Tobacco Documents Online

3 with, inter alia, no discussion of control of confounding variables, publication bias or bias due to misclassification of active smoking. All that has been demonstrated is that the evidence tends to show a weak negative association. P N Lee 12.10.99.

2 100 ml lower than those in the lowest cotinine quintile. The relationship was stronger for males than females and for ex-smokers than never smokers. None of the associations noted for never smokers were statistically significant. (iii) Longitudinal analyses relating change_in FEV, from HALS1 to HALS2 to change in living with a smoker (Table 4) Compared to those not living with a smoker at either HALSI or HALS2 there was no evidence that change in FEV, was any different for those living with a smoker at HALS 1 only, at HALS2 only or at both times. (iv) Analyses relating FEVI to parental smoking No significant relationships were seen. (v) Meta-analyses of data from 21 studies relating FEV, to ETS in adults ETS exposure was estimated to be associated with a decline of-1.7% in FEV, (95% CI -2.8% to -0.6%). While it is possible that ETS exposure may result in a small decrease in lung function in adult nonsmokers, the analyses of data from HALS 1 and HALS2 do not provide strong evidence that it does. One should note: (i) None of the relationships seen in never smokers were statistically significant; (ii) It is difficult to control fully for past smoking in ex-smokers (who are not usually considered in analyses of ETS effects); (iii) The control for potential confounding factors is rather limited, surprisingly so since I have a copy of the data on my computer and there is extensive information on a whole range of potential confounders. It would also have been useful to know the effect that adj ustment for the confounders they did consider (social class, region and pack-years) had on their estimates. It is also notable that the section on meta-analyses of adult studies of ETS is far too brief,

~(!?- 4 -3 3 I REVIEW 1072 CONFIDENTIAL Subject ref 8c "The effects of environmental tobacco smoke exposure on lung function in a longitudinal study of British adults" IMCareyetal Epidemiology (1999), 10, 319-326 This paper describes analyses relating ETS exposure to lung function based on the UK Health and Lifestyle Survey in which 9003 adults were interviewed initially in 1984 and 1985 (HALS 1) and then followed up seven years later (HALS2). Data on FEV, were available at both times, as were data on smoking habits and ETS exposure. Salivary cotinine was available for HALS2. Attention was restricted to 1623 subjects who satisfied the following criteria: age t 8-73 at HALS1, satisfactory FEV1 data at both times, never or ex-smokers, cotinine data available, cotinine not >14.7 ng/ml (i.e. not active smokers) and in the middle 99% of the distribution of change in FEV,. FEVI was adjusted for age and height. Social class, region of residence and pack-years of past smoking were taken into account as potential confounding variables. The main fmdings from this very clearly presented paper were as follows: (i) Cross-sectional analyses relating FEV~ to living_with a smoker (Table 21 At both HALSI and HALS2 the overall data showed essentially no relationship. However this concealed some contrasting patters with, at both surveys, living with a smoker being associated with a marginally significant increased FEVt (by about 80 ml) in female never smokers and with a close to significant decreased FEV, (by about 140 ml) in male ex-smokers. Smaller decreases in male never smokers and female ex- smokers were not close to statistically significant. (ii) Cross-sectional analyses relative FEV, to cotinine (Table 3) Based on HALS2 data, there was a significant regression of FEV, on log cotinine (-26, 95% CI -44 to -8) with subjects in the highest cotinine quintile having FEV, levels about