Tobacco Documents Online

:watq~:.; Passive smoking and cardiorespiratary health in a general populatiion, in the west of Scotland West of S'cottand Cancer Surveillance ,Unit,,Ruchill I Hospital+ Glasgow G20!9NB', David J'Hole, etsc,, stnrirrioion Charles R Gillis, atn; director t)epattment lof' Epidemiology, Schoollof Public Health, University, ofMichigan, Ann Atbor, Michigan,l7nited States Carol Chopra, rtsearcb studear Victoi IN' Hawthorne, ,Mm, Drotessor Correspondenceandl requests for reprints to: Mr Hole. n.ardy 1929:m4234 David J Hole, Charles R Gi1lIs; Caro1 Chopra, Victor M Hawthorne Abstract Objective-To assess the riskofcardiorespiratoryk symptoms and mortality in non-smokers who were passively exposedito environmental smoke. Duign-Prospeetive,study of cohort from general population firstiscreened between 1972 and 1976 and followed up for an average of'11-5 years, with linkage of data from participants in the same householdi Setting.-Renfrew and',Paisley, adjacent b'ttrgbs in urban west Scotland. Subjecrr-1'S399 Men and women (80%' of aill those: agcd' 4S-6+i1 resident in Renfrew or Paisley) comprised the original cohort; 7997' attended for multiphasic screening with a cohabitee. Passive . smoking, and' control groups were defined on the basis of a lifelong non-smokihg index case and whether the cohabitee had ever smoked or never smoked. Main outcome measure-Cardiorespiratory signs and symptoms and mortality. Results-Each of the,eardiorespiratory symptoms examined produced relative risks >14(though none were significant) for passive smokers compared with controls. Adjusted forced expiratory volume in•one second was significantly lower in passive smokers thani controls. All', cause mortality was higher, in passive smokers than controls (rate ratio 1-27 (95% confidence interval 0-95 to 1-70)), as were all causes of death related to smoking (rate ratio 1-30 1(0-91 to 1•85))'and mortality from htng cancer (nte,ratio 2141<' (0-45 to 12-83)) and ischaemic heart disease (rate ratio 2.01 (1 21 to 3-35)). When passive smokers were divided into high and low exposure groups on the basis of the amount smoked by their cohabitees those highly exposed had higher rates of symptoms and death. Conclusion-Exposure,to environmental tobacco smoke cannot be regarded as a safe involuntaryy habit. Introduction Though cvidenec has accumulated about the risk to healthof involuntary, or,passive, exposure to environ- mentalitobaeco smoke, further information is required from cohort studies to confirm these observations. Deleterious effects on the respiratory systemof infants and children have been observed" ' as have chronic effects on lung function in adults," but these findings have:been criticised on methodological grounds.''tAn overview of 10 ase-control and three cohort studies estimated a rclative risk of I-35' for lung, cancer in people passively exposed compared'with non-exposed l controls.' Three studies have: reported increased' (though norsignificant)Irisks of ischaemic heartidisase in non-smokers with partners who smoke."•'Problems in, interpreting these findings include lack of an objective measure bfdoseor exposure, failure to adjust for confounding,variables, inappropriate methods of statistical ianalysis„andlailure to masure other poten- tially important variablcs." This,rcport is basedon the Renfrew-Paisley survey,, which was arricd'out in an area with a high iincidence BMJI VOLUME 299' 12 AUGUST 1989 of lung,anccr; it overcomes many of these criticisms: The survey prospectively studied a general population aged. 45-64 years, and' the eollected data allowed participants from the same household to be idetati6ed: The measure of exposure to environmental tobaoto was obtaitted directly from cohabitea anddid'not rely on ~self reporting. I)ata on prevatetttes of symptoms of' , respiratory and catdiovasrnlar disase, forced iapiTatory trolume in one second', mortality, and incidence of. cancer are all' available for this population. Tltee findings reported here update an earlier report; it adds 567 further deaths to the previous findings" atvd extends the range ofbaseline inasurements to indude forced expiratory volume in onesecond. Confounding variables:sueh as social class, blood pressure, eholes- terol concentration, body mass index, and social class have been.allowed for in calculating relative risks for passive amokcrs. Subjects and methods This general population i cohort comprises 211 men , and women aged 45-64 years resident inithe towns of Renfrew and Paisley in1 the west of',Scotlutdibetween 19721and 1976,"Eligibilitywasestablishedlbyadoorto door census of all households in the two towns. Everyone,who met the age and residency criteria was invited to atrtcnd one ofl 12 temporary centres for a multiphasic ardiorespiratory scrccning examination." Between 1972 and 1976, 15 399' ttsidents (an 80% response) completed a standardised self administered I questionnaire that included questions on smoking, behaviour, and' was checked by experienced inter- viewers when subjects attended for screening. Rapira- tory symptoms were assessed with the Medical ResearchCounciRs bronchitis questionnaire. By identi'• fying participants from: the same, household it was possiblc to study varying exposures to tobacco smoke in a subsample of 3960 men and 4037 women and to calculate relative risks for a range of ardiorespiratory , variables including mortality. Four groups, in which the index ase was aged 45-64 at the time of', the survey, were defined i based! on the index ase and!on the cohabitees ever or never having smoked. (1)Control= the index case hadinever smoked and lived at the same address as another subject'who had never smoked. No one else in the household who attended for screening was a smokeror ex-smoker. (2) Passive smoking; the index case had never smoked and lived iat the same address as a subject wl. had. (3) Single smoking: the index case was a smoker or ex.smoker and lived'at the same address as a subjecC who had never smoked. No one else in the household, who aMended for screening was a smoker or ex- smoker. (4) Double smoking: the index case was a smoker or,ex-smoker whoiived at,thesame address as a subject who was also a smoker or ex-smoker. If'the indcx' cases were ex-smokcrs they were classified as single,smokers or double smokers depend- ing on whether the cohabitees had never smoked or 423'

ever smoked. lf'the'eohabitees were ex-smokers the index cases were classiAed' as passive'smokers if they had never smoked or as double smokers if they had, ever smoked: Thus the controls represent, a group: whose passive exposure was'u low as possible within the constraints of the studydesign: Results for the two aetive'smoking'groupsihavc been',includcd to gi ve'some indicatiomof dose-response'and provide a'perspectin•c 'Tor any' dilTcrences' found between the control and', passive smoking'groups: A'eohabitcewas defined as a respondent sharing the same household'i environment and examined at the same time in the survey as the index case. Some households contained cohabitees of the same sex. Some of the subjeets i who: were examined were above or below the age range eligible for inclusion in thc:study. These subjects were not analysed as index cases but information ion their smoking behaviour as cohabitees waslused as the measureofpassitro exposure for eligible index cases. Mortality data was obtained. from the National Health Servieecentral register and the General IRegister Td at.a'I-Cotepositiml'oJ jroaps exposed ro citarette snloke. No(Ei')ormcn (iuukz eues) Nb.(%)o(sromen. (iideY casas) Taal Qonttnls (nekbar indeacr.e nor enh.bitee e.er smnkaW)'. 421(1d8) . 489(12'J). 917 t'wi.esnuoking(onycotubitotrvarrnoked). ' 243. (6•1). 1295i(32•1). 1338 SingkmmoM ing(mlpiitdn eau'evet,amoked) Ikmlik smok(ng (bahundls c.xasd cohab6te c+ermmok'ed) 1420(35-9), 1'869 (47.2) 33(i (8-2). 1751 1922 (47-6) . 3791 . Tmal 3960(100). 4037(100)7997' TAslE,tt-Sorial clau o(manin groups expared io cigarettt smoke. FigurtJ in parentkeus are percenrages. Exposure Staup' Socfalclass. Conmroll~~s P,astiee smoking~g Singlc'. smoking~g Doubk~ smoking~~. 1 2s (54) IS (s3) 61 (4•3)i 79 (t+2) 11, 83.(19+9) 37.(133) 225(ITrB) 235'.(1t'6). tllurm-l 63.(14a7) 23 (9+5). 197(I3•9) 20t'(10•9). 111'manal 1571(36•J): 96'.(39,5) 538(3719) 771(41•3) IV. 80(IS-7). . S9(24•3): 3.15(2Y.2). 4HS(23•4). V. 17' (4-0)~. 111 (4•S)~ 68 0.8): 122'. (¢S)~~ Inw(Kcienuin(ormuion 3'. (0•7)', 4I (1•6)~. 16~. (1•I): 21 (l-1) Taal 428 (i100'~I). 243°(W9) 1420(100). 1869(100) OfCtce.for Scotland. incidence of ancer'mas obtained, through, the cancer registry system and used to verify that the classi8cation on the death certi5ate was the same as that'received by thcregistry: Data presented are eomplete to thc end'of Deeember 1985, an average follow up of 11• S•ycars. Pncvalcnces for respiratory and ardiovascularsylnp- toms were standardised for, age and sex using the agp, and sex distribution of the whole cohort as standard. Similarly, mortality was standardiicd for age and,sex using life tables to cstimate'sun+ii+al at I1 years of follow up."' Mean forcediexpiratory, volumes in one.second'for the four exposune groups wereadjusted'foragc,,height, and sex~ by determining the best fit sct' of ' parallel regressionmodelsior forced expintoryvolume in one' second as a linear function of age and height for men and' women separatelj+ in each group. The mtan adjusted forcedlexpiratory volume in one'smond'for each group was'then calculated for the'avenge ageand heighr,of ineniand women separately,,and a weighted, average{(:orresponding to the proportion, of men andd women) was computed. Probability' values were obtained from the analysis of variance. Estimates of celative'risk and 9594 confidence inter- vals for passive'smokers compared with controls were adjusted for'age, sex, social class, diastolic blood, pressure, serum cholesterol concentration and body mass index (weight (kg)l(tieight (m)Yx 100) using,the logistic regression model" for ardiorespiratory'symp-' toms and I Cox's proportional hazards model for mortality." L.evela of significance were derived from the partial likelihood function " Thebiomet6cal data processing programs (BMDP) package was used to compute estimates of risk and levels of probability."' A supplementary questionnaire in two' of the 12 centres in, which, the survey was arried'. out asked subjects the extent to' which they were exposed to cigarette smoke from any other person in the house- hold, irrespeetive;ofiwhether these peopk were eligible for or attended the survey,, and also in, their work environmenti ResIIlts' The number of men and women in the four exposure groups is shown in dable l. Passive smokers comprised TAatt'tn rSmoking habit olcokabktes in,passins,snwktng and,doub'k smokiaggroups. Fipcnesart'percentagS S(aumben) Noofaig.rettcs anakd per day byoolubitee. 1•14~ aIb 13-24' iZ3. Ensmoker. In.kxrafe Aten Women Prviresmolimi8rnuP Doubk smok'in8 irouR P.sfi+'e amoking group . noubk smokin8 stroup 31~r3 (76)) 4671(1112)~. 30v(361): s2-P(9as) 13•1(J96); 4PS(64p 11•4 (219). 36-2UDS0). 424(102). 46•9(g5g) 30r8(397) 37r1(713) 22•K~. (}S)', 4-1! (10)~. 6,8(J27). 17r3(323) 43+t(}58) 1.H0(141) 32'4 (623) 19r1(367) 'rAtO.['av-Ate cnd'sex'standardised'rates of rnpirotorv und cardiooasculcr symptoms related to exposure to ei=arette smoke. Ntnnbers oj'indts' easa goitk syneptorns are given in parentheses . l:.oo:wz s,nup Caa.oh Passirasnnking. Sieglewnoking. Doubkwmking (n-9t7). (n-7568). (p~t731) (n-079I). Respiratory symppoms: Infcated spuwwn PerpueM sputum Dyspom Hypenaecuon Card'wroCular'rymptoms: Antrim I . Ataior aEnorma6tT found on sleetroardiogram Me.n foettd xspia.torpr.te in one,+ecmd(1): 2-3(22) 3•3 (44) 10•3(1s9), 10.5 (396y 7rE1721 9'9(l22) 2s-0(S41) 2H7(1079) i 10•1(93) I2;b(197). 13r4(229) 16-6' (618), 3-3(48) 6,9 (al) )7h6(327) Ig•3 (6a1p 4,603)~. 4•7' (74)~. 77(165) 9-1 (334). 1•0 (8)'~ 19 (13)~ 1-4 (31)~ 1~rS (119)~ Un.diuued 1 2•32 Adjusted 2•31 424 23~1 2•12 21.04' 243 '~ 2•12~~ 2'07 BMJ vot.v(ttt: 299 12AUCUSr 1989 BN

tAtltl< v-Ate aed iex edjraed tiaortaluy per 10000 per ycarby catetnry oJexpesrre m rrtercee tnnRe. Figrms iin yarnetkaes art aaral itnnben~ef demAt Ginvots tb>ti.e anokinc si314k nnokint t3ouEk tandunj AWe.un. a3•1(99). 97,4!(164). 160•0(41.0). IS5,61(734). Lunaanaer 1,6 (2) 5•0 (7) 23•a (S4)' 21•4 (93): I.eh.emicAemdiune 27.3(30) 47•7 (54) 61 OCI71)' 60•4(260)'. AUcausete(denhteMedtomtokitts 60;U(7J) 7r•2(1104) 130•4(362)1 129•9(S92)' Tru>U vt-Abr adjrued precwleecs of respiostory and cardiovatexlar ipmpwm: aed.ter ttasdardrtad :tormliy pa 100100 p,ery oar far eamnetr ia'tanfrol and passiar'nnok'rnt grotrpt. Fijrra ie ipmredbtra atr raaabas of acbwl cara NAn(re anoken Cmnroh . (nr4d9) t.awexpnaKe (n-734) High ecpo.ute. (n-S41)~ !'rmrkwa . R ' ept r.tnry q mpnns: Inkatedqma 2•1(10) 2J(ilr) 3•J I(17J Ptr>bteetspanna 6-4(31) 5.e(45) a4(46) Dyspooea 12•7(60) 113(64) 16•2(re) Hyperseaaioa 4 1(19) l+a (29). 5•7(30) Card;o.nea)rr rympamsc Anem 3•6(07) 4,1i(32). S•t (3q Atajorrt.wrnWityfoundnnekcarocardio6nm0•4 (2). 1•1 (6) 0•S (2). Afurtaliq Aacwot . Sa3(32). 64400), 17.1(54) . iwtqe.nnr 3+2 (iI). 2•S. (2) S•7 (3). lf(h.emiehtsndisene 6,1 (3): 14•2(14) 2s.a(16): A'Uewnrso(derhimhteditn.smok'ins 34,9(17)' 35-2(39) 47•3 (30)', 6'1P~'.' (24313960) of men and 32•1% (129514037) of women. Of the cohabitecs, 91-6% (7325) were of the opposite sex. Tho composition oCthe groups by social class isshown in tablc II l The extent of' passive' exposure experienced by passive smokers in,rclation to subjects in the double smoking group is shown in table III. In alli 46'r 1%(112) men and 4l'•8% (541) women,in the passive smokingg group lived in households where the cohabitee was smoking 15 or mere'cigarcttcs a day. This'eompared with 57•7°!0 (985) men iand 56-2% (1080) womeninthe double smoking'group. Ex-smokers were more common in households in, which the index case had' never smoked. The prevalence of signs and'symptoms for the four exposure groups is shown in table IV. For each of the four respiratory measures (infected sputum, persistent sputum,,dyspnoea,and'hypersecretion) Rhe'ntes in the control', group were' Jowcr than those in , the passive smoking group and considerably' lower than in the single and double smoking', groups. The rates for angina and major abnormalities found on electrw cardiography were similar,in'the control and passive smoking,gnoups andilower than in the'active smoking gpoups: Mean forcedi expiratory volumes in one, second adjusted for sex, age, and height: were significantly higher'(p<0•01) in controls than in those passivcly exposed' te cigarrctte smoke and rret~e sigai6aandy, higher than among active'satokets. Mortality adjusted forage and sex in the'four groupa is' presented~ in , tabl!' V. Total mortality w as higher among pmsivesmokersthaneontmis. This was re9octed in the catcgary of all lausesof death related to smoking and was highest for ischaetnie heart disease. Lung cancer mortality was higher'atnong passive stttokers' than controls, but the number of deaths involved was: small. The supplementary qu-stionnaine on exposure to cigarrtte'smoke at:home and work allowed reheck to be made of'the smoking habits of other household mcmben wlio were not parr of the survey. A regular smoker living in the same household'was repot'tediby 5% (2Y44) of controls compared with 69% (27139) of' passive smokers. Of!9ramen„2'1%' (13/62) of controls Gved in i housoholds with a regttlar smok'er compared with 63%' (l'25/197) of passive amokers. . VVotltcn reported that most of their passive expOwte was'at home tathcrthanarwork, which'suggested that they were the appropriate group in which to tsamine whether there was a dose-response relation. A high exposure passivesmoking group'was'therefore deftaed as women whose cohabitee was smoking 15 or more cigarettes daily,, and the' remaining female passive smokers werrdcfined as a ~Irnr exposurrgroup. Table VI.p,rescnts the age,standardised rates for'respiratoryr and cardiovascular symptoms and i mortality for, the control and the low and high exposure passive smoking groups. For each of the four respiratory symptoms'the highly exposed passive smokers had rates that were higher than those in passive'smokers'whose exposure was low and those in the controls. There' were no • consistent dit7erences' between the low passivee exposure group and ;t1te'controls. A similar pattern was found for angina but noC' for major abnormalities detected byeltctrocardiography. The adjusted forced expiratory voltune' at one second was signiGcantlylower in passive smokers with high exposure'compared with those with 1ow'exposure (mean',1 •83I v 1'•891; p<0•05). A1o signi6cantdillbrencr was found between'passive smokers with low ecposuroe and eontrols(l -891'a 1-881): Ageadjusted'mortality wass increased lfor the passive smokers with high exposure' compared with low and' with controls for' all cause mortality, all cause mortality related to smoking, ischacmic heart disease, and lung,cancer:, Table VII shows the adjustedi relative risks for passive and active smokers compared with controls. For each variable' the relative risk associated with: passive smoking was >1-0. The confidence intervall included' 1'•0 except for ischaemic heart~ disease; for which the estimate of risk was significantly different from.unity (p=0•008). Table VIII shows the relative risks for double smokers compared with singlestnokers afteradditiotul adjustment for quantity smoked. Dyspnoea was signi- rASt>: vn - Relatrat'rifif asroriated witA paurtr smoking adjvtted for ure,,rex, and social class and /ortardiooatcrlar variables, diauolic blood presrurc, rermn cJmlerteral ronctntrmranj and body mau index Retiti.e risk (pmi.e.stmokersaos"red withicontroh)h Respintorysymqoms: Infeela! aputum 1 •34 Peniltent vutum 119 . tJy;pmea, . 1.09. Hypenoaction 1,21 Cy!diovncvlar symqnms:. AeNa 1•II Maiorabnorma)ieiksifoandinnekettocandiaerun1•272.4bttditr:: . AU ksuus 3.27 All lonaevordeuh rehted to wnok'ina 1•30. t.chaemic:.hcandism,c 2•01. . t.ungcaneer 2-41 BMJ votuMe 299 12 AUGUST 1989 9S%Con6dencr inmerval pVdue Itdui.e ruk (.ai.eanokeneampned '.itheaaud.) 0-76 to ~ 2,36 : 0-3 ~ 4•53 ~. 0'ZE.to 1467 ~ 0.3 4,49~ 0-a2 te 1'4S ~ 0-5. t+60 ~. 0•811to 1•82 0;3 3+77' 0•73ito 1•70 0•6' 1•t9 0-Ato3•3S. 0.6 1.51 0!9S to~. 1•70~. 0•10 2•07~ 0'91 to~. I•tSl 0•tsl 2•33 Ir21 to ~ 3dS'~. 0~-U08 ~. 217 0,433 to 12+s3; 0.3 10,64 ' 425

. T.auswn-ita.darrwab,JerBteawekenoa.,o.nti!eel+'k'dselr+awoake.r,.d'rwdlrar+ee,.M.iwanW fotniit prerfottslr;' and the risks forapng cancer aro. '.etmt,'"+ me'a dar''"d/°' awd;°"°,eal°r°'"'6Ie' 1"°uolir e'r°md yrrnr.ti'n'ar dal`a'e.ot' • consistent with thosrin the overview by Wald'eral.• oairamariaN, oNd 3edy awuiindez I ki d' I t d F ttcladre.riik 95%GonGkna im~c+rd~ p Natbe ~~. t «y symt,mon: w+~ 0•96 o-79Ynt•16~ 0165 . Peninem ,twmm 1-06 0192Yod•21 ON5. ~~ 1•25 1-02 If151etJ9 M~s7kot•20 0-02 0-75. ea,die..xvlar rympams: . I~r17~ 0•95 1n t•44 o•t5'. A ~t+nMmiltun irnund inn dn,rocu+t~ I-IIt o•60 101-79. O•h5'. !>tonilnr. , Aa bw.ee 1.01 0•67i1o Pta. 0,9 M Muse, mridath' rc6sod to unoki~ .. 0'.99 0•Nno1-16' 0.9 Gch.omic~ha.n idifd+c o~d9 0•72iof•tl 0-3 t;nn<~r 1•11 0-79na,1,63' 0-5 ficantly more common among double smokers (p= 0-02)1 and though none of'the other variables was significant, six had,risks,> 1 •0. o car wvaseu ar ata rc ate passtve smo ng ew disease, but a relative risk as high ias 2-2 for mortality from ischaemie hart disease in passive smokers has been quoted.' Our risk of 2-0 seems large, in; com- parisoni with!that found for active smokers, and the possibility that chance has in8ated this risk cannot be excluded, but as the lower 95% confidence limit forr the relative risk is greater thanone it would appear than chance alone is not responsible for the exQtss. When investigating,rusks close to unity it is impor tant to consider the effect of potential biases. Biases may operate at the time data arr collected. Between. 1972 and' 1976, however, passive smoking was not an issue. Subjects reported their own smoking habits and no self reporting of passive exposure was undertaken. It was not until 1983'thar subjects within the same household were linked, and this was carried out without ,any reference to the measures of outcome examined subsequently. Therc is no directmeasurc a vailableto prove that'the Discussion passive smokers received a higher environmental dose Whether inhaling other people's tobacco smoke is a of tobacco smoke than the controls, but in, the risk factor for lung cancer and other diseases related to supplcmcntaryquestionnaire that covered the smokingg smoking is now under serious scientific considerationl habits of'household members irrespective of whether Studies of the concentrations of cotinine in the urine they attended the original survey only 5'9frof controls and saliva of passive smokers suggest that tkd'dosr said that there was a current smoker in the household, received may be equivalent to smoking up to three compared, with 63°'% of passive smokers. Greater cigarettes a day." Though sidestream smoke contains exposure to tobacco smoke at work supported the idea different proportions of chemical constituents than that passive smokers were more liltelythan controls to does mainstream smoke and the same dose received be in contact with environmental tobacco smoke passivelymi'ghtnottranslatedirectlytothesameriskas outside the home. This was measured by Wald and in active smokers,, the risks expected, for passive Riichie;' who showed that non-smoking,husbands of smokerswilliprobabiybe~ofasimilarmagnitudetothose smoking wives had higher urinary,cotinine concentra- found in active smokers of up torthree cigarettes daily; tions than, non-smoking husbands of non+smoking consequently, only vcry'large studies will have sufficient wives : • Our definition of categories of ' exposure is power todetect such ~risks. A' meta-analysis is currently comparable with that of other studies and would seem the only way to establish preciseestimatesof'risk, andlit to identify groups with different mcan levels of passive is essential'that all,studies are included: exposure. The high level' of heavy smoking in, our 71his paper updates a previous publication" with, eohort*, might also indicate that this difference is mortalitynow oxtended to an average follow,up time of greater than that found'in other studies. 11 •5'yars and the,control and passive smoking groups The problem of smokers deliberately classifying redefined to exclude those who smoked only pipes or themsel ves as non-smokers° is a far less serious bias in cigars and those who smoked cigarettes i'rregularly. cohort studies than in icase-control studies, because at. The original qucstionnairc in its roded form did not theintcrviewstagethereisnoundiationwhichsubjects distinguish pipe and cigar smokers and those who will! subsequently diea The likelihood of differential l smoked fewer than!five cigarettes a day from non- misclassificationi rates-that is, higher in the passive, smokers. Written information on the questionnaires smoking,thaninthecontroligroup-isdebatableasthis allowed! this to be clarified, and'these additional data implies that someone in the double smoking group is were added to the eomputeofiles, more likely to pretend to be a non-smoker than The sample, size in this study does not provide someone in the single smoking group. When the sufficient statistical' power to detect risks ofl the cohabitee is a smoker the reverse may be more likely to magnitude expected. Thus the lack of significance, be true. should not' be the sole criterion of whcther, a genuine It has been suggested that nonsmokers who marry effbct may be present. Several findings should be borne smokers may be different from non-smokers who in mind when interpreting these results. Firstly, for marry non-smokers." A higher proportion of passive each of'thr 10 measures examined, from respiratory smokers were insociallclisscs lII manual„IV,,and V, symptoms to auses of monality,,the relative risk was but no differences werrfound for other possible risk consistently larger than unity. This remained'so after factors, such as occupation, raised blood pressure, adjusUng,i'or intervening risk factors such as age, sex, cholesterol concentration, or body mass index. In anyy social class, blood pressure, cholesterol eoncentrationi case the final analysis, which estimated the relative and body mass index. Secondly, the one measure for risks, adjusted fonach ~of'ahese factors. which sufl5cientstatisticai power wasavailable-that is, The effect of passive smoking on those who already forced' expiratory volume in one second-gavt a smoke is fhr, harder to isolate: The dose received by significant result. Thirdly, when a group of'passive active smokers from smoking ranga widtly,"~' and smokers with high exposure was dcfined',there was an adding a small extra component due to passive ex- increase in the dose-response relation for nine of the,10 posure may not lead to much of a difference in mean variables: Fourthly,,in comparison,with the relative doses for double smokers compared with single risks found for the two aeti ve smoking groups, each smokers. Hence, the increased risk for double smokers increased risk was biologially plausible, with the relative to single smokers may be substantially less possible exceptionof that for ischaemic heart disease. than that for passive smokers relative to eontrols: Thus 71te findings for respiratory symptoms areximilar to the statistical power of a single study is an importantt those of'otherstudies: adecrased forced expiratory consideration and in the absence of other published'' volume in one second in passive smokers has been, data on this aspect it is difficult to interpret our results 426 BMJ VODUME'299 12 AUGUST I!989' F

I I . IDcpartment iof Medicine . and ProteiniReferenee Glhit, ,Roysl'Hiflamshire Hospital, Sbet6eld S(ll 2JF A'Kapur, eMED6ct, madiccl indent G' Wild, ltse, tenior rcientirt A Milford-Ward, FRCrATH, director of yroteinnjerence unit D R Triger, tner, readenoe , msdiairtr Correspondtnceto: DrTiriger. Rr Ar.d J.1911 for'the effects of passive smoking on uttokas. There- fore the main emphasis of this'paper'is an estimation of the risks of passive smoking in lifelong non,smokers; data are presented for the active smoking groups to provtde.an estimate of dose-response. Our results arrbased on a general populationcohortn study carried out in an area with a high level'of diseases related to smoking. A' eonsistent increase in risk was observedlin'passive smokers for each of the'Ip variables measured covering, respiratory symptoms, forced ex- piratory, volume in one second, cardiovascular symp- toms, and subsequent,mortalitys including lung,eancer andiischaemic heart disease. A dose-response relation was' seen, and the risks were' biologically plausible in relation to the size of:the risks found for the active smokers. These three factors taken together increase our concern that exposure to other people's tobacco smoke cannot be regarded as a safe involuntary practice:, I Caniey JR?. Hallud VN. Goak3till RT~ trduaKed padre saukift rut i parenW ip6tepn un iAleunwnia and bmndudis i.n arly sh84hoe6.i Lare N74¢iaaJ14. 2Wciss ST. Tya la. ,SprixT FE: Ranaa. tleneaa iwAeac-Ns Wnie. ro~ mydmarr il4iev. ti6arrne rnokin2~ and kal~dipuMa.ary fw¢uee n a taFalafiue.samFk in eMldeew. Aw Rrs ReyinAb Ifli;1i.•6M•52. , ): RM'he JR. FmeA iHF. SnuR air.ars dy,fannids G.. aensmokaa clwauitdlp. eapard wMroe6:w~oEe. N F.aRll,Ned 191trJ62:720.2. 1',KuuRmmn F, TeoierJF..Orid W.,Aduh pmi~e mwkasi.lbehanccnrinumnem: a tisk facmr der.afiumnk rrao. SunYniae. Aw.? Eriib.id 19111;117:2M.110: s IxAo.i~:61D.ln/luenad(sd.e.uiwR'in~~m.putA~e~wyJunaien:a,a.ey. 2tinMd 19H:12i61343. " 6W.Id NJ..Nanchdal K. Thonqo. SG. cacSJe HS. Das (weMllint a3ia, peopk'aab.c66smorc nu.e I" iaeted9.Md;r I9t6Q9):1217-22. 7. Garland ie, I(xnutamer E; Sr.rcs L. n d. E(kas d raui.c.smokin<an . ixharmic 4KaM dittue aertalinyy d; non-fmoken: a prarenive stud}'. A.1 . F•pidi+.ie),19tY ltI:6i5•50.. a Htnj...T.lw+e4p.r.6dJWCii°'- Mramat6•m~12. !. f.end.ef K~Hj R.av LH. M.eI. )rU;OcRi+eJK.FJracadp.d.e.mRiq Y As .Mtipk Ad racta ina.cstida aid. A.1 s!Wembl .11l7a2R . 763+9f. lo US ~ Drpnanw of HaM . awd l Hiwr sa.4es. . R.yirwy. .&* of we.hwny ~.+eal. fq.w.r. .p:Je.:dqir .uliia R.p.r d. w66i/~.1J . Nq 11111. ' EenAeda. Maryl.ed: NatirwJilmtinwes diHdU'.1M3.. H GiqiaiCR. Huk,Dl. HnnYarne~VM. Ea,k P. •r)t e(kat.('e..kmuneeat tahau anoke In nn irbr cenunaritin iil t6e .ea d~Saed.nd. Er~ R61.illn ~I961;63'.(l•tTl'.12)):J 21 4. 12 Hawdiane VM, GiaiY CI(. Muka. DS. Mbnnaine hnlHi iu SsadMd, , JArJ'ERi4mid It+73;t:Mf•'74... IS Haxt3wrneVM.Gen.esDA.ah.enDG.abatpsameiI.a m,M.I7 H7t;1:660•7.. 16 Ky+tanEL.&kiet'.NaMramnrieniunnianRaninavny4eusbser.atieas: Jw.tiAifidkAwe.kaawnarirdAnawiw a15r:32:2s7-r1. I3 Cet iDR. TAt,.a.fj.b.fJiwyLna. t.andan: Mahoca. t170. . 16Co. DR. . Reenatwm modds rd 7ik u(9n.1-d~.J Ih Ruyd'Sw/iakd .. SK+~7 (R) I lf 72',76:1 rr.220. 17 Ce. DR. tSvtid tikdiMod. Biu~ ilY)Sf2:269.76: It `Diao.WJ; aienw Ma; Enslmsa, l..erd.~ Ruwefirdla.paePiv/nywwtSi.Kafb( bJlw.r,)9tS. tw. iAnRkc Uei.eniydCdtm.i. t1es.171s. , If . Maak.n s: Tatnin.a T. Xi(atr N. e rt~. F.fknrdie..Yw~eenul us6ne. roke ae . .ri.ay aaMiae ewwia. .1 ~.e..wmla.. N. Eqlt 7Araf 19NOtUt2l.2p. 20 . Hdfnbr D. aeun.nma. tCD. AtlMa jn. nd. tndoorpeEwio~ by,roE.as '.rotca: iaa11( qud'KS mlre sptake ~.bY.o.snolrn,.2k.aerda.d t, e6. l aba .ir....Iw% ),nder iw.k:q. l.nierGrt ~ aL Aw.:K ./ilewd.p. Vell2. Srocktwlnt: Swc" Gou,nd for au"" Rs.e1d1.JlH913a. (I4aiaedinpd ilie Jed :qasea/iasr itadncna reindonr rr 9uaGyed e6ftule: suppl D 17.)'. 21 iak . N. Recfte G'ViGJalio. d snd'us on Arq onaer i. .o.~l.dao. .anied tr~aoke.. Lnrt/ Hllylal667i . 22 GdFsCR:HekDJ.HmAotneVM.($artlsn.nki+Kandauk3>.6icaiaa, iwara d.vy Aigb YKkeen. B: Ropat ds {cnerY ipqmlnio. catien nwyY t/ e oew of Saaland.JE'bJmw (~!'Ha" ts66;42:6iY. . 22 Ise PH. Mndaadcatioe, ai a faaa, I. ~~ pr.kv wmkiaR','6k:g t...re . N66-.:iK7: N aacf FRJ. Fw1.er.oh6K ,Y~ udeHhooJ and aeca d.k. A.;I'Ejik.:.t 19t6;122:%1-t 23. WaIUNJ. Dorshrn Jiftky A. Rkcsir C, ~Hadda. JE. Knishl'G. UAi~ eainisr r a imarker. d'bMCaAis6 ahet peopk's lab.no ! amlm.. Lara 19t1;ii2BQ I ~: . (ptrcpnJ2J2dq19d9) Carbohydrate deficient transferrin: a marker for alcohol abuse A Kapur„G W'ild, A Milford-Ward, D R Triger Abstract Objective-To assess the value of serum carbohydrate deficient transferrin as detected by isoelectric focusing,on agarose as an indicator'of alcohol abuse.. Design-Coded analysis of serum samplertaken from, patients with carefitlly defined alcohol intake both with and'without liver disease. t;,,omparison,of carbohydhte deficient transferrin with standard laboratory tests for alcohol abuse. S'etting-A teaching hospital unit with an interest in general medicine and liver disease. Patients-22 "Self confessed" alcoholics admitting,to a daily alcohol intake of at least 80 g for a' minimum of three weeks; 15 of the 22' self confessed alcoholics admitted ito hospital for alcohol withdrawal; 68'patients with alcoholic liver disease confirmed by biopsy attending outpatient clinics and claiming to be drinking less than S0'g alcohol ;daity; 47 patients with non-alcoholic liver disorders confirmed by biopsy; and 138 patients with disorders other than of the liver and no evidence of excessive alcohol const3mption, Itttervention-Serial studies performed on thr. 15 patients undergoing alcohol withdrawal in, hospital. Main outcome measure-Determination of relative value ofl techniques for detecting alcohol abuse. Results -Carboh'ydt•ate deficient transferrin was detected in 19 of the 22l(g6°/6) self confessed lalcohol abusers, none of'thr47 patients with non-alcoholic BMJ VOLUME' 299 12 AUGUST 1989 liver disease, and one of the 38I (3%) controls. Withdrawal of alcohol led to the disappearance of carbohydrate deficient transferrin at,a'variable sate, though inisome subjects it remained detectable for up to 15 days. Carbohydrate deficient transfersin was considerably superior to the currendy available conventional markers for'alcohol abuse. , Conclusion-As' the techniarye is fairly simple, sensitive, and inexpensive we suggest that it may be valuable in detecting alcohol abuse.. Introduction The medical and' sociai consequences of alcohol abuse are major' problems throughout the world. Although'many people readily ackt2owledge the extent of their'alcohol consumption, others'attempt to conceal it, and we lack,reliable objective means of identifying surreptitious alcohol iconsumptiom Currently available laboratory marken have considerable limitations, being insensitive, non-specific, or dependent on liver damage. The mean corpuscular volume rises in patients with ahyroid disease, folic acid deficiency, and liver discase; whereas' serttm y-glutamyltransferasr aetivity'is affected by drugs'that induce micrasomali enzymes as well as rising in all forms of obstructive liver damage.'' Serum aspartate aminotransferase', activity is more commonly raised in alcoholics than alanine aminotransferase activity is, and whereas a ratio of aspartateao alanine aminotransferasc'activity of greater than 2:1 is stnongly suggesti ve ofalcoholic'liver disea,te this is of little value in subjects in whom the ~