Vineis et al report a prospective case-control study examining 123 479 “healthy” never smokers or former smokers. Over 7 years of follow up, cases were defined as “respiratory cancers” or deaths from COPD. Controls were well matched. Information about tobacco smoke exposure and a thorough list of confounders was obtained by questionnaire at recruitment. This strengthens the study by cutting out recall bias. Follow up data were from multiple sources. A subset of subjects and controls were tested for the presence of genetic polymorphisms implicated in carcinogenesis.

Exposure to tobacco smoke was associated with increased risk of death from respiratory cancers/COPD (hazard ratio (HR) 1.30, 95% CI 0.87 to 1.95) and lung cancer alone (HR 1.34, 0.85 to 2.13). The effect was significantly greater for exposure at work (HR for lung cancer 1.65, 1.04 to 2.63) than at home (HR 1.03, 0.60 to 1.76) and for former smokers than for never smokers. Work exposure may have been higher than at home but exposure was not quantified and only measured at one point in time. The authors postulate that the greater effect in former smokers may be due to genetic mutations already accrued. The case for causality is strengthened as genetic polymorphisms increase the susceptibility to smoke: odds ratio (OR) for lung cancer increased from 1.33 to 2.86 between the presence of one or two versus three or more risk polymorphisms. Self-reported exposure to smoke in childhood showed a clear dose-response effect in never smokers. Daily exposure for many hours as a child increased the HR for lung cancer in adulthood from 1.0 to 3.63 (1.19 to 11.11).

McGhee et al report a retrospective case-control study. Cases were identified from all deaths reported amongst never smokers, and those reporting the death completed a questionnaire about the deceased. Second hand smoke exposure was classified by the number of smokers who lived with the deceased (0, 1, 2 or more). There was no attempt to further quantify exposure. This retrospective nature risks recall bias and the only confounder controlled for was education, a proxy of social class. There was a 34% increase in all cause mortality with any exposure to second hand smoke. Importantly, there was a significant dose-dependent association between second hand smoke and mortality from lung cancer (OR 1.74, 1.06 to 2.86 comparing no exposure to living with >2 smokers). There was a similar association between exposure and mortality from COPD (OR 2.51, 1.22 to 5.18), stroke (OR 2.08, 1.33 to 3.25) and ischaemic heart disease (OR 1.68, 1.05 to 2.68).

These studies add impetus to the demand for restrictions on smoking in public places. In the Hong Kong study deaths from poisoning were also analysed, as a control measure, and showed no relation to tobacco smoke exposure. Perhaps now there is an argument for reclassifying some of the other environmental tobacco related deaths as “poisoning”?