Background: We have previously shown that the inverse social gradient in risk of ischaemic heart disease (IHD) was not explained by self-reported smoking habits. We pursued the issue in a follow-up study 15 years later, where use of tobacco was validated by serum cotinine.
Methods: Some 3216 men aged 53-75 years were included in a study on the association between self-reported tobacco use and serum cotinine concentration. The men had their morbidity and mortality recorded over 4 years. Some 2833 men without overt cardiovascular disease were included in the incidence study. Potential confounders examined were serum lipids, serum selenium, alcohol consumption, physical activity, hypertension, blood pressure, and body mass index.
Results: There was a strong positive correlation between serum cotinine level and self-reported tobacco smoking: r = 0.68, P < 0.0001. The misclassification rate of smokers as non-smokers was apparently higher in low social class. However, a larger proportion of men in low social class were users of chewing tobacco or snuff, and, when taking this into account, there was no social gradient (i.e. trend) in the estimated misclassification rates from social class I to social class V: 1.0%, 3.8%, 3.2%, 2.0%, 2.3%, P = NS. After validation of use of tobacco with serum cotinine measurements, compared with social class I, social class V had an overall significantly increased risk of IHD, relative risk = 4.5 (95% confidence interval: 1.6-12.9), P < 0.01, which was slightly higher than when no validation was performed.
Conclusions: We conclude that, (i) social differences in use of tobacco validated by measurements of serum cotinine did not account for social inequalities in risk of IHD in middle-aged and elderly men, (ii) no significant social differences existed in the misclassification of smokers as non-smokers, (iii) reclassification of self-reported non-smokers should not be done without due consideration of the use of chewing tobacco and snuff.