Mortality and morbidity of potentially misclassified smokers

Misclassification of smokers as non-smokers may bias estimates of the excess morbidity and mortality associated with smoking. The issue has been given little, if any, attention in prospective epidemiological studies. This study examined characteristics of potentially misclassified smokers with respect to mortality, morbidity, and risk factors.

A prospective study (within The Copenhagen Male Study, Denmark) used serum cotinine as an objective marker of use of tobacco. A serum concentration of 100 ng/ml was regarded as a relevant threshold for active smoking. In all, 3270 males aged 53-74 years who reported their previous and current tobacco habits, including the use of chew tobacco and snuff, were included. Incidence of all causes of mortality (ACM) during 9 years and death due to ischaemic heart disease (IHD) during 8 years of follow-up were the main outcome measures.

Overall cumulative incidence rates of ACM and IHD were 19.1% and 4.3%, respectively. Of 1405 men who reported being non-tobacco users, i.e. no current smoking and no use of chewing tobacco or snuff, 1377 had levels < 100 ng/ml, 28 men (2%) had levels equal to or above this threshold value and were considered potentially misclassified smokers. They had significantly higher mortality rates, 35.7% versus 14.7%, P < 0.001, than other self-reported non-tobacco users, and a slightly higher prevalence of tobacco-related cancer, and a highly significant higher prevalence of myocardial infarction, P < 0.001. Compared to non-tobacco users with low cotinine, age-adjusted relative risks (95% CI) were 2.4 (1.3-4.5), P < 0.01, for ACM, and 5.3 (95% CI : 2.1-13.4), P < 0.001, for IHD.

Potentially misclassified smokers deviated strongly from other non-smokers with respect to mortality and morbidity. The importance of this reporting bias when estimating the risk associated with active or passive smoking is discussed.