Background
When low smoking rates were observed among doctors in the 1960's they were attributed not only to better awareness of the hazards of smoking within the medical profession but also to an unidentified factor related to academic education. Thus Lynch [
1], for example, found low smoking rates among employees in both medical and non-medical faculties, and Brown and Gunn [
2] reported low rates among employees in a university where no medical faculty existed. The low prevalence of smoking among the educated people is well-known nowadays [
3,
4], but few studies have looked for variations within educated communities such as universities. It would seem likely that the scientific career pursued in a university, or the particular academic culture in which one works, could affect smoking habits through social learning. We conducted a smoking survey in the University of Tartu (UT), Estonia, soon after the disruption of communism (1992) and found low but still widely varying smoking rates among the staff and students in different faculties, the prevalence of daily smoking ranging from zero to 30% [
5]. Smoking was most common in faculties having the closest connection with the contemporary changes in society, such as the faculty of theology, and among female employees in the faculties of law and economics, while the rates were very low in the faculties of exercise & sports and mathematics, where physical or mental performance are of particular significance.
Since Estonia's transition to a western economy may have modified smoking patterns, the UT employees were re-surveyed in 2003, two years before the enforcement of the Estonian Tobacco Act. In general, we expected to see an overall decline in smoking rates throughout the faculties-despite persistently high national rates. This could be predicted from innovation diffusion theory, which presupposes that new trends in smoking are first adopted by the most educated people [
6]. In particular, we tested the hypothesis that smoking differences between the faculties still exist after the first transitional decade and are similar to those observed immediately after the disruption of communism. Any departures from the previous pattern would suggest that the changes in Estonian society during the 1990's had interfered with smoking trends in different ways depending on the faculty. Information on smoking trends among the highest educated professionals would also add to our understanding of the natural course of the smoking epidemic in transitional societies.
Discussion
It was found in the first national health survey conducted in Estonia (1990) that 50% of working-aged men smoked daily and that 40% of men with an academic education did so [
10]. We surveyed the UT staff at the same time [
5] and found the men's smoking rates to be as low as 20%, although they varied widely from one faculty to another. This led us to the hypothesis that the declining phase of the smoking epidemic among this highly educated community is at a very advanced point, but the course of the epidemic is likely to be affected by differing professional cultures and may perhaps be modified by concurrent changes in society. In this re-survey conducted after Estonia's first transitional decade we show not only that smoking rates among the UT staff have further declined but also that differences still exist between the faculties which are not entirely dissimilar to the previous patterns. The findings can be interpreted in terms of a heterogeneous diffusion of non-smoking behaviour into the most educated segment of the country's population. It remains unclear, however, how well the present findings are applicable to other academic communities.
The validity of our data can be considered to be reasonable. Smoking was elicited using identical questions in the two surveys and the response rates were acceptable, although we cannot rule out biases possibly introduced by selective participation. Thus lower response rates in some faculties, for example, may have led to underestimates of their smoking rates. Since non-responders were not examined, there is a possibility of non-response bias. Smoking is known to be slightly more prevalent among non-respondents [
11], but the resulting bias is reportedly small [
12]. Underreporting of smoking is a possibility, especially among women [
13,
14], but it would seem unlikely to have introduced any major flaws. The smoking questions used in the national surveys were slightly more restrictive than the present questions on daily smoking, but the effect of this on smoking rates is considered to be small [
8]. Similarly, the lower response rates in the national surveys (63% in both 1992 [
7] and 2004 [
8]) compared with those in UT can hardly introduce any marked flaw, since the differences in smoking rates were large.
A further limitation is that the smoking differences reported here cannot be interpreted as faculty-specific effects for the university community alone, since 93% of the male respondents and 80% of the females had started their smoking before the age of 25 years, and a half had quit before that age [
15]. The results can be understood partly in terms of different mixes of workers entering the faculties or some other kind of selection, and partly as reflecting the effects of the individual's working community. The quit ratios are informative in the latter respect, but as they are not based on an actual follow-up of individual respondents, they are only crude indicators of the propensity to quit. Since we were unable to conduct separate analyses for detailed occupational categories, the results only apply to entire faculties or other workplaces. Differences in occupational composition between the workplaces could have confounded the results, but this factor was allowed for in the analysis.
Even though our sample is larger than those in most other smoking studies conducted among university employees [
1,
2,
16‐
20], the statistical power was low in some faculties, e.g. the faculty of law. Therefore, the results regarding the smallest strata should be approached with caution. Similar studies performed in other academic communities with pooling of the data would be desirable.
To our knowledge this is the first study to compare smoking patterns between employees in a wide range of university faculties, and especially the first one to assess recent changes in such patterns in a former socialist country. Previous studies have compared smoking between people working in medical and non-medical faculties [
1], in combinations of faculties [
2], or using only small samples [
17,
18].
It is frequently observed that members of a medical faculty [
1,
17‐
20], doctors in university hospitals [
21], and physicians outside academies [
22,
23] smoke less than employees in other faculties or professions, or people in the general population. Even so, smoking among doctors is regarded as disproportionately common bearing in mind their medical knowledge and their significance as role models [
22]. We actually noted one of the highest prevalences of smoking among the men of the medical faculty, although the contrary was true of the women. Only one previous study comparing several faculties has reported a high smoking rate in a medical faculty, but the sample was small and there was no breakdown by sex [
16]. In our sample, the relatively low quit ratios among the men in the medical faculty may indicate a poor propensity for behavioural changes or a failure to accept the medical information even though it is readily available. This may not apply to the women in the medical faculty, who showed one of the highest quit ratios in the second survey.
The high prevalence of smoking in the faculty of philosophy (effectively humanities) is a new finding, although students of the humanities and fine arts are said to be smokers [
24,
25]. The relatively high percentages of daily and ever-smokers in both surveys and the relatively low quit ratios among men and women apparently indicate some life style-related resistance to abstinence from smoking and place this faculty among the last ones to conform to the general decline in smoking.
Another novel finding was the high prevalence of men's smoking in the faculty of biology and geography. Contrary to other faculties having initially high smoking rates, the men in this faculty had greatly reduced their smoking, as also indicated by the increase in the quit ratio from 7% to 64%. The trend is not easily explained, but one might conceive that the growing interest in environmental issues which was a catalyzing factor in the country's independence movement in the late 1980's might have influenced the mix of people entering this faculty and introduced some health-based selection.
Smoking was rare in the faculties of sports & exercise, mathematics and physics & chemistry. Although no similar findings have been reported elsewhere, our findings comply well with the knowledge that physical exercise and smoking are negatively associated [
26] and that sportsmen and sportswomen have low smoking rates [
27,
28]. No such information exists for mathematicians, physicists or chemists, however, except that it has been conjectured that students of mathematics may have low smoking rates [
25]. We assume that the harmful effects of smoking on physical and cognitive performance [
29‐
31], perhaps also reflected in poorer school results [
32,
33] and educational achievements [
34] among smokers, have prevented people in these faculties from smoking.
The faculty of law was the only one where women smoked more than men in both surveys, and a difference of a similar kind was seen in the faculty of economics in the second survey. These faculties have had the closest connection with the recent changes in Estonian society, and the reversed sex ratio probably indicates the women's intensified drive for equality, and perhaps also coping-related stress under rapidly changing conditions. No other study has analysed sex ratios in smoking among an academic population in a transitional society.
Conclusions
Smoking differences at a given point in time are best interpreted as a snapshot from a longitudinal process, the smoking epidemic [
6]. The declining phase of the epidemic has started only recently in Estonia as a whole [
35], but it may have started some decades ago within the country's academic community [
36]. We have shown here that even within this highly educated segment of the population a non-smoking form of behaviour is adopted in a heterogeneous manner, and paradoxically, not necessarily in the medical faculty first, but rather in faculties where physical or mental performance is of particular significance. We assume that the smoking epidemic in UT is now approaching its terminal phase, some of its faculties being almost smoke-free, which is in sharp contrast to the persistently high smoking rates in the population as a whole. Local anti-smoking actions are still needed and should be directed at male employees in the faculty of philosophy, which now lags behind, and also at men in the medical faculty, which should lead the way to a smoke-free academic community. As UT is the intellectual and cultural hub of Estonia, it serves as a major centre of innovation [
6], and successful efforts within this community could speed up the diffusion of non-smoking throughout the country.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
The study was conceived by PHK, RH and JK. The data analysis was performed by SN and RH, and the manuscript was first drafted by RH and revised and finalized by SN, JK and PHK. All the authors have read and approved the manuscript. PHK and SN are the guarantors of the study.