Maternal and Personal Cigarette Smoking Synergize to ... - ATS Journals

3 downloads 0 Views 97KB Size Report
Nov 20, 2002 - structive pulmonary disease (COPD) risk. In 1998 we reported an inverse association between maternal smoking (prerecorded) and. FEV1 in ...
Maternal and Personal Cigarette Smoking Synergize to Increase Airflow Limitation in Adults Mark N. Upton, George Davey Smith, Alex McConnachie, Carole L. Hart, and Graham C. M. Watt Department of Social Medicine, University of Bristol, Canynge Hall, Bristol; Departments of General Practice and Public Health, University of Glasgow, Glasgow, United Kingdom

Susceptibility of the lungs to cigarette smoke is poorly understood. It is not known whether maternal smoking increases chronic obstructive pulmonary disease (COPD) risk. In 1998 we reported an inverse association between maternal smoking (prerecorded) and FEV1 in adults. Because FEV1 and FVC are strongly correlated, it is unclear whether the association in question reflects a link with lung volume, airflow limitation, or both. We extended our original analysis to investigate whether maternal and personal smoking synergize to increase airflow limitation. We estimated residual FEV1 to express FEV1 variation that was not associated with FVC. Maternal smoking was inversely associated with FVC and FEV1 irrespective of personal smoking. It was inversely associated with FEV1/FVC, forced midexpiratory flow rates (FEF25–75 [mean forced expiratory flow during the middle half of the FVC], FEF25–75/FVC), and residual FEV1 in current smokers but not in never or former smokers (heterogeneity p ⫽ 0.016, 0.024, 0.021, and 0.016, respectively). We tested the clinical relevance of findings in ever smokers without asthma: 10 cigarettes/day maternal smoking increased prevalent COPD by 1.7 (95% confidence interval: 1.2–2.5) after adjustment for covariates. Maternal smoking impairs lung volume irrespective of personal smoking and appears to synergize with personal smoking to increase airflow limitation and COPD. Keywords: chronic obstructive pulmonary disease; disease susceptibility; smoking; passive smoking

Chronic obstructive pulmonary disease (COPD) is forecast to be the third leading cause of death in the world by 2020 because of a global increase in smoking (1). More knowledge about risk factors is needed to guide research into the biological mechanisms of COPD and to target smoking cessation (2). An early origin for COPD is supported by observations of an inverse association between childhood lower respiratory tract illness (LRTI) and adult FEV1 and FVC (3–6). Interpretation has focused either on possible causal roles for early viral or bacterial LRTI in the etiology of COPD (7, 8), or alternatively, how the associations might be confounded by a lifelong constitutional tendency to respiratory illness (7, 9). There has been less discussion of the nature of the deficit in adult lung function associated with childhood LRTI. It is surely relevant that most (3, 5, 6) (but not all [e.g., Reference 4]) prospective studies find no association between childhood LRTIs and adult airflow limitation, indicated by lower FEV1/FVC. The role played by exposures that lower FEVs (FVC,

(Received in original form November 20, 2002; accepted in final form November 20, 2003) Supported by the Wellcome Trust, the National Health Service Research and Development Programme, and a Department of Health Public Health Career Scientist Award. Correspondence and requests for reprints should be addressed to Mark Upton, M.B.B.S., B.Sc., M.Sc., Woodlands Family Medical Centre, 106 Yarm Lane, Stockton-on-Tees TS18 1YE, UK. E-mail: [email protected] This article has an online supplement, which is accessible from this issue’s table of contents online at www.atsjournals.org Am J Respir Crit Care Med Vol 169. pp 479–487, 2004 Originally Published in Press as DOI: 10.1164/rccm.200211-1357OC on November 20, 2003 Internet address: www.atsjournals.org

FEV1), rather than FEV1/FVC, in the etiology of smoking-related airway obstruction is not straightforward. FVC and FEV1 are highly correlated, with correlation coefficients lying between 0.80 and 0.90 (10). If a childhood exposure were predominantly to be (inversely) associated with lung volume rather than airflow limitation, it would not be surprising if that exposure were associated with FEV1. Children exposed to maternal smoking are at increased risk of LRTI and asthma and have lower FEVs and forced expiratory flows (FEFs) (11, 12). In 1998 we reported graded inverse associations between maternal smoking and FEV1 in adults, irrespective of personal smoking, but did not investigate whether the associations reflected a link with lung volume or airflow limitation (13). The present study involved an extended analysis in our original population, rather than a new cohort. Our 1998 report was a research letter with a single table of data and assessed only FEV1 as outcome (13). To test whether maternal smoking affects lung volume, airflow limitation, or both, this study included FVC, FEV1, FEV1/FVC, FEF25–75 (mean FEF during the middle half of the FVC), and FEF25–75/FVC as outcomes. We estimated “residual FEV1” (RESFEV1) to express FEV1 variation that was not correlated with FVC. First, we investigated effects of personal smoking on spirometric outcomes. We compared associations between maternal smoking and outcomes with those of personal smoking, examining never, former, and current smokers separately. We tested effects of former and current smoking on FEV1/ FVC across the range of exposure to maternal (and paternal) smoking. We tested the clinical relevance of findings using prevalent COPD as the outcome. We assessed effect modification by sex and asthma and the potential for confounding of maternal smoking by personal smoking, paternal smoking, current environmental tobacco smoke (ETS) and socioeconomic position in some detail, including use of serum cotinine.

METHODS The eligible population comprised 3,202 adult offspring of couples who themselves participated when aged 45 to 64 in a population study conducted from 1972 to 1976. Methods for parental (14–16) and offspring (17–19) populations have been reported previously. In 1996, 2,338 offspring aged 30 to 59 from 1,477 families completed questionnaires and underwent examination (response, 73%) (18, 19). We recorded FVC maneuvers on a Fleish pneumotachograph using American Thoracic Society standards to define spirogram acceptability (20). The spirometer’s calibration was checked before every session. To improve spirometric quality, technicians were given performance feedback (19). A total of 2,294 subjects attempted spirometry, of whom 2,195 (96%) provided at least two American Thoracic Society acceptable curves (19, 20). Between-visit coefficients of variation for FEV1 and FVC were 3% (19). Serum was analyzed for cotinine by gas–liquid chromatography (21).

Coding of Parental and Personal Smoking From 1972 to 1976, parents were asked to report smoking habit and age at which they had taken up (or quit) smoking (14). Mothers were not asked about smoking in pregnancy. We used parental data to code offspring’s exposure to maternal and paternal smoking. There is uncer-

480

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 169 2004

tainty about critical periods for exposure to parental smoking, but prenatal exposure to maternal smoking is important (22). Therefore we specified that offspring had to be exposed to mother’s smoking before birth to be coded as exposed to maternal smoking and to father’s smoking at least before school age (5 years) for paternal smoking. Offspring whose mother or father had reported being a former smoker during the period 1972 to 1976 were coded as unexposed to that parent’s smoking if the parent had quit more than 1 year before the child’s birth. We excluded 164 offspring whose mothers took up smoking after the child’s birth because numbers were insufficient to estimate effects of postnatal maternal smoking, and coding such offspring as not exposed to maternal smoking may have contaminated the baseline (unexposed) group. Similarly, we excluded nine offspring whose fathers took up smoking after the child reached school-going age. We coded intensity of offspring’s smoking in cigarettes per day (cigs/day), and duration in integer years.

Analysis We used STATA software (23). We predicted FEV1 for an individual’s FVC using data from healthy never smokers without asthma in the 1996 cohort to establish their relation by sex-specific regression of FVC on FEV1. We subtracted predicted FEV1 from measured FEV1 to give RESFEV1, which measures FEV1 variation not associated with FVC. We used linear regression, adjusted for covariates as indicated in the text, to estimate effects of personal and parental smoking on lung function. We estimated the effect of maternal smoking on prevalent COPD in ever smokers without current asthma. We defined COPD cases by British Thoracic Society criteria (24): percent predicted FEV1 less than 80% and FEV1/FVC less than 70%; controls by percent predicted FEV1 higher than 85% and FEV1/FVC higher than 75%. We estimated odds ratios by unconditional logistic regression. We obtained predicted FEV1 from linear regression in healthy never smokers without asthma: Men: predicted FEV1 (L) ⫽ 5.34 · height (m) ⫺ 0.023 · age (year) ⫺ 4.51 Women: predicted FEV1 (L) ⫽ 3.39 · height (m) ⫺ 0.028 · age (year) ⫺ 1.47 All analyses were adjusted for clustering of offspring within families. Further details about population selection, methods, and analysis are shown in the online supplement.

RESULTS There were 884 men and 1,116 women, of whom 398 (19.9%), 1,161 (58.1%), and 441 (22.1%) were aged 30 to 39, 40 to 49, and 50 to 59 years, respectively. In men and women, respectively, there were 396 and 553 never-, 269 and 283 former, and 219 and 280 current smokers. Pearson correlation coefficients between raw FEV1 and FVC in men and women were 0.89 and 0.91 in never-, 0.88 and 0.90 in former, and 0.84 and 0.89 in current smokers; correlations between FEV1 and FEV1/FVC were 0.37 and 0.26 in never, 0.29 and 0.42 in former, 0.57 and 0.52 in current smokers (all p ⬍ 0.0001). Baseline Variables

In all, 50% of offspring were exposed to maternal smoking and 74% to paternal smoking (Table 1). Men and women did not differ in exposure to parental smoking. There were differences between never, former, and current personal smokers in exposure to parental smoking (explored further for maternal smoking in Table 2). In offspring exposed to maternal smoking, mothers of former smokers smoked fewer cigarettes per day than mothers of never or current smokers. Personal smoking was not associated with ever or current asthma but was associated with differences in age, height, body mass index, parental and personal social class and tenure. There were no differences between former and current smokers in intensity of smoking or age at starting to smoke. Smoking intensity was greater in men, but smoking duration did not differ between men and women.

Potential Confounders of Maternal Smoking

Maternal smoking was not associated with never–ever personal smoking status. However, in offspring who had taken up smoking, maternal smoking intensity was associated with a number of personal smoking characteristics. It was negatively associated with former smoking, positively associated with current smoking, and positively associated with duration before quitting. Irrespective of personal quitting, maternal smoking intensity was positively associated with personal smoking intensity. It was also positively associated with paternal smoking intensity. There were no associations between maternal smoking status and intensity and ever or current asthma. Offspring exposed to maternal smoking were more likely to be manual workers and have manual worker parents and were less likely to own their home and have parents who owned their home. There were age, height, and body mass index differences by maternal smoking. Effect of Personal Smoking

In linear regression models to estimate effects of status, duration, and intensity of personal smoking on lung function, duration and intensity were centered at 25 years and 20 cigs/day, respectively, and set to zero in never smokers (Table 3) (25). Although duration differed between former and current smokers and intensity differed between men and women, adjusting smoking status by centered duration and intensity allowed ␤-coefficients for status to be interpreted as lung function differences between never smokers (baseline) and former or current smokers with identical smoking histories in both sexes. Coefficients for RESFEV1, FEV1/FVC, FEF25–75, and FEF25–75/ FVC generally did not differ between former and current smokers, except for greater FEF25–75 impairment in female current smokers. Former and current smokers differed in coefficients for FVC and FEV1, whereby only current smokers had lower FVC and FEV1 compared with never smokers. Results suggest that RESFEV1, FEV1/FVC, FEF25–75, and FEF25–75/FVC estimated previous smoking by former and current smokers, whereas FVC and FEV1 estimated mainly “acute” effects of current smoking. Consistent with this, duration was inversely associated with RESFEV1, FEV1/FVC, FEF25–75, and FEF25–75/FVC. There was good agreement between men and women in the strength of coefficients and pattern of results for smoking status and duration. However, heterogeneity tests showed that smoking intensity was inversely associated with FEV1/FVC, FEF25–75, FEF25–75/FVC, and RESFEV1 in women but not in men. There was no effect of smoking duration on FVC, which supports use of FVC to estimate expected FEV1, before estimating RESFEV1. Sex-specific “smoking” models in Table 3 were adjusted for height, age, age-squared, body mass index, asthma, parental and personal social class and tenure, and clustering in families. Continuous variables for maternal and paternal smoking (cigs/day) and binary variables that coded whether mothers and fathers subsequently quit were added to smoking models. To increase power for an analysis that could be stratified by personal smoking, data for men and women were combined. Fitting an interaction between sex and personal smoking intensity to improve modeling of sex differences shown in Table 3 made little difference to results for maternal (and paternal) smoking that follow. Effect of Maternal Smoking

In the total data set, 10 cigs/day maternal smoking was associated with ⫺0.062 (0.015) L FVC, ⫺0.053 (0.013) L FEV1, and ⫺0.052 (0.024) L/second FEF25–75 (Table 4). Overall, maternal smoking was not associated with RESFEV1, FEV1/FVC, or FEF25–75/FVC. When interactions were fitted between maternal and personal

Upton, Davey Smith, McConnachie, et al.: Maternal and Personal Smoking

481

TABLE 1. BASELINE VARIABLES AND LUNG FUNCTION MEASUREMENTS BY PERSONAL SMOKING STATUS IN MEN AND WOMEN (n ⫽ 2,000) Personal Smoking Status Male (n ⫽ 884) Never (n ⫽ 396) Maternal smoking, % Intensity, cigs/d‡ Paternal smoking, % Intensity, cigs/d‡ Ever asthma, % Current asthma, % Manual social class, % Parents Offspring Owner occupier, % Parents Offspring Age, yr Height, m Body mass index, kg/m2 FVC, L FEV1, L RESFEV1, L§ FEV1/FVC, % FEF25–75, L/s|| FEF25–75/FVC, s⫺1|| Personal smoking intensity, cigs/d Age started, yr Duration, yr Pack-years

Female (n ⫽ 1,116)

Former (n ⫽ 269)

Current (n ⫽ 219)

Never (n ⫽ 553)

Former (n ⫽ 283)

Current (n ⫽ 280)

43.1 14.8 (0.6) 77.7 20.2 (0.7) 6.7 4.1

53.9 16.7 (0.7) 74.9 21.1 (0.7) 4.6 2.7

47.9 15.6 (0.4) 71.8 21.1 (0.5) 9.0 6.9

49.5 14.1 (0.6) 72.4 21.0 (0.6) 7.1 5.0

55.4 16.2 (0.5) 79.6 20.4 (0.6) 8.2 5.7

65.7 31.1

68.0 41.6

68.5 56.2

70.2 15.7

59.7 23.3

19.4 92.7 43.5 (0.3) 1.76 (0.003) 26.6 (0.2) 4.94 (0.04) 3.82 (0.03) ⫺0.04 (0.01) 77.4 (0.3) 3.46 (0.05) 0.70 (0.01)

19.0 90.3 45.6 (0.4) 1.75 (0.004) 27.3 (0.20) 4.79 (0.04) 3.64 (0.03) ⫺0.12 (0.02) 76.1 (0.3) 3.14 (0.05) 0.66 (0.01)

15.5 68.5 45.1 (0.4) 1.74 (0.004) 25.8 (0.3) 4.68 (0.05) 3.47 (0.05) ⫺0.21 (0.02) 74.0 (0.5) 2.86 (0.07) 0.61 (0.01)

17.2 89.7 44.5 (0.3) 1.61 (0.002) 26.2 (0.2) 3.45 (0.02) 2.69 (0.02) ⫺0.01 (0.01) 78.2 (0.2) 2.51 (0.03) 0.73 (0.01)

15.2 91.9 46.4 (0.3) 1.62 (0.002) 26.7 (0.3) 3.48 (0.03) 2.68 (0.03) ⫺0.05 (0.01) 76.9 (0.3) 2.39 (0.04) 0.69 (0.01)

50.0 15.5 (0.4) 70.5 20.5 (0.5) 6.6 4.0

– – – –

20.6 16.8 15.1 18.1

(0.9) (0.2) (0.5) (1.1)

19.1 17.0 28.1 27.7

(0.7) (0.3) (0.5) (1.1)

– – – –

14.1 17.9 15.1 12.8

(0.6) (0.2) (0.5) (0.8)

p Values for Difference by Smoking Status Male*

Female*

All†

0.085 0.13 0.076 0.72 0.55 0.67

0.13 0.023 0.072 0.67 0.57 0.49

0.040 0.004 0.038 0.99 0.65 0.56

74.3 36.1

0.82 ⬍ 0.001

⬍ 0.001 ⬍ 0.001

0.018 ⬍ 0.001

9.3 66.4 44.9 (0.4) 1.61 (0.004) 24.8 (0.3) 3.31 (0.03) 2.49 (0.03) ⫺0.12 (0.01) 75.0 (0.4) 2.10 (0.05) 0.63 (0.01)

0.49 ⬍ 0.001 ⬍ 0.001 0.022 ⬍ 0.001 0.007 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001

0.012 ⬍ 0.001 ⬍ 0.001 0.76 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001

0.011 ⬍ 0.001 ⬍ 0.001 0.19 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001 ⬍ 0.001

0.23 0.61 ⬍ 0.001 ⬍ 0.001

0.14 0.65 ⬍ 0.001 ⬍ 0.001

0.94 0.94 ⬍ 0.001 ⬍ 0.001

15.1 17.6 27.3 20.9

(0.5) (0.3) (0.4) (0.8)

Definition of abbreviations: Cigs ⫽ cigarettes; FEF25–75 ⫽ mean forced expiratory flow during the middle half of the FVC; RESFEV1 ⫽ residual FEV1. Values are expressed as mean (SEM) for continuous variables. * p Value, adjusted for age, compares never, former, and current smokers, except for personal smoking variables where former and current smokers only are compared. † p Value adjusted for age and sex. ‡ Intensity of maternal or paternal smoking in those exposed. § RESFEV1 expresses FEV1 variation that is not correlated with FVC (see METHODS). || FEF25–75 measurements missing in three individuals.

smoking, there was evidence of heterogeneity of maternal smoking by never–former–current personal smoking status for FEV1, RESFEV1, FEV1/FVC, FEF25–75, and FEF25–75/FVC with p values equal to 0.044, 0.016, 0.016, 0.024, and 0.021, respectively. A stratified analysis showed that heterogeneity was largely explained by differences between never and current smokers in RESFEV1, FEV1/FVC, FEF25–75, and FEF25–75/FVC, with p values equal to 0.015, 0.009, 0.039, and 0.019, respectively. There were no differences between never and former smokers in coefficients for maternal smoking. In current smokers, 10 cigs/day maternal smoking was associated with ⫺0.067 (0.024) L FVC, ⫺0.087 (0.024) L FEV1, ⫺0.042 (0.016) L RESFEV1, ⫺1.01 (0.40)% FEV1/FVC, ⫺0.119 (0.041) L/second FEF25–75, and ⫺0.022 (0.010)/second FEF25–75/FVC. Splitting the data set further, to stratify by sex or asthma, increased SEs, but Tables E1 and E2 (see online supplement) show no evidence that the pattern or strength of coefficients for maternal smoking differed by sex or asthma. There were no associations between mother’s (or father’s) quitting smoking and lung function (data not shown). Effect of Paternal Smoking

Paternal smoking was not associated with FVC or FEV1 (Table E3, online supplement). However, it was weakly associated with RESFEV1, FEV1/FVC, FEF25–75, and FEF25–75/FVC in ever smokers but not in never smokers (heterogeneity by never–ever smoking status p ⫽ 0.025, 0.022, 0.040, and 0.063, respectively). Unlike

maternal smoking, results were consistent between former and current smokers. The strength of association between paternal smoking and airflow limitation was between one-third to one-half that of maternal smoking in current smokers. Synergy between Personal and Parental Smoking

Figure 1 shows coefficients for personal smoking on FEV1/FVC analogous to those in Table 3, after dividing former and current smokers into four groups defined by maternal smoking. Never smokers (baseline) were not subgrouped in this way because maternal smoking was not associated with FEV1/FVC in never smokers. As maternal smoking increased, the deleterious effect of current smoking on FEV1/FVC increased (trend p ⫽ 0.007), whereas the effect of former smoking on FEV1/FVC was unchanged (trend p ⫽ 0.89). The analysis was repeated for paternal smoking (Figure 2). Unlike maternal smoking, there was evidence of synergy between paternal and personal smoking in current (trend p ⫽ 0.03) and former (trend p ⫽ 0.10) smokers, but synergy was weak, and more obvious when former and current smokers were combined as ever smokers (trend p ⫽ 0.02). Clinical Relevance

In ever smokers who did not have current asthma, there were 56 prevalent cases of COPD versus 542 control subjects (Table 5). Compared with those not exposed to maternal smoking, odds ratios for COPD were 0.6 (95% confidence interval, 0.2–2.1),

482

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 169 2004

TABLE 2. ASSOCIATIONS BETWEEN MATERNAL SMOKING AND POTENTIAL CONFOUNDING VARIABLES (n ⫽ 2,000)

Personal smoking variables Never, % Former, % Current, % Intensity, cigs/d In former smokers In current smokers Age started, yr In former smokers In current smokers Duration, yr In former smokers In current smokers Pack-years In former smokers In current smokers Paternal smoking variables Current, % Intensity, cigs/d Ever asthma, % Current asthma, % Manual social class, % Parents Offspring Owner occupier, % Parents Offspring Males, % Age, yr Height, m Body mass index, kg/m2

Current Maternal Smoking (cigs/d)

Maternal Nonsmoking (n ⫽ 1,008) (a)

1–14 (n ⫽ 417) (b)

15–24 (n ⫽ 496) (c)

25⫹ (n ⫽ 79) (d)

48.2 29.4 22.4

46.8 30.0 23.3

47.2 22.6 30.2

43.0 24.1 32.9

Trend p Value* (Continuous) 0.47 0.008 0.001

All Current Smokers (n ⫽ 992) (b) ⫹ (c) ⫹ (d) ⫽ (e) 46.7 25.8 27.5

Binary p Value* Nonsmoking vs. Current 0.283 0.221 0.012

16.4 (0.8) 16.0 (0.6)

15.3 (1.0) 16.0 (0.8)

20.8 (1.1) 17.7 (0.8)

21.9 (2.8) 21.8 (2.2)

⬍ 0.001 0.006

18.2 (0.8) 17.5 (0.6)

0.017 0.063

17.8 (0.2) 17.7 (0.3)

17.1 (0.3) 16.5 (0.3)

16.8 (0.3) 17.3 (0.4)

16.1 (0.7) 16.9 (0.9)

⬍ 0.001 0.26

16.9 (0.2) 17.0 (0.3)

0.001 0.065

14.5 (0.5) 27.4 (0.5)

14.8 (0.8) 27.9 (0.7)

16.2 (0.8) 27.7 (0.6)

19.4 (1.9) 28.5 (1.6)

⬍ 0.001 0.26

15.8 (0.6) 27.8 (0.4)

0.020 0.066

14.3 (0.9) 22.3 (1.0)

13.7 (1.4) 22.7 (1.3)

18.8 (1.5) 25.4 (1.4)

22.9 (3.9) 32.5 (4.0)

⬍ 0.001 0.001

16.6 (1.0) 25.1 (1.0)

0.011 0.014

64.6 20.1 (0.4) 7.6 4.9

77.7 19.5 (0.5) 8.2 6.0

85.5 21.9 (0.4) 5.9 4.6

98.7 24.3 (1.3) 8.9 5.1

⬍ 0.001 0.003 0.73 0.85

83.3 21.2 (0.3) 7.1 5.2

⬍ 0.001 0.14 0.56 0.75

65.0 27.2

66.9 29.0

74.4 35.9

69.6 49.4

0.008 ⬍ 0.001

70.9 34.1

19.3 86.9 44.8 45.4 (0.2) 1.68 (0.003) 26.0 (0.1)

15.8 85.1 41.7 44.1 (0.3) 1.67 (0.005) 26.3 (0.2)

11.5 82.5 44.6 44.6 (0.3) 1.67 (0.004) 26.7 (0.2)

11.4 78.5 46.8 44.6 (0.7) 1.68 (0.010) 26.2 (0.6)

⬍ 0.001 0.006 0.75 0.022 0.039 0.023

13.3 83.3 43.6 44.4 (0.2) 1.67 (0.003) 26.5 (0.2)

0.030 ⬍ 0.001 0.002 0.037 0.44 0.002 0.029 0.147

Definition of abbreviation: cigs ⫽ cigarettes. Values are expressed as mean (SEM) for continuous variables. * p Value adjusted for age and sex.

Serum cotinine was used to investigate smoking deception, to validate reported current ETS exposure in nonsmokers, and to investigate smoking intensity in current smokers (Table E4 in the online supplement). We assessed effect modification of personal smoking by current ETS and confounding of maternal and paternal smoking by each other, by socioeconomic position (data not shown), and by personal smoking and current ETS (Tables E5 and E6 in the online supplement).

in adulthood because studies have reported that childhood LRTIs are inversely associated with adult FVC and FEV1 but not with FEV1/FVC (3, 5, 6). We found inverse associations between maternal smoking and FEVs (FVC, FEV1) irrespective of personal smoking, whereas maternal smoking was inversely associated with measurements of airflow limitation (FEV1/FVC, FEF25–75, FEF25–75/FVC, and RESFEV1) only in current smokers. Findings were similar in men and women and in those without and with asthma. The effect of 10 cigs/day maternal smoking on airflow limitation was numerically equivalent to 10 years personal smoking. We tested clinical relevance using a widely recommended case definition for COPD (percent predicted FEV1 less than 80%, FEV1/FVC less than 70%, and no asthma) (24, 26). In ever smokers, risk of prevalent COPD increased by 1.7 (1.2–2.5) per 10 cigs/day maternal smoking. As far as we know this is the first report, based on prerecorded exposure information, to indicate that maternal smoking may increase COPD risk. Unlike maternal smoking, paternal smoking was not associated with FVC but was inversely and weakly associated with FEV1/FVC, FEF25–75, FEF25–75/FVC, and RESFEV1 in former and current smokers. There was no evidence that the association between paternal smoking and airflow limitation was clinically important.

DISCUSSION

Study Design

Our starting point was doubt about the relevance of childhood exposures that affect FEV1 to smoking-related airflow limitation

It is difficult to mount studies to test interactions between exposures acting 20 or more years apart on the risk of chronic disease

2.3 (1.0–5.3), and 6.0 (1.8–20.5) in ever smokers whose mothers smoked 1 to 14, 15 to 24, and more than or equal to 25 cigs/ day, respectively (adjusting for sex, age, body mass index, duration, and intensity of personal smoking, paternal smoking, and clustering of offspring in families). With 10 cigs/day maternal smoking, COPD risk increased by 1.7 (1.2–2.5) before and 1.6 (1.1–2.5) after additional adjustment for parental and personal social class and housing tenure. In sex-specific analyses, COPD risk in men and women increased by 1.6 (1.0–2.8) and 1.8 (1.1– 3.1) (heterogeneity by sex p ⫽ 0.91) before and by 1.6 (0.9–2.9) and 1.9 (1.1–3.2) (heterogeneity by sex p ⫽ 0.97), respectively, after additional adjustment for socioeconomic position. Paternal smoking did not increase COPD risk overall or in either sex. Results were not affected by adjustment for former asthma or exclusion of offspring with former asthma. Additional Analyses

Upton, Davey Smith, McConnachie, et al.: Maternal and Personal Smoking

483

TABLE 3. PERSONAL SMOKING IN MEN AND WOMEN: ADJUSTED* EFFECTS OF FORMER AND CURRENT SMOKING COMPARED WITH NEVER SMOKING AND DURATION AND INTENSITY OF PERSONAL SMOKING ON LUNG FUNCTION (n ⫽ 884 MALE, 1,116 FEMALE) Smoking Status† Sex

Former

Current

Within-Sex p Values, Former–Current

Between-Sex p Values, Global Status

Smoking Duration (per 5 yr)

Smoking Intensity (per 10 cigs/d)

p ⫽ 0.140

0.018 (0.017) ⫺0.008 (0.013) p ⫽ 0.037

⫺0.043 (0.020)‡ 0.013 (0.022) p ⫽ 0.139

p ⫽ 0.398

⫺0.007 (0.015) ⫺0.025 (0.011)‡ p ⫽ 0.089

⫺0.037 (0.017)‡ ⫺0.037 (0.020) p ⫽ 0.731

p ⫽ 0.907

⫺0.019 (0.008)‡ ⫺0.020 (0.006)§ p ⫽ 0.684

⫺0.007 (0.010) ⫺0.045 (0.010)¶ p ⫽ 0.005

p ⫽ 0.850

⫺0.44 (0.19)‡ ⫺0.64 (0.16)¶ p ⫽ 0.460

⫺0.13 (0.23) ⫺1.25 (0.29)¶ p ⫽ 0.002

p ⫽ 0.628

⫺0.046 (0.028) ⫺0.070 (0.021)§ p ⫽ 0.225

⫺0.034 (0.031) ⫺0.124 (0.036)§ p ⫽ 0.048

p ⫽ 0.686

⫺0.012 (0.006)‡ ⫺0.022 (0.006)¶ p ⫽ 0.201

⫺0.002 (0.007) ⫺0.034 (0.009)¶ p ⫽ 0.005

FVC, L M F p||

0.047 (0.057) 0.071 (0.046) p ⫽ 0.309

⫺0.117 (0.050)‡ ⫺0.131 (0.040)§ p ⫽ 0.287

p ⫽ 0.016 p ⬍ 0.001

M F p||

⫺0.049 (0.047) ⫺0.047 (0.039) p ⫽ 0.361

⫺0.194 (0.051)¶ ⫺0.221 (0.034)¶ p ⫽ 0.535

p ⫽ 0.015 p ⬍ 0.001

M F p||

⫺0.084 (0.026)§ ⫺0.096 (0.019)¶ p ⫽ 0.781

⫺0.115 (0.032)¶ ⫺0.128 (0.017)¶ p ⫽ 0.688

p ⫽ 0.397 p ⫽ 0.159

M F p||

⫺1.83 (0.59)§ ⫺3.12 (0.52)¶ p ⫽ 0.570

⫺2.59 (0.75)§ ⫺3.32 (0.48)¶ p ⫽ 0.927

p ⫽ 0.352 p ⫽ 0.751

M F p||

⫺0.231 (0.088)§ ⫺0.304 (0.070)¶ p ⫽ 0.714

⫺0.359 (0.087)¶ ⫺0.479 (0.062)¶ p ⫽ 0.483

p ⫽ 0.273 p ⫽ 0.040

M F p||

⫺0.059 (0.019)§ ⫺0.108 (0.018)¶ p ⫽ 0.621

⫺0.070 (0.018)¶ ⫺0.106 (0.016)¶ p ⫽ 0.397

p ⫽ 0.654 p ⫽ 0.925

FEV1, L

RESFEV1, L

FEV1/FVC, %

FEF25–75, L/s

FEF25–75/FVC, s⫺1

Definition of abbreviations: Cigs ⫽ cigarettes; FEF25–75 ⫽ mean forced expiratory flow during the middle half of the FVC; RESFEV1 ⫽ residual FEV1. Values are expressed as ␤-coefficient (SE). * Adjusted for height, age, age-squared, body mass index (⬍ 20/20–24.9(baseline)/25–29.9/⭓ 30 kg/m2), asthma (never/former/current), parental and personal social class (nonmanual/manual) and housing tenure (owns/rents), and clustering of offspring in families. † In former and current smokers, duration was centered at 25 yr, and intensity centered at 20 cigarettes per day. In never smokers, centered values of duration and intensity were set to zero. Smoking status therefore compares never smokers with former or current smokers who smoked 20 cigarettes per day for 25 yr. Differences between former and current smokers provide an estimate of the acute effects of smoking, less any differences between former and current smokers before taking up the smoking habit. ‡ p ⬍ 0.05. § p ⬍ 0.01. ¶ p ⬍ 0.001. || p Values for heterogeneity by sex.

(27). In most longitudinal studies that have originated in childhood, participants are not yet old enough to test the clinical relevance in adulthood of early exposure to parental smoking. We linked prerecorded information about parental smoking to spirometric measurements at mean age 45 (range, 30–59). Strengths of our study include its general population base (14, 16–18), absence of offspring recall bias for parental smoking, serum cotinine to validate offspring’s nonsmoking and current ETS exposure, and performance feedback to technicians to improve spirometric quality (coefficients of variation for FEV1 and FVC of 3%) (19). Selection of participating families from the general population did not bias parental smoking status or relationships between traditional explanatory variables and parental FEV1. Smoking deception was present in 1 and 8% of participants who reported never or former smoking, respectively, consistent with reported low levels of smoking misclassification in population studies (28). No biomarkers of maternal or paternal smoking were measured, but reports from the original Midspan study support the validity of self-reported smoking from 1972 to 1976. For example, there were graded positive associations between smoking intensity and respiratory, cardiovascular, and all-cause mortality (15). Mothers reported smoking status and age at which they took up smoking when aged 45 to 64, on average 23 years (range,

6–39) after their pregnancies. It is reasonable to assume that midlife smoking is a marker for smoking earlier in adulthood: 93% of mothers and 95% of fathers who smoked took up smoking before age 25. Because parents reported their age at taking up smoking (or quitting) in 1972 to 1976, there is some potential for recall bias. However, at the time of the 1972 to 1976 survey, neither investigators nor participants knew that offspring would be surveyed 20 years later. We excluded offspring of mothers who started smoking after children were born, but the absence of information about smoking in pregnancy is a weakness. Nevertheless, offspring were born before there was widespread recognition of the hazards of smoking in pregnancy and passive smoking in general. Confounding

Maternal and paternal smoking main effects were adjusted for each other and for sex, age, age-squared, height, body mass index, duration and intensity of personal smoking, former and current asthma, parental and personal social class and housing tenure, current ETS exposure, and clustering of offspring in families. As with studies in children (12), there was little evidence that associations between parental smoking and lung function were confounded by socioeconomic position. In the online supplement, we argue that main effects of maternal and paternal

484

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 169 2004

TABLE 4. EFFECT OF MATERNAL SMOKING IN COMBINED DATA FOR MEN AND WOMEN STRATIFIED BY PERSONAL SMOKING: ADJUSTED* EFFECT OF 10 CIGARETTES PER DAY INCREASE IN MATERNAL SMOKING ON LUNG FUNCTION (N ⫽ 2,000)

FVC, L FEV1, L RESFEV1, L FEV1/FVC, % FEF25–75, L/s FEF25–75/FVC, s⫺1

Never (n ⫽ 949) (a)

Former (n ⫽ 552) (b)

Current (n ⫽ 499) (c)

Ever (n ⫽ 1,051) (b) ⫹ (c)

Total (n ⫽ 2,000) (a) ⫹ (b) ⫹ (c)

⫺0.070 (0.021) p ⫽ 0.001 ⫺0.053 (0.018) p ⫽ 0.003 ⫺0.005 (0.009) p ⫽ 0.595 0.04 (0.23) p ⫽ 0.872 ⫺0.047 (0.034) p ⫽ 0.160 0.000 (0.008) p ⫽ 0.955

⫺0.054 (0.025) p ⫽ 0.029 ⫺0.029 (0.019) p ⫽ 0.140 0.008 (0.012) p ⫽ 0.498 0.21 (0.29) p ⫽ 0.473 ⫺0.004 (0.041) p ⫽ 0.930 0.005 (0.010) p ⫽ 0.623

⫺0.067 (0.024) p ⫽ 0.006 ⫺0.087 (0.024) p ⬍ 0.001 ⫺0.042 (0.016) p ⫽ 0.008 ⫺1.01 (0.40) p ⫽ 0.011 ⫺0.119 (0.041) p ⫽ 0.004 ⫺0.022 (0.010) p ⫽ 0.026

⫺0.057 (0.018) p ⫽ 0.002 ⫺0.056 (0.017) p ⫽ 0.001 ⫺0.018 (0.011) p ⫽ 0.103 ⫺0.42 (0.27) p ⫽ 0.119 ⫺0.060 (0.031) p ⫽ 0.050 ⫺0.009 (0.007) p ⫽ 0.221

⫺0.062 (0.015) p ⬍ 0.001 ⫺0.053 (0.013) p ⬍ 0.001 ⫺0.011 (0.008) p ⫽ 0.133 ⫺0.20 (0.19) p ⫽ 0.293 ⫺0.052 (0.024) p ⫽ 0.027 ⫺0.004 (0.006) p ⫽ 0.46

Heterogeneity Test p Value (a) ⫺ (b) ⫺ (c)

(a) ⫺ (b)

(a) ⫺ (c)

0.614

0.599

0.692

0.044

0.568

0.073

0.016

0.866

0.015

0.016

0.841

0.009

0.024

0.627

0.039

0.021

0.970

0.019

Definition of abbreviations: FEF25–75 ⫽ mean forced expiratory flow during the middle half of the FVC; RESFEV1 ⫽ residual FEV1. Values are expressed as ␤-coefficient (SE). * Adjusted for sex, height, age, age-squared, body mass index, smoking status (except where stratified), duration and intensity of personal smoking, asthma, paternal smoking intensity, mother quit smoking, father quit smoking, parental social class and housing tenure, personal social class and housing tenure, and clustering of offspring in families.

smoking were distinct from each other (data not shown) and not confounded to an important degree by personal smoking (Tables E5 and E6). However, the possibility of residual confounding by imperfectly measured smoking is a concern in all studies that investigate smoking susceptibility. A longitudinal design would be better placed to assess the potential for residual confounding by variable smoking intensity earlier in life or subtle differences in duration of exposure to personal smoking.

Figure 1. Maternal smoking. Difference (95% confidence interval [CI]) in FEV1/FVC ratio in former (diamonds and dotted lines) and current smokers (diamonds and solid lines) compared with never smokers across the range of exposure to maternal smoking (none, 1–14, 15–24, and ⭓ 25 cigarettes [cigs]/day), adjusted for sex, age, age-squared, height, body mass index, former and current asthma, intensity and duration of personal smoking, paternal smoking, mother quit smoking, father quit smoking, parental and personal social class and housing tenure, clustering of offspring in families. Trend test by maternal smoking: p ⫽ 0.89 for former smokers; p ⫽ 0.007 for current smokers.

Effects of Personal Smoking on Lung Function

A randomized study of smoking cessation in adults aged 35 to 60 years identified acute and chronic effects of cigarette smoking on rate of change of FEV1 by analyzing results separately between 0 to 1 year and between 1 to 5 years (29). In our cross-sectional data, when we standardized duration and intensity of personal

Figure 2. Paternal smoking. Difference (95% CI) in FEV1/FVC ratio in former (closed diamonds and dotted lines) and current smokers (closed diamonds and solid lines), compared with never smokers across the range of exposure to paternal smoking (none, 1–14, 15–24, and ⭓ 25 cigs/ day), adjusted for sex, age, age-squared, height, body mass index, former and current asthma, intensity and duration of personal smoking, maternal smoking, mother quit smoking, father quit smoking, parental and personal social class and housing tenure, clustering of offspring in families. Data for former and current smokers were combined to show effects in ever smokers (open diamonds and dashed lines). Trend test by paternal smoking: p ⫽ 0.10 for former smokers; p ⫽ 0.032 for current smokers; p ⫽ 0.022 for ever smokers.

Upton, Davey Smith, McConnachie, et al.: Maternal and Personal Smoking

485

TABLE 5. PREVALENT CHRONIC OBSTRUCTIVE PULMONARY DISEASE IN EVER SMOKERS WITHOUT CURRENT ASTHMA: ODDS RATIOS FOR CHRONIC OBSTRUCTIVE PULMONARY DISEASE FROM MATERNAL AND PATERNAL SMOKING

Sex Male Female Age, yr 30–39 40–49 50–59 Trend, 5 yr Maternal smoking, cigs/d 0 1–14 15–24 ⱖ 25 Trend, 10 cigs/d Paternal smoking, cigs/d 0 1–14 15–24 ⱖ 25 Trend, 10 cigs/d Smoking duration, yr ⬍ 25 25–29.9 30–34.9 ⱖ 35 Trend, 5 yr Smoking status, intensity, cigs/d Former 0 Current 1–14 15–24 ⭓ 25 Trend, 10 cigs/d Body mass index, kg/m2 ⬍ 20 20–24.9 25–29.9 ⭓ 30 Trend, kg/m2 Sex-specific multivariate OR (95% CI) for maternal smoking Trend, 10 cigs/d

Univariate

Multivariate

Control Subjects, n (%)

Cases, n (%)

(n ⫽ 542)

(n ⫽ 56)

OR

95% CI

p Value

OR

95% CI

245 (45.2) 297 (54.8)

25 (44.6) 31 (55.4)

1.0 1.0

(0.6–1.9)

0.94

1.0 0.8

(0.4–1.6)

0.51

108 (19.9) 333 (61.4) 101 (18.6)

2 (3.6) 29 (51.8) 25 (44.6)

1.0 4.7 13.4 2.0

(1.1–20.3) (3.0–58.7) (1.5–2.7)

0.0001 ⬍ 0.0001

1.0 2.7 6.0 2.0

(0.6–11.9) (1.2–30.4) (1.3–3.0)

0.056 0.001

285 111 129 17

(52.6) (20.5) (23.8) (3.1)

22 5 21 8

(39.3) (8.9) (37.5) (14.3)

1.0 0.6 2.1 6.1 1.7

(0.2–1.6) (1.0–4.3) (2.0–18.8) (1.2–2.4)

0.001 0.003

1.0 0.6 2.3 6.0 1.7

(0.2–2.1) (1.0–5.3) (1.8–20.5) (1.2–2.5)

0.006 0.007

137 86 216 103

(25.3) (15.9) (39.9) (19.0)

13 (23.2) 4 (7.1) 25 (44.6) 14 (25.0)

1.0 0.5 1.2 1.4 1.2

0.428 0.080

1.0 0.4 1.0 1.2 1.2

(0.1–2.4) (0.5–2.2) (0.5–3.1) (0.9–1.8)

0.71 0.20

378 87 60 17

(69.7) (16.1) (11.1) (3.1)

13 18 11 14

⬍ 0.0001 ⬍ 0.0001

1.0 5.3 2.9 9.9 1.5

(2.1–13.1) (1.0–8.4) (2.5–39.7) (1.2–2.0)

0.0009 0.004

⬍ 0.0001 ⬍ 0.0001

0.8 1.5 1.2 1.2

(0.3–2.1) (0.6–3.6) (0.4–3.9) (0.8–1.9)

0.52 0.33

(2.1–13.7)

0.004 0.09

5.3 1.0 0.8 0.9 0.9

(0.1–1.9) (0.6–2.7) (0.6–3.2) (0.97–1.6)

(23.2) (32.1) (19.6) (25.0)

1.0 6.0 5.3 23.9 1.9

323 (59.6)

16 (28.6)

1.0

93 (17.2) 90 (16.6) 36 (6.6)

7 (12.5) 22 (39.3) 11 (19.6)

1.5 4.9 6.2 1.9

(0.6–3.7) (2.4–10.1) (2.4–15.8) (1.5–2.4)

22 15 9 10

3.5 1.0 0.7 0.9 0.9

(1.6–7.8)

207 213 95 27

(38.2) (39.3) (17.5) (5.0)

(39.3) (26.8) (16.1) (17.9)

Sex-specific multivariate OR (95% CI) for paternal smoking Trend, 10 cigs/d

(2.8–12.7) (2.3–12.6) (9.4–61.0) (1.5–2.4)

p Value

1.0

(0.3–1.4) (0.4–2.1) (0.8–1.0)

(0.3–1.8) (0.4–2.2) (0.8–1.0)

0.003 0.08

Men (n ⫽ 270) Women (n ⫽ 328)

1.6 1.8 Heterogeneity by sex, p ⫽ 0.91

(1.0–2.8) (1.1–3.1)

0.074 0.027

Men (n ⫽ 270) Women (n ⫽ 328)

1.3 1.2 Heterogeneity by sex, p ⫽ 0.52

(0.9–2.0) (0.7–2.0)

0.21 0.52

Definition of abbreviations: cigs ⫽ cigarettes; CI ⫽ confidence interval; OR ⫽ odds ratio. Definitions of cases and control subjects in ever smokers without current asthma: cases ⫽ percent predicted FEV1 ⬍ 80%, FEV1/FVC ⬍ 70%; control subjects ⫽ percent predicted FEV1 ⬎ 85%, FEV1/FVC ⬎ 75%.

smoking (25), former and current smokers (and men and women) showed the same degree of airflow limitation (30, 31), indicated by similar deficits in FEV1/FVC, FEF25–75, FEF25–75/FVC, and RESFEV1, whereas FEV1 and FVC were lower in current smokers. Our interpretation is that the former four parameters measured the effect of previous smoking on airflow limitation in former and current smokers, whereas differences in FEV1 and FVC between former and current smokers reflected acute smoking. A caveat, assessed in the online supplement, is that differ-

ences between former and current smokers may also result from health selection before smoking initiation or quitting. Effects of Maternal Smoking on Lung Function

In children, maternal smoking is associated with diminished levels and rates of growth of FEVs and FEFs (12). Together with our earlier study in this population (13), this analysis confirms that detrimental effects of maternal smoking on FEVs persist into adulthood. In addition, in current smokers, but not in never or

486

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 169 2004

former smokers, we found an adverse effect of maternal smoking on airflow limitation that was clinically relevant. Joint effects of maternal and personal smoking on FEV1/FVC were consistent, whether interpreted from the perspective of maternal smoking (Table 4) or personal smoking (Figure 1). The pattern of association between maternal smoking and RESFEV1 also differed by offspring’s smoking and was remarkably similar to that for FEV1/FVC, FEF25–75, and FEF25–75/FVC. Interaction tests offered limited support for heterogeneity of maternal smoking on FEV1 between never and current smokers (p ⫽ 0.073), stronger support for heterogeneity on RESFEV1 (p ⫽ 0.015), but none for FVC (p ⫽ 0.69). RESFEV1 did not give additional information about airflow limitation to FEV1/ FVC, FEF25–75, or FEF25–75/FVC. However, it made explicit that there are two general physiologic components underlying FEV1 variation at the population level, lung size and airflow limitation, that may be differently affected by exposures. Does Synergy Influence Acute or Chronic Effects of Personal Smoking?

There was no evidence that susceptibility to airflow limitation from personal smoking differed between former and current smokers (before taking account of maternal smoking). Therefore, our finding of no association between maternal smoking and airflow limitation in former smokers seems to suggest that the statistical interaction between maternal and personal smoking on lung function arises from a biological interaction linking maternal smoking to the acute effects of current smoking. On the other hand, the patterning of spirometric variables associated with maternal and personal smoking is consistent with synergy having a chronic effect on airflow limitation. As already mentioned, differences in FEV1 and FVC between former and current smokers seemed to reflect acute effects of personal smoking, whereas in current smokers, increasing exposure to maternal smoking lowered FEV1/FVC, FEF25–75, FEF25–75/FVC, and RESFEV1, indicators (Table 3) of chronic rather than acute effects of personal smoking. Without longitudinal data with which to assess acute and chronic effects of personal smoking, it is difficult to choose between these interpretations and impossible to know whether any chronic effect (if it occurs) of synergy is mediated through growth or decline of lung function (32). In the online supplement we argue that there was no evidence that synergy was mediated through asthma, although we did not collect prospective information about childhood asthma. A possible, but speculative, explanation for the lack of effect of maternal smoking on airflow limitation in former smokers may have been differences in exposure to maternal (but not paternal) smoking during childhood in the families of those destined to become former smokers, compared with never and current smokers. There is scope for this because we inferred early exposure to parental smoking from reports many years later. Table 1 shows that mothers, but not fathers, of former smokers reported smoking fewer cigarettes per day compared with parents of never or current smokers. If such differences had been accompanied by other variation in smoking behavior, for example, less maternal (but not paternal) smoking inside the home, then this might have accounted for our findings. Paternal Smoking

We found synergy between paternal and personal smoking on airflow limitation in former and current smokers, although the effects were weak. Nevertheless, the patterning and similarity of the associations in question between former and current smokers supports a chronic effect of synergy between paternal and personal

smoking on airflow limitation. We did not find an effect of paternal smoking on FVC or FEV1, unlike a study of children in a rural Chinese population where it was uncommon for women to smoke (33). Our paternal smoking findings are consistent with postnatal exposure to parents’ smoking in the home being relevant to synergy with personal smoking. Differences in the strength of effect between maternal and paternal smoking suggest either that prenatal exposure is also relevant to synergy or may simply reflect the fact that children generally are in closer proximity to their mothers. Interestingly, there was no evidence of synergy between personal smoking and current ETS exposure on airflow limitation. Conclusions

Maternal smoking adversely affects offspring’s lung function in at least three ways. It (1 ) lowers lung volume irrespective of own smoking, (2 ) is associated with greater smoking intensity and less quitting in those who have taken up smoking, and (3 ) appears to synergize with personal smoking to increase airflow limitation in adults. Longitudinal studies are needed to test whether maternal smoking modifies acute or chronic effects of personal smoking and whether any chronic effects operate through growth or decline of lung function. It is a concern that replicate surveys found that the prevalence of maternal smoking in pregnancy remained constant throughout the 1990s (34). Conflict of Interest Statement : M.N.U. has no declared conflict of interest; G.D.S. has no declared conflict of interest; A.McC. has no declared conflict of interest; C.L.H. has no declared conflict of interest; G.C.M.W. has no declared conflict of interest. Acknowledgment : The authors are grateful to participants of both Midspan studies, to Victor Hawthorne who initiated the original Renfrew-Paisley (Midspan) study, to the office staff and research nurses who sent invitations, checked questionnaires and carried out lung function tests, to Colin Feyerabend for cotinine assays, and Neil Pride and Dan Teculescu for commenting on the data. Thanks also to Yoav Ben-Shlomo, Dan Dedman, Anne McCarthy, David Strachan, Neil Thomson and Erika von Mutius for commenting on early drafts of the manuscript, and to the anonymous reviewers for helpful suggestions for revising the manuscript.

References 1. Murray CJ, Lopez AD. Alternative projections of mortality and disability by cause 1990–2020: global burden of disease study. Lancet 1997;349: 1498–1504. 2. Buist AS. Risk factors for COPD. Eur Respir Rev 1996;6:39253–39258. 3. Barker DJ, Godfrey KM, Fall C, Osmond C, Winter PD, Shaheen SO. Relation of birth weight and childhood respiratory infection to adult lung function and death from chronic obstructive airways disease. BMJ 1991;303:671–675. 4. Shaheen SO, Barker DJ, Shiell AW, Crocker FJ, Wield GA, Holgate ST. The relationship between pneumonia in early childhood and impaired lung function in late adult life. Am J Respir Crit Care Med 1994; 149:616–619. 5. Johnston ID, Strachan DP, Anderson HR. Effect of pneumonia and whooping cough in childhood on adult lung function. N Engl J Med 1998;338:581–587. 6. Shaheen SO, Sterne JA, Tucker JS. Florey CD. Birth weight, childhood lower respiratory tract infection, and adult lung function. Thorax 1998;53:549–553. 7. Weiss ST. Early life predictors of adult chronic obstructive lung disease. Eur Respir Rev 1995;5:31303–31309. 8. Retamales I, Elliott WM, Meshi B, Coxson HO, Pare PD, Sciurba FC, Rogers RM, Hayashi S, Hogg JC. Amplification of inflammation in emphysema and its association with latent adenoviral infection. Am J Respir Crit Care Med 2001;164:469–473. 9. Rasmussen F, Taylor DR, Flannery EM, Cowan JO, Greene JM, Herbison GP, Sears MR. Risk factors for airway remodeling in asthma manifested by a low postbronchodilator FEV1/vital capacity ratio. Am J Respir Crit Care Med 2002;165:1480–1488. 10. Higgins MW, Keller JB. Seven measures of ventilatory lung function. Am Rev Respir Dis 1973;108:258–272. 11. Cook DG, Strachan DP. Summary of effects of parental smoking on the respiratory health of children and implications for research. Thorax 1999;54:357–366.

Upton, Davey Smith, McConnachie, et al.: Maternal and Personal Smoking 12. Cook DG, Strachan DP, Carey IM. Parental smoking and spirometric indices in children. Thorax 1998;53:884–893. 13. Upton MN, Watt GCM, Davey Smith G, McConnachie A, Hart CL. Permanent effects of maternal smoking on offsprings’ lung function. Lancet 1998;352:453. 14. Hawthorne VM, Watt GCM, Hart CL, Hole DJ, Davey Smith G, Gillis CR. Cardiorespiratory disease in men and women in urban Scotland: baseline characteristics of the Renfrew/Paisley (Midspan) study population. Scott Med J 1995;40:102–107. 15. Watt GCM, Hart CL, Hole DJ, Davey Smith G, Gillis CR, Hawthorne VM. Risk factors for cardiorespiratory and all cause mortality in men and women in urban Scotland: 15 year follow up. Scott Med J 1995; 40:108–112. 16. Hole DJ, Watt GCM, Davey Smith G, Hart CL, Gillis CR, Hawthorne VM. Impaired lung function and mortality risk in men and women: findings from the Renfrew and Paisley prospective population study. BMJ 1996;313:711–716. 17. Davey Smith G, Hart C, Ferrell C, Upton M, Hole D, Hawthorne V, Watt GCM. Birth weight of offspring and mortality in the Renfrew and Paisley study: prospective observational study. BMJ 1997;315: 1189–1193. 18. Upton MN, McConnachie A, McSharry C, Hart CL, Davey Smith G, Gillis CR, Watt GCM. Intergenerational 20 year trends in the prevalence of asthma and hay fever in adults: The Midspan Family Study surveys of parents and offspring. BMJ 2000;321:88–92. 19. Upton MN, Ferrell C, Bidwell C, McConnachie A, Goodfellow J, Davey Smith G, Watt GCM. Improving the quality of spirometry in an epidemiological study: The Renfrew-Paisley (Midspan) Family Study. Public Health 2000;114:353–360. 20. American Thoracic Society. Standardization of spirometry: 1987 update. Am Rev Respir Dis 1987;136:1286–1296. 21. Feyerabend C, Russell M. Rapid gas-liquid chromatographic determination of cotinine in biological fluids. Analyst 1980;105:998–1001. 22. Stick SM, Burton PR, Gurrin L, Sly PD, LeSoue¨f PN. Effects of maternal

23. 24. 25.

26.

27. 28. 29.

30. 31. 32. 33.

34.

487 smoking during pregnancy and a family history of asthma on respiratory function in newborn infants. Lancet 1996;348:1060–1064. StataCorp. Stata Statistical Software: Release 7.0. College Station, TX: Stata Corporation; 2001. British Thoracic Society. Guidelines for the management of chronic obstructive pulmonary disease. Thorax 1997;52:S1–S28. Leffondre´ K, Abrahamowicz M, Siemiatycki J, Rachet B. Modeling smoking history: a comparison of different approaches. Am J Epidemiol 2002;156:813–823. Pauwels RA, Buist AS, Calverley PMA, Jenkins CR, Hurd SS. Global strategy for the diagnosis, management and prevention of chronic obstructive lung disease: NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD): workshop summary. Am J Respir Crit Care Med 2001;163:1256–1276. Kuh D, Ben-Shlomo Y. A life course approach to chronic disease epidemiology. Oxford: Oxford University Press; 1997. Rebagliato M. Validation of self reported smoking. J Epidemiol Community Health 2002;56:163–164. Anthonisen NR, Connett JE, Kiley JP, Altose MD, Bailey WC, Buist AS, Conway WA, Enright PL, Kanner RE, O’Hara P, et al. Effects of smoking intervention and the use of an inhaled anticholinergic bronchodilator on the rate of decline of FEV1: The Lung Health Study. JAMA 1994;272:1497–1505. McFadden ER, Linden DA. A reduction in maximum mid-expiratory flow rate. Am J Med 1972;52:725–737. Macklem PT. Workshop on screening programs for early diagnosis of airway obstruction. Am Rev Respir Dis 1974;109:567–571. Weiss ST, Ware JH. Overview of issues in the longitudinal analysis of respiratory data. Am J Respir Crit Care Med 1996;154:S208–S211. Venners SA, Wang X, Chen C, Wang B, Ni J, Jin Y, Yang J, Fang Z, Weiss ST, Xu X. Exposure-response relationship between paternal smoking and children’s pulmonary function. Am J Respir Crit Care Med 2001;164:973–976. Owen L, McNeill A, Callum C. Trends in smoking during pregnancy in England, 1992–7: quota sampling surveys. BMJ 1998;317:728.