Trends in smoking behaviour among Estonian physicians in 1982–2014

Background Smoking surveys among physicians have proved useful in highlighting the importance of physicians as healthy life style exemplars and role models in tobacco control and smoking cessation. The aim of this study was to give an overview of smoking behaviour among Estonian physicians from 1982 to 2014. Methods Three cross-sectional postal surveys using a self-administered questionnaire were carried out among all practising physicians in Estonia. The number of physicians participating in this study was 3786 in 1982, 2735 in 2002, and 2902 in 2014. Data analysis involved calculating the age-standardized prevalences of smoking, prevalences of smoking by age group and mean age of smoking initiation. A non-parametric test for trend was used to assess significant changes in smoking over time. Results Age-standardized prevalence of current smoking among men was 39.7% in 1982, 20.9% in 2002, and 14.3% in 2014 and among women 12.2%, 8.0%, and 5.2%, respectively (p < 0.0001 for trends). From 1982 to 2014, the biggest decline of current smoking among men and women was in age groups under 35 (from 55.2% to 16.7% and from 16.7% to 2.8%, respectively) and 35–44 (from 47.1% to 8.3% and from 19.5% to 5.1%, respectively) (p < 0.0001 for trends). Mean age of smoking initiation decreased from 20.4 to 19.3 among men and from 24.5 to 20.4 among women over the study period. Conclusions In 1982–2014, smoking prevalence among Estonian physicians declined substantially. This may influence the willingness of society to recognize the health consequences of smoking which could give a support to the decline of the smoking epidemic in the country. Differences between smoking among male and female physicians persisted over the study period, but mean age of smoking initiation decreased. A further decline in smoking among Estonian physicians should be encouraged by special efforts targeted at physicians.


Background
The tobacco epidemic is one of the biggest preventable public health threats, killing around 6 million people a year [1]. Smoking cessation counselling by physicians has been effective in increasing cessation rates, but their own smoking habits are known to influence their effectiveness in this role [2,3]. Physicians are often viewed as healthy lifestyle models by their patients and the communities in which they live [2]. Early age of initiating smoking has an important role in becoming a smoker [4]. Moreover, the prevalence of smoking among physicians may reflect the 'maturity' of the smoking epidemic in a particular country. When the prevalence of smoking among physicians falls below that of general population, the country's smoking epidemic can be considered to be 'mature' [5].
Tobacco smoking by physicians in different countries has been the subject of epidemiological studies for a long time [6][7][8][9]. These studies have shown that smoking prevalence among physicians in high income countries such as Great Britain, the United States, Australia and Scandinavian countries, has decreased during the last decades. Also, smoking prevalence among the total population in developed countries has decreased during the last decades [10].
In Estonia, data concerning smoking among 16-64years-olds has been available since 1990 and shows a decreasing trend. While in 1990 the prevalence of daily smoking was 48% among men and 15% among women [11], by 2014 the rates were 31% and 15%, respectively [12]. In 1990-2014, the highest prevalence of smoking was in 1994 when 50% of men and 21% of women were daily smokers [13]. In Estonia, surveys concerning smoking among physicians have been carried out four times (1978,1982,2002,2014), but trends in smoking habits among Estonian physicians during the last three decades have not been reported.
The aim of this study was to analyse trends in smoking behaviour among Estonian physicians from 1982 to 2014.

Study population and design
The present study was based on three smoking surveys among Estonian physicians in 1982, 2002, and 2014. Unfortunately the data of the first survey in 1978 were not available.
The methods used in the surveys in 1982, 2002 and 2014 have been extensively described elsewhere [14][15][16][17][18][19]. In short, all practicing physicians who were permanent residents of Estonia were eligible for sampling in all study years. Cross-sectional postal surveys using similar, self-administered questionnaires were used to collect information on smoking in 1982, 2002, and 2014. In 1982, all practicing physicians were identified from the database of the Ministry of Health in Estonia. In 2002, all practicing physicians were identified from the database of Estonian Health Insurance Fund. In these two study years questionnaires were mailed to the workplace address of the physicians. In 2014, the sample of all practicing physicians was based on data from the Estonian Health Care Professionals Registry and questionnaires were mailed to the home address of physicians. To receive home addresses, data from the Estonian Health Care Professionals Registry were linked with the Population Registry in Estonia.
The number of respondents was 3792 in 1982, 2747 in 2002, and 2903 in 2014 ( Table 1). The crude response rates were 80.7%, 66.3%, and 52.0%, respectively. The corrected response rates (excluding the physicians who were unavailable, retired, had an incorrect address, had left Estonia or died) were available only for the last two study years, being 67.8% in 2002 and 53.1% in 2014.
Age of initiation of smoking among current smokers was measured in full years.
Smoking status was determined by combining answers to several questions concerning daily, occasional, past and never smoking and classified as current (daily and occasional) smoking and non-smoking (past and never smoking).

Statistical analysis
Data analysis involved the determination of prevalence estimates and the corresponding 95% confidence intervals (CI). Calculation of age-standardized prevalence of smoking using European standard population [20] was performed. Mean age of respondents with standard deviation (SD), minimum (min) and maximum (max) value, and mean age of smoking initiation with 95% CI were calculated. A non-parametric test for trend was used to assess significant changes in smoking over time [21].
In 1982 six physicians, in 2002 twelve, and in 2014 one physician did not answer the questions concerning smoking and were excluded from the analysis. A total of 3786 questionnaires from 1982, 2735 from 2002, and 2902 from 2014 were used in the analysis. Questionnaires with missing information concerning age of smoking initiation (1.8% of questionnaires over the study period) were excluded from the calculation of mean age of smoking initiation.
Age-standardized prevalence of past smoking among men increased from 24

Smoking by age groups
In 1982-2014, the biggest decrease of current smoking among men and women was in those aged −34 (55.2% and 16.7% for men, 16.7% and 2.8% for women, repectively) and in age group 35-44 (47.1% and 8.3% for men, 19.5% and 5.1% for women, respectively) (p < 0.0001 for trends) (Fig. 2). Current smoking decreased slightly but significantly among 45-54-and 55-64-year-old men (p = 0.0006 and p = 0.0064, respectively), but increased slightly among men in the oldest age group (65+) over the study period Among women, current smoking decreased in the 45-54 age group (p = 0.0089), was stable among 55-64year-olds (p = 0.7509), but increased slightly in the oldest age group (65+) (p = 0.1563) (Fig. 2). While in 1982 and 2002 the highest prevalence of current smoking among women was in the age group 35-44 (19.5% and 12.3%, respectively) followed by the age group −34   Table 3). Daily smoking among women was the highest among those aged 35-44 in 1982 (17.0%) and 2002 (7.2%), but among 55-64-year-olds in 2014 (8.0%). Daily smoking was the lowest among men and women in the oldest age group in all study years. In 2014, the lowest daily smoking prevalence was among 35-44-year-old men and in the youngest age group of women.
Prevalence of occasional smoking was less than 5.0% among men and women in all age groups in 1982 and 2014 (Table 3). In 2002, the prevalence of occasional smoking was somewhat higher being greatest among under 35-year-old men (12.9%) and 35-44-year-old women (5.2%). While in 1982 and 2002 the prevalence of past smoking among men was higher in older age groups, then in 2014 the highest prevalence of past smoking was among 45-54-year-olds. Prevalence of physicians who had never smoked was in the youngest age group of men 31.5% in 1982, 50.0% in 2002 and 79.2% in 2014 (among women 71.9%, 78.0% and 85.6%, respectively).

Discussion
The main research question of this study was to analyse trends in smoking behaviour among Estonian physicians.
Before discussing the results some limitations of this study should be addressed. First, response rates in the three physician surveys declined over time, as is the case with postal surveys in most populations worldwide. The non-responders may have revealed different patterns of smoking compared with those who returned the questionnaire. At the same time, the respondents were representative of the overall physician population by men-women ratio in Estonia, with slightly higher proportion of respondents among women. Second, self-reports of smoking tends to slightly underestimate the actual amount of smoking among physicians. As physicians know more about the devastating effects of smoking than the general population, they may be especially prone to self-deception or understatement. Third, as the data of the first study were collected in the Soviet era, and some minor changes were made in study design and questions between studies, the findings of trends in the study need to be interpreted with caution. Despite these caveats, several inferences can be drawn.

Trends in smoking
The findings highlight several trends in smoking behaviour in Estonia.
First, the age-standardized prevalence of current smoking among physicians declined by nearly three times from 1982 to 2014 in Estonia. In 1982, about one fifth, but in 2014, less than one tenth of physicians were current smokers. Most developed countries (e.g. Great Britain, Australia, United States, Finland, and Japan) appear to have experienced a similar steady decline in smoking prevalence among physicians over the past decades [6-9, 16, 23, 24].
Second, despite of overall decline in smoking, the prevalence of current smoking among physicians in Estonia was still higher than in most developed countries. For example, the prevalence of smoking among physicians in Estonia in 2014 was comparable with smoking among physicians in Australia and the United States in the 1980s [7,25]. However, smoking prevalence in some developed countries (e.g Japan, Italy, and France) have remained high with prevalence of current smoking over 25% [8,26,27].
Third, the age-standardized prevalence of current smoking in Estonia declined among male as well as among female physicians over the study period. In 1982,   than two-fold among women over the study period. In 1982-2014, the biggest decline in current smoking was among male and female physicians in younger age groups (−34, 35-44) in Estonia. The same trend was seen in age-standardized prevalence of daily smoking among physicians in Estonia. While in 1982 more than one third of male and one tenth of female physicians were daily smokers, by 2014 the age-standardized prevalence was more than three times lower among male and nearly three times lower among female physicians (11.7% and 3.9%, respectively). The age-standardized prevalence of past smoking among men increased 1.2fold in 1982-2002 and thereafter slightly decreased, but among women increased 1.5-fold over study period. The age-standardized prevalence of never having smoked increased 1.5 times among men, but was stable among women physicians in 1982-2014. At the same time, the biggest increase in the prevalence of never having smoked was among male and female physicians in younger age groups which means that non-smokers entering to the profession are also a source of change in smoking prevalence among physicians. Thus, the decline in smoking among physicians in Estonia was the result of an increase in quitting behaviour and possibly in the increasing number of nonsmokers becoming physicians. Similarly, the decline in smoking among physicians observed in the United States has been the result of an increase of both quitting behaviour and in the number of nonsmokers entering the occupation [25].
Fourth, compared to the first study year, Estonian physicians, especially women, started smoking at a younger age in the last study year. It is well established, that early age of starting smoking is one of the determinant factors of nicotine dependence and smoking temptation [4,30]. One reason why earlier onset did not appear to influence the declining trend in smoking among physicians in Estonia, could be explained by the increasing number of nonsmokers entering the occupation. Based on the results of Estonian Health Interview Surveys in 1996 and 2006, the trend in onset of smoking among adult population seems to be the same among the adult population in Estonia [28,29]. One explanation of the earlier start of smoking could be the heavy marketing by tobacco companies flooding Estonia after the collapse of the Soviet Union in 1991.
Fifth, almost all currently smoking male and female physicians in Estonia were daily smokers as agestandardized prevalence of occasional smoking varied from one to 3 % in 1982 and 2014 being somewhat higher in 2002 (5.9% among men and 3.3% among women). Conversely, only one third of current physician smokers in Finland were daily smokers [16]. The high percentage of daily smoking among current smokers in Estonia could be explained by the decreasing age of onset of smoking over the study period. According to the worldwide literature there is much evidence that those who start smoking earlier are more prone to become heavy smokers than those who start later [4,30].
Sixth, despite the overall decline in smoking, the difference between men and women persisted in Estonia being about three time higher among men than women.
Smoking among the general population is similar with much higher smoking prevalences among men in Central and Eastern Europe than in Western European and Scandinavian countries [10]. Smoking among physicians by gender differs by country. In Armenia [31] and Japan [32], smoking is more common among male physicians, while in Italy [33], female physicians smoke at higher rates, but in Australia [6,7] and the United States [6,9], smoking prevalence is almost the same between the sexes.
Seventh, the prevalence of smoking among physicians was much lower than that observed in the general population and in the highest educational bracket of the total population in Estonia. These findings are consistent with previous studies [16,17] indicating that physicans are far less likely to smoke than the general population in Estonia. Thus, smoking among physicians in Estonia as in most developed countries [7,23,25], is comparable with the "mature "smoking epidemic [5]. Nevertheless, in Italy the overall smoking prevalence among Italian physicians was similar to that of the general population in 1995. While male physicians in Italy smoked less frequently than the male general population (25% vs 34%), female physicians smoked more frequently than the female general population (23% vs 17%) [26].

Tobacco control in Estonia
Tobacco legislation and control in Estonia have become increasingly powerful.
In Estonia, there was no tobacco legislation in the 1980s during the Soviet era and the first years after the collapse of the Soviet Union in 1991, when tobacco policies were based on guidelines from the Ministry of Health. Since 2000, Estonia has witnessed considerable changes in tobacco control. The first Tobacco Act in Estonia was enforced in 2001 [34]. The revised Tobacco Act came into force in June 2005 and brought Estonia in line with the European Union directive on tobacco and the WHO anti-tobacco convention [35]. This law banned smoking completely in catering establishments. In 2014, the Green Paper on Tobacco Policy was published in Estonia to give an overview of necessary measures to be implemented to reduce the prevalence of smoking and tobacco-related health effects [36].
Since 2000, Estonia has belonged to the International Network of Health Promoting Hospitals and Health Services (HPH) [37]. A key strategy of HPHs was creating a smoke free environment in hospitals. In 2015, about 20 hospitals (around 30%) in Estonia joined HPH [38]. In 2005, the Estonian Network of Health Promoting Hospitals and Health Services (Estonian HPH Network) joined the European Network of Smoke-Free Hospitals (ENSH-Global). Since 2015, six leading hospitals in Estonia have become smoke-free [39].

Conclusion
In conclusion, this study is the first to describe smoking among Estonian physicians over three decades. Overall, it can be clearly seen that from 1982 to 2014 smoking declined among physicians in Estonia. At the same time, physicians started to smoke at an earlier age from 1982 to 2014. The difference in smoking prevalence of male and female physicians persisted over the study period.
It is important that smoking trends among physicians continue to decline so that all physicians can become good role models for the general population. A further decline in smoking among Estonian physicians should be supported by special efforts targeted at physicians.