Use of tobacco and alcohol by Swiss primary care physicians: a cross-sectional survey

A postal survey was conducted during the spring of 2002 among 2'756 Swiss primary care physicians identified through the membership database of the professional organisation of Swiss physicians (Federatio Medicorum Helveticorum) [17]. First, all physicians certified as general practitioners, general internists, paediatricians and physicians without a specialty qualification were identified (7'711 members). Second, a random, non stratified sample of 3'000 primary care practitioners was selected. After exclusion of 3 deceased doctors, 8 doctors with incorrect addresses, 15 doctors who did not practice clinical medicine and 218 doctors who did not practice as primary care doctors, 2'756 (91.9%) physicians remained eligible for the survey, representing 36% of all Swiss primary care physicians. The main objective of the survey was to measure the well-being of primary care physicians and its determinants. The 2'756 primary care physicians received a questionnaire with questions addressing subjective mental and physical health, burnout, medical care use, work-related satisfaction, socio-demographic and work-related characteristics, and smoking and drinking behaviours.

Two questions were used to measure smoking behaviour among physicians: "Do you smoke regularly (at least 1 cigarette per day)?" and "In the past, have you been a regular smoker (at least 1 cigarette per day for at least 6 months)?". Physicians were categorised as current smokers if they answered "yes" to the first question, former smokers if they answered "no" to the first question but "yes" to the second one, and never smokers if they answered "no" to both questions.

To measure alcohol use, we used the three items of the AUDIT-C score: "How often did you have a drink containing alcohol in the past year? (a "drink" to be a can or bottle of beer (2.5 dl), a glass of wine (1 dl), or one cocktail or a shot of hard liquor like scotch, gin, or vodka (0.25 dl)" (response options were never (0 points), monthly or less (1 point), 2 to 4 times a month (2 points), 2 to 3 times a week (3 points), 4 or more times a week (4 points)), "How many drinks did you have on a typical day when you were drinking in the past year?" (response optionswere 0 to 2 drinks (0 points), 3 to 4 drinks (1 point), 5 to 6 drinks (2 points), 7 to 9 drinks (3 points), 10 or more drinks (4 points)) and "How often did you have 6 or more drinks on one occasion in the past year?" (response options were never (0 points), less than monthly (1 point), monthly (2 points), weekly (3 points), daily or almost daily (4 points)). These questions have been validated for the screening of alcohol problems in men [18] and women [19]. The cut-offs of AUDIT-C score we chose in our study for the diagnosis of at-risk drinkers correspond to the optimal operating point in terms of sensitivity and specificity (≥ 5 points for men: sensitivity 91.2% and specificity 95.2%, and ≥ 4 points for women: sensitivity 81.4% and specificity 93.1%) [20, 21]. Based on the answers to these 3 items, the participating physicians were categorised as non drinkers (AUDIT-C score 0), drinkers (AUDIT-C score 1 to 3 for women and 1 to 4 for men), and at risk drinkers (AUDIT-C score ≥ 4 for women and ≥ 5 for men). Distinction between at-risk consumption and dependence could not be made, because the three items of the AUDIT-C reflect only alcohol consumption [18‐21]. Answers to the third question of the AUDIT-C were also used to categorise physicians as at risk drinkers (any mention of 6 drinks or more per occasion), because similar data was available for the general population (cf. below).

Socio-demographic (sex, age, living alone, economic role in the household) and work-related characteristics (medical specialty, type of practice, location of practice) were used as independent variables of smoking and drinking behaviours in an exploratory way.

The initial questionnaire was developed in French and pre-tested among a small group of physicians for readability and acceptability. When validated items in German or Italian were not available, three independent translations were made by bilingual physicians and professional translators and a final version obtained by consensus. The translated items were pre-tested before their use. The study was approved by the research ethics committee of the Institute of Social and Preventive Medicine at the University of Geneva.

Prevalence of smoking and drinking behaviours among primary care physicians were calculated using frequency tables. Associations with socio-demographic and work-related determinants were tested using cross-tabulations, χ² test and linear trend test where appropriate. Agreement concerning categorization of alcohol consumers by AUDIT-C score and the third question of AUDIT-C was tested using kappa statistics. Statistical analyses were performed with SPSS (Statistical Package For Social Sciences, version 11.0).

Overall, 12% were current smokers and 22% former smokers. 66% were drinkers and 30% at risk drinkers according to the AUDIT-C score, respectively 62% and 34% according to the first and third question of AUDIT-C (Table 1). Only 4% were never smokers and non drinkers. Never smokers were more often non drinkers (6%) than current smokers (3%) or former smokers (2%) and non drinkers were more often never smokers (81%) than drinkers (71%) or at risk drinkers (51%). Finally, current smokers and former smokers were more often at risk drinkers (39% and 44% respectively) compared to never smokers (23%), whereas at risk drinkers were more often current smokers (16%) compared to drinkers (11%) or non drinkers (8%).

Smoking was more frequent among male physicians, aged 45 to 55 years, practicing in the French-speaking part of Switzerland, living alone and being the principal source of outcome in the household (Table 2). Proportion of never smokers decreased with age. At risk drinking was more frequent among men, aged 45 to 65 years, of the French-speaking part of Switzerland, living alone and being the principal contributor to the household (Table 2). Proportion of never drinkers decreased with age, then increased after 65 years. Similar trends were observed for both the AUDIT-C score and the third question of AUDIT-C, excepted for differences between geographical regions of the country and living alone. The agreement between the two measures was satisfactory (Kappa 0.76, SE 0.02) (Table 1).

We found distinctive patterns of tobacco and alcohol consumption among primary care physicians. Primary care physicians were two to three times less likely to be current smokers when compared to the Swiss general population (12% vs. 30%).

The prevalence of smoking among physicians has decreased during the last decades in many places [8, 13, 28, 29]. In these countries the population appears to be well informed about the dangers of smoking and physicians seem to respond by giving up smoking more frequently than other individuals, resulting in this important drop in prevalence. In several European countries and in the USA the prevalence of smoking among physicians is now lower than 20% (Table 3). When the prevalence of smoking among doctors falls effectively below that of the general population, the country's smoking epidemic is considered by some experts as "mature" [1, 25, 29]. By contrast, the prevalence of smoking among physicians remains very high in several southern European countries and countries with transitional economies [23, 24, 26, 30].

A reverse pattern was found for alcohol consumption. At the time of the survey Swiss primary care physicians drank more than the general population and only 4% were abstinent (vs. 22% in the general population). In addition, those who drink consume more: 30% of the primary care physicians were classified as at risk drinkers vs. 15% for the general population.

In spite of differences in the definition used to define at risk drinker, studies conducted in other countries (Table 4) have also found high prevalence of alcohol consumption among physicians [7, 13‐16] Some authors suggest that the higher prevalence of alcohol use among physicians could be explained by their higher socio-economic status [13, 16]. Others advocate that medical institutions, medical schools and hospitals lack efficient health promotion programs to prevent alcohol use and abuse among future and current physicians [31]. Finally, the higher use of alcohol among physicians could be directly related to the commonly accepted evidence showing that drinking a small quantity of alcohol can protect against cardiovascular diseases and is therefore not a risky behaviour for health [11, 12]. Hence the results of our study point out clearly that primary care physicians not only drink more frequently, but drink more alcohol when they drink, which increases the risk of future alcohol abuse and dependence. However the different studies conducted over the last 20 years among American physicians have shown an increase of abstinence, indicating that the perception of the hazards of alcohol consumption has possibly changed among physicians in this country.

Several limitations of the current study must be kept in mind when considering the results. Firstly, the comparability with the 2002 Swiss Health Survey and international studies listed in Tables 3 and 4 may be altered for several reasons. The comparison between primary care physicians and the Swiss general population may be biased, because data had not been collected in the same manner during both surveys. Self-administered questionnaires identified by an anonymous number were used for physicians, whereas personal telephone interviews were performed for the other survey in the general population. Both methods raise the issue of underreporting and measurement error. For the physicians' survey, we believe this was unlikely, because the main aim of the study was to assess physicians' subjective health and well-being and not desirable lifestyle behaviors. In addition the questions used, especially for alcohol, were not strictly identical which raises the issue of measurement bias. In the 2002 Swiss Health Survey the questions used for smoking status were "Do you smoke?" and "Have you been a regular smoker for at least 6 months?" and for alcohol drinking "What is your usual alcohol consumption ?", "Do you drink several times during the day?" and "During the past year, how many times did you have 8 drinks (for men)/6 drinks (for women) of beer, wine, or other alcoholic beverage ?". Compared to the items we used (cf Methors section), we believe that the questions were similar enough to limit this type of error. However, several other differences may also alter the comparability between these studies, mainly the medical specialty (see Tables 3 and 4), and sexe and age variation across studies. Secondly, due to the cross-sectional nature of the survey, we cannot draw definitive causal conclusions about the observed relationships between socio-demographic characteristics and current smoking and at risk drinking. Finally, as we could not differentiate between at risk drinkers and dependant drinkers, one could argue that our results are not relevant and subject to classification bias. However, several authors consider that it is essential in term of public health to consider drinkers and at-risk drinkers, because at this stage most consequences of alcohol use are still reversible [18‐21].

Our study included a large sample of primary care physicians and the global participation rate (65%), corresponding to 23% of the Swiss practitioners, can be considered as excellent for a postal questionnaire consisting of 16 pages. This response rate compares favourably with other similar studies performed among healthcare professionals [13, 15]. Participation rate was particularly high for general practitioners, probably reflecting their high interest for this survey on their own health and well-being. Finally, the sample was also sufficiently large to allow exploration of even weak associations between variables.