Document kgryE5JERYKe20Xq0GgDXJm0

"J= I b n i i i 1 i r - r 3 AYLUCAN JOLRWAoL? EmDcmDuw;v C d r c 1989 bv The Jobm HopLwu Cniwrnity School d H All nghu mend v .ndPublic H d t h Letters to the Editor RE:'BENZE.VE .4ND LELKEMIA: A REVIEW OF THE LITERATURE AND A R ~ C ~ In their recent paper, .\ustin et ai. I ) conclude A Krial point in the review by Austin et al. ( 1\ that: "In the aggregate. the epidemiologic evidence which we take iWUe is their dismissal of the -+,Ic suggests a link between benzene and leukemia" ( 1. p. etiologc asdociation between benzene and hemato- 428). Although we hold the opinion that Austin et al. poietic malignancies other than acute mono. are unnecessarily tentative in this conclusion. we con- myelogenous leukemia. TWOstudies cited by Austin- cur with it. We are concerned. however. by several Arp et al. ( 3 )andCheckowayet ai. 14)--foundelevatcd aspects of their review. and particularly by their dis. relative risks for mortality from lymphocytic Ieuke. cussion of epidemiologic risk assessment. Austin et al. mia. However. Austin et al. dismiss those findings I criticize our group for having used a risk assessment 'not persuasive" or as possibly due to erpowm to model based on a nested case-control study. They 'solvents" other than benzene. Austin et ai. not suggest that in our estimate of the likelihood of leu- consider the poaible etiologc association kemra in persons e x p o d to benzene we should in- benzene and lymphoma. Three recent epidemioloDc stead have used data based on a stratification of studies-our own analysia (5) and those by Wong (61 'I person-years from our entire cohort; indeed, Austin et and DeCoufle et al. (7)-have, however. found ex- al. utilize such an approach in their own risk IWCUI- mortality from multiple myeloma and other f o w ment. lymphoma in persons exposed 10 benzene. Wong ( 6 ) We consider their criticism of our approach to have argue8 that mortality from leukemia and lymphom several shortcomings. One problem is that the expo- should not be conri&red separately in popuktiom sure categoria in a sundarttized mortality ratlo exposed to benzene. because of the diagnostic o M r b (SMR) analysis are inherently a h i t r a m vanation in between the two categories of dircuc. the well known the definitions of the bundana of r h m categoria transition or propemion of many lymphomu into can have a profound effect on the c d c u l a t d risks. leukemia and the poor concordance between hum ThL effect is especially likely 10 be evident when there pathologc and either immunologc or cytogcneucc h - are few death. sfication schema. We suggest that a formal amm- By contrast. the casr-concml approach to risk as- ment of the possible etiolodc m i a t i o n b t m n sesmcnt provides an elegant m e w for modeling cu- benzene and lymphom is rrqurrcd. mulauve exporurr with the use of an internal cornpar- iron group. Thir u a more appropnate referent roup RlWEUNCES 1u thur the United States population and is far leas dupararc in wrrm of theu gcnerrl health. A b . in a study of no h t d poplktlon. such as our cohon. whrch mght have encount4redmmc occupauod cxmumother than bennne. the cue-contml approach I. Aurcin H. D c h U E. Cok P. Benzrne ud leukcmu: 1 WYI of the liurature md a ruk -merit. Am J Epidcmiol ISl3&127:419-39. 2. B m b w NE. Lubm JH.Muek P. et d. Multlplruuve d l r d cohon udyw. J Am Sue Awoc 1983.iB:I- prowdea an opporrrrmty to balance out thorn expo- 12. SUM: both c a m and conrroh should have had equal Ilkellhood for e x p o w c ta other matcnala. such as solvenu. well aa for pnor expomure to in arean of the factones wuide the depanmenu under study. Finally, it hu been demonstrated (2) that nsk aucrrment uama properly chosen sample of controls wdl mult in atmates of mk that are nearly i d e n t l d to those whxh mult from u.c of the full cohon, and 3. Arp EW. Wolf PH. C h e k m y H. Lympbocyw leukernu mdexpaurea to k n z r m mdotherd w n u in rhr nrbbcr induty.J Occup Mcd 198325:598602. 4. Checkorav H.Wikcaky T,Wolf P. et d.An evduauon O f the - l l t i O N Of hh- M d N b b indurvv -1- vent rrpaum. Am J lnd Med 1 9 8 1 : 5 : W 9 . 5. Riruky RA. srmrh AB. Hornurq R et 11. bnmm d l d e m u : M epderniologc ruk LucymenL N End J Mod 1987.316:1OU-50. this approach is much more coat efficient than n s k assessment hudon the full cohort. Austin et al. ( 1) have attempted to utilize our SMR analysis. which &vides the cohon into four categona 6. Won6 0. An indusuy 4 m o d i c y study of chemical workers a c u p a u o d y e x p a d to k n n m Br J Ind Yed 1987.~:363-95 Bhttner WA. A yOmhey Imong I of cumulative benzene expoawe. and to w the mid- ~ v ~ f ~ r : ~ ~ , ; ~ ~ ' e Otoher .Icnu points of these caugones to estmate leukemia nsk. This methodology is not optimal. &CAUSEthe penon- year Qstnbution in each category is highly skewed f~ the lower exposure. a more appropnate procedure Robert A. Rinskv Diouwn of S u r v e k e . Hazard Eualuatwns and would have been to use the person-yean weighud Field Stdres average e x p u r e for each category. With such an .Vatmnal Institute for Occupatwd approach. Austin et al. would have obmned a nsk SoIpty and H e d h estimate closer to ours. CincinMtL OH U226-1998 AND A RISK stin e t ai. 1 \ with sal of the Possible ene and hemato. in acute mono. cited by AustinIb-found c h a t & mphocl;ric leukethose tindings as e to exposures to stin et al. do not miation between tnt epidemiologc hose bv Wong 16) wer. found excess nd other iorms of mzene. Wong ( 6 ) a and lymphoma y in populations hagnostic overlap E. the well known I. iomas into e . $ e n hlsro.cytogenetic clas- a formal assessx u t i o n between Y and kukcmra. a -menc 4m .I a1 .wultrpllcarlvc k t w lga3.xi - n p k a w l r itukemra \ m uIn tht ruobrr AIAn c\aluAIion mer i n a u i n -I .JM9 a1 &nunc and ..rnent \ Engl J .cudv of cnemtcal m e Br J Ind Med Uonahtx amoni and other aiCnlS THE AUTHORS REPLY_ - In a recent Letter to the E d m r of the New Englond Journal of Medlcine 1)pchluning to a n s k naaessment by finsky et d.(2).we identified a hscrepnncv be- tween the srandardutdmohality ratioa (SMFb) they tatimted and the results of a nested case-control study they b d on the same cohort of phoofilm workers. We also porntad out that theu use of the neated cane- control analyam wan mefficient because benzene ex- p o r u m had k n eatmated for all cohon members and therefore there WM no r e w n for nratnctmg their nsk asaunment to nme leukcmu cmea and 90 con- trols. We .Lo noted in t h t l e t m m d m our m e w article (3) that Rmrky'r mk amemment d t war considerably tugher th.n d othen. including o w . We rgnrnth Ruulryetal. (4) that a a d cuc- control should y d d result. n w l y icbntrul to thoee of a full cohort d y u r of the MIM data aet. Incbsd tbrr uthe renoo we were n r r p d t h t theu SMR and caamxntrol d y w a were ao dlpurtc. Their letter mphea emncourly that we hehew that the m k umument rhould have been byd on theu SMEL. Although we Qd w e the SMEL. we did 10 tmcauae SMR. were the only result^ mrkbb on the enure cobon. Cox's proporuond huucb d l (5). or other utable m e t h h l ~sh, ould b v a beon u d to d y z e the complete cohort. k k y a rL contend that thew cmwxmrol study u ' e l w t " bsum 1) it dorr for m&Uing of eumuhuve e r p o w c whrle controfhq for potential confoundem 2) it pamu m d w u o n of efm modifiat~~nan. d 3) UI m u r d compmmn parpan b e d However. thaoobpc- tlvm could bve b a n r t u d mom efficl.ntly by appropruu d y a m of the cornplat8 cohon lady. It u not 'elegant" to ducud mfomuon nwdlcrriy. We nquartbd that either M rrulyur of the full cohort be mrud or that UI erpLarnon bc p m & d u u, why the d y s m rp)rutncwdu, a rubwr of t& cohon (1). Neither w u p m d d (6). Instend. they contend thrt ow SMR d y u r would b v e yddmd a ruk . u r u r m n t d t c l m r to what they M in the u w c o n t r o l mdy d r r , hd used difemut ICQ- for the expowe categonu. Thu pomt u largely urel- want and a d y u not an adequate explannuon for the diapu~tybetween the d t . of the cohon and cubcontrol d y u r . Funhennore. the conLuntlon rn not nrb.untuud rnth d n a although such data d d have been p m d e d eudy by Rmsky et al. Our m e w evaluated awed camcontrol stu&ea of leukcmu amonc nrbbcr workers. Two stud~esre- ported a pouuvc rekuon ttetween knrcne and lym- phocytic leuhemu (7, 8). H a v e r , the relative risks per~lruryto other solvenu IO both studin were nearly identical w thw for benzene. We pomted out that the mvestmton of one of these mdm .&o e r p d the concern that the amxiation betateen bcnzcne and lymphocytic hkernia may have been confounded by other solvenu t8t. The International Agency for Research on Cancer also has recently s a t e d that 'mixed expoeurc pattcrna" in t h w ckpe- control studits has rendend interpretation of the studiea difficult ( 9 ) .Thue. Rinaky et d.'s a p m n t acceptance of a Caud rehtion between benzene and lymphocytic Ieukemia in credulous; we are not sw- p d t b t they judge our conaidemtion of thir h u e u *cavalier." We did not svduate the epidemiolmc data peruin- ing to bmnoand dignarit neoplumr of lymphatic -and hermtopoietic k,other t b leukemia. nor d i d n inchdathw adigmncim in o w rirtauesn- ment Thathzw rtudiacited in Riruky'rletterincludc 16 k t h s fmm thw other nmbgamt n e o p h n u with .bout 11.8 exputad (SMR = 1 3 6 , s per cent cod- i n w d 77-220). Such a wsak and imprecim -tion QII not wurant be- included in a riak asmeamant. Thcu other argument^ for including these 0th . also uc not compelling.We point out th.t Rinrty a at. did not k l u b these other in their om risk amesmment (2). lafrnu (a coauthor of Rinlity'r on the o r i w report on the pliofilm cohort (10)) and DiSruio b v e ragntly dirurd benzene and leukemia (11).and in so doing, tbsy uwd a riak -nt rrrult m d y identsui to aua. Tby tht their crri?llcltr of the effwt of be- on leukemia monrlity u rimikr to t h t of alat othem accaptd by th occup.rioMl s.fccy .rdH d t h AdrmninMion. k k y et Il'r rhk r a t yi.ba d t four timro,high I t il our bkf,udq'rprrwtly thrt of molf othsr* that Rinrlcy e t d.'r d uy..msat Mdt u to0 hgh md CiMOt bed IARC. 1387. 10. Infanu PF.Wagoner JK.Rinrky RA,et d.Lcutvmu in benzene worten. h n r c t 1977:2:76-6. 11. Infante PF. DiSuuo MV. Ocrupariond benzene crpo- H u W Awtm W t h Deltrll Philip Cole Department of Eprdcrnldogv - - S c M o/ Public Hcolth Unwemty of Alabama at Birmmngham Birmcngham. AL 35294 RE: 'MORTALITY OF LEAD SMELTER WORKERS" The provocative account by Sclevan et d.(1) of the (6).The s o l u b i l i d Lead ions are absorbed through the epickrnlolqpc relauon existing between expoare of thin walk of capillanea which invade into the area of smelterworkem to Led and the manner of theu fatal injury during the phasc of inflammatory rewr.Some mue well documents one of the most common sources of thcrc wmeh pe- for long p e d . Thus. over an of plumbum. narc engroomed in the field8 of forenaic extended time, small amounts of lead h v e the em- m d m e and pnthology have. on ofc..lon. become hsddsd minile and enter the circulation. u) travel a w m of another. albeit u n d . encounter rnth thu wi&ly within the body nnd reach diaMt vulnexable ekment. Over loo yeam ago,a u r a waa dcrnbedof 8 man rho had m i v a d n gunal!tot wound to che knee and who succumbed 12 ymm later mth rymptomatdog, c b l y mauckmgchronrc i4dintomemuon (2). S i organa such u the kidney and b m . The m i d t a t cliniul .Mipathologic a l ~ m t i moa~y, over a span of yun,becoma u davaruting as choaeotaed in the organa of l e d smelter rorkem then. them bve been OWDOIDUI re!pom of sum& cam, nil bvmg ID common the long uw retenon ofludrmrula d y in .nduwndthelugepmu of the exuwmua (3). k n t c W d u n o n of aome h.r revealed hematologic fin- of l a a r m ~ rmculoryw~lad hrophrlie s t i p p b g o red blood ah.d CONUlltlDt With 1 4 pomlllg. M O M Of blood lad lenh up to 5 ryldl are recodad. The nor- mrl level u la than 0.6 d d l . clind findtng r e k to~ c h r o w I K p h t l 8 d amate r c d f h M oftenpresent tqetber mthIsd eacrpblop.thy. th4 p a p i u t m g catme of death. The Lnrr u mubd uucomK.uy by Ccnhrd rn- -d hernutlon. and brun atem hemorrhqw (4.5). The I w p . m and rend tubukrcelh mveddchu- lc~crutleintrnnuctclumcluuanr.Tbe rrrpvrtoyryr- tern u not p m n u d y a f f d UI t h e be- caw the slirr of quund Id u not e r p n r l y CUKd arth the gundot. Of F . t rcleVumto th.hzud of Ruladbad u the epibmiohgy of ciiruul I d tomcity. L e d bulku s t n t m g d e w bo^ of an extremty, am frequently frymcnted- The pmcea may entcr mto &e purt rprc where they u e e r p a c d to lryMNul flud that ponema the nbility to drvoive 1 4 slowly. but conmtmtly Edrrrprf n y D~r. s.lcoon and hu cv-admnWR gawn thr opportunity to rrrpond to thu letter. oemrdin# to Journal poky, & C h o r nu# to do 80. RE: 'CONFOUNDIP.3 CONFOUNDING' In a recent uticle. Grayson (1) argued t h t a &fi- muon of confoundmg bend on c h a p in a purmcter's value upon c o v w t e control [noncobpsabllty) w u fundamental. We thmk that both C n n o n ' s de- Sceond like wveral earlier autbon (2). Grayson faded to r c ~ o g l l u athat an oddr ratio a n be an unbiased mawlbof effect even if it ir not collaprible (1.e.. even if the crudeoddr rat10diffen from the strntum~spcafic