Results

The mean levels of serum thiocyanate were 157.2 μmol/l in men and 152.5 μmol/l in women. As in other studies, the relation between serum thiocyanate and cigarette consumption was curvilinear, tending to asymptote at about 15–20 cigarettes per day, reflecting smokers tendency to titrate their intake at higher consumption level (data not shown).

There was a clear educational gradient in cigarettes per day and in serum thiocyanate among men but not among women (table 1). Men in the lowest educational category smoked, on average, six cigarettes per day more, and their plasma thiocyanate was 26.8 μmol/l higher than that of those in the highest category. Controlling for the average number of cigarettes per day reduced the social gradient in thiocyanate (the difference between the highest and lowest category was reduced to 16.5 μmol/l), and the trend became non-significant. Educational differences among women were smaller. Neither car ownership nor crowding were associated with serum thiocyanate in men or women.

Discussion

We found that serum thiocyanate concentrations were inversely associated with education among men. The absence of this association among women may partly be caused by the small number of women with university education, but it is also consistent with relatively small social differences in smoking in Czech women.6 There seems to be a cohort effect among women; the gradient has the “western” direction (higher prevalence of smoking in women with lower education) among younger women (< 35 years old) but the opposite among older women. The absence of an association between serum thiocyanate and car ownership and crowding is consistent with previous studies of socioeconomic differences in health in the Czech population; education appears to be the most important social indicator while factors reflecting material conditions are less important.6 9

Although thiocyanate is a less accurate indicator of smoke intake than cotinine, the results clearly demonstrated that among men in this population, the educational gradient was not limited to the prevalence of smoking, but was also pronounced among smokers. Higher intakes can be achieved by different ways, including smoking more cigarettes or inhaling more smoke from each cigarette. In either case, the smokers choose the level of intake they prefer. In our data, the educational differences in serum thiocyanate were only partly attributable to differences in the average number of cigarettes per day. This suggests that smokers with lower socioeconomic status also inhaled the cigarettes they smoked more intensively. Socioeconomic differences may thus have contributed to the racial differences in plasma cotinine observed in the USA.2-4

There are two main potential confounders: diet and passive smoking. The contribution of diet, particularly beer and some foodstuffs, to thiocyanate concentrations is well known. Information on beer consumption was available in our dataset, but controlling for this variable made no difference to our finding. We were not able to adjust for consumption of vegetables such as brassicas, but it is unlikely that this could account for our findings. In this population, we have previously found no socioeconomic differences in consumption of different types of vegetables (M Bobak, unpublished data). It could be argued that smokers from lower socioeconomic groups would also be more heavily exposed to other people's tobacco smoke, and that passive smoking could therefore be a confounder. Again, this is unlikely. Previous work has shown that thiocyanate is not sensitive to passive smoking exposure.10 Data on self reported passive smoking were available, but adjustment for this again made no difference to our results.

The data support the hypothesis that nicotine addiction may be stronger among smokers in lower socioeconomic groups. This may contribute to the lower effectiveness of anti-smoking interventions among smokers in lower social strata. Nicotine addiction may also help to explain the observation that most of the current social gradient in smoking is caused by low cessation rates in lower social groups although they seem to be similarly motivated to quit.5 11If this hypothesis proves to be true, policies to reduce the socioeconomic gradient in smoking may need to be social group specific and take into account the higher levels of nicotine dependence in lower social groups.