16.01.2013 Views

ets exposure, lung cancer - Legacy Tobacco Documents Library

ets exposure, lung cancer - Legacy Tobacco Documents Library

ets exposure, lung cancer - Legacy Tobacco Documents Library

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

http://legacy.library.ucsf.edu/tid/wyk81f00/pdf<br />

310 I,F.VOIS AND UYAftU<br />

marriagc to a smoker as the b`TS <strong>exposure</strong> indgx . The resulu of each study were<br />

adjusted for smoking status misclassification-currtat or exsmoketa falsely declaring<br />

themsalves to be never-smokcrs-betorc bcing comhined in th : rnetn•analysis . An<br />

adjustment for background <strong>exposure</strong> (<strong>exposure</strong> from soutYxs other than the spouse)<br />

increased the summary rclativc risk to 1 .59 (95% confidencc interval, 1 .02-2.90) . The<br />

relative risk from bttckground ETS exposurc for women married to nonsmokers w-.u<br />

estimsted to be 1 .34 (95% confidencc interval, 1 .01-2 .08) . On the basis of the U .S .<br />

relativc risk estimate of 1 .19 before background adjustment, which was judged to be<br />

statisticafly significant, ond on the basis of the rosults of foreign studies as well as<br />

consideration ofthe biological plausibility of the association, the EPA repon concluded<br />

that there was sufficicnt evidcnce from human studies to ciassify F.'rS as a Group A<br />

(knawn humtin) carcinogen in terms of the EPA Guidelines for Cnrcinogen Risk<br />

Assessment (U .S . EPA, 19$6) .<br />

However, there are problems with the k.PA's risk determination . Its meta-analysis<br />

of U .S . studies omitted two studies which were published before the EPA document<br />

was rcleased ; Brownson et u! . (199' Z) and Stockwcll er ul . (1992) . Using the L'?A's<br />

methods and assumptions, we have calculated a summary relative risk of 1 .07 from<br />

a mera-analysis of 13 U .S . female spousal smoking studies, including these two recent<br />

studies . This relative risk, with 95% conticJence intt :rval of0 .95-1 .21, is not atatistically<br />

significant. Furthermure, iheru is a highly significant (P < 0 .001) inconsistency between<br />

the summary relative risks for six country-specific groups of female spousal smoking<br />

studies considered by the EPA (Table 1) . Thcse relative risks sre all either only wcakly<br />

clevated or, in the atsz of China, below 1,0 . In Section 2 we point out that e likely<br />

explanation for this geographic inconsistency is bias and confounding introduced by<br />

using spousal smoking as an exposurt surrogate . In Section 3 we cxaminc the epidemiologic<br />

data on workplace ETS <strong>exposure</strong>, which were ignored by the EPA but which<br />

overall exhibit no risk elevation . The inconsistency between the spousul and workplace<br />

studies supports the hypothesis that small risk elevations rcponed in some spousal<br />

studies are due to uncontrolled bias and confounding .<br />

2. BIAS ARISING FROM "I'fiE SPOUSAL SMOKING STUDY DES(GN<br />

Grcc= 2 .01 1 .33 3 .04<br />

Honj Kont 1 .49 1 .16 1 .88<br />

lapan 1 .42 1 .15 1 .75<br />

Western Europc 1,16 0,7E 1 .73<br />

United States 1 .07 0 .95 1 .21<br />

China 0 .34 0 .71 1 .00<br />

Wii ST :ZB--`t66i'TZjSLL :-v6biSjb E) A='MIX01 NI S1N01'T1SNO :)<br />

r<br />

:<br />

non<<br />

evitl :<br />

ine .>r<br />

A<br />

:LtT o:<br />

1 ;10r<br />

nftL<br />

smus<br />

be~ ;,<br />

ciatic<br />

in frL<br />

cou ;<br />

rc c .<br />

ofe :<br />

sn•,c<br />

Stcr<br />

an c!<br />

spou<br />

evar•<br />

clcri~<br />

a~cr<br />

A<br />

thru<br />

tobe<br />

13eca<br />

c : nc<br />

in sG<br />

clas`<br />

was .<br />

et a/<br />

The only explanntion offered by the EPA report for the geographic inconsistency a 2 . :<br />

in the spousal smoking studies was that it could he due to d'tffcrences in the intensity is qt<br />

of spousal <strong>exposure</strong> between countrics or to diffrrcnces in the ratio of spousal to nte .<br />

0 .89<br />

et a,<br />

TARLE I<br />

misc<br />

MezA•AN&I.YSCS 6(' SKUtsAL St.tORtNG CrtoYMiotAGiC SttJUtES<br />

clmc<br />

fo rr i<br />

Country Rclntivc risk 95% Cnn6dcnco intcrra7 O coul<br />

N ~ relnt A<br />

tion .<br />

~ dosc :<br />

~ ing-<br />

~'°~ rclat

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!