Study design

The data were drawn from Estonian physicians’ cross-sectional smoking surveys conducted in 2002 and 2014. The surveys were carried out as postal questionnaires initially involving all practicing physicians in Estonia (n=4140 in 2002, and n=5666 in 2014; crude response rates were 66.3% and 51.9%, while corrected response rates were 67.8% and 53.1%, respectively). The methods have been described thoroughly elsewhere^17,19.

The current study sample was restricted to under 65-years-old ‘current smoker’ physicians, who gave answers regarding desire to quit smoking in 2002 (n=322) and in 2014 (n=189).

The surveys in 2002 and 2014 were approved by the Research Ethics Committee of the University of Tartu (Decisions No. 87/1 and 235/T–12, respectively). An informed consent form including a description of the study design and how the collected data would be used was sent to the recipients with the questionnaires. The form explained that participation in the study constituted consent, thus additional written consent was not obtained.

Variables

The main outcome indicator was desire to quit smoking (yes, no, cannot say). Factors associated with desire to quit smoking were: current smoking frequency, factors related to smoking behaviour, background characteristics, and study year. Smoking frequency was based on answers to several questions and classified as: 1) daily smokers (those who currently smoke every day), and 2) occasional smokers (those who currently smoke but not every day).

The following questions were used regarding factors related to smoking behaviour: concern (not concerned, concerned) about harms of smoking; number (none, 1–2, 3–4, ≥5) of previous quit attempts; main reason (stress, strong withdrawal symptoms, alcohol consumption, weight gain, smoking in proximity, other) to restart smoking ; and motivation (personal health problems, material stimulus, wish to set a good example, other) to quit smoking. To determine reason to restart smoking and motivator to quit smoking, the respondents were required to check one answer from the list provided.

Background characteristics used in this study were the following: gender; age group in years (≤34, 35–44, 45–54, 55–64); ethnicity (Estonian, non-Estonian); place of residence (Tallinn, other city, rural); medical specialty (family physicians, specialist doctors, dentists); study year (2002, 2014).

Data analysis

Distributions of physicians by desire to quit smoking, smoking frequency, factors related to smoking behaviour and background characteristics were calculated along with the mean age of respondents and standard deviation. To determine association between desire to quit smoking and factors related to smoking behaviour, multiple logistic regression analysis was used. Dichotomized desire to quit smoking (yes; no: no and cannot say) was used as dependent variable and all other factors as explanatory variables in logistic regression models. Logistic regression analysis was performed with two models, Model 1 (adjusted for gender, age and year) and Model 2 (adjusted for all explanatory variables). Odds ratios (OR) and 95% confidence intervals (CI) were calculated.

Questionnaires with lacking smoking status values (n=11) and data concerning desire to quit smoking (n=12) were excluded from the analysis. In total, 511 questionnaires were included in the final analysis. Questionnaires without information concerning factors related to smoking behaviour and background characteristics were excluded from the logistic regression models.

Characteristics of the sample

Number of respondents was 322 in 2002 and 189 in 2014. Age range of physicians was 27–64 years in 2002 and 25–64 years in 2014, with mean age 45.4 ± 9.3 and 49.7 ± 10.4, respectively. In both study years, more than two-thirds of smoking respondents were females (67.1% in 2002, and 65.1% in 2014).

Overview of background characteristics is provided in Table 1. In 2002, the highest proportion of respondents was in age group 35–44 (36.0%), while in 2014 it was in age group 55–64 (40.2%). Most of the respondents were of Estonian ethnicity (82.9% in 2002, and 78.3% in 2014) and nearly half of respondents (43.8% in 2002, and 43.4% in 2014) were residents of other cities (excluding Tallinn). More than half of respondents were specialist doctors (58.4% in 2002, and 56.6% in 2014).

The prevalence of desire to quit smoking was 55.3% in 2002 and 52.9% in 2014 (Table 2). In 2002, almost a third (30.1%) remained undecided whereas in 2014 this proportion was 22.2%. Among ‘current smokers’, the majority were daily smokers in both study years (64.3% in 2002, and 77.8% in 2014). Majority of physicians (77.3% in 2002, and 79.4% in 2014) were concerned about the harms of smoking (Table 2). In 2002 the percentage of physicians who had 3–4 previous smoking quit attempts was 21.4%, while in 2014 it was 16.4%. Among reasons to restart smoking, stress was most prevalent in both study years (30.7% in 2002, and 35.4% in 2014), followed by smoking in proximity and weight gain. Two-thirds of physicians (63.0% in 2002, and 63.5% in 2014) stated that personal health problems could motivate them to quit smoking. Wish to set a good example was the second most prevalent motivator for quitting.

Association between desire to quit smoking and factors related to smoking

According to the results of the fully adjusted logistic regression model, desire to quit smoking was significantly associated with being concerned about the harms of smoking, number of previous quit attempts, wish to set a good example, and medical specialty (Table 3). After full adjustment, daily smoking appeared not to be associated with the higher desire to quit smoking but the wish to set a good example was.

Compared to the physicians who were not concerned about harms of smoking, odds for desire to quit smoking were significantly higher among physicians being concerned about harms of smoking (OR=9.06; 95% CI: 4.15–19.74). Compared to physicians who had not tried to quit smoking, odds for desire to quit smoking were significantly higher among physicians who had 1–2, 3–4, and ≥5 attempts to give up smoking. Compared to physicians who agreed with the statement that health problems are the best motivators to quit smoking, odds for desire to stop smoking were significantly higher among physicians who agreed that setting a good example was the main reason to quit smoking (OR=2.38; 95% CI: 1.12–5.09). Compared to specialist doctors, desire to stop smoking was significantly higher among dentists (OR=2.42; 95% CI: 1.25–4.69).

No significant associations were found with any factors listed under main reason to restart smoking, with almost all factors listed as motivators to quit smoking (except wish to set a good example), with sociodemographic factors (gender, age, ethnicity, and place of residency), and with study year.

Prevalence of desire to quit smoking and factors related to smoking behaviour

According to the results of the study, more than half of smoking physicians indicated a desire to quit smoking in 2002 and 2014. These proportions are comparable, for example, to those of Poland in 2004 where about 60% of physicians wished to quit, but lower than those of the Czech Republic in 1995 where about 75% of smokers intended to quit²⁰. Although the proportion of physicians who had a desire to quit smoking was similar in both years among ‘current smoker’ physicians in Estonia, the prevalence of daily smokers and the proportion of those who did not want to quit smoking was higher in 2014 than in 2002. The higher prevalence of daily smokers might indicate smokers with a stronger nicotine dependence who were not determined to quit.

Studies concerning quitting smoking usually measure intention to quit, not desire to quit. However, both reflect the general attitude towards quitting⁸. Desire to quit has been stated as an important factor in smoking cessation as it reflects motivation but, on the other hand, is different from actually intending to quit and has been merely seen to predict quit attempts and not the success of cessation^7,21. The questionnaire used in this study explored only desire to quit smoking because the survey focus was on smoking behaviour in general.

In this paper, stress, smoking in proximity and withdrawal symptoms were reported to be the most important reasons for restarting smoking. These findings are in accordance with previous results²². Smoking in proximity was the second most mentioned reason but the prevalence of this declined notably from 2002 to 2014. The decline might be due to smoking decline in general but also because in Estonia smoking in health care establishments is prohibited. The current Tobacco Act, enforced in 2005 and revised several times, is in accordance with EU and WHO tobacco policies and places extensive restrictions to smoking in public and in establishments like schools, hospitals and restaurants.

In our study, personal health problems and wish to set a good example were the most prevalent motivators for quitting. Previous findings also suggested that concern for personal health and wishing to set a good example are the most important reasons among medical staff and the general population^23,24.

Previous studies have also shown a connection between nicotine dependence and smoking cessation^22,23,25. In the Estonian physicians’ smoking surveys, nicotine dependence was measured only in 2014, and so it was not possible to include these data in the present study.

Associations between desire to quit smoking and factors related to smoking behaviour

Desire to quit smoking was found to be significantly associated with being concerned about harms of smoking, number of previous quit attempts, and wish to set a good example. These findings are in accordance with previous studies^10,25. Attempts to quit have been shown to be associated with intention to quit and are considered an important predictor of cessation success^12,25. In the present study, compared to respondents who had no quit attempts, physicians who had 1–2, 3–4 or ≥5 quit attempts, had higher odds for desire to quit. Similarly, to previous findings, Estonian physicians who were concerned about the harms of smoking were more likely to desire to quit^10,25. These associations could indicate that a person is aware of the damaging effects of smoking, is motivated to quit, has tried but failed and so is in need of cessation support.

None of the reasons for relapse was significantly associated with desire to quit smoking in this study. In previous studies, alcohol consumption has been found to be associated with intention to quit smoking. Smokers who had higher number of drinks per week were less likely to intend to quit smoking¹⁰. The questionnaire that was used to collect data for this study did not measure alcohol consumption.

In motivators for quitting, only wish to set a good example was significantly associated with desire to quit smoking among physicians in Estonia. As doctors are considered role models, it was expected that setting a good example would be a significant finding among Estonian physicians, as well. Material stimulus was not significantly associated with desire to quit in our study. This was somewhat expected as physicians can be considered as having higher socioeconomic status. However, in contrast, among the general population, material incentives have been described as important factors in quitting smoking, thus increasing the prices of tobacco products should be prioritized when designing tobacco control programs^26,27.

The association between desire to quit smoking and medical specialty showed that compared to specialist doctors, dentists had significantly higher odds for desire to quit smoking. The association was not significant for family physicians. While this last aspect was unexpected, the finding for dentists was predictable as they are in close contact with the patient and therefore aware of the devastating effects of smoking on oral health. The lack of a notable difference between family physicians and medical specialists could be explained by the distribution of specialties among physicians — only about 10% of medical specialists were surgeons in this study. It has been shown elsewhere that, among male physicians, surgeons had the highest prevalence of smoking compared to other specialities²⁸.

No association was found between desire to quit smoking, gender and age among physicians in Estonia. There is little information about whether gender is associated with desire and intent to quit smoking worldwide but the findings concerning successful cessation tend to show that women are less likely to stay abstinent^23,25,29,30. It is possible that among women who desire to quit, less have serious health conditions that would otherwise motivate them to abstain from relapses. However, as men tend to be less active in seeking medical help, their health problems could be more serious and so motivating them to remain abstinent. Older age has also been shown to be a predictor of successful cessation, with increased cessation rates observed among those who were aged 45 years or older²³. However, intention to quit has been shown to decrease with increasing age²⁵. These findings might reflect a relationship between health issues and quitting smoking, since it has been shown that smoking physicians tend to believe that many people have smoked all their lives to an old age and did not become ill, hence smoking is not as dangerous as experts declare¹⁸.

In this study, no association was found between desire to quit smoking and study year. This finding was unexpected since between 2002 and 2014 several major legislative changes occurred in terms of tobacco policy in Estonia. Also, most of the major hospitals joined the network for tobacco-free health services³¹ and now promote smoking cessation among their staff. However, smoking in health care establishments has been prohibited by the Tobacco Act in force from 2005.

Limitations and strengths

First, this study relied on self-reported data, thus self-representation bias should be considered since smokers tend to underestimate the amount smoked³². Second, the response rates of 67.8% in 2002 and 53.1% in 2014 may indicate underrepresentation of smokers. Third, as smoking prevalence has been declining, the sub-samples of smoking physicians were small. This may have led to limited statistical power in detecting differences. Despite these limitations, Estonian physicians’ smoking surveys provide excellent opportunity to analyse smoking behaviour as physicians can be considered as a homogenous sample and the survey methods were similar in both years.