Tobacco Documents Online

Epidemiology May 1999, Volume 10 Number 3 MALES NOT LIVING WITH A SMOKER (nA554) 8 6 14 Geome[rIc Mean = 0.723 nyJml ETS, COTININE, AND VENTILATORY FUNGTiON 321 Is 16 4 FEMALES NOT LIVING A SMOKER (n=711) CeomeMC Menu m 031'! o81m1 ,,. A 95 k Con06ence Interval - (0.413.036"!) ,.j 95Y.ConpAen<eInlevNa(0.658,0.794) C 6 C. 0 1 2 3 4 5 6 7 8 9 10 I l 12 13 0 1 2 3 4 5 6 7 8 9 10 11 Cetlnine (nqfmi) CuOniue (ng/mp MALES LIVING VVITH A SMOKER (a=129) a 2 a . FEMALES LIVING WITH A SMOKER (n~29) 18 16 14 GmmetrlcMevn=2]64nyJm1 12 . GeometrleMenn=1534np/ed 95'AConfdenceInterval~(1.970,2.184 ~ 10 95%CnnOdenceln4rv.l-(Id38,1.158) a 6 4 J AL 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 0 1 1 3 4 5 6 9 8 9 10 11 12 33 . Cotlnine (ng/mt) Cotinlne (nglmq FIGURE 1. Distribution of cotinine by gender and household ETS exposure at foIlow•up (N = 1,623). TABLE 1. Household Environmental Tobacco Smoke Exposure Rates and Cotinine Levels by Selected Factors Number % Living with a Smoker at HALSI • Totals 1,623 30 Smoking sratua Never-smoker 962 31 Ex-smoker 661 28 Sez Male 683 26 Female ~ 940 . 33 Age group (years) 46-459 ~ 585 397 t 28 33 60-73 301 23 on . Wales , 72 31 North 96 - 32 North West ~ 199 30 Yorkshi umbecside 134 36 Wesx M" )ands 148 32 East M id lsnds 148 24 _ g Easr outFt Wlese S South East 131 366 26 33 Greater London 120 28 ~ tlcand ass 136 31 1 134 15 I 477 25 I I N-m M 35 V 1 206 36 V Other 49 47 Parenral smokitig None 222 21 Mother only 97 31 Father only 767 29 Both parents 520 36 History of respiratory illness No 611 28 Yes 1,012 32 'HALS1 and HAIS2 = x<alth md Li(atyte smveys I and 2. % Living with a Smoker at HAIS2' Geometric Mean CotiNne at I-3AIS2' 95% Confidence Interval 22 0.76 .. 0.72-0.81 22 0.75 0.70-0.81 22 0.78 0.71-0.86 19 0.90 0.83-0.99 24 0.67 0.62-0.73 22 0.90 0.79-1.02 26 0.73 0.66-0.81 23 0.78 0.69-0.88 13 0.68 0.58-0.78 24 0.93 0.70-1.44 25 0.98 0.77-1.24 23 0.72 0.60-0.86 30 1.09 0.88-1.34 20 0.75 0.63-0.91 18 0.67 0.54-0.81 16 0.58 0.46-0.74 22 0.58 0.46-0.72 22 0.65 0.58-0.74 21 0.78 0.61-0.99 21 1.19 0.97-1.45 10 0.51 0.41-0.62 18 0.62 0.56-0.69 24 0.79 0.66-0.93 23 28 0.86 0.96 0.77-0.96 0.80-1.15 35 1.32 0.96-1.83 50 1.25 0.72-2.16 18 0.67 0.57-0.80 22 0.85 0.67-1.08 19 0.71 0.65-0.77 28 0.88 0.79-0.98 23 0.78 0.70-0.86 22 0.76 0.70-0.82

322 CAREY ET AL TABLE 2. Cross-Sectional Effect on FEVi (in m1) of Household Environmental Tobacco Smoke Exposure (Yes/ No) at Health and Lifestyle Surveys I and 2 Smoking Status All Subjects Males Females HALS I All subjects Subjects expased/coral 489/1623 179/683 310/940 Mean adjusced difference in -4 -90 54 FEVI (ml)y 95% confidence limits -57, 49 -190, l0 -5, 113 Never-smokers Subjects exposed/roral 302/962 91/329 211/633 Mean adjusted diff'erence in 32 -69 78 FEVi (ml)• 95% confidence limits -37, 101 -210, 72 5,151 Ex-smokers Subjects exposal)total 187/661 88/354 991307 Mean adjusced difference in -61 -125 -18 FEV, (m!)" 95% confidence limits -141, 19 -256, 6 -122, 86 HALS2 - All subjecrs Subjects exposedltotal 358/1623 1291683 229/940 Mean adjusced difference in -5 -87 39 FEV. (ml)• 95%confidertce limitt -58, 48 -189, 15 -18, 76 Never-smokers Subjects exposed/total 210/962 6I/329 149/633 Mean adjusted difference in 61 -2 80 FEV, (ml)• 95%confidence Ifmits -6, 128 -149, 145 9, 151 Ex-smokets Subjects exposed/toral 148/661 68/354 80/307 Mean adjusted difference in -95 -155 -32 FEV, (ml)" 95% confidence limits -183, -7 -298, -12 -132, 68 • Di(f mnce In F5V, residaal (adjusted for social class, region, and packyears) between exposcd and unexposad sabjea. published studies of adult ETS, calculating a point esti- mate based on a random-effects modelz' to take account of the large heterogeneity between studies. Results D[STR[nll-RON OF COTLNINE At HALS2, living with a smoker was associated with higher cotinine levels for both males and females (Figure 1). Table 1 shows rhe determinants of reported house- hoLd ETS exposure and cotinine level. Household ETS exposure decreased between studies (from 30% to 22%). At both studies females were more likely to be living with a smoker, as were younger subjects and those of tower social classes. There was ho notable variation across regions or with past smoking status. Cotinine levels were higher in males,•despite fewer men than women living with smokers. Trends for coti- nine with age and social class were similar to those for household ETS exposure, but there was variation across regions, with the highest values seen in Scotland and YorkshirefHumberside and Lowest seen in East Anglia and the South West. Adjusting cotinine level for all factors in the table did not remove the heterogeneity within any of these regions. The geometric mean coti- nine level for each of the 11 regions was also positively correlated with both the regional prevalence of ETS exposure in the home (Pearson's correlation coefficient, Epidemiology May 1999, Volume 10 Number 3 E 4 Nv i <~ 00

w N ? TABLE 5. Cross-Sectional Studies of Adult Environmental Tobacco Smoke on FEV. Ex - Effects on FEV; First Author Year of Publication Country Age (years)• Numbert Population Smokers Included?# Exposure Used for ETS Effecc Estimate Beta§ Standard Errodl Main Pulmonary ResultsY Schilling's 1977 United States Parents 252 Families in three U.S. towns Yes Spouse smoked vs not Negative N/A No effect of spouse's smoking status on FEVi in never•smokers Whites 1980 United States Middle-aged 800 Recruits from a physical fitness course in San Diego, CA Yes Smokey environment for 20 years vs not -5.5 1.5 Largest differences seen for FEFss_,s and FEFts-as Comstock° 1981 United States 20+ 418 Residents in Maryland No 1+ smokers in household vs not Negative NJAz RR = 1.42 for males for FEV, <80% of predicted Kauffmannrs 39 1983 France 25-59 2,898 Seven French cities No Lives with a current smoker vs not 0.55 0.9 Effect present in women age >40 years (stronger for FEFrsrs) Jones 1983 United States 20-39 (F) 205 Community in Michigan Yes Smokers in household vs nor Positive N/As OR = 0.76 between high FEV, and low FEVi Lebowitr30 1984 United States Parents 271 Sample of the Tucson area Yes Spouse smokes vs not NJA N/A+t No effect of spouse's smoking status BranekreeP° 1985 Netherlands 40-b0 (F) 57 Rural subsample of a larger population study Yes > 10 cigarettes/day at home vs none -0.8 3.4 Larger effect seen for PEF with present exposure; no relationship with FEVi decline Svendsen° 1987 United Smtes 35-57 (M) 676 Eighteen U.S. cities (Multiple Risk Factor Intervention Trial) No Wife smokes vs not -2.8 1.3 Effect of FEV, larger at baseline and nor when summed over all visits Masin 1988 Canada 15-35 293 Recruits from schools and banks in Montreal No Person-years cumulative exposure Negative NJAx Regression coefficient strongest for home exposure on FEFs in men Mzsjedi'r 1989 Iran 18-65 288 Healthy hospital workers and visitors in Tehran No ETS at home or work vs none -2 1.5 Effects of exposure on FV~, FEV;, and FEFr,-75 stronger in men Kauffmann't 1989 United States 25-69 (F) 1,211 Five U.S. cities No Husband smokes vs not Negative NJAx Very small effects of both FEV, (negative) and FVC (positive) Hole16 1989 Scotland 45-64 1,295 Two communities in West Scotland No >15 us <15 cigarettes per day by cohabitee -3.2 1.6 Comparison given for high w low exposure; no difference between low and none Ng's 1993 Singapore 20-74 (F) 739 Population sample in Singapore No >1 heavy smoker at home vs none -3.8 1.7 Small effects of FEVi in all subgroups Xul' 1995 China 40-69 502 Residential area in Beijing No ETS at home or work vs none -6.1 2.5 Reductions in both FEVi and FVC that are dose dependent for home exposure Frette's 1996 United States 51-95 651 Southern California communiry ' No Lived in household with a regular smoker vs not -2.5 1.8 Small reductions in FEVi and FEFts,5 found with household exposure •(M) or (F) indicates that only males or females were included in smdy. T Number included in estimate calculation. ; Yes(No whether ewsmokesss included in estimate calcularion (we tried to use neversmokers wherever possible). § Pacentage deficit in the exposed group (positive or ttegative indicaees that we were only able to obtain direction of effect). II Standard errors calculated wherever possible (in the case of Hole et u!° we assumed that the lower 95% confidence limit was exactly 0 to derive id. 9 PEV = forced expimtory volusne; FEF = forced expiratay flow; RR e risk ratio; PEF = peak expirsrnry flow; FVC = forced vital capaclry; subscript = seconds. r Calculation of a standard error was not possible, bue the 95% confidence interval is known tu contain 0 but of unknown width. LE098990SZ ( (

The Effects of Environmental Tobacco Smoke Exposure on Lung Function in a Longitudinal Study of British Adults lain M. Carey, Derek G. Cook, and David P. Strachan Small effects of environmental tobacco smoke exposure on lung function have been demonstrated in many studies of children, but fewer studies have examined adults in this re- spect. We examined these relations in a 7-year longitudinal study of 1,623 British adults, age 18-73 years, who were nonsmokers throughout. Outcome was measured by forced expiratory volume in I second (FEVI) adjusted forsex, age, and height. Exposure was assessed by asking subjects whether they lived with a smoker (at both the initial and the follow-up studies) and by salivary cotiNne measurements (follow-up ty only). Cross-sectionally, subjects exposed at home wed tiny FEVt deficits at both studies of -4 ml [95% 0 fidence limits (CL) = -31, 23] and -5 ml (95% CL =-32, 22), resiectivelp. Cotinine adjusted for potential confounders showed a stmri,,~r association with FEV,, with the highest quintile showing a-105-m1 deficit (95% CL =-174, -37) in comparis,n with the lowest. Longitudinally, no clear relation was appaicnt benrcen change in FEVr and average exposure or change in cxpt*'ure. These resulta indicate that environmental tobacco smokc is associated with small deficits in adult lung function, .'onsisrrnrwith our meta-analysis estimate of a 2.7% deficit in expes,el nonsmoking adults. The relations seen with cotinine but n.rt with household exposure may reflect the importance of exposure outside the home. (Epidemiology 1999110:319-326) Keywords: adult, corfnine, environmental tobacco smoke, forced expiratory volmne, cohort study, meta-analysis. Exposure to environmental tobacco smoke (ETS) has . been shown to be associated with poor respiratory health in sevet: 1 studies of children." Our recent meta-analy- sis of ETS exposure on lung function in children con- cluded that maternal smoking is associated with cross- sectional deficits in forced expiratory volume in I second (FEVt) of 1-2%° at school age. Much of this difference may be dur~to matemal smoking during preg- nancy. Fewer studies have assessed ETS exposure and lung function in adults. These stuaies are inherently difficult owing to the dominant effect of active smoking (past and present) and the possibility of reporting bias, which m result in active smoker's being classified as "passive" c Nevertheless, exploring the link between exposure andlung-ftirtctionmay be informative, as small deficits can be early indicators of chronic obstructive pulmonary disease.s Two recent reviewss•6 of adult studies concluded that exposure to ETS was associaced with small deficits Fivm rf.a Department ot Public Heabh Sciences. St Geurge's HoapiraL Medicsl School, Lm.don, unircd Kingdam. Addreas coneapwrdence m: Ivn Carey, Dcparmr<nc of Public Health Sciences. St r3ewge'a Hmpinl Medical School, Crenmer Tenace, landon SW l7 ORE, utc Submitted July 22, 1998; final verslon accepted )snuary 6, 1999. C 1999 by Epidemiology Raourccs tac n in lung functioti 7-1s All studies used questionnaire mea- sures to nteasure exposure- The evidence for longitudinal effects is even weaker, on the basis of two studies finding small effects on FEVI decline.lo•'9 In this paper, we use longitudinal data from a national survey of Bririsli adults to assess both cross-sectional and longitudinal effects of ETS exposure in the home on FEVt. We also use salivary cotinine measurements that were taken at follow-up to assess recent ETS exposure more precisely. Subject, and Methods THE I"IEALTH AND LIFESr1'LE SVRvEy The first health and lifestyle survey (HAI:SI) consisted of a random sample of 9,003 adults resident in England, Scotland, or Wales, who were interviewed and measured initially during 1984 and 1985. A follow-up study (HALSZ) 7 years lacer interviewed and remeasured 5,352 (59.4%) members of the original sample. Further details of the sample selection and data collection for both studies have been published.ZO•zt The structure and method used in both surveys were similar. First, an interview carried out in the respon- dent's home addressed a wide range of information, including socioeconomic status, self-reported health, di- etary habits, and smoking history. Subjects were classed as having ever smoked if they reported ever having "smoked at least one cigarette a day for as long as 6 319

320 CAREY ET AL months." ETS exposure was assessed by positive re- sponses to the question, "Does anybody else in this household smoke regularlyt"; this was a question asked at both studies. Parental smoking was assessed by asking, at baseline, "Have either of your parents ever smoked?" Individuals were assigned to a household socioeconomic group on the basis of the 1980 Registrar General's clas- sification.22 The interview was followed by a home visit by a nurse who collected a saliva sample and carried out physiological measurements, including height and respi- ratory function. SPIROMEiRY Standing height was measured with a portable stadiom- eter. FEVI was measured with an electric turbine spi- rometer (Micro Medical Instruments, Rochester, En- gland). Calibration of each instrument was carried out at the start of each wave of data collection and rechecked at the end by placing it in series with a Vitalograph spirometer (Buckingham, England). Measurements were discarded in the few cases in which significant calibm- tion drift had occurred. After suitable instructions and a practice attempt, each subject performed three forced expiratory maneu- vers. The maximum value of FEVi attained is used in the analysis here. Subjects with an acute respiratory infec- tion had their measurements discarded, as did any whose tests were deemed unsatisfactory by the nurse.EO As pre- viously noted?} there was probably a systematic under- estimation of forced vital capacity (FVC) within HALS1 and HALS2, so we do not present any analyses of FVC here. COTININE Dental rolls were used to collect saliva for cotinine measurements. Respondents were asked to place the roll in their mouths for 3-5 minutes until it was saturated with saliva, and the roll was then placed in a small specimen tube. These were either dispatched immedi- ately by first class mail to the laboratory or kept in a domestic freezer until forwarded. Cotinine concentra- tion was measured by gas-liquid chromatography as de- scribed elsewhere.Z9 EXCLUSIONS We restricted our sample to Caucasians who were be- tween 18 and 73 years of age at the first survey (HALSl) with satisfactory spirometric measurements at both ex- aminations, who had not smoked between studies (and were subsequently classed as either never-smokers or ex-smokers), and who had salivary cotinine measure- ments. Data were too sparse among subjects age 74 or more years at entry. We also excluded subjects (N = 13) with cotinine levels greater than 14.7 ng/ml (consistent with active smokingzs) and those for whom change in FEV, (AFEV) lay outside the middle 99% of the distri- bution (N = 17). After these exclusions, we had 1,623 subjects for analysis. Epidemiology May 1999, Volume 10 Number 3 STATISTICAL METHODS We used a "two-step" approach to our regressions, which was identical to a previous analysis on these data.L6 First, we adjusted FEV/ and AFEV measurements (both in (iters) for the effects of age (in years) and height (in centimeters):.To do this adjustment, we used a set of prediction equations that were based on the never- smokers with no history of respiratory illness throughout (246 males and 497 females). Cross-sectionally, these were the following: Males (1984): FEVI =-2322 +(0.012 X age) + (0.036 X height) -(0.000024 X age X heigtt)-[0.C0041 x(age)Z] Females (1984): FEV, =-2.779 + (0.051 X age) + (0.036 x height) -(0.00030 x age x height)-[O.C0032 X(age)z] Males (1991): FEV, = -4.013 +(0.014 X age) + (0.048 X height) -(0.00012 x age x height) - [0.00024 x(age)z]r Females (1991): ~/ FEV, = -2.462 + (0.029 x age) + (0.036 X height) -(0.00023 x age x height) -[0.00020 x(age)2] Longitudinally, we used a model with annual change in FEV, (i1FEV/a1AGE) as the dependent variable, as some subjects had 6 or 8 years between studies. The model contained terms for mean age (AGEM) and height at first examination, which is compatible with the form of the cross-sectional model with height as- sumed constanr.Z6 The longitudinal equations were: Males: AFEV/DAGE _ -204.6 - (0.578 x AGEM) + (1.164 x height) Females: OFEV/GAGE = 70.16 - (0.618 X AGEM) - (0.403 x height) We used the regression coefficients from the above equations to calculate a set of predicted values and residuals (observed minus predicted values). The mean residuals for FEV~ for the 1,623 subjects were -106 ml in ,~ 1984 and -82 ml in 1991 (nonzero and negative, as th _e model has-been extended to include ex-smokers). In the- longitudinal model, for the purposes of presentation, we . multiplied the residuals by 7 to represent the predicted loss over the 7-year interval between examinations.. To look at the effects of ETS on lung function we regressed the FEV, residuals on both the dichotomous home ETS exposure variable and (natural) log cotinine and the OFEV residuals on exposure profile (a four- factor category representing all permutations of exposure at baseline and follow-up) using the OLM procedure in SAS (SAS Institute, Cary, NC). We included the fol- lowing as potential confounding variables: social class (six household socioeconomic groups plus an `bther" category), region of residerice (nine English re- gions, Scotland, and Wales) and pack-years (estimated as the average consumption times years of smoking given at baseline). We then carried out a meta-analysis of

Epidemiology May 1999. Volume 10 Number 3 WhSte F---~'-~ i + Kauffmaun (Fr) B k f i i i i l i rune ree ~ ~ Svendsen /---0--~ di i Masje i Hole i~---fl----i ~ N g ~ x o~ ; u . Frette F-¢-`-1 Fizad K-1 Ravanm F-"-1 -12 -6 0 6 12 PercentDiRerence(FEV,) FIGURE 2. Meta-analysis of the effects of ETS on FEV, in nonsmoking adsil[@. , home, such as a the workplace or public areas, may be of importance in addition to home exposure as a source of exposure to tobacco smoke. We have previously demon- strated the importance of:, exposure outside the home in -tennining cotinine levels in children?t This exposure ~rfiay explain why subjects who were not currently exposed at home, but who had lived with a smoker 7 years ago, had higher cotinine levels than those not exposed throughout. Nondomestic exposure is also a likely explanation for males having greater cotinine levels despite a smaller proportion with current home exposure. F.ELAT[ONS WfTH LUNG FUNCC[oN There was no substantial effect on FEV, of living with a smoker at either study, in contrast to the consistent reduc- tion in FEV1 with increasing cotinine level at HALS2 (which persisted in never-smokets and in those not ex- posed at home). The half-life of cotinine is 20 hours, and thus it is largely a measure of exposure during the past 48 hours.r^ Possible reasons why we see a relation for FEV, with cotinine and not with household ETS exposure could be that (1) cotinine is a more precise measure, as it is able ETS, COTININE, AND VENTILATORY FUNCTION 325 to quantify level of exposure; (2) recent exposure is most influential in lowering FEV,; and (3) exposure from outside the home is more important than domestic sources either as a determinant of overall ETS exposure or in causing a deficir in FEV,. It is likely that all three reasons have had some influence in HALSl and HALS2. Tredaniel et al suggested that, if ETS exposure causes small airways dysfunction, theff-measurements that "de- pend on total airway resistance and elastic recoil of the lung," such as FEV, and FVC, will not pick up these de$cits.6 Instead, mid- and end-forced expiratory.flow rates (FEFz~75 and FEF7,85), which have been shown to produce larger deficits in a meta-analysis of children's studies,° are thought to be better for assessing small-ainvay dysfunction. Unfortunately, these measurements were not made in HALS1 and HALS2. The inverse association between cotinine and lung func- tion wass greater in ma[es and ex-smokers. The gender difference may be attributable to level of exposure, as males were seen to have much greater cotinine levels. The larger deficits in ex-smokers, noted by Schilling et 47 may indi- cate that some have continued to smoke intertnittentiy during the follow-up. We excluded subjects with raised cotinine levels consistent with regular active smoking to minimize this problem, however. The deficits may also represent lost lung function from their smoking years that is not regained, but this would only bias our results if ex-smokers had had greater cotinine levels or domestic exposure rates, which they did not (Table 1). Longitudinally, there was a weak relation between changes in household ETS exposure and changes in FEV, which suggests small benefits of others quittingg in the home. There was little evidence of more rapid de- cline in subjects exposed at both studies, arguing against any cumulative effect of living with a smoker. CHILDHOOD ExPOSURE AND EFFECTS The role that ETS exposure during pregnancy and child- hood plays in determining adult lung function is still unresolved. We know that this exposure causes small deficits in childhood FEV i,4 which may reflect impaired lung growth. Whether this carries over to the subsequent adult level and decline is unknown. Parental smoking as recalled (retrospectively) by adults in HALSI and HALS2 showed a weak positive relation with adult FEVi. This finding implies that any long-term effect of childhood exposure is subtle and difficult to detect in a study of this size with such crude indicators of parental smoking. Our meta-analysis of adult studies indicated a deficit in the range of 1.7-2.7% among passively exposed non- smoking subjects, a result that is consistent with the U.S. Environmental Protection Agency's conclusion of a 2.5% deficit in those exposed to ETSJZ This result compares with a reduction of 1.4% in a recent meca- analysis of the effect of parental smoking on FEV i in childhood.9 We conclude that the effects of ETS on lung function in adults are likely to be similar in magnitude to effects

Epidemiology May 1999, Volume 10 Number 3 EI-S, COTININE, AND VENTILATORY SUNCTION 323 TABLE 4. Cotinine Level and Change in FEV, (AFEV) by Household En- ETS exposure did not appear to influ- vironmental Tobacco Smoke between Studies ence the rate of FEV, decline at alL; Lives with Smoker at HALS2` Lives wirh Smoker at I-IALSI" No Yes No Number 1038 96 Geometric mean cotinine at HALS2 (ng/ml) 0.57 1.15 95% confldence limits 0.53, 0.6Z 0-96, 1.37 Difference in AFEV (ml)t Baseline -43 95% confidence limits -114, 28 Yes Number 227 26Z Geometric mean codnine at HALS2 (ng/ml) 0.76 2.10 95% confidence limirs Difference in AFEV (ml)f 95% confidence limits 0.65, 0.89 +14 -35, 63 1.83, 2.40 +25 -20, 70 ' Health and Lifestyle Surveys I and 2. Y DilF rence in dFEV between selccted ezpmed and nonexposed groups adjusted for.ocial cla.u, region, and pack.years (all at baseline). , r = 0.52) and the regional prevalence of active smoking calculated from all the HALSI data (r = 0.74). I.ROSS-SEGTIONAL EFFECt3 OF ETS Among all subjects, living with a smoker was associated with tiny deficits in lung function at both studies (Table 2) of -4 ml [(95% confidence limits (CL) _-31, 231 and -5 ml (95% CL =-32, 22), respectively. Only among male ex-smokers was there any suggestion of a larger deficit in FEV, among the exposed group [-61 ml (95% CL = -141, 19) and -95 m[ (95% CL = -183, -7) at HALS1 and HALS2, respectively]. , Cotmine (measured at HALS2 only) showed a stronger relation with FEV, (Table 3). There was a deficit of -105 ml (95% CL =-174, -37 ml) between the top and bottom quintiles of exposure: This deficit was not ex- clusive to ex-smokers, as it was apparent among never- smokers [top-bottom quintile deficit = -76 ml (95% CL =-160, 8 ml): The effect appeared stronger in men, as there was a 166-ml (5.3%) deficit in FEVI between top and l5ottom quintiles in men compared with 68 ml for women (2.8%). The magnitude of effect increased by restricting the analysis ~Gsubjecs (N = 1,265) with no reported household EIS exposure at HALS2 (not shown). The top-bottotn quintile deficit was -147 ml (95% CL = Z36, -57). LONGITUDINAL EFFE(.`I5 OF ETS Cotinine level (at follow-up) was related to home ETS exposure profile over both studies (Table 4). Among those exposed at HALS2, levels were higher if they had also been exposed at HALS1. Similarly, among those not reporting home ETS exposure at HALS2, cotinine levels were higher if they had been exposed at HAL91. These trends were seen in males and females separately. There was no clear relation between change in expo- sure and change in FEV, (Table 4)_ The "newly" ex- posed subjects did show a greater fall in FEV, albeit modest, than those who never lived with a smoker throughout [-36 ml (95% CL =-114, 28). Persistent those..living with a smoker at both studies showed the smallest fall in lung function between studies. EFFECrs OF PA$ENFAL SMOKRJG We looked at the effects of parental smoking habit (none, mother only, fa- ther only, and both) on current lung function and found no evidence of a negative relationship with FEVI. At both studies, current nonsmokers whose mothers had been smokers had on average higher FEV, than those whose mothers had not smoked, de- spite their having higher cotinine lev- els at HALS2. This difference was 64 ml (95% CL = 14 to 114) at HALS1, falling to 47 ml (95% CL = 2 to 92) at HALS2, with the decline of FEVI between studies being only weakly related with maternal smoking [-25 ml (95% CL = -58, 9]. MEPA-ANALYSIS OF ADULT STUD[ES OF ETS Using the 21 studies included in two recent reviews of the literature on ETS and adult lung function,s4 we conducted a meta-analysis of the effect of ETS on adult FEV, (Table 5 and Figure 2). Among 15 population studies that looked at FEV, cross-sectionally,7-`&2sao only 9 gave sufficient information to be included in our calculations,atatl.ls.ls-is.zs and these were generally from the studies that had found statistically positive results. Conse- quently, our random-effects estimate of a -2.7% (95% CL = -4.1%, -1.2%) deficit in FEV, is likely to be biased. By estimating the standard error for the remaining six studies from the existing ones, we can recalculate the random-effects deficit on the conservative assumption that on average the six excluded studies found no association. The pooled random-effects estimace for the deficit in FEV, is then -1.7% (95% CL = -2.8%, -0.6%). Discussion In this national longitudinal study of British adults, we explored the relation between lung function and ETS, assessed by whether each subject lived with a smoker and also by measurement of salivary cotinine. Our results indicate that any effect of ETS on FEV, is small, but is stronger when cotinine is used as an index of exposure. DETERMINAN15 OF ExPOsuRE This is the first report to explore adult cotinine levels among a national sample of nonsmokers in the United Kingdom. We found marked variations between regions, with the northern regions and Wales having the highest levels. Although this variation was partly due to home exposure, it was also strongly related to active smoking prevalence in each region as calculated from all the HAIS1 data. Thus, we believe that sources outside the

326 CAREY ET AL seen in childhood and chat sources of exposure away from the home are important determinants of this effect. We found no evidence to link domestic ETS exposure with more rapid tung function decline in adult life. Acknowledgments We thank the Economlc Social and Reseazch Council Data Archive for provid- ing us with the dara, Brian D. Cox and the numerous research workers who conducted both health and lifestyle aurveys, and Martin Jarvu and Colin Fey- erabend for the cotinine assays. References 1. Strachen DP, Cook DG. Healrh effecrs of passfve smoking. 1. Parental smoking and lower respiratory illness In infancy and early childhood Thoraz 1997;52:905-914. 2. Cook DO, Srrachan DP. Health effccrs of pacsive smoking. 3. Parenml smoking and prevalence of respimtory symptoms and esthma in school age children.Thorax 1997;52:I081-1094. 3. Snachan DP, Cook DO. Health effects of passive smoking. 6. Parental smoking artd childhood asthma: longitudinal and case.control studies. Tho- rax 1998;531204-212. 4. Cook DG, Snachan DP, Carey IM. Parenmi smoking and spirometry in childrem Thorax 1998;53:884-893. 5. Coulrzs DB. Passive smoking and the risk of adulr asthma and CDPD: an update.Thorax 1998;53:381-387. 6. Tredaniel J, Boffetta P. Saracci R, Hhsch A. Exposure to environmental robac<o smoke and ruk of lung cancen the epidemiological evidence (Re- view). Eur Revpir J 1994;7:1877-1888. 7. Schilling RS, lztzi AD, Hui SI., Beck G), Schoenberg JB, Bouhuys A. Lung function, respiratory disease, and smoking in familics. Am J Epidemlol 197706:274-283. 8. White JR. Froeb HF. Small-alrways dysfunction in nommokers chronically exposed m tobacco smoke. N Engl J Med 1980;302:720-723. 9. Comsrock OW, Meyer MB, He6ing KJ, Tockman MS. Respimtory effects on household exposures to tobacco smoke and gas cooking. Am Rev Resyir Dis 198L;124d43-148. 10. Bmnekreef B, Fucher P, Remijn B, Van DL Schouten J, Quanjer P. Indoor air pollution and la effect on pulmonary function of adult non-smoking women. Lll. Passive smoking and pulmonary function. Int J Epidemiol 1985;14:227-230. 11- Svendscn KH, Kuller LH, Martin M). Ockene JK. Effecrs of passive smoking in the Multiple Risk Factor Inrervenrion Trial. Am J Epidemiol 1967:126: 783-795. 12. Masi MA, Hanley JA, Emat P, Becklake MR. Environmenral exposure to cobacco smoke and lung function in young adults. Am Rev Respir D'u 1988:138:296-299- 13. Masjedi MR, KazemL H, Johmon DC. Effects of passlve smoking on the pulmonary function of aduls. Thorex 1990;45:27-31. 14. Kauffmann F, Dzkcry DW, Speixr FE, Ferrts BC Jr. Respirarory symptoms Epidemiology May 1999, Volume 10 Number 3 and lung function in relation to passive smoking: a compamtive smdy of American and French women. Inc ] Epidemiol 1989;18:334-344- 15- Hole DJ, Cill;s CR, Chopra C, Hawthorne VM. Passive smoking and cardiorespimmry health in z gcnenl population in the wessofScodand. BMJ 1989:299:423-427. 16. Ng 7P, Hui KP, Tan WG Respimmry symptoms and lung function effecte of domaric exposure to tobzcco smoke znd cooking by gas in non-smoking women in Singapore. J Epidemiol Commun Heafth L993;47.454-458. 17. Xu X, Li B. Expnsure-response relationship between passive smoking and adult pulmonary function. Am ] Respir Crit Care Med 1995;151:41-46. 18. Frene C, Barrett-Csnnor E;Hlausen ]L. Effect of active and passive smoking on venrilatory function in elderly men and women. Am J Epidemioi 1996; 143:757-765. 19. Jzzkkola MS, Jaakkola JJ. Becklake MR, En+st P. Panive smoking and evolution of lung function in young adults: an B-year longitudinal study. J Clin Epidemiol 1995;48d17-327. 20. Cox BD, Blaxter M. Buckle AL). Fenner NP, Golding JF, Gore M, Huppert FA, Nickson ), Roth M, Stark J. Wadsworth MEJ, Whichelow MJ. The Health and Lifeuyle Survey. London. The Hcafth Prornotion Research Trust. 1987. 21. Cox BD, Huppcrt FA, Whichelow MJ- The Health and Lifesryle Survey: Scven Years On. Aldershot. U.K.: Dartmouth Publishing. The Health Pro- motion Research Trust, 1993. 22. Office of Population Censuses and Surveys. ClassiAcacion of O<cupacions and Coding Index. London: Her Majesty's Stationary Offrce- Office of Population Censuses and Surveys, 1980. 23. Semchan DP, Cox BD, Erainclioglu SW, Walters DE, Whichelow MJ. Venrilsrory function and winter fresh fruit consumption in a mndom sample of British adulrs Thorax 1991;46:624-629. 24. F<yerabend C, Rusreli MA. A rapid gae-liquid chromacographic method for ~° the decenninacion of cotinine and nicocine in biological fluids. J Phacm Pharmscol 1990;42:450-452. 25. McNeill AD, Jarvis MJ, West R, Russell MA, Bryanr A. Saliva cotlnine as an indicator of cigarette smoking Ln adulescenu. Br 3 Addict 1987;82:1355- 1360. 26. Carey IM, Stra<han DP, Cook DO. The effeccs of changes in fresh fruit consumption on ventLlatory function in healchy Britvh adulrs. Am J Respir Crit Care Med 1998;158:728-733. 27. Hardy RJ, Thompson SO. A likelihood approach to mem-analysis with mndom effects. Smt Med 1996;15:619-629. 28. Kaut7mmsn F, Tessier JF, Ortiol P. Adult passive smoking in the home environmene a risk factor for chronic air(low limiwtion. Am J Epidemioi L983;117d69-280. 29. ]ones JR, Higgins IT, Higgins MW. Keller JB. Effecrs of cooking fuels on lung function in nonsmoking women. Arch Environ Health 1983:38:Z19- 222. 30. Lebowitt MD, Knudson RJ, Buaows B- Famify aggregation of puln»nary function measuremenrs. Am Rev Rcspir D'u 1984;129:8-11. 31. Cook DO, Whincup PH, Jarviz MJ, Svachan DR Papawsra 0, Bryant A. Passive exposura to tobacco smoke in children aged 5-7 years: individual, family, and community facmrs. BMJ 1994;308:384-389. 32. US Envimnmental Pmecction Agency. Respiratory Health Effects of Passive Smoking: L.ung Cancer and Other Disorders. Washington, DC. Office of Research and Development, 1992.