The incidence of smoking and risk factors for smoking initiation in medical faculty students: cohort study
© Senol et al. 2006
Received: 13 December 2005
Accepted: 10 May 2006
Published: 10 May 2006
Skip to main content
© Senol et al. 2006
Received: 13 December 2005
Accepted: 10 May 2006
Published: 10 May 2006
Medical education requires detailed investigation because it is a period during which the attitudes and behaviors of physicians develop. The purpose of this study was to calculate the yearly smoking prevalence and incidence rates of medical faculty students and to identify the risk factors for adopting smoking behaviour.
This is a cohort study in which every student was asked about their smoking habits at the time of first registration to the medical faculty, and was monitored every year. Smoking prevalence, yearly incidence of initiation of smoking and average years of smoking were calculated in analysis.
At the time of registration, 21.8% of the students smoked. At the end of six years, males had smoked for an average of 2.6 ± 3.0 years and females for 1.0 ± 1.8 years (p < 0.05). Of the 93 medical students who were not smokers at the time of registration, 30 (32.3%) were smokers at the end of the 6 years of the course.
The first 3 years of medical education are the most risky period for initiation of smoking. We found that factors such as being male, having a smoking friend in the same environment and having a high trait anxiety score were related to the initiation of smoking. Targeted smoking training should be mandatory for students in the Medical Faculty.
Smoking is a major public health problem all around the world, especially in developing countries. It is estimated that by 2025, 75% of early deaths in developing countries will be due to smoking-related illnesses . One third of the global population aged 15 years and over (approximately 1.1 billion people) are current smokers. Most of them live in developing countries (800 million) and most are males (700 million) . In Turkey, 44% of adults are smokers (60–65% in males, 20–24% in females) . Among European countries, Turkey is second only to Greece in cigarette consumption per person. The percentage of smokers is steadily increasing in Turkey, in contrast to developed countries .
Physicians have an important role in the struggle against smoking. Besides dealing with smoking-related health problems, they are also expected to represent a role model for the population by adopting a non-smoking lifestyle . Their dedicated opposition to smoking may have a huge effect on society [4, 5]. However, current evidence shows that most physicians in Turkey smoke [6, 7] and would probably not influence the population positively in this respect. The time during medical education seems to be the ideal period to establish such a positive influence. In order to assess the importance of this period, many investigators have studied the smoking behavior of medical students [8–12]. However, most of these studies have been cross-sectional and do not suffice to elucidate the adoption of smoking behaviour. In our opinion, it is important to define the characteristics of students who started to smoke during medical education in order to identify high-risk groups and to plan effective preventative strategies. A subsequent prospective follow-up study, allowing smoking incidence to be calculated for each year of medical education, would be useful in determining the riskiest period for initiation of smoking.
The purpose of this study was to investigate the risk factors for initiation of smoking among medical students and to determine the annual prevalence and incidence rates of smoking.
This cohort study was carried out among students of Akdeniz University Faculty of Medicine, established 30 years ago.
Students smoking ≤ 1 cigarettes per day were considered regular smokers . Any student who had had ≤ 1 alcoholic drink per week during the past year was considered an alcohol taker, while those who had not, or had drunk only on special occasions over the same period, were classified as alcohol non-takers. The students were asked to rate their nutritional status as sufficient if they had breakfast regularly and consumed vegetables, fruit and meat adequately. Students who took part in physical exercise for more than 20 mins at least twice per week were accepted as exercisers. The participants were also asked to complete the General Health Questionnaire-12 (GHQ), Spielberger State-Trait Anxiety Inventory (STAI) 1–2 and Beck Depression Inventory (BDI) to determine their mental health status [14–16].
The 12- item GHQ was chosen because of its well-established validity in student samples . Each item in the GHQ has a standard 0-0-1-1 scoring method (maximum score of 12). Depression was assessed using the BDI, a 21-item self-reporting questionnaire in which each item is scored zero to three . Higher scores indicate a higher level of depression. The total STAI score was obtained to determine anxiety. The STAI has a 20-item form for state anxiety and a 20-item form for trait anxiety and is scored from one to four . Each form allows a minimum score of 20 and a maximum score of 80.
Smoking prevalence, yearly incidence of initiation of smoking and average years of smoking were calculated.
Incidence was defined as the number of new cases occurring during a given time interval divided by the population at risk at the beginning of that interval . In accordance with this definition, the incidence was calculated as follows:
We calculated yearly incidence rates (for each year) as well as total incidence (for the entire 6 years). Any student included in the calculation of incidence of initiation in a particular year was excluded from calculations of incidence in the following years.
Prevalence is defined as the number of individuals with a given behaviour at a given moment divided by the population at risk at that moment .
In addition, Total prevalence was the number of students who smoked at least once during the 6 years of medical education divided by the total student population.
In order to calculate this variable we added the smoking duration of each student and divided this number by the whole study population. Years of smoking before the entering the medical faculty were included in this calculation.
In order to investigate the relationship between mental health status and smoking behavior, the six-year mean scores on the three different scales were compared between those who started smoking and those of non-smokers.
Statistical analyses of the variables related to initation of smoking involved only the 93 students who were non-smokers at initial registration. Chi-square tests, Fisher's exact chi-square tests and logistic regression analyses were used for statistical evaluations. To determine the relative importance of each predictor variable on the initiation of smoking, multivariate analysis was performed by logistic regression. The Forward Conditional model was used. Mental Health Scales were entered in the model as "continuous" variables. Gender, family location, smokers among friends, smoking family members, tea consumption, alcohol use, coffee consumption, boyfriend/girlfriend, sports and nutrition were entered in the model as "categorical" variables. In addition, a replacement-of-missing-value operation was undertaken for the 7 students whose data showed deficits in some years, without affecting the data distribution or mean values. The SPSS "replace" function was used to replace the missing values by the means of all values. The order of the subjects was randomized. The replacement operation was performed only for regression analysis. Statistical significance was accepted when p ≤ 0.05.
Of the total 119 medical students included in the study, 78 (65.5%) were male and 41 (34.5%) were female. A total of 26 (21.8%) students were already smokers at the time of registration.
At the end of six years, the smoking duration in males (2.6 ± 3.0 years) was significantly higher than that in females (1.0 ± 1.8 years) (p < 0.05), but the age at which smoking began showed no significant difference between the genders (for males 17.0 ± 2.6 years; for females 17.7 ± 1.7 years).
Annual smoking frequency of students and percentage of new starters.
% (%95 CI)
% (%95 CI)
Variables affecting initiation of smoking (new smokers) during 6 years of medical education.
Other smokers among friends
Smoking family members
Mental health scales of students who started or did not start smoking during the 6 years of medical education.
Mental State Scale
Started smoking (n = 30) Mean ± SD
Not started smoking (n = 63) Mean ± SD
41.9 ± 7.1
38.8 ± 8.2
45.7 ± 6.0
42.2 ± 7.3
9.8 ± 5.6
7.5 ± 5.7
4.6 ± 2.2
4.2 ± 2.4
Odds ratios (OR) for the factors affecting initiation of smoking among medical students.
Other smokers among friends
Trait anxiety score
The present study had both advantages and limitations. As a prospective study, it allowed us to calculate the incidence of smoking for each single year and thus enabled us to determine the most risky period for initiation during medical education, and the best period to implement available anti-smoking strategies. On the other hand, the exclusion of several students decreased the research population, and missing data for some years were major limitations. Fortunately, the number of students omitted from the study was relatively low (10) and their characteristics were similar to the general distribution (Figure 1). Although we included all students registered, the sample size may still not be large enough to determine the effects of all independent variables on initiation of smoking.
Tobacco use is frequent among medical students. Although the annual smoking prevalence rates found in our study were between 21.8% and 33.6% (Table 1), similar to those reported in previous studies [12, 19–21], the total prevalence rate during medical education (47.1%) seems higher, possibly because of the methodological difference between total and cross-sectional prevalence calculations. The smoking rate was 22.5% in the Comenius University of Slovakia , 29.0% in the Zagreb Medical Faculty  and 31.7% in seven British Medical Faculties . It was reported to be 23.6% for the first year and 27.3% for the fifth year for medical students in Australia . Similar rates for students in the first (26.7%) and fifth (24.4%) classes were noted at Columbia Medical Faculty . As these studies were all cross-sectional, they reflected only the rates for the year of the study. It is not possible to ascertain the exact rates for new and/or existing smokers, since some people may stop smoking. If primary prevention against initiation of smoking is taken into consideration, then it is obvious that everyone who has ever tried smoking must be considered within the risk group. Therefore, cross-sectional calculations seem to have some limitations in reflecting the actual situation.
If the present study had been designed cross-sectionally, we would have erroneously concluded that the risk was highest in the fourth year of medical school (Table 1). Actually, the highest incidence of initiation occurred in the first year, perhaps because of adaptation to a new educational atmosphere, difficulties in relationships with new friends and the intense anxiety experienced during the first year. Previous studies have shown that university life increases the stress levels of students [23, 24]. Our findings, like those of previous studies [12, 25, 26], suggest that medical education does not prevent or reduce smoking; moreover, tobacco use increases over time.
The factors affecting cigarette consumption were defined in detail in previous studies [12, 26–28]. We found that factors such as being male, presence of a smoking friend in shared accomodation (e.g. home or hostel), insufficient nutrition and having a high trait anxiety score (Tables 2, 3, 4) were significantly associated with initiation of smoking. In contrast to others [27–31], however, we found that a smoker in the family, alcohol intake and dating had no significant effects on the adoption of smoking behaviour. The smoking-family relationship emphasized in several studies was probably based on family-children interactions in the pre-university period. We excluded the smoking students at initial registration, so our findings did not reflect the effects of interactions with family prior to medical education. It is well known that friends have more influence than family members on initiation of smoking during education .
Smoking frequency was higher among males than females in our study. However, there was no significant difference between males and females regarding the age of onset of smoking. Also, boys showed a 2.88 times greater risk of initiation of smoking than girls. Traditional and/or cultural norms may play a role in the low starting rates observed in females .
In agreement with the finding that people eating a balanced diet with vegetables/fruits and having regular breakfasts are less likely to smoke, while those with an irregular eating routine are more likely , we found a significant relationship between nutrition and initiation of smoking. This relationship may have been affected by a third variable that we did not take into consideration.
Another major finding of the present study was the relationship between initiation of smoking and the trait anxiety score, which reflected the mental state of the student (Tables 3 and 4). Loss of self-confidence, a socially withdrawn character, a depressive personality and anxiety are frequent among smokers [34–36]. In addition, some studies have reported that smoking is perceived among young people as beneficial for improving social relations and coping with anxiety, and this perception may well encourage them to consume cigarettes . It has also been suggested that mental health state and smoking may have a common genetic predisposition . Despite current evidence for a relationship between mental health state and smoking, the nature of this interaction remains unexplained . We found that the risk of initiation of smoking increased with the trait anxiety score. Given that medical education causes anxiety, which negatively affects the mental health of students, and that the mental health scores of students increase dramatically during the first year of this education period , medical education may possibly have an indirect negative effect on smoking.
Our findings suggest that any preventative anti-smoking measures must be implemented in the first year of medical education with more focus on males, since male students with high trait anxiety scores and with at least one smoking friend are at high risk of starting smoking. Both social and psychological support starting in the first class of medical education would be effective in the struggle against smoking. Precautions such as increasing cigarette prices and banning tobacco advertising, and supporting activities such as peer education on smoking, as well as offering relevant courses in the first year curriculum at medical school, may be effective for young people at high risk of adopting smoking behaviour. However, it would not be adequate to implement such programs only for faculty years. Enlargement of both scope and area of implementation would be needed to increase the success rate. Low smoking rates in both the general and student populations in countries with rigid preventative methods [3, 38] prove that effective outcomes may be obtained with the aforementioned interventions.
However, it should be kept in mind that childhood through adolescence is a very important age for development of attitues towards smoking, so preventative measures should be implemented as early as elementary school years. Education in medical faculties only partially influences attitudes. In medical education, greater attention should be paid to the treatment of smoking dependence and ways of preventing smoking in order to prepare the students, as future health prefessionals, for their role in prevention.
As a in conclusion, the first 3 years of medical education are the most risky period for initiation of smoking. Male gender, presence of a smoking friend in shared accomodation, insufficient nutrition and having a high trait anxiety score were the significant factors on smoking initiation in our study. Interventions including both social and psychological support starting in the early years of medical education are needed.
The authors of the study would like to thank to Dr. M. Kemal Alimoglu and Kerry Bingham for their constructive comments and excellent help in preparing the manuscript in English.
This study was supported by Akdeniz University Scientific Research Products Management Unit.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.