyuplte, fcroommpmanoirees tahnadt othneeirpreoxdpuecrtts.13attempt to show that exposures to asbestos from their product are somehow "different" or even irrelevant Manufacturers argue that exposures to asbestos from their products are quantitatively or qualitatively unlike other asbestos expo sures.13 For example, asbestos in their products is `bound" or "degraded" or "decomposed" or "results in too low an exposure" to cause mesothelioma.13Unfortu nately, some courts have adopted aspects of this reason itnypge.17thInattihneducasstreyosfticllhrcylasiomtisledaosebsesntoost--catuhseeomnleysofitbheer lioma--the argument that the particular type of asbestos in a company's product is different from other types is now generally recognized as the "amphibole hypothe sis."18-23 It is important to note that product manufac btuyrcelrasimwihnoguthseadt thameroeswitaesfiinbseurffmicaiednetseipmidileamr iaorlgougmiceenvtis dence to establish that amosite caused mesothelioma.24 MANIPULATION O F SCIENTIFIC DATA-- META-ANALYSIS AND EXPOSURE DATA MISREPRESENTATION Manipulation of scientific data and misuse or re interpretation of standard scientific reasoning is ubiq uitous in the automobile industry-sponsored asbestos studies. The most common are: selection of inappro priate studies for re-analysis, selective presentation of study data, non-differential exposure-determination bexiapso, seidnacdoehqourattwe itshamanplienapsipzreo,prcioamtepcaorinstoronl gorfouapn, athnedsemsitsruatseegioefs,ciollnufsidtreantecde winitthersvpaelsc.ifiHceerxeamweplepsrefsroenmt the Bendix, Ford, General Motors, and Daimler Chrysler studies. Selective Presentation ofExposure Data In his article, Wong outlines the "lack of exposure" argument as described in the previous section.6 In doing so, however, he fails to report studies indicating that work with asbestos brake linings leads to exposures that have resulted in asbestos cancers in other occupa tional and non-occupational settings.25Wong not only ignores exposure studies whose results contradict his industry position; he also misrepresents the results of tshheowstnudthieast bhreakdeoelisnrinepgodrut.stHceonsttaaitness ltihttalet "rsetcuodgiensizhaabvlee cclhariymsotthilaetfcibuerrrse"natnedxpeoxstruarpeoslaartees"ferxotmrema efelywlsotwu.d"i6eIsttios true that only some studies have examined decomposed d(uusreindg) btrhaekessa;ntdhiengheaanvdiesgtreinxdpionsguroeftnoeawsbbersatokseso.c25cuOrsf course, current exposures do not reflect past exposures or risk. Studies of workers' previousjob activities reveal exposure levels hundreds of times higher than current permissible limits.25Wong cites some "data"from a 1987 Finnish study by Kauppinen and Korhonen by listing average fiber concentrations as <0.05 fibers/cc.26 This study also found, however, that the grinding of brake linings produced asbestos concentrations as high as 125 fibers/cm3. Kauppinen and Korhonen reported expo sures up to 8.2 f/cm3during the cleaning of passenger car drum brakes using a compressed air jet to remove brake dust.26 (Unfortunately this is still a common prac tice in many nations, including the United States.) Aver age concentrations were between 0.1 and 0.2 f/cm3for ttrourcekpoanrtdfrboums rtehpisaisrt.udTyhreepornelysednatetadpaoveinratgWe eoxnpgocshuorsees during passenger-car repair. Contrary to Wong's mis leading data selection and analysis, this study demon strated that average asbestos concentrations in Finland mexapyocsoumrepsloyfwteintheTxWceAedleimdittsh,ebOutSaHlsAo esxhocuwresdiotnhlaitmpiet.a26k Wong also selectively excerpts information from a paper by Agudo et al., whom he quotes stating, "Over all, 37% of cases of mesothelioma in our population are attributed to a sure (or almost sure) occupational exposure to asbestos and this proportion comes up to 62% when occupations with any probability of expo sure are included." In turn, Wong concludes that "this observation underscores the high proportion of malig nant mesothelioma cases not related to occupational exposure."6 Wong makes this assertion but omits Agudo et al.'s subsequent qualifying sentence, which sptoapteusl,a"tAiornouconudld40b%e doufecatosecsaoufsems eostohtehreltihoamnaoicncuopuar otitohnearl oecxcpuopsautiroenatol exaspboessutoress, ianltnhoonu-gtrhadpitoisosnibalilliytyhaozf ardous scenarios cannot be ruled out." M anipulation ofMeta-analysis: Inappropriate Study Selection and Selective Presentation ofD ata Industry researchers can often influence results by careful selection of the epidemiologic studies they decide to evaluate. Studies favored by the industry per spective (those with negative results in regard to the relationship between exposure and cancers) are included, and others are ignored or derogated and then excluded. Goodman et al. and Wong reanalyzed a series of epidemiologic studies, claiming that they fianiclreedastoedfiinndciadneanscseocoifamtioensobtehtewlieoemnab.5r'a6kBeowtho,rhkoawnedvearn, idnidsrivegidaurdal nsutumdieersouthseymecthhoosdeoltoogiinccoprrpoobrlaemtesininthtehier analyses and ignore study limitations noted by the orig ionmalit akuetyhoirnsf.orFmuartthioenrmtohraet,utnhdeesremainuethsotrhsesireleccotnivceluly sionGsoaonddmeavnideetnacle..first attempt to evaluate the quality of 11 studies with potentially relevant information 362 Egllman, Billings w w w .ljoeh.com . INT J OCCUR ENVIRON HEALTH about the relationship between brake work and mesothelioma. The authors use a self-created 11-point scale to evaluate the studies (the rating system allowed for negative scores). The highest ranked study--one of two purported "studies" based on "personal communi cations" by Goodman and Hessel themselves--garners a 5, out of 9. The scores for the rest of the studies i4n,c1l,udanedd i2n, othuet mofe9ta.-aGnoaolydsmisaanreetunali.mepxrcelsusdiveed: s3t,u4d,ie1s, that scored minus 1 (the rating system allowed for neg ative scores) but none of the included studies achieved a passing grade of 60% based on normal academic stan dards. Goodman et al. seem to believe that evaluating the studies in such a way somehow mitigates their underlying methodologic errors and inadequacies. Unfortunately, revealing errors is not the same as fixing them, and these studies, which were never designed to assess mesothelioma risk to brake-exposed workers, are beyAofntedrrtehpisaier.laborate quality evaluation, Goodman et athl.apt eWrfoonrmg ehdada mreevtiae-waneadlystihsreuesinygeatrhse bseafmoree.stuTdhieeys excluded two cohorts. The first of these was a study of l6i8o5maaus.t2o7mTohbeilesemcoecnhdanfoicusntdhaat rsetpatoisrttiecdalltywosimgneisfoicthanet excess of mesothelioma in mechanics with a reported rateFurartthioe.r2m8 ore, Goodman and his colleagues present the data entirely out ofcontext. The original authors of the underlying studies did not design them to examine the specific issue of exposure to asbestos during brake work and subsequent incidence of mesothelioma. The extraction, therefore, of occasional study participants who happen to be automobile mechanics not only lacks sctoanticsrtiectael epvoiwdeenrcbeupt rcoavninnogt tbhee ianbdseepnecnedoenf talyreulsaetdioans ship between a complicated exposure and a potential outcome. Due to the inherent difficulty in determining asbestos exposure (especially in the case of brake work pdauret-ttiomtehberalakregwe onrukmthbaetrcaonf eaamsialyteguorubnrnaokteicjoedbswahnend dcaeltleyrmdeisniignngeedxptoosaudrde)recssohtohret sistusudeiesofmmusetsboethseplieocmifia among those exposed to brake dust. Until such studies are conducted, it is wrong to conclude that epidemio logic studies fail to establish a relationship between brake work and asbestos-induced cancer. The converse, however, is not true. To paraphrase Karl Popper, find ing a thousand white swans does not prove that all swans are white; the discovery of a single black swan dis proves that hypothesis.29This is particularly true when a researcher is relying on studies of all bird species (all SIC codes) to determine the color of swans (risk in borfakbelawckorkswerasn)s. AisnLtehmeencohnatsexshtowofn,atshbeersetoasr-einpdluecnetdy ccaasnecsero;fhmeeshoatshdeloiocummaeinntuedsemrsoorfefrtihcatinon22p0ropduubcltiss.h25ed The automobile manufacturers have spent millions of dollars to reanalyze studies that were never designed to determine mesothelioma risk in brake mechanics. On the other hand, they have access to a cohort of their own and their dealers' workers, whose work involves sthiteuaintesdtaltloatuiosne aonthderredpeamiroogfrbarpahkiecsa.llTyhseiymairlearpeemrfpelcotlyy elaetsiownh. oInwomraknayt cthaesesse tshaemyehfaavceiliaticecsesass atocomnetdroiclaploapnud workers' compensation records that would allow them to efficiently conduct such a study. Why haven't they donWeosnog?*and Goodman et al. both exclude evidence from other epidemiologic studies that contradict their conclusions. Hansen, for example, examined the mor tality of automobile mechanics and found an excess mortality from mesothelioma.32 The study reported a single case of mesothelioma over a ten-year period among approximately 20,000 automobile mechanics, yet, because of the rarity of the disease, the authors cdounricnlgudthede rtehpalta, c"eamsbeensttoosfebxrapkoesulrineinisgsk,naonwdnthtoe soicncguler case of pleural mesothelioma is an indication that this exposure has not been negligible." Differential Exposure-determination Bias In his meta-analysis Wong relied primarily on casefceornetnrtoilalsteuxdpieossuthrea-tdweteerremfrinauatgiohnt wbiitahs.pArosbbleesmtoss oefxdpiof sure is difficult to determine since many individuals are unaware of their exposures, especially in secondary cases in which an individual might inhale asbestos fibers from the clothing of a spouse or parent return ibnrgakheowmoerkf,rommanaysbinesdtiovsi-druealalstedenwgaogrke. iInnptahret-tcimasee oorf amateur mechanic work, yet otherwise engage in other cdaerteeermrsiannadtiomnayisthbearseefdoreonbeaclcalsassisfiiecdalasepuindeexmpioosleodgiicf determination of "regular occupation." Of course the direction of any potential or actual bias is a key issue in Ranaanlydzoimng ietsxppoosteunretialmeifsfcelcatssoifnictahteiornesublitasseosf aresstuudlyts. tcoawsea,rdbsecthauesneuellx.pHooswureeveirs, masowree doifstceunssmbieslsoewd, iinn tthhies exposed cohort, the misclassification is likely to lead to an inverse relationship between exposure and disease. In the case of mesothelioma, in which the time between diagnosis and death is usually a matter of months, retrospective study designs such as case-con *Or perhaps their lawyers have already cond ucted such a study and kept it secret. As has been noted, in tobacco litigation, company lawyers often contract with experts to conduct studies. If the studies are unfavorable the company does not have to produce these in liti gation. The brake manufacturers' lawyers have funded and actually helped conduct some studies related to brake exposures.30'31 VOL 11/N O 4, OCT/DEC 2005 w w w .ljoeh.com Abuse of Epidemiology 363 trol studies often require the interviewing of surrogates (spouse, children, etc) to determine exposure. Direct interviews of cases or controls are far more likely to uncover prior brake exposure than are interviews with surrogates. Individuals who did not work primarily as automobile mechanics are likely to be misclassified as unexposed if surrogate interviewees are unaware of any part-time or amateur brake work done earlier in life. Since mesothelioma cases are more likely to be deceased at the time of interview than controls, a dif ferential exposure-determination bias may show that the "exposed" have less risk than the unexposed. Nextof-kin interviews are particularly subject to this result, since brake exposures often occur among "shade tree," occasional, or even high school mechanics in automo bbielerereppoaritredcobuyrsepsa.tTiehnetssetheexmposseulvreess athreanmboyrerleilkaetliyvetso, pthaertirceuselaarrlcyhcehrsilddreetenrmofipnaetdieenxtsp.oIsnurme afrnoymofdtihreecsttiundtieers view of controls, in comparison with greater reliance on interviews with survivors of mesothelioma patients. Next-of-kin interviews were therefore used more often to estimate exposure in mesothelioma patients than in controls. Since direct interviews are more likely to elicit a history of exposure to brakes than next-of-kin inter views, brake exposures are systematically more likely to be recorded for controls than cases. Goodman et al. acknowledge the problem of next-of-kin interview bias bouf tthoims bitiaasnyisdaisrceussuslitosnhoofwthinegfathcat tthbartaktheewcoornksepqruoetencctes astgraaitnesdt bmyetshoethfealciot mthaa.t Tmhoisst pohfetnheomsteundoiens Wis odnegmaonnd Goodman et al. selected had rate ratios ofless than one. Wintoenrgv'aslo0v.e6r6a-ll1p.2o3i)n,tweshtiilme aGteowodasm0a.9n0e(t9a5l%. dceotnefrimdeinnecde that exposure to brake dust produced a statistically sig nificant and important reduction in the rate of mesothelioma of 0.67 (95% confidence interval 0.53-0.84) in all studies combined.t The finding that bsyrastkeemwatoicrkerrerdourcienstthhise mrisektao-afnmaleyssoist,hbeleicoamusaeiseaxpsiogsnuroef to asbestos brakes cannot reduce the risk of developing mesothelioma by 33%. That is, however, what these selected meta-analyzed data showed. Other Bias Issues Tinh2e0s0t0u,ddyecmoonndsutcratetedsbhyoAwgduidffoeraenndtiaolthexerpso,spuurebldisehteerd mination can bias results when exposure histories of cdairseecstly(w) haore amreoredeliakdelyortotboeodseitcekrmtoinebde biyntseurrvrioegwaetde interviews than are exposure histories for controls.33 JTheyfound combined RRs of 0.92 (95% Cl 0.55-1.56) in Tier I studies and 0.81 (95% Cl 0.52--1.28) Tier II studies. Forty-four percent ofcases were deceased at the time of the study, and thus researchers determined their expo sures through family-member interviews. Among the control group, however, the researchers directly inter viewed all but one participant Gases, therefore, may have been less likely to be classified as exposed, which would bias the resulting risk ratios to be less than one. In addition, the researchers considered only occupa tions that had been held for at least six consecutive months, meaning that workers with exposures from part-time or amateur brake jobs were omitted. This random misclassification biased the results of the study towards the null. The authors also assumed without ques tion that brake work puts an individual at risk for asbestos exposure, by placing brake mechanics in the "risk of exposure" category in their analysis. The rate reartsiowfaosretlheivsaetendti,reatca2t.e5g9o, rbyuotftahsebeasntoasly-esixspoosfeda wsmorakll number of individuals within the category of at-risk occupations (i.e., the brake mechanics) cannot lead to anyTshoeunstdudcyonpculubsliisohnesd. in 1994 by Spirtas and others provides additional evidence that exposure-determina tion bias is a real threat to study validity.34In the study, younger cases had a higher attributable risk for mesothelioma from asbestos exposure than did older cases. This is likely to have resulted from more accurate exposure assessment among younger individuals, who were better able to remember more recent exposure esevsenotfs.thSepimrtaesmebtearls. doifdthneotcionhcolurdt we ihnodreeppeonrdteednt"abnraalkye lriantiengowf omrkulotirprleepaasirb"es(t1o0s5 eoxfp7o4s1u)rbesecianustehiosf tghreouhpig.3h4 Unontleikeththeseeaunthoonr-sbroafkteheeoxrpigoisnuarlesstuodyr, Wthoengorfaigilisntaol authors'exclusion of these workers, and relies on these rejeTchteedaudtaotma ionbhilies mmaetnau-afancatluysriesr.s6paid Hessel et al. to "update" the Spirtas cohort and reanalyze its brakeworker data in 2004.4 The results of the analysis pre sented by Hessel and colleagues reveal the mthod ologie problems with the use of the NIH data set. First, the analysis of Hessel et al. indicated that non-occupa tional brake work prevented workers from developing mesothelioma (regression beta was -0.184). In addi tion, their analysis found that higher levels of exposure to asbestos among workers performing occupational brake work were more protective (regression beta was -r0es.3p0o7n)s.eInreloatthioenrswhoiprdsb,ettwheeseen aeuxthpoorssurfeoutnodaasbdeostsoes-lcionuerdseb, rbarkaekse wanodrkrceadnuncoetdrerdisukceowf omrkeesrost'hreislikofmoar.coOnf tara"cdtoinsge''m-deespoetnhdeleinomt ma,aannnderi.t Tcehretaiunnlybeclaienv'at bdloe tdhoisseinrseusfpfeornesde rferolamtiosnysshtiepmiastficurtehrreorre.vHideesnsceel tahnadt thheisstucodly leagues also examined six other job categories of 364 Egilman, Billings www.IJoeh.com . INT J OCCUP ENVIRON HEALTH asbestos-exposed workers, all ofwhom have been found to have developed asbestos-induced mesothelioma or iontghewroarskbeersstoesx-preolsaeteddtodisaesabseestoins ointshuelratsitoundiiens,fiunrcnlaucde or boiler installation or repair,35'36 building demoli tion,37,38plumbing or heating repair,39,40elevator instal lpartoiodnucotirornepoafipr,a41peprropdroudctuicotns.4o2f-45asIbnesctoonstrtaexsttitloesp, raenvdi ous cohort studies, Hessel et al. found no increased risk of mesothelioma in workers performing these jobs. They found elevated odds ratios for shipbuilding and repair work (OR 6.04, Cl B.74r-9.75) and insulation work (OR 3.38, Cl 2.20-5.17). While elevated, these odds ratios are orders of magnitude lower than those found in cohort studies of these same two groups. For example, Selikoff found that about 10% of insulators died from mesothelioma, giving an approximate rate ratio of 1,000.46 Thus, unless one is willing to believe tthhaatt pbrreavkioeuws coorkhoprrtostteucdtisesaogfaiontshtemrjoesbotchaeteligoomriaesaanrde wthraotngth, eHiresmseeltheot dal.bihaasvese rpersouvltisdetdo rtehseultnsuilnl daicnadti/nogr reverses actual effects of exposure, turning harmful expToessucrhekseinetot baeln. eufisceidal oanneso. ccupation-classification scheme that may also have biased results towards the null.47 The authors stated, "In addition, grouping of occupations was likely to result in non-differential misclassification, usually biasing risk estimates to the null value."Wong omitted this critical fact. The study is also another example of differential exposure-determina tion bias--the authors determined exposure histories for 13.6% of controls' histories of exposure were obtained through next-of-kin interviews, compared with 33% of cases. The study also had a higher non-par ticipation rate among controls, which could lead to selection bias, and interviewers were not blinded to the case-control status of the interviewee, possibly leading to iTnhteervWiewoietorwbiiatzs. and Rodelsperger study focused solely on the issue of a relationship between automo bstiuledymiescohfannioc wvaolruke ainnddmeteesrmotihneilnigomthae48rPealarttioonfsthhiips between any asbestos exposure and mesothelioma because the authors used lung cancer patients as con trols. Exposure status is likely to be similar among the control group and the cases, which would lead to little discernable difference in risk between the two groups, despite any real association that may exist between brake work and mesothelioma. Since asbestos exposure is a well-known risk factor for lung cancer, any mesothe lioma case-control study with lung cancer patients as athteorcboenltorowl tghreonuupllw. iTllhaislmstousdtycderetmaionnlystprarotedsutcheisreesfufelctst waselwl,esllinacseltuhnegrcesaenacrecrheprastiuensetsd a"spocpounltraotilos.nTchoentorodldss" ratio (OR) for "hospital controls"was 0.75, almost half that for population controls, 1.32. Rather than report both results, Wong presented an average OR of 0.87. Goodman et al. published both ORs but included only the lower average in their meta-analysis.3,6 McDonald and McDonald conducted a case-control study, published in 1980, examining mesothelioma cases from Canada and the United States.49 Like the WanoiitnoawpiptzroapnrdiaRteodcoelnstpreorlggerrosutupd;yc,otnhtirsosltusudbyjreecltisewd eorne patients in cases diagnosed by pathologists in which "pulmonary mtastass were present from a non-pul monary malignant tumor."48,49 For example, the con trol group may have included patients who had died from mesothelioma or other asbestos-induced malig nancies such as laryngeal or colon cancer.50,51 The choice of such controls biased results towards the null. In addition, the study reviewed occupational status only ten years prior to death. Since mesothelioma has a min inmout mnelcaetessnacryilyperaidoddreosfsabanoyut stuebnjeyceta'rss,mthoisststruedleyvdaindt leuxnpgo-sfuibreesr.buMrdcDenosnbaledtweetenalc. osnutrbosleqaunednmtlyescoothmelpiaormeda patients and found no statistically significant difference blietetw.52eeTnhitsheisafmuortuhnetrs oevficdhernycseotitlhea,tamcoonstirtoelsorwcerroecindoo-t adequately chosen, since they were clearly exposed to as much asbestos as the 99 mesothelioma patients. This misclassification biases the results to the null. thaBt orethveWalomngeasontdheGliooomdma rainskettoal.birgankoe-rme oanthuefracsttuurdiinegs workers. Since these authors base their argument for lack of risk on fiber length and the general issue of whether or not chrysotile can ever cause mesothe lioma, they are clearly relevant And while they are also compromised by numerous mthodologie and other problems, they do show increased rates of mesothe lioma in brake manufacturing workers.53,54Dodson has critiqued the "short fiber" argument elsewhere.55 The rqiuskesttoiobnraokferi-smkatnoubfraacktuer-irnegpawirowrkoerrkseriss,rseilnecveantht etofitbheer used in the manufacture of brakes is the same length as that in new brakes. If there is relevance to the short- fbirbaekresa.rgBurmakeen-rteaptaairllw, iotrrkeelrasteasretoexepxopsoesdurteosffibroemrs fursoemd both new and used brakes from tooling parts as well as bloIwn-o1u9t8o2f, bMrackDeo-wnaeladr danudst Fry published results of a study of a Connecticut chrysotile asbestos brake-manu facturing factory and concluded that, `The failure to feimndplaoysienegsleofmtehseoltahregleiostmaanadmooldnegst14a6sb9edsteoasthfascitnor4y73in8 the state suggests that chrysotile in the manufacture of friction products was not responsible for the prob lneemct.i"c54u t I n p laa nstt,udMy coDfotnharlede ept l ant al. s rienpcolurtdeidn g"othnely "Co0n.4 ccahsreyssotoilfe amsbeessottohseflaioctmoraiespoevrer1t,h0e00couinrdseivoidfuaaplpsroixn VOL 11 /NO 4, OCT/DEC 2005 w w w .ijoeh.com Abuse of Epidemiology 365 imately 40 years [1/100,000 or 10 per million per year]. They then asserted that, "This finding supports much other evidence that amphibole are mainly responsible for mesothelioma whereas chrysotile has little or no mesothelioma-producing potential."54 The authors present this evidence, however, without discussing the expected population rate for comparison. Teta et al. report an expected rate of 2.2 cases per million each year for the total population and 4 per million for Con necticut.56 Comparisons with expected population rates are also problematic, since at least half of all mesothelioma cases are caused by occupational expo hsuigrehsertothaasnbe1stopse.rTmhielliuonne.5x7pFoosreda etrxupeecctoemd praarteisoisnntoo be drawn, it would have to be with only those cases that were not occupational, since the risk of mesothelioma among automobile mechanics should not be compared with the risk of workers in other occupations known to cause the disease, but with those who were not occupa tionally exposed. The rate reported by McDonald et al. is therefore at least more than double the expected rate. McDonald's data thus support the claim that chrysotile asbestos causes mesothelioma. The McDonalds'research suffered from other, more serious problems. In 1983, Teta et al. described find ings from the state of Connecticut cancer registry and reported three mesothelioma deaths in the same Con necticut factory and during the same time frame used itnionthoefTMectDa'sopnaaplders,tiundy1.95684O, MnecDyeoanraaldfteetratlh. epupbulbislhicead thhoewirev"efri,ndali"d rneoptorretpoonrttthhee pfulallndt.atTahfirsom"finthael"orreipgionrat,l rceophoorrtte, dwdhaictah firnocmludtehde bmoatlhe mpoerntioanndofwtohmeecno, hbourtt only.58 Two "missed cases" occurred in women who were members of McDonald's cohort.59 McDonald et aml.esreopthoerltieodmtahiant tthhiesyphlaandtfaonudndfaniloedt atosimngelneticoansethoef cases Teta cited from the state of Connecticut registry.53 In their 1982 paper, McDonald and Fry claimed their data were reliable, in large part due to the existence of the very same registry. They stated, "The opportunity for recognition of mesotheliomas was also very good at the chrysotile friction materials plant, as the state of Connecticut has maintained a renowned cancer regis theard."54chCelcekaerldy tthhies rweoguisltdryhoplrdiotrrutoe othnelypiuf bthliecaatuiothnoorsf their paper; unfortunately, it appears that they had not ddiosncerespoa.n59c'6y0Tanedtaient1a9l.8c6laMimcDthoenyalidnftoriremdetdotehxepmlaionf tthhee omission away.59McDonald stated that she and her col leagues omitted the cases because mesothelioma had not been indicated on two of the deceaseds'death cer stitfuidcayt,easn, dththeatptrhimeyaeryxclmuedaesdutrhee othfirodutccaosembeecianustehehier had terminated work prior to the beginning date of their study.59However, in 1986, by the time McDonald had commented on these discrepancies, at least two other cases of mesothelioma in the plant had been reported on death certificates (see Table 1, cases 1 and 2). Both fit the initial cohort definition and the first left employment before there was any alleged amphibole use. These cases occurred after the termination date of the McDonald study. We are now aware of at least six additional mesothelioma cases in workers from the plant, four ofwhom had no other possible exposures to asbestos (Table 1). At least four of these cases are known to the main sponsors of the McDonalds' research, since members of the Quebec Asbestos Mining Association (QAMA) were sued by three of them. Therefore, there were a total of at least nine mesotheliomas in this factory, which used chrysotile until 1957 only.]: Perhaps by unfortunate coincidence, all of these cases occurred before or after the begin ning and ending dates of the McDonalds' studies or the diagnoses were not reported on the death certificates. Death certificates, in general, are often inadequate means ofdetermining a mesothelioma diagnosis, due to the strong possibility of other related conditions' becoming the primary causes of death (e.g., cardiac arrest) and because mesothelioma is frequently misdi agnosed. Mesothelioma was not added as a specific ICD code until 1968.61 The study was conducted in the 1970s, mostly examining data from prior decades, in which mesothelioma was a lesser-known disease that was frequently not accurately diagnosed. In addition, the Connecticut registry was an excellent source of data for a comparison of expected and plant rates. To make the ddaettaercmoimnepatthibelem, ethseotrheegliisotmryasrhaoteusldfohravtehebepelannutswedortko emrsis.sHedadtthheeytwdoonemseos,oMthceDlioonmaald ceatsaels. wthoautldoncoctuhrarevde during their selected observation period. Of course, the most important result of the exposure to chrysotile at this plant is the death of nine workers from mesothe lioma in a population of about 4,500 workers. Although the McDonalds et al. stressed the importance of this cohort in asserting that chrysotile did not cause mesothelioma, these nine cases provide additional evi denDceespthitaettchheryfasocttitlheactaTuesetas ims eascooth-aeulitohmora.of one of the brake-manufacturer--funded papers and works for the bcoramkpea-mnyantuhfaatctguerneer-rfautenddedthrreeeseaorfchtheerssefapilaepdertso, uthsee asunryeotfothchersyessottuildei-elsinoerddbartaakteos.a5s'6seOssthtehreihnadzuasrtdryo-ffuenxdpeod researchers have also failed to use this information in acsosmespslinetgecdharyrsisoktilaessriesskssm. Renetcefonrtlya,sbCersutomspfoarntdheBeErPmAa.6n3 To some extent this was an update of a previous assess- }Some anthophyllite was used beginning in 1957, and about 400 pound of crocidolite were used experimentally in the laboratory from 1964 to 1972.53 366 Egilman, Billings w w w .ljoeh.com INT J OCCUP ENVIRON HEALTH TABLE 1--Mesothelioma Case in a Chrysotile Brake Manufacturing Plant Date of Birth Date of Diagnosis Raymark Job Work Dates Description Other Exposures Case 1 5/20/1908 5/21/1980 1926-1937 O ffice worker Shipping clerk None: no known home repair, friction, or additional occupatio nal exposures. Death Certificate/ Autopsy Left lobar p n e u m o n ia , mesothelioma Case 2 7/7/1914 9/3/1985 1940-1985 Unknown Unknown M a lig n a n t mesothelioma Case 3 4/21/1930 3/24/1994 1983-1984 Unknown Was also exposed to asbestos In th e Navy, a n d h a d other occupational exposures Including EB, A m erican Linen Supply, Upjohn and U.S, Rubber Respiratory failure, mesothelioma Case 4 9/15/1918 9/16/1996 1937-1962 1970-1968 Plant worker None: w orked In security and sales for other occupations; no known hom e repair or friction exposure Respiratory failure, mesothelioma Case 5 10/12/1991 5/24/1999 1940-1941 C a rd in g / spinning room laborer wNoonoed:wtorurkcek rdfroivreorthaenrd occupations Mesothelioma Case 6 9/23/1930 3/14/2000 1965-1985 M achine operator None: no known home repair, friction or additional occupatio nal exposures Lung cancer Autopsy report: mesothelioma Based on Lewlnsohn's62Connecticut expected rate of 4 mesothelioma c.ases per million, the expected number o f cases Is 1.4, observed number Is 9; therefore, the RR Is 6,4 and p < 0.05. ment they had done for the Asbestos Information Asso ciation, an asbestos industry lobby and public relations front, in 1984.64Berman and Crump ignored the pub lished mesothelioma cases from the friction plant that McDonald had acknowledged in 1986, and based their dose-response analysis on the assumption that there were no cases in that plant, Pbeetrwhaepens ththee19m86osKt Linvtaelrueesteisntgimaotfestahned cthheancguers ruecntst pKlLanvtailnueCeosntinmeactteicsuitnv(MolvceDsotnhaeldfreicttiaol.n1p9r8o4d). Arelstphoonusgehwaas eresltaimtivaetelyd fsrmomallt,hipsosstiutidvyeinexthpeos1u9r8e6HtheaatltthhisEifsfeescstsenUtipaldlaytae,nethgeatbiveeststcuudryre(nnot eesxtcimesastreisiks attributable to asbestos) 63 TCertuamspe'rsvpeadpears, aanndoafflitchioaul grhevtihewe eormoitfteBdecrmasaens waenrde noted in 1983, along with the need for further follow-up on this cohort, Teta recommended that the EPA adopt the Berman-Crump model before the cohort follow-up winacsoprpeorfroartemtehde.s6e5 T19h8e3Bdeartam, aann-dCwruemhpavme roedpeolrdteidd onuort partial follow-up of this population in this paper. BQeArMmAan satnuddieCsruambopuatlswo hreiclihedthoen daoustheoersstimthaetmessefrlovems sstiaotned[,o"fItparertmicaliensdodsoeubetsftuiml,ahteosw, etvheer, ownhlyetmheeracsuorneveorf exposure in the QAMAstudies] to a fibre equivalent can have much epidemiological validity."66 Inadequate Sample Size Because of the broad nature of most of the studies, examining mesothelioma risks among many different occupational groups, the numbers used by Wong from each study (i.e,, the number of garage mechanics in each study) is usually less than 25. Hessel et al. (the high easnt-OraRtedfosrtubdraykienwGoorkoedrms abnaseetdaol.n'sorne-lyantawlyosicsa)serespaoftretsr a"cnodnotrnolalisnigng"lfeocraisnesaufltaetriocnonatnrodllsinhgipfbourialdllinogtheexr ppoostuerne tial asbestos exposures. Any conclusions relative to safety based on these tiny sample sizes are reckless. Use ofInappropriate Control Group It is extremely difficult to find a comparable control group for hospitalized mesothelioma cases. Due to the rarity of the condition, mesothelioma cases are often rtreoflesrwrehdo tdoolanrogtehmaveed"ircaarlec"ednitseerass.esSowmillenhoot srpeiftlaelctctohne true demographics of the desired denominator popu lfarotimon.ruPraatlieonrtssuwbiuthrbmanestoothuerbliaonmhaoasrpeitaolfstefnorredfiaegrrneod smisenotrs tarreeatlmikeenlyt.toPabteienotvserfrreopmrestheensteeduirnbatnheecnovnirtoronl VOL 11/NO 4, OCT/DEC 2005 w w w .ljoeh.com Abuse of Epidemiology 367 group and are more likely to have been exposed to brakes than the rural or suburban neighbors of the cases (controls that more accurately reflect the demo graphics of the cases). This effect helps explain how all but one of the studies Wong and Goodman et al. ana lyzed found that brake work reduced the risk of the development of mesothelioma. Misuse of Confidence Intervals Ainltmerovsatlsaltlhaotf itnhcelusdteudaiersatreevriaetwioedgrreeaptoerrt tchoannfiodneen,c5e,6 Wpohinent ersistikmraatteiosscaarnenboet ibnegccoanlcsuidlaetreeddopnarstimcualllarslaymapclceus, rate, and examination of confidence intervals is more important in determining risk. Wong and Goodman et al., however, mention these confidence intervals only in passing and base their entire analysis on point estimates. D IS C U S SIO N Despite the clear methodologic problems with the abovemeteanlt.ioannedd Wstuodnigespurseesdeinnt tthheeimr maesta-uannqaulyessetsi,oGnaoboldemeavni dence of the absence of an association between brake work and mesothelioma.5,6None of the studies used was designed specifically to examine this association, bring iantgpliany.thDeepsopsitseibtihliitsy, Wthoant ugnpkrnesoewnntsctohnefsotuunddieesrsucnoduelrdthbee heading of "Epidemiologic Studies of Mesothelioma among Auto Mechanics" and Goodman et al. tide their Vpaepheicr,le"MMeecshoathneicliso: mAaMaentad-aLnaulnygsisC."aAnscwereahmavoensghoMwont,oar wstuodrkyethxaptoissunroetisdeliskieglnyetdo sbpeecfrifaiucaglhlyt wtoitdhebteiarms dinuee btoratkheecomplexity of exposure determination. The World Trade Organization issued a ruling on the matter in September 2000, after extensive expert testimony led to a conclusion that chrysotile asbestos does produce a real risk of mesothelioma to mechan ics.67The ruling allowed European countries to ban the use of asbestos-containing brakes. Both papers omit this information as well. licaAtifotenropfutbhlaictaotifoGnooofdWmoannge'stpala.,pceornbsuidt ebreafbolreeetvhiedepnucbe emerged from studies conducted in Australia by Leigh et al. supporting an association between automobile m6,0e0c0hainnidciwviodrukalasnidnmAeussottrhaelilaio'smeax.teOnfsitvhee ampepsrootxhiemliaotmelya mdeaantyh, rtehgeistorny,ly1.s2o%urwceeroefauastobmesotobsileexmpeocshuareniwcsa.s68-b7r0aFkoer lining repair.68,69 In a number of individuals examined, only chrysotile asbestos fibers were found, indicating that chrysotile exposure alone can cause malignancy.69 quaOtenlythaedwdrheoslse,thWeoinsgsuaendofGtohoedrmelaantioent sahl.ipfaiblettowaedeen brake work and mesothelioma. The small numbers of individuals pulled from larger studies, from which Wong reports risk ratios to support his argument, lack statistical power as well as epidemiologic meaning due to the large amount ofpotential bias. The meta-analyses begin to the address the problem of a lack of statistical power. However, since the data used for the analysis are handpicked from the literature to the exclusion of con trary evidence, and those that are included are likely to snuoftfererlfiraobmle.vUarniotiulsptryopsepsecotfibvieasc,othhoerrtesstuuldtisersesppoerctiefidcaalrley designed to determine asbestos exposure from brake lcionnincglusiinosntasllaabtioount tahnedmraegpnaiitrudareeocfothnidsurcistekdc,adnenfoint ibtee made. It is clear that a risk exists.25 Moreover, the entire argument addressed by these analyses, regarding the risks of chrysotile, is specious in light of the following undisputed facts: 1. Chrysotile causes lung cancer.72 2. Chrysotile causes asbestosis.72-74 3. Chrysotile used in the United States is contami nated with tremolite--an amphibole.75 4. Products made from chrysotile contain tremolite.78 5. Chrysotile is concentrated in the pleura.77 6. Populations with mixed exposures to amphiboles and chrysotile have double the rate of mesothe lioma of those exposed to amphibole alone.78-80 7. Basrbaekseto-rseepxapirosuwroerskferrosm ggertindsiunbgstbarnatkiaelanadirbclourtnceh parts and blowing out dust from brake assemblies.25 Aamlmpohsibt oallel bwriatkhe, owroarskweresllhaasv,echhraydsostoilme.eTehxupso, seuvreentiof csyhnryersgoitsitliecaallloynwe iitshnoamt aphciabuosleesoftomeisnoctrheealsieomthaeitiancctis dence of this cancer. C O N C LU SIO N The methods utilized by the asbestos and related indus tries--the redefinition of scientific criteria for deter mining cause-effect relationships and the manipula tion of scientific data and scientific inferences--have aonbdscmureesdotthheelitorumea.piIcntucroenrtergaastrdtoingthaesibr eesxtopseretxsp' oospuirne ions, the relationship between exposure to asbestos athned bmraekseo-tihnedluiosmtrya ehxaescluotnivgessiwncheobreeetaninreedcoGgnoiozdemd abny eadt darle.ssantodbWrakonegm.5,a6nIunfachtiusre1r9s7, 3IkeprWiveaateverp,rtehseidCenhatiiarl man of the Asbestos Study Committee of the Friction Materials Standards Institute, an industry trade organi zation, noted the importance of this phenomenon: Ithkantohwis forficntioonwpayroadnuyctosf, ruesgacradnlebses oafbwsohluettehleyr sthuerey are sold as original equipment or on the replace 368 Egllman, Billings w w w .ijoeh.com INT J OCCUP ENVIRON HEALTH mopeenrat timoanrskeotr, waliltlernaotiot nbse isnubthjeectefideldto thadadt itcioounladl rtoesauilrtbinorenxeceasssbiveesteoxspfoibsruer.e Iohf awvoerkbeeresnorabpypsatlalneddetros lheaasrnococfuarrneudm, baenrdosfoinmsetanocfesthwesheerecatsheiss pinrvoobllveemd pkenoopwlebethttaetr.c.e.rt.ainly might have been expected to Ioffthmisakjoinr dOoEf th[ionrgigoincacluresqiunipfamberincta]tiocnusotopmereartsio, nist napepeedafrosrtoedmuceattihoenraelcmaneabsuernesotaorgreudmuecnetcahbaonucet sthoef ucuntnteincgesospareyraetixopnos.suTroe tdhuorsienwg hgorianrdgiuneg,thdartilllainbge,linogr omreontthmerattyeprieaslsofbwecaarunsiengfanbereicdatnoorst aoprpalypptolierresplhaacne dinltienrgmriettpelnatclye,mIesnaty qemuapnhtiattiiecsalalyretheixs jpuossteadinre'tlanteivceelsy smaraijloyr spoo!teLnatriaglehvazoalurdms,earnedpelavceenmsmenatlljuosbersshporpessceannt nifeoepdelersastliyonexspaoreseppeerofoprlme teodhwigithhfoiubtrecocontnrcoelsn.t8r1ations Despite the secret industry acknowledgement of risk uanladteheigphideexmpioosluorgeisc, sGcioeondcme aton ceotnacl.luadned tWhaotnegxmpoasnuirpe troelaatsebdesmtoasligfrnoamnciebsr.akAess wdeoeasndnootthienrdsucheaveasbneostteods,absence of evidence is not evidence of absence; in other words, failing to find an association does not pro vide any evidence that an association does not exist.82 Once it is generally accepted that a substance can cause disease in humans, all that is required to establish cau sation is documentation of exposure to that substance. For example, the OSHA benzene standard did not require epidemiologic evidence of cancer risk in every sector covered by the standard.83 In the case of vinyl chloride, chemical companies themselves removed the compound as a propellant in hair sprays based on animal studies alone because of the risk of unlimited liability.84 The asbestos industry and the automobile industry, rather than hiding behind a "dust cloud" of masibselestaodsining bscraieknecsew, sohrloduwldidde.o the same and stop using References 1. JDuadvgiedsDona,viMds.onR.e5: .CInautsheeNDoi.str2i0c0t4C-0o3u9rt6,4HAasrrbiesstCoosulnittiyg,aTtieoxnas, 11th Judicial District, in re: Asbestos litigation Brake Hearing, 2. MLaanyg5e,r2A00M3,. McCaughey WT. Mesothelioma in a brake repair 3. wPaourkseter.nbLaacnhceDt.J,1R98ic2h;2t:e1r1R01O-,3.Finley BL, Sheehan PJ. An evalua btiroankeodfutsht.eAhpisptloOriccaclupexEpnovsiurorensHoyfg.m2e0c0h3a;1n8ic:7s86to-80a4sb. estos in 4. bHreaskseelmPeAc,hTaentiacsM: Ja,nGeoxopdamndaendMa,naLlayusisEo. fMaecsaosthe-eclioonmtraolamstuodnyg. 5. RGioskodAmnaaln. M20,0T4;e2t4a:M54J7,-H52e.ssel PA, et al. Mesothelioma and lung cOacnccuepr Hamygo.n2g00m4.otor vehicle mechanics: a meta-analysis. Ann 6. Wamoonngg Oau. toMmaleicghnaannitcsm: aepsoprthaieslaiol mofascainendtifaicsbeevsitdoesnceex.pRoseugruel 7. PTaouxsicteonl bPahcahrmDaJ,cFoiln. l2e0y0B1;L3,4L:1u70E-T7,.Brorby GP, Sheehan PJ. Envi rporensmenecnetalofanadsboecstcouspiantiobnraakl ehelainltihnghsazaanrddspaasdssoc(i1a9te0d0wtoithprtehse Renevt).:2a00"4st;a7t:e2-5o-f8-0th. e-art"review. J Toxicol Environ Health B Crit 8. wWhoongwoOrk. wEpitihdebmraikoelos.gyProepf amreedsoftohreAliollmiead Samignoanlg(Binednidviixdu) ailns 9. tEhgeilmmaatnteDr ,ofKjiomhnJ,FB. KikolerontkMo.. OPrcotvoibnegr 1ca6u, s1a9t9io8n.: the use and aroboumse--oafn mepedidiceamlioalnodgisstc'siecnrtiitfiiqcueevoidfetnhceejuindsicidiael itnhteerpcoreutrat 10. Jtoionneso,f tDhaeyD, aRuebveisrt raunldingP.oFgouoed. DCruogrpLoarwatJe. 2A0c0t3iv;5it8y:22P3r-o5j0e.ct. 11. O<hStHtpA://BtoebnazcecnoedSoctaunmdaerndts2o9rgC/FlaRnd1m91a0n.1/3072857(5c)h, tm(el)>.(21090759.). <http://w w w .osha.gov/SLTC/benzene/standards.htm l>. 12. IDnefcaenmtebPeFr.1B0e,1n9ze8n7e.: an historical perspective on the American aMnadcGEaurrvoipneMan, eotccaul p(aetdios)n.aLl asteettLinegs.soInns: HfraormremEaorelys WP, aGrneiengDs:, TrehpeorPt,reEcEaCu.ti2o0n0a1r.y Principle 1896-2000. Environmental issue 13. D5-a5v-2id0s0o3n. IMn.tRhee:DCiasutrsiectNCoo.u2r0t,04H-a0r3r9i6s4C; oRuen:tAy,sTbeesxtaoss 1li1titghaJtuiodni. 2ci0a0l4D. istrict, in re: asbestos litigation brake hearing. January 20, 14. Fnaosr,AApupsetilnle,eTsX: L, Ryn. nOe. LFaibuelkrt,oH, oHuosutostno,nT,XT,Xa,nHdeFcotroAr Hpp.eCllaarndtse: RInidcihvaidrduaHlly., Caanrddeansasn,exAtusftriinenTdXo. fVNirigcihnoilaasHBerrniaanndFerziasFrainasd, TStheovmenasSiImsaoanc FFrriiaass,, CGaerolorgseJeAsunsthForniays,FCrihasri,sJteinssaeMVeajliean, tainndFJroiasse, MFriaansu, eAlpGpealllvaannts, va.sAretlparnetsicenRtiachtifvieeldoCf tohmepaensyta, tLeyoonfJdeeslluPseVtraolcehnemtini 2ca0l0C5.om1p0a4nyS,.Wan.d3dLy9o2n5d;el2l-0C0it3goTRexef.inAinpgp.CoLmEpXaInSy L41td1.5, ,ACppoeullreteos.f 15. AGprepeenallsanodf TS.exRaesl,aFtioounrotefepnrtohbaDbiisltirtiycto,fHcoauussatotino.nMtoayre1la5t,iv2e00r3is.k asoncdiadl opurobblilnegm.dAosme:JaPumbelitchoHdeoallotghi.c1e9r9r9o;8r 9t:h1a1t66h9a.s become a 16. SteorxwHorCth. -MHAeiBnMemCaHnKnI,M19. 9M8.edical Decision Making. 1st ed. But 17. LAEusXtiInS u43K1e7r]r,-M5-c1G6e-e20R0e0f.. NCoor.p.06[2959-0S0.W08.73-dC2V8,0C; o20u0rt0oTfeAxp. pAepapls. 18. MofoTsesxmaasn, SBiTx.thMDeicshtraincti,smTesxoafrkasabneas.tJousncear2c9i,no2g00e0n.esis and toxic ity: the amphibole hypothesis revisited. Bi J Ind Med. 1993; 19. 5S0ta:6y7n3e-r6L. T, Dankovic DA, Lemen RA. Occupational exposure to chhyrpyostohteisleis.aAsbmesJtoPsuabnlidc cHaenaclethr.ri1s9k9:6a;8r6e:v1i7e9w-8o6f. the amphibole 20. --CuglolenneMbRu.t Tnhoet faomrgpohtitbeonl.eAhmypJoPthuebsliiscoHfaesabltehs.to1s9-9re6l;a8t6e:d15c8a-n9c.er 21. MboolessmhyapnotBhTe,siGs.eTehJBe .hAypsboethsteossi-sreislastteidll csaunpcpeorrtaenddbtyhescaiemntpishtis 22. WanadgsnceireJnCti.fiAcsdbaetsat.oAs-mrelJaPteudbcliacnHceeraaltnhd. 1th99e7a;8m7p:6h8i9b-o9l0e. hypoth Heseisa.ltThh. e19f9ir7s;t87d:o68c7u-m8.entation of the association. Am J Public 23. bLoalnegherypAoMth,eNsios.laTnhRePa.mApsbheibstoolse-rheylaptoedthceasinsc:enreaitnhderthgeoanme pnhoir 24. Pfoitrtgsbouttregnh. CAommJinPguCbolircpoHraetaiolnth, .A1p9p9e7ll;a8n7t:,68v.8-C9e.cileJ. Walters, indi vDioduuaglllaysaFn.dWaaslpteerrss,odneacleraesperde,sAenptpateilvleeeo, fNthuemhbeeirrs1a3n-9d7-e7s5ta3t-eCVof. A[1ppSe.Wals.3odf7T5e9x;a1s,9T99hiTretexe.nAthppD. iLsterxicist, 5C8o3r6p]u. s8C-5h-1ri9s9ti9..ACuoguurstt o5f, 25. L19e9m9.en RA. Asbestos in brakes: exposure and risk of disease. Am 26. JKIanudpMpineden. 2T0,0K4;o4r5h:2o2n9e-n37K. . Exposure to asbestos during brake mInadinHteyngaAnscseocoJf.a1u9t8o7m;4o8t:i4v9e9v-5e0h4ic,les by different methods. Am VOL 11/N O 4, OCT/DEC 2005 w w w .ijoeh.com Abuse of Epidemiology 369 27. aGnudstaevxspsoonsuPr,ePtloatodiNes,eLl iedxsthraoumstEaBm, oHnoggsbtuedstgCa.raLguengwocarknecresr. ScandJ Work Environ Health. 1990;16:348-54. 28. fMoralkpeleruHrSal, MmceLsoatuhgehlliionmJKa ,iMn aSlwkeerdBenK,, e1t96al1.-O79c.cJupNaatitolnCalarnicskesr Inst. 1985;74:61-6. 29. lPaonpdp: eHruKtcRh.inTshoen,L1o9g5i9c. of Scientific Discovery. London, Eng 30. fWibeeirrsFrWele, aMseedradzuLriBn.gMgroinrpdhinogloagnicdadl rcilhlainragcotefrfirsitcictisonofpraosdbeuscttoss. 31. ABlpapkleOCcLc,uvpaEnnOvirrdoennHDygR.,2B0a0n1a;1s6ik:1M14,7H-9a.rbison RD. Airborne pasebrmesitsossibcloenecxenptorsautiroenlifmroimt. Rbreagkuel cThoaxnicgoinlgPdhoarems ancootl.ex2c0e0e3d; 32. H38a:5n8s-e7n0.ES. Mortality of auto mechanics. A ten-year follow-up. 33. AScgaunddoJAW, aolrekz ECnAv,iBrolendHa eMaJl,the.t 1al9.8O9;c1c5u:4p3a-t6io. n and risk of malig nInadntMpeledu. r2a0l0m0;e3s7o:1th5e9l-i6o8m. a: a case-control study in Spain. AmJ 34. lSiopmirtaa:s aRt,trHibeuintaebmleanriEskF,oBfearnsbsetesitnosL,exept oals.uMrea. liOgcncaunpt mEensvoitrhoen Med. 1994;51:804-11. 35. Danielsen TE, Langard S, Andersen A. Incidence of cancer a1m99o6n;5g3:2N31o-r4w. egian boiler welders. Occup Environ Med. 36. KmuorrutamlitaytasntuidNy,oNfwatoorrkieYr,s MexipzoustaenditRo,asebteaslt.oAs inhiastroerfiictatilncgoshhoiprt yard. Ind Health. 1999;37:9-17. 37. BtioronwonfSaKsb. eAsstobse-sctoems eenxtpocslaudrebduuilrdiinnggsr.enAomvaItniodn Hayngd AdsesmocolJi. 38. Y19e8u7n;g48P:,4R78o-g8e6r.s A. An occupation-industry matrix analysis of rmoensoHtyhge.li2o0m01a;1c6a:s4e0s-4in. Australia 1980-1985. Appl Occup Envi 39. MEnegdluLnadv.A1.99R5e;8c6e:n4t3d5-a9t.a on cancer due to asbestos in Sweden. 40. pClaunmtobrerKsPa,nSdopnitpagefJiMtte,rsH. AeidmJMIEndPMatteedr.n1s9o86f;m10o:7rt3a-l8it9y. among 41. ABrsebsensittozsE-rAel,aGteidlmraandiMogJ,rGaprhacicelyabEnJ,oArmiroallidtiieJ,sViongeellEev, GatoetrtecroWn. 42. sBterurrcytioGn, Nweowrkheorsu.sAe mMLR,eWv RagesnpeirrJDCi.s.M1o9r9t3al;i1t4y7f:r1o3m41-a4l.l cancers oirfoansbMesetdo.s2f0a0c0to;5ry7:w78o2r-k5e.rs in east London 1933-80. Occup Env 43. BtoenreryalGm, edseoKthleelrikomNHas, iRnefiodrAm,eert mal.inMerasliagnndanmt pillleeursraolfacnrdocpiedroi- l2i0te04a;6t 1W;eli4tt.enoom, Western Australia. Occup Environ Med. 44. Dasebmesetonst JMtex, tBilerowwnorDkePr,sO: kcuonhoAr.t Fmololortwa-liutpy satunddy ocafsceh-rcyosnottriolel 45. MancaDlyosensa.lAd AmDJ,IFnrdyJMSe,dW. o1o9l9le4y;2A6J:,4M3McD7.onaldJC. Dust exposure aanndd mcroorctiadliotlyitieniannmAaminelryictaenxtfialectmoraynuusfiancgtucrher.yBsortJileI,nadml oMsietde,. 1983;40:368-74. 46. lSaetiloiknofwf Iofr,kSeersidimnanthHe .UAnsibteesdtoSs-taatsessocaiantdedCdaenaatdhas,am19o6n7g-1i9n8s7u. 47. ATensnchNkYe AKc, aMdoSregia. n19M91S;,6C43h:e1c-k1o4w. ay H, et al. Mesothelioma sur vHeeilallatnhc.e19to97lo;8c8a:t1e6s3o-8u.rces of exposure to asbestos. Can J Public 48. Wmeocithoawniictzs?HAJn,nROocdceulpspHeyrgge. r19K94. ;3M8:e6s3o5t-h8e. lioma among car 49. NMocrDthonAamlderAicDa., CManccDero.n1a9ld80J;4C6.:1M65a0l-i6g.nant mesothelioma in 50. mFraulmigknianncHy,reBveierwlinaJn.dAmsbeteas-taonsaleyxsipso.sAurme JanInddgMasetrdo.in1t9e8s8ti;n14a:l 51. S79m-9it5h. AH, Handley MA, Wood R. Epidemiological evidence i1n9d9i0c;a3t2e:s49a9s-b50e7st.os causes laryngeal cancer. J Occup Med. 52. MofclDunognailndmADs,oMthcDlioalnatuldmJoCu,rPsoionleNyoFrDth. AMmineerriacla.fiAbrnencOonctceunpt Hyg. 1982;26:417-22. 53. MancdDmonoartladlAityDi,nFraynJSA,mWeorioclalenycAhJr,yMsoctDileonaaslbdeJsCto.sDfruicsttieoxnpporsoudre ucts plant. BrJ Ind Med. 1984;41:151-7. 54. AMmcDeroicnaanldaAsbDe,sFtorsyJfaSc.tMorieesso--thperlieolimmainaanrdytrheepfoirbte.rStcyapnedinJ Wthroerek Environ Health. 1982;8 suppl 1:53-8. 55. rDeoladtseodn toRFp,oAtetknitniaslonpaMthAog, eLneicviitny:JLa. cAristbiceastlorsevfiiebwe.r AlemngJthInads 56. MTeetda.M2J0,0L3e;4w4i:n2s9o1h-7n.HC, MeigsJW, Vidone RA, Mowad LZ, Flan nanerdygJeTo.gMraepshoitchealsisoomciaaitinonCso.JnnOecctciucputM, 1e9d5.51-918937;72.5O:7c4c9u-5p6a.tional 57. Htioilnlearidaanl dG.loMwesdootsheelieoxmpao:sucraesse.s OascscoucpiatEednvwiriothn nMoned-o.cc1u9p99a;- 56:505-13. 58. aMncdDmonoartladliAtyD,inFraynJSA,mWeorioclalenycAhJr,yMsoctDileonaasbldeJsCto.sDfruiscttieoxnpopsruorde 59. uMcctsDpolnaanltd. BArDJ.InMdeMsoethde. l1io9m84a;4i1n:15C1o-7n.necticut. J Occup Med. 60. T19e8ta6;M28J,:8L0e8w-9i.nsohn HC, MeigsJW, Vidone RA, Mowad LZ, Flan n19e8ry6;2J8T:8.08M-9e. sothelioma in Connecticut. J Occup Med. 61. HU.eSa.ltDh eSpearrvtimceenHteaolfthHReaesltohu,rcEedsuAcdamtioinnisatrnadtioWn.elCfaorme,paPruabbliilc itthyeoIfnMteornrtaatliiotynSaltaCtilsatiscssifficoartitohne oSefvDeinsethasaens,dUEnigithetdhSRtaetveiss,io(nVsitoalf atinstdicHs eSaylstthemS;tantiost.ic6s:6S).erOiecsto2b, eDrat2a5,fro19m75t.heDNHaEtWionpaul bVliitcaaltSiotan 62. HLeRwAin7s6o-h1n34H0.C, MeigsJW, Teta MJ, FlanneryJT. The influence ionfcoidcecnucpeatoiofnmaalliagnndanetnmviersoontmheelniotaml aaisnbeCstoonsneexcptiocusut,reIAoRnCtShcei Publ. 1980;655-60. 63. mBeernmt afnorDaWP,roCtorcuomlptoKA.sFseinssaAl dsbraefstt:osT-ereclhanteicdaRl iSsku.pEpPoArt#D93o4c5u. 64. ACpruriml 6p, 2K0.03C. rEuPmA.p's slides used during Testimony for the Adoscbkesetto. sJuIlnyfo9r,m19a8ti4o.n Association. Exhibit 237B OSHA H-033C 65. WEaosrtkersnhoRpestoeaDrcihscGusrsouapP, Irnocp.oRseedpoPrrtootoncothletoPeAesrsCesosnAsusblteasttioosnRelated Risk. EPA Contract No. 68-C-98-148. May 30, 2003. p<dhftstp/a:s//bweswtows_.ereppao.grto.pvd/fs>u.p e rfu n d /p ro g ra m s /ris k /a s b e s to s / 66. BMocgDovosnkaildP J(Ced.).ABsiboelsotgoiscisal iEnffecchtrsyosofAtilsebemstoinse;sPraoncededminilglss. oIfna: Wtioonrsk,in1g97C3.onference. Lyon, France, IARC Scientific Publica 67. iRnegpoasrbt eosftothseapnadneals.bEesutroosp-ceoanntaCinoimngmpurnoidtiuesc--ts.MWeaTs/uDreSs1a3f5fe/Rct. 68. WLeoigrlhdJT, rDadaeviOdsrognanPi,zHatieonnd,rSieepLt,emBebrerry1D8,. 2M00a0li.gnant mesothe 69. LlioeimghaJin, DAruissctroallliaT,. 1M9a4l5ig-2n0a0n0t.mAemsoJthIenldioMmeadi.n2A00u2s;t4ra1l:i1a8,81-924051. 70. H20e0n2d.eIrnstoJnODcWcu,JpoEnnevsiMroLn,HdeeaKltlhe.rk20N0,3e;9t:a2l0.6T-1h7e. diagnosis and raettproibrut toiofnthoefAasdbeelastiodse-rWeloartekdshdoipseoasnesAisnbeasntoAs-uRsetrlaaltieadnDciosneatesxest:, 71. OMcctDoboneral6d-7JC, 2. 0A0s0b. eIsntotJs OanccdupluEnngvciraonnceHr:eahlaths. t2h0e04c;1a0se:40b-e6e.n 72. pBreocvkelank?eCMheRst,.F1o9u8r0n;i7e8r:-3M74a-s6s.ey G, Rossiter CE, McDonald JC. LQuunegbefcu.nActricohnEinnvcirhornysHoteialelthas.b1e9s7to2s;2m4:4in0e1-a9n. d mill workers of 73. LchidadneglelsDo,veErys2s0enyeGa,rsTihnomrealsatDio,n MtocDcohmrytsldotiCle, Rexapdoiosluorgeicianl 74. QRousesbiteecr.CInEh, aBleridstoPlarItJ., 1C9a7r5t;i4erPPtH2:,7e9t9-a8l1. 3R.adiographic changes iEnnvcihrroynsoHtielealtahs.be1s9t7o2s;2m4i:3n8e8a-4n0d0.mill workers of Quebec, Arch 75. RSeebspasirtiaetnoryP,cManccDeorninaldchJrCy,soMtilceDtoenxatillde AanDd, mCaisneinBg, iHndaurlsetryieRs:. 4e6x:p1o8s0u-7re. inferences from lung analysis. Br J Ind Med. 1989; 370 Egilman, Billings w w w .ijoeh.com INT J OCCUP ENVIRON HEALTH 76. COoraulrtd, eHpoasrirtiisonCoofunWtyi,lliaTmexaLson1g1ot.h 5J-u2d3-ic20ia0l5.DIinstrtihcet, Dinistrriect Asbestos Litigation Chrysotile Hearing. ' 77. mSuazluigkniaYn,t YmueesnothSeRl.ioAmsab.esItnods Htiesasulteh.b2u0r0d1e;n39:s1tu5d0-y60on human 78. Seidman H, Selikoff IJ, Gelb SK. Mortality experience of taom4o0styteeaarssbeasfttoesr foancstoertyowf oshrkoerrts-:tedromsew-roerskpeoxnpseosruerlea.tiAonmshIipIsnd5 Med. 1986;10:479-514. J 79. Selikoff IJ, Hammond EC, Seidman H. Mortality experience of AinnsnulNatYioAncwaodrSkceir.s1i9n7t9h;e33U0:n9i1te-1d1S6t.ates and Canada, 1943-1976' 80. Acheson ED, Gardner MJ. Mesothelioma and exposure to mix- Htuereasltho. f19c7h9r;y3s4o:2ti4l0e-2a.nd amphibole asbestos. Arch Environ 81. IAndddursetrsys, ogfivIeWn aWt tehaevAer,nn"uAaslbMesetomsbaenrsdhipthoef FthreicAtisobnesMtosatTereixal mtileeeItninsgti,tu2t0e05o.n June 27, 1973, reported in the minutes of the 8832.' OSHA Benzene Standarfde, v2i0d0e5n.ce is not evidence of absence84. MticasrkoofwInitdzuGstEri,aRl oPsonlleurtiDo.n.DBecerekitelaenyd, CDAe:nUianl.ivTehrseitDy eoafdClyalPifoolri nia Press, 2002. VOL 11/NO 4, OCT/DEC 2005 w w w .ijoeh.com Abuse of Epidemiology 371