Marital status and cardiovascular risk in French ... - Wiley Online Library

Journal of Internal Medicine 2001; 249: 315±323

Marital status and cardiovascular risk in French and Swedish automotive industry workers ± cross sectional results from the Renault±Volvo Coeur study L . K U M L I N 1 , 3 , G . L A T S C H A 2 , K . O R T H - G O M EÂ R 3 , L . D I M B E R G 4 , C . L A N O I S E L EÂ E 2 , A. SIMON5 & B. ERIKSSON6 AND THE COEUR STUDY GROUP*

From the 1Health Department, Volvo Aero Corporation, TrollhaÈttan, Sweden; 2Service Medical, Renault Automobiles, Billancourt, France; 3 Department of Preventive Medicine, Karolinska Institute, Stockholm, Sweden; 4The World Bank, Washington DC, USA; 5 Broussais Hospital, Paris, France; and 6Nordic School of Public Health, Gothenburg Sweden

Abstract. Kumlin L, Latscha G, Orth-GomeÂr K, Dimberg L, LanoiseleÂe C, Simon A, Eriksson B, the Coeur Study Group (Health Department, Volvo Aero Corporation, TrollhaÈttan, Sweden; Service Medical, Renault Automobiles, Billancourt, France; Department of Preventive Medicine, Karolinska Institute, Stockholm, Sweden; The World Bank, Washington DC, USA; Broussais Hospital, Paris, France; and Nordic School of Public Health, Gothenburg Sweden). Marital status and cardiovascular risk in French and Swedish automotive industry workers ± cross sectional results from the Renault±Volvo± Coeur study (Original article). J Intern Med 2001; 249: 315±323. Objectives. To compare the coronary risk pro®les in a sample of the French and Swedish automotive industry employees who were married/cohabitant, divorced or single (never married). Design. A cross-sectional study comparison from biological and questionnaire data between the French and Swedish samples. Setting. Occupational health departments at Renault (employees from the north-west of France) and Volvo (employees from the south-west of Sweden). Subjects. Two random samples of males aged between 45 and 50 years were examined in 1993, from Renault 1000, and from Volvo 1000. Main outcome measures. Biological data including cholesterol, blood pressure as well as the Framingham risk index. Self reported information regarding marital status, smoking, exercise, alcohol habits, and work stress assessed by the Karasek method,

private social support indices, and type A behaviour according to the Bortner scale. Results. More employees were married/cohabitant and fewer divorced or single at Renault. Apart from waist/hip ratio being marginally lower in Swedish single men, compared with married and divorced, no signi®cant difference in biological cardiac risk factors (total cholesterol, blood pressure or Framingham risk index) was seen between the subgroups from any of the two countries. Compared with married/cohabitant men, it was shown that in men living alone smoking was more prevalent at Renault and Volvo. These men also showed less type A behaviour, a lower work control and a lower work support and fewer close friends. Alcohol consumption was reported in smaller amounts for Volvo employees living alone compared with married or divorced employees. Married/cohabitant and divorced staff showed similar values regarding all measured variables when compared within each country. Conclusions. Employees living alone in both France (Renault) and Sweden (Volvo) automotive companies seem to have increased nontraditional cardiac risk factors pertaining to life style and social network compared with married or divorced men. These results, in combination with the ®nding that more Volvo than Renault employees were living alone, suggest a higher risk for coronary heart disease amongst Volvo employees. This hypothesis will be evaluated in the 5 and 10 years follow up study. Keywords: cardiovascular risk, French paradox, marital status, psychosocial factors.

* Members of the Coeur Project Group are listed in the Appendix. ã 2001 Blackwell Science Ltd

315

316

L . K U M L I N et al.

Introduction In the late 1950s, large geographical differences in mortality from coronary heart disease (CHD) amongst European countries were described particularly between southern Mediterranean countries like France and the Nordic countries Finland and Sweden [1]. These differences could not be explained by differences in standard risk factor levels including dietary habits and fat intake. In particular the high fat diet cholesterol ± CHD theory did not seem to be relevant in the French population where CHD rates were low in spite of a high fat diet. This is often referred to as the French paradox [2]. Other explanations for this paradox such as higher consumption of alcohol (wine) in France [3] were suggested. More recently, in 1990, the MONICA studies [4] have shown that the French middle aged males have only half the CHD event rates of Swedish males of the same age. A comparison of the populations of Lille, in northern France and Gothenburg, south-western Sweden yielded age standardized CHD event rates of 227 per 100 000 inhabitants and 406 per 100 000 inhabitants, respectively [4]. The Renault±Volvo-Coeur-study is a prospective study of cardiovascular risk factors pro®le amongst representative groups of automotive industry employees in France and Sweden. One aim was to study different aspects of the s.c. French paradox, assuming that any responsible cultural factors, would be present in these subpopulations and made available for in depth study through the occupational health setting. In 1993,1000 male employees aged 45±50 years, from each company, were studied by using blood samples, physical measurements, ECG and questionnaires. This was the ®rst phase, or baseline. In 1996±1997, 90 high risk and 90 low risk subjects were selected in each country for an in-depth examination to determine connections between baseline data and certain possible pathogenetic mechanisms. This second phase covered ultrasonographic measurements of the heart, aorta and carotid arteries, certain blood and hormone analyses and in-depth analysis of diet. The third phase in 1998 involved follow-up on the following end-points in the form of cardiovas-

cular complications: thrombosis, arterial constriction (claudiocatio and angina pectoris), myocardial infarction and death after 5 years. This paper presents data from the baseline survey pertaining to aspects of marital status and cardiovascular risk. The diet analysis is part of the second phase and will be presented elsewhere. The results from the baseline comparison of weighed traditional risk factors [5] showed very little overall differences between the two cohorts. These risk factors included age, male gender, systolic blood pressure, total to HDL cholesterol ratio, smoking, diabetes and ECG signs of left ventricular hypertrophy. This leaves the possibility for other explanations of the lower level of CHD in France than in Sweden. In the Coeur study we have also collected information about non traditional cardiac risk factors such as work stress, type A personality, depression, diet and marital status. It is known from demographic statistics in many countries that men living alone have a higher age adjusted mortality than married or cohabitating men. In Sweden, men who were divorced, widowers, or single/never married had twice the age adjusted mortality rate of those married or cohabitating [6]. In Swedish men, the lack of social support and having a poor social network have been associated with increased age adjusted mortality from CHD, and mortality from all causes [7]. It was hypothesized that the relative protection provided by social and family ties might differ between Sweden and France. If a higher proportion of French employees were married, this could indicate a higher stability of family and social life providing relative cardiovascular protection in the French workers. The aim of this study was to compare the French and Swedish employees who were married, divorced, or single and never married, regarding both their traditional cardiac risk pro®le as well as their life style, individual personality, and environmental factors. It was hypothesized that men living alone had a higher level of both traditional and nontraditional risk factors which then in the long run could contribute to CHD. If it were found to be in the Swedish cohort compared with the French, more employees living alone and having higher risk factor levels, this could explain a portion of the French paradox.

ã 2001 Blackwell Science Ltd Journal of Internal Medicine 249: 315±323

MARITAL STATUS VERSUS CARDIAC RISK

Methods In 1993, 2333 Caucasian men aged 45±50 years, 1189 subjects at Renault (Paris and Normandy region) and 1144 subjects from Volvo (Gothenburg region) were randomly selected by occupational physicians. The goal was to recruit 1000 men from each country. This ®gure was calculated needing to detect the difference of risk factor proportions in the two cohorts with high statistical power, at the same time as it would be feasible to manage in the occupational health settings. As there were more men available in the age group to be studied (men born between 1943 and 1947) a segment of birthdays, for instance men born on the 15th to the31st any month, were selected to give a list of well over 1000 men proportionately recruited from the participating units in each country. These men were then sorted in alphabetical order, as they otherwise would have been grouped by department and not randomly covering the whole plant, and were recruited by going down the list asking each individual until a complete number of 1000 in each country had been selected. The subjects were contacted by letter, if they did not reply by telephone. A total of 2000 participants were included in the study and amongst them 1946 could be categorized according to marital status: married/cohabitant, divorced (and living alone) or single (never married). A total of 189 employees in France and 144 in Sweden declined to participate, yielding a participation rate of 84% for Renault and 87% for Volvo. For each of the parameters described below more than 90% of the total subjects of each sample were included in the analyses. Missing values varied from 3 to 9%. Base-line data of the employees laboratory ®ndings included total-, HDL- and LDL-cholesterol, blood pressure as well as Framingham risk index [8] (based on age, smoking, diabetes, systolic blood pressure, total/HDL cholesterol ratio and left ventricular hypertrophy). These methods have been previously described [5]. All employees completed a self-administrated standardized questionnaire on demographic and life style risk factors. Marital status was categorized as married/cohabitating, divorced and living alone, or single and never married/cohabitating. Furthermore, information about smoking, alcohol habits and exercise was also gathered [5]. The smoking questions were part of a French published algorithm for calculating the

317

number of pack-years smoked [9]. Smoking habits were compiled by the question about current smoker (yes/no) de®ned as `Do you smoke at least one cigarette per day for the last 3 months?' Information about alcohol habits was also collected through a questionnaire and total alcohol was calculated and adjusted by the size of glass as follows: for Swedes 1 glass of beer ˆ 6 g, 1 glass of wine 12 g and 1 glass of distilled spirit equals 13 g. The corresponding ®gures for the French men were 10, 14 and 16 g. The basis for this calculation comes from differences in national data; for beer in Sweden there are three different alcohol levels, mean level is 3.2%, in France beer alcohol level is 5%. This is the reason for the somewhat higher alcohol gram levels for beer in France. Similar calculations were made for wine and spirits. The question about exercise was categorized as: up to 1, 1±3 or >3 h per week. These questions were validated versus resting pulse rate as a marker of ®tness. In both countries taken together we have calculated that men exercising 3 h week±1 a pulse rate of 65.2 bpm (P < 0.001). Similar correlations were found between exercise and pulse rate within each country. Type A behaviour was measured according to the Bortner scale which represents extroversion, time urgency and over commitment [10±12]. All employees also completed a self-administrated questionnaire concerning job stress as assessed by the Karasek method [13] on decision latitude and psychological demands at work. Social support at work was measured using questions identi®ed from the Swedish Survey of `Living Conditions' [14]. Social support outside work was assessed using seven items in the questionnaire which described structure and function of the social network contacts. This global social network support index was modelled after the Alameda County measure of social support. Although not directly validated across French and Swedish populations, this social network model has been applied in a variety of populations, including several studies in the US, Finland, Sweden and Netherlands [15]. The seven questions in the questionnaire were coded as follows: `Ever engaged in social leisure activities?' yes ˆ 1 p; `Social leisure activities every week?' yes ˆ 1 p; `Engaged in hobbies (one or more)?' yes ˆ 1 p; `Belongs to an association?'

ã 2001 Blackwell Science Ltd Journal of Internal Medicine 249: 315±323

318

L . K U M L I N et al.

yes ˆ 1 p; `Number of close friends?': zero friends ˆ 0 p, 1±2 friends ˆ 1 p, 3±5 friends ˆ 2p, >5 friends ˆ 3 p; `Meets friends every week?' yes ˆ 1 p, `Meets friends every month?' yes ˆ 1 p. A maximum score of 9 from this social support index could then be obtained. The number of close friends were also used as a separate variable. Statistical methods To study the associations between marital status and cardiovascular risk factors we used multiple regressions models. The independent explanatory variables were: age, blue/white collar status, Renault/Volvo. Furthermore living single was de®ned as answering yes to the question `Do you live alone?' and no to the questions `Are you married/ cohabitating?' and `Are you divorced?' Being divorced was de®ned as answering yes to the questions `Are you divorced?' and `Do you live alone?' and no to the question `Are you married?'. The dependent, outcome, variables are described in the methods section above. A binary logistic regression analysis was used for the smoking variable. Standard multiple regression was used for the other variables. Regression models were studied (i) for the entire baseline material and (ii) separately for each of the countries as there were different structures in the countries. The motive for this was to investigate if the same relation would be found in both countries.

Table 1 Distribution of marital status in middle-age men, 45±49 years at Volvo and Renault Category Marital status

Renault

Volvo

n

%

n

%

P-value

886 26 67

90.5 2.7 6.8

739 107 121

76.4 11.1 12.5