Tobacco Documents Online

REVIEW 730, CONFIDENTIAL Subject ref 8b "Chapter 6: Passive Smoking" from the "Auckland Heart Survey" R T Jackson. PhD Thesis, Uhiversity of Auckland,, Aucklan&„ New Zealand, 1989, pp 157-1J2' Recently I was in correspondence with Dr K Brown in~connection with a forthcoming,review he is preparing for OSHA on ETS and heart disease. Material he sent me referred inter al!ia to a thesis by Jackson on the Auckland Heart Survey, which I was previously unaware of. Accordingly I asked Brown for a copy of this. He actually sent me only the chapter concerning passive smoking, together with copies of correspondence between Jackson~„ Glantz and Wells which clarify some of the details. The whole of the material sent is attached.. The ETS analyses are based on~ a case-control study involving the following numbers of never smokers: Men Women MI = non-fatal myocardial infarct cases 28 11 MIC = controls for MI 123' 112' CD - coronary death cases 21 9 CDC = controls for CD 61 62' No details are given in Chapter 6 (presumably they were given in earlier chapters) of how the various cases and control groups were defined. It seems from p158 that the subjects were aged 35-64 and that

-2- the controls for the MIi cases were population controls, and it is probable that the controls for the coronary death cases were also decedents (else they would only have had one control group) but these are really only educated guesses. Note that Jackson does point out there is some overlap between,the MI and'CD cases, as there i.s no restriction that decedent cases should not previlosly have been i'nclud'ed in the l'iving, cases earlier when alive. For the purpose of analysis relating to ETS exposure, subjects were divided into those exposed at home, at work, at home and at work, or neither. For home exposure, which appeared to be based on whether a cohabitant smoked or not, subjects were also classified as having high exposure if more than one cohabitant smoked or there was exposure to more than seven cigarettes per day at home. The analyses related to ETS exposure excluded any subject with a past admission to hospital for CHD or angina as diagnosed by the Rose questionnaire. Thils did not apply to the numbers of never smokers cited above. Thus the true numbers of never smokers used in analysis are even lower thanithe smalli numbers ciited in that table. The relative risks with 95% confidence intervals are summarized below for convenience:

Men 3 Women Crude MI CD MiI CD Home only - 0.6(0.06-5.9) 2.4(0.4-14.1) 23.5(2.8-199.0). Work only 1.5(0,6-3'.8) 1.2(0.4-3.6) 0.9(0.09-8.0) 5.2(0.9-30.1). Home and work 2.7(0.60-12.1) 1.6(0.3-9.6) 3.6(0.4-49.5) - Home +_ work 1.1(0,3-4.3)' 1.0(0.2-4.5) 2.8(0.6-13.6) 7.8(1.3-48.0). Work ± home 1'.7(0.7-4.0) 1.3(0.5-3.6) 1.4(0.3-8.2) 3'.6(0.7-2-0.1). Adlusted for aFe and':social class Home + work 1.03 Work + home 1.8 1.1 2.7 5.8. 1.8 1.1 2.2' MI + CD MI + CD Crude Home + work 1.1(0.4-3.0) 4.0(1.35-13.1) Work + home 1.5(0.8-3.0) 2.2('0:.7-7'.4). Home - low 1.3(0.4-4.2), 2.1(',0.4-11.0) Home - high 0.9(0.4-4.3), 7.5(1.8-30',5) Points to note from these results are as follows: (i) None of the results for males are even close to being statistically significant. (ii) Although the relative risks associlated with at home exposure inn females are moderately elevated for MI cases they are again not close to being statistically significant. (iiii), The only significant increases noted are in relation to at home exposure in females for coronary deaths. However, although the magnitude of the increase l!ooks impressive, one should realize: (a) it is totally implausibly large, much larger than RRs associated with active smoking,;. (b) it has huge variability, based on only nine coronary deaths at most; (c) it becomes non-significant after adjustment for age and social class (although confidence intervals are not given for the adjusted relative risks, one can deduce they are not

4- significant); (d) no attempt has been made to adjust for any other coronary risk factors; (e) no correction to significance levels has been made for multiple testing - with multiple indices of exposure, endpoints and sexes it is scarcely surprising some relative risks come out si~gnificant; (f) there is a possibility of recall bias, if awareness of disease affects reporting of ETS exposure. (iv) Jacksorn himselif states he never published the results in a paper because he felt the numbers were too small. All in all', the results from this study add little to the overall evidence. Note that in the correspondence Wells refers to a review of passive smoking and heart disease he is doing for the American, College of Cardiology. Also Jackson refers to another case-control study they are doing which shoul~d be available mid 1i994. P Ni Lee 24.5.94