Obesity and myocardial infarction – vulnerability related to occupational level and marital status. A 23-year follow-up of an urban male Swedish population

Authors


 Dr Bo Hedblad, Department of Community Medicine, Division Epidemiology, Malmö University Hospital, 205 02 Malmö, Sweden (fax: +46 40 33 62 15; e-mail: bo.hedblad@smi.mas.lu.se).

Abstract

Abstract.  Hedblad B, Jonsson S, Nilsson P, Engström G, Berglund G, Janzon L (Lund University, Malmö University Hospital, Malmö, Sweden). Obesity and myocardial infarction – vulnerability related to occupational level and marital status. A 23-year follow-up of an urban male Swedish population. J Intern Med 2002; 252: 542–550.

Background.  People, who are single, have a blue-collar job or low income have an increased cardiovascular risk. This study on myocardial infarction sought to explore whether the socio-economic pattern of disease has any relationship with obesity.

Methods.  In the cohort are 20 099 middle-aged men of whom 9150 were manual and 9190 nonmanual workers and 1759 were self-employed. A total of 4081 were single, 16 018 cohabiting. The body mass index (BMI) cut-off values for overweight and obesity were 25–30 and ≥30 kg m−2, respectively. Local and national registers were used to monitor incidence of events over 18 years.

Results.  Obesity was associated with an increased incidence of coronary events and deaths in each occupational group. Being single significantly increased the risk associated with obesity. After stratification for civil status the risk associated with obesity was limited to those who were single and who either had a blue-collar job or were self-employed. The multivariate-adjusted relative risk (RR) of coronary events and deaths in obese manual workers who were single was 1.91 (95% confidence interval: 1.21–3.02) and 2.54 (1.74–3.69), respectively, times higher than it was amongst those who were cohabiting. Amongst those who were self-employed, the corresponding age-adjusted RRs were 4.79 (1.69–13.57) and 3.80 (1.62–8.93).

Conclusions.  Adjusted for lifestyle and biological risk factors, the increased risk of coronary events and death for obese men with manual jobs was applicable only to those who were single. It is concluded that being single significantly increases the cardiovascular risk associated with obesity.

Introduction

Between groups defined in terms of level of education, occupation, annual income and civil status there are marked differences of the incidence cardiovascular disease (CVD) and death [1–6]. In general, people who are single have a blue-collar job or low income have an increased risk. Some of these circumstances are associated with an increased prevalence of overweight and obesity [7–9].

The independent influence of obesity on the occurrence of CVD has been documented in several cohort studies, including a long-term follow-up of the male population in the Swedish Malmö Preventive Project (MPP) [10–14]. Whether and how the risk associated with obesity is modified by socio-economic circumstances has received less scientific attention.

The objective in this prospective cohort study on the incidence of myocardial infarction and death has been to explore whether, there are differences of the vulnerability associated with overweight and obesity between groups defined in terms of occupation and civil status.

Methods

The MPP was organized and carried out by the Department of Medicine at Malmö University Hospital [15]. The project has been described in detail previously [15–17]. In short, the aim of the project, which started in 1974, was to invite the adult population to screening in order to identify high-risk individuals for CVD. All male subjects living in the city of Malmö– born in 1921, 1926–42, 1944, 1946, 1948 and 1949 – were invited by letter to participate in a broad health screening programme, including a physical examination and a panel of laboratory tests. When the unit was closed for men in 1984, 22 444 men had attended the examinations with an overall attendance rate of 71%[16, 17]. Information on nonparticipation in the MPP has been reported previously [17]. In the present study, we excluded men with history of myocardial infarction and stroke (n = 233), those who were unemployed (n = 108), pensioners (n = 414), students (n = 268) and men having occupations that did not match any socio-economic index (SEI) code (n = 145). Another 1177 men had to be excluded because of missing information with regard to body mass index (BMI) or civil status. This resulted in a study cohort of 20 099 men who at the baseline examination were between 27 and 61 years of age. The health service authority of Malmö approved the screening programme of the age cohorts, and all participants gave their written consent.

Baseline examination

In the mailed invitation were instructions to abstain from food, alcohol and tobacco for 12 h prior to the examination [15]. Height (m) and weight (kg) were measured whilst the subject was wearing light indoor clothing. BMI was calculated as weight/height2 (kg m−2). In accordance with the World Health Organization weight classification [7], the men were categorized as overweight when BMI was 25–30 kg m−2 and obese when BMI ≥ 30 kg m−2.

Blood pressure (mmHg) in supine position was measured twice, after 10-min rest, in the right arm by use of a sphygmomanometer and a rubber cuff of appropriate size. The recorded pressure is the average of the two measurements. Men with systolic or diastolic blood pressure equal to or exceeding 160/95 mmHg and men who used blood pressure lowering medication were considered to have hypertension [18]. Total cholesterol, triglycerides and glucose were analysed in blood samples drawn after an overnight fast and were analysed with standard methods at the laboratory of the Malmö University Hospital [15–17]. Prevalence of diabetes is based on the number of men who either had history of the disease or whole blood glucose ≥6.70 mmol L−1[19].

Information on smoking habits and physical activity during leisure-time was assessed from a self-administered questionnaire [15–17]. The following smoking categories were used: nonsmokers, former smokers (smokers who had quit smoking at least 1 year before the examination) and current smokers, e.g. daily consumers of at least 1 g of tobacco [16, 17]. Two categories were used for the classification of leisure-time physical activity, e.g. sedentary or not. The prevalence of problematic drinking behaviour is based on a modified version of the Michigan Alcoholism Screening Test (Mm-MAST) [20, 21].

Civil status and occupational level

Occupational level and civil status was obtained by data linkage with the Swedish national population census carried out in the years 1970, 1980 and 1985 (Folk-och Bostadsräkningen, FoB). Classification was carried out in accordance with the census that was closest to date for the health examination. The influence of civil status on the incidence of events is based on a comparison of those who were single with those who were cohabiting. Occupational status, assessed by answers to questions concerning job titles and work tasks, formed the basis for classification into SEI groups, according to methods used by the National Bureau of Statistics, Sweden [21, 22]. This classification system takes into account the educational background needed to qualify for a particular job, any additional prerequisites, the level of responsibility the job entails and the specific tasks to be performed. In this study, the different SEI categories have been combined into three occupational groups: nonmanual workers (i.e. business executives, engineers with university degrees, physicians, college teachers, secondary school teachers, office assistants, sales people), self-employed (i.e. professionals with and without employees, entrepreneurs, farmers), and manual workers (i.e. auto mechanics, metal workers, construction workers, factory workers, waiters, cleaning staff).

Retrieval of end-points

Information on morbidity and mortality following the health examination was obtained by record linkage with the national inpatient register [23], the Swedish Causes of Death Register [24] and the Malmö Heart Infarction register [25]. Death certificates used for the classification of the underlying cause of death and hospital diagnosis had been coded in accordance with the ninth version of the International Classification of Diseases, Injuries and Causes of Death (ICD) [26]. Information on emigration was retrieved by data linkage with the emigration register at the Swedish National Bureau of Statistics. Each of the participants in the cohort was followed for mortality and incidence of coronary events until date of emigration or 31 December 1997. In the numerator used for computation of incidences of coronary events are those who had acute myocardial infarction and those who; according to the death certificate, died of chronic ischaemic heart disease (ICD codes 412–414). Only the first event was counted.

Statistics

SPSS (version 10.0) was used for all statistical analyses. A general linear model (for continuous variables) and logistic regression (for dichotomous variables) were used to assess the age-adjusted relationships (by test for linear trends) between BMI categories and civil status, occupation status and other risk factors for CVD. Incidence of coronary events and deaths in relation to BMI categories were expressed in terms of age-adjusted number of events per 1000 person-years of follow-up. Cox's proportional hazards model was used to estimate the influence of BMI categories, occupational level and marital status after adjustment for age. In a second model, further adjustment was made for differences with regard to lifestyle factors (i.e. smoking habits, leisure-time physical activity and history of problematic drinking behaviour). In a third model, further adjustments was made for biological risk factors (i.e. hypertension, diabetes, total cholesterol and triglycerides). In a final model, additional adjustment was made simultaneously for BMI categories, civil status and occupational level. Age, cholesterol and triglycerides (log-transformed) were modelled as continuous variables; hypertension, diabetes, smoking habits, physical inactivity during leisure-time, history of problematic drinking behaviour, BMI categories, civil status and occupational level were fitted as categorical variables. In an additional analysis, the possible influence of over-adjustment of the impact of obesity on coronary heart disease has been explored by removing diabetes, hypertension and triglycerides from the final Cox's proportional hazards model. A stratified analysis of the event rates in BMI categories by civil status and by subgroups of occupational level was also performed. Emigrants (n = 401) were followed up until the date of emigration and were thereafter censored in the analyses. As information on leisure-time physical activity was missing in 1995 men and in order not to loose power in the stratified analyses, missing values on physical activity were coded into the models as a dummy variable.

Possible interactions between obesity and civil status and between obesity and occupational level were evaluated by including interaction terms in the final Cox's proportional hazards model. Interaction was also expressed in terms of a synergy index (SI) by methods described by Hallqvist et al.[27]. Values above 1 show a positive interaction or synergy. Gauss-approximation was used to calculate the confidence interval of the SI.

Results

Risk factors in relation to categories of BMI

With the exception of smoking, the age-adjusted mean values and prevalence of cardiovascular risk factors increased in a stepwise fashion with increasing body weight (Table 1). Between groups defined in terms of occupation and civil status, the prevalence of overweight ranged from 29.3 to 38.8%, and the prevalence of obesity from 4.1 to 7.9%. Both conditions were more common amongst self-employed or blue-collar workers than it was amongst those with nonmanual jobs, and more common amongst single men than it was amongst men who were cohabiting (Table 2).

Table 1.  Age and age-adjusted mean values (95% confidence interval) or prevalence of cardiovascular risk factors at baseline by body mass index (BMI) categories
 Normal weightaOverweightaObesityaP for trend
  1. a Normal weight is defined as a BMI up to 24.9; overweight 25.0–29.9; and obese, at least 30.0 kg m−2. bAge are given as mean (SD). cBased on 18 144 men.

No. of men11 81471231162 
Age (years)b42.9 ± 6.644.5 ± 6.345.2 ± 6.0<0.001
Age adjusted
 Systolic blood pressure (mmHg)124.4 (124.2–124.7)129.6 (129.2–129.9)136.4 (135.6–137.2)<0.001
 Diastolic blood pressure (mmHg)83.5 (83.3–83.6)87.4 (87.2–87.6)92.0 (91.5–92.6)<0.001
 Blood pressure medication (%)1.64.29.5<0.001
 Hypertension (%)12.725.446.9<0.001
S-triglycerides (mmol L−1)1.29 (1.27–1.31)1.69 (1.67–1.72)2.20 (2.14–2.26)<0.001
S-cholesterol (mmol L−1)5.51 (5.49–5.53)5.73 (5.70–5.75)5.79 (5.73–5.85)<0.001
Diabetes mellitus (%)1.72.47.6<0.001
Current smokers (%)51.444.042.0<0.001
Ex-smokers (%)14.718.818.8<0.001
Sedentary leisure-time activity (%)c17.919.624.0<0.001
Problematic drinking behaviour (%)15.516.918.40.221
Living alone (%)20.818.222.8<0.001
Nonmanual worker (%)49.242.533.1<0.001
Self-employed (%)8.09.111.6<0.001
Manual worker (%)42.848.455.3<0.001
Table 2.  Age-adjusted prevalence of overweight and obesity in relation to occupational level and civil status
Occupational levelCivil statusBMI category
Normal weightaOverweightaObesitya
  1. BMI: body mass index. aNormal weight is defined as a BMI less than 25; overweight 25.0–29.9; and obesity, at least 30.0 kg m−2.

Nonmanual workerCohabiting, n = 758862.533.44.1
Single, n = 160265.629.35.1
Self-employedCohabiting, n = 144754.937.27.9
Single, n = 31255.737.66.7
Manual workerCohabiting, n = 698354.338.86.9
Single, n = 216758.833.57.7

Mortality

A total of 2126 men died, 819 (39%) of CVDs, 762 (36%) of cancer and 538 (25%) of other causes. Cause of death, which was based on autopsy in 50%, was unknown in seven cases. Obesity, having a manual job and living alone were all – after adjustment for potential confounders – associated with significantly higher risk of dying [relative risk (RR): 1.34; 95% confidence interval (CI): 1.14–1.57; 1.15: 1.05–1.26 and 1.68: 1.52–1.85, respectively] (Table 3).

Table 3.  Adjusted relative risk of all-cause mortality by body mass index (BMI) category, civil status and occupational level in 20 999 men, the Malmö Preventive Project
VariableAge-adjustedModel 1aModel 2bModel 3c
  1. a Adjusted for age, smoking habits, sedentary leisure-time physical activity and history of problematic drinking behaviour. b In addition to covariates included in model 1 adjusted for hypertension, diabetes, serum cholesterol and triglycerides. c In addition to covariates included in model 2 adjusted simultaneously for BMI categories, civil status and occupational level. BMI categories were defined as normal weight (BMI <25.0), overweight (BMI 25.0–29.9), and obesity (BMI at least 30.0 kg m−2).

BMI category
 Normal weightReferentReferentReferentReferent
 Overweight1.07 (0.98–1.17)1.15 (1.05–1.26)1.02 (0.93–1.12)1.03 (0.93–1.13)
 Obesity1.69 (1.46–1.96)1.84 (1.58–2.13)1.37 (1.17–1.61)1.34 (1.14–1.57)
Civil status
 CohabitingReferentReferentReferentReferent
 Single1.88 (1.71–2.07)1.69 (1.54–1.87)1.70 (1.54–1.87)1.68 (1.52–1.85)
Occupational level
 Nonmanual workerReferentReferentReferentReferent
 Self-employed1.19 (1.02–1.39)1.13 (0.96–1.32)1.08 (0.92–1.26)1.06 (0.91–1.25)
 Manual worker1.36 (1.24–1.49)1.23 (1.12–1.35)1.22 (1.11–1.33)1.15 (1.05–1.26)

Incidence of coronary events

A total of 1526 men had myocardial infarction and 289 of these events were fatal. Another 212 men died of chronic ischaemic heart disease. Men with obesity, manual jobs and men who were living alone had an increased incidence of coronary events. These risks remained statistically significant after adjustment for potential confounders (RR, 1.24: 1.02–1.50; 1.15: 1.03–1.28; and 1.26: 1.11–1.43, respectively) (Table 4). The RRs associated with obesity was, as expected, much higher when hypertension, diabetes and triglycerides were removed from the model (RR, 1.76: 1.47–2.11).

Table 4.  Adjusted relative risk of coronary events by body mass index (BMI) category, civil status and occupational level in 20 999 men, the Malmö Preventive Project
VariableAge-adjustedModel 1aModel 2bModel 3c
  1. a Adjusted for age, smoking habits, sedentary leisure-time physical activity and history of problematic drinking behaviour. b In addition to covariates included in model 1 adjusted for hypertension, diabetes, serum cholesterol and triglycerides. c In addition to covariates included in model 2 adjusted simultaneously for BMI categories, civil status and occupational level. BMI categories were defined as normal weight (BMI <25.0), overweight (BMI 25.0–29.9), and obesity (BMI at least 30.0 kg m−2).

BMI category
 Normal weightReferentReferentReferentReferent
 Overweight1.22 (1.10–1.36)1.33 (1.20–1.48)1.09 (0.98–1.22)1.09 (0.98–1.22)
 Obesity1.72 (1.44–2.05)1.93 (1.61–2.31)1.26 (1.04–1.53)1.24 (1.02–1.50)
Civil status
 CohabitingReferentReferentReferentReferent
 Single1.35 (1.20–1.53)1.27 (1.12–1.44)1.28 (1.13–1.45)1.26 (1.11–1.43)
Occupational level
 Nonmanual workerReferentReferentReferentReferent
 Self-employed1.16 (0.97–1.40)1.11 (0.93–1.34)1.06 (0.88–1.28)1.05 (0.88–1.27)
 Manual worker1.30 (1.17–1.44)1.19 (1.07–1.32)1.18 (1.06–1.32)1.15 (1.03–1.28)

Interaction between obesity, occupational level and civil status

Adding interaction term between being single and obese in the Cox's proportional hazards model – simultaneously adjusting for possible confounders in the final model – showed a significant interaction (P = 0.014) for incidence of coronary events. The corresponding interaction term for all-cause mortality was borderline significant (P = 0.062). There were no significant statistical interactions between obesity and occupational level on the incidence of coronary events or on the all-cause mortality. After stratification for occupational level, the significant statistical interactions between obesity and being single on coronary events, and all-cause mortality was limited to those who either had blue-collar jobs or who were self-employed (P = 0.033 and 0.057, respectively, P = 0.017 and 0.063).

In each of the occupational groups, the age-adjusted incidence of coronary events and deaths was highest for those who were obese (Tables 5 and 6). The increased risk remained after adjustments for smoking habits, leisure-time physical activity and history of problematic drinking behaviour. After further adjustment for biological risk factors, the increased risk associated with obesity was limited to those who were single and who had a blue-collar job. The interaction between being single and being obese on the incidence of coronary events and deaths corresponded to a SI value of 3.33: 1.18–9.40 and 1.85: 1.13–3.20, respectively. The age-adjusted RR of coronary events and deaths amongst single obese men having a blue-collar job was 1.98 (1.26–3.09), and 2.72 (1.89–3.92) times higher, respectively, than it was amongst those who were cohabiting. This risk increase remained after controlling for potential confounders (RR: 1.91: 1.21–3.02; and 2.54: 1.74–3.69; respectively). Amongst those who were self-employed, the corresponding age-adjusted RRs were 4.79 (1.69–13.57) and 3.80 (1.62–8.93). No further adjustment for potential confounders was possible because of limited number of events amongst obese self-employed men.

Table 5.  Age-adjusted rates and relative risk (RR) of mortality in relation to body weight, level of occupation and civil status
Occupational
level
BMI
categorya
Civil status
CohabitingLiving alone
No. of deaths
(no. of men)
Deaths/1000
person-years
RR (95% CI)
Model 1b
RR (95% CI)
Model 2c
No. of deaths
(no. of men)
Deaths/1000
person-years
RR (95% CI)
Model 1b
RR (95% CI)
Model 2c
  1. BMI, body mass index; CI, confidence interval, NMW, nonmanual worker; SE, self-employed; MW, manual worker; NW, normal weight; OW, overweight; OB, obesity. aNormal weight is defined as a BMI <25; overweight 25.0–29.9; and obesity, at least 30.0 kg m−2. bCovariates in model 1 included age, smoking habits, sedentary leisure-time physical activity and history of problematic drinking behaviour. cCovariates in model 2 included age, hypertension, diabetes, serum total cholesterol, triglycerides, smoking habits, sedentary leisure-time physical activity and history of problematic drinking behaviour. Cohabiting men with NW served as the referent group for each analysis. dSignificantly different from all other groups in, respectively, occupational level.

NMWNW359 (4700)4.16ReferentReferent96 (1080)5.121.6 (1.3–2.0)1.6 (1.3–2.0)
OW273 (2578)5.061.2 (1.06–1.5)1.1 (0.9–1.3)59 (447)7.522.0 (1.5–2.6)1.7 (1.3–2.3)
OB52 (310)9.572.0 (1.5–2.6)1.4 (1.01–1.9)11 (75)8.282.3 (1.3–4.2)1.5 (0.9–2.8)
SENW67 (784)4.64ReferentReferent23 (178)7.411.6 (1.01–2.6)1.7 (1.03–2.7)
OW63 (545)6.311.4 (0.9–1.9)1.3 (0.8–1.8)15 (115)7.521.7 (0.9–2.9)1.6 (0.9–2.8)
OB15 (118)7.501.8 (1.02–3.2)1.7 (0.9–3.0)9 (19)25.777.8 (3.9–15.9)7.3 (3.4–15.8)
MWNW372 (3780)5.48ReferentReferent188 (1292)8.531.6 (1.3–1.9)1.6 (1.4–1.9)
OW279 (2724)5.761.0 (0.9–1.2)0.9 (0.8–1.1)123 (714)9.971.8 (1.5–2.3)1.6 (1.3–2.0)
OB71 (479)8.331.4 (1.1–1.9)1.1 (0.8–1.4)51 (161)20.143.5 (2.6–4.8)d2.7 (2.0–3.7)d
Table 6.  Age-adjusted incidence and relative risk (RR) of coronary events in relation to body weight, level of occupation and civil status
Occupational
level
BMI
categorya
Civil status
CohabitingLiving alone
No. of events
(no. of men)
Events/1000
person-years
RR (95% CI)
Model 1b
RR (95% CI)
Model 2c
No. of events
(no. of men)
Events/1000
person-years
RR (95% CI)
Model 1b
RR (95% CI)
Model 2c
  1. BMI, body mass index; RR, relative risk; CI, confidence interval; NMW, nonmanual worker; SE, self-employed; MW, manual worker; NW, normal weight; OW, overweight; OB, obesity. aNormal weight is defined as a BMI <25; overweight 25.0–29.9; and obesity, at least 30.0 kg m−2. bCovariates in model 1 included age, smoking habits, sedentary leisure-time physical activity and history of problematic drinking behaviour. cCovariates in model 2 included age, hypertension, diabetes, serum total cholesterol, triglycerides, smoking habits, sedentary leisure-time physical activity and history of problematic drinking behaviour. Cohabiting men with NW served as the referent group for each analysis. dSignificantly different from all other groups in, respectively, occupational level.

NMWNW270 (4700)3.17ReferentReferent48 (1080)2.591.1 (0.8–1.4)1.1 (0.8–1.5)
OW228 (2578)5.901.4 (1.2–1.7)1.1 (0.9–1.4)38 (447)5.041.7 (1.2–2.4)1.4 (1.01–2.0)
OB32 (310)6.061.7 (1.2–2.4)1.2 (0.8–1.7)9 (75)6.982.8 (1.4–5.5)1.7 (0.9–3.4)
SENW58 (784)4.15ReferentReferent13 (178)4.061.1 (0.6–2.1)1.2 (0.7–2.2)
OW37 (545)3.820.9 (0.6–1.4)0.8 (0.5–1.2)14 (115)6.891.9 (1.04–3.4)1.6 (0.9–2.8)
OB10 (118)4.571.3 (0.7–2.6)0.9 (0.5–1.9)6 (19)21.555.6 (2.4–13.3)4.7 (1.9–11.4)
MWNW270 (3780)4.04ReferentReferent97 (1292)4.471.2 (0.9–1.5)1.2 (0.9–1.5)
OW239 (2724)5.001.3 (1.07–1.5)1.1 (0.9–1.3)71 (714)5.781.6 (1.2–2.0)1.3 (0.9–1.7)
OB56 (479)6.801.7 (1.3–2.3)1.1 (0.8–1.5)30 (161)12.013.0 (2.1–4.4)d1.9 (1.3–2.8)d

Discussion

Eighteen years after the baseline examination, 18.0% of the obese men and 9.4% of the men with normal body weight were no longer alive, and 12.3 and 6.4%, respectively, of them had been treated for an acute coronary event. The increased risks associated with obesity are in line with the results from many other cohort studies [10–13]. Yet, within the obese population some are obviously more vulnerable than others. It is concluded that between groups defined in terms of civil status and occupation, there are significant differences of the cardiovascular risk associated with obesity.

Rupture of plaque with subsequent thrombus formation is the final event of the disease processes causing myocardial infarction [28]. Associated risk factors may hence either promote the development of plaque or increase the probability of complications. Obesity is associated with the establishment of coronary ischaemia as well as with the occurrence of myocardial infarction [8, 12]. The true mechanisms behind these associations are not fully understood. The prevalence of cardiovascular risk factors showed a dose–response relationship with body weight. In terms of causes of obesity there seems to be several different subgroups within the obese population, i.e. inadequate exercise, overconsumption of foods and alcohol, metabolic disorders and genetic abnormalities. Previous studies have shown that the risk associated with obesity varies between groups defined in these terms [29, 30]. Several studies have suggested that the distribution of fat influences the risk and that the risk associated with obesity in fact is related to the amount of fat in the abdomen and upper body [31, 32].

According to other studies it seems that the influence of obesity on cardiovascular risk is related to weight increase in early adulthood [33]. Our results show that between groups defined in terms of civil status and occupation – although this categorization is crude – there are significant differences of the influence of obesity on the incidence of coronary events and death. Men with manual jobs were exposed to an increased risk if they were single. The risk increase remained after adjustment for lifestyle and biological risk factors. In the absence of these circumstances, obesity had no statistically significant influence on the event rates. It should, however, be remembered that the biological factors (hypertension, diabetes and lipids) could form part of the causal chain from obesity to coronary events and adjusting for these factors could conceal the association under study.

The underlying biological pathways by which socio-economic conditions can promote CVD are not clearly understood. So far no studies have presented evidence suggesting that having a manual job or being single would influence the atherosclerotic process per se. Amongst the obese there was, however, an over-representation of single men with manual jobs. It is still possible that these men, although being apparently healthy, on average have more severe atherosclerotic lesions. To what extent this circumstance contributes to the increased risk in that group remains to be evaluated.

Maintenance of ideal body weight has been recommended as an important measure to reduce the incidence and mortality from CVD [34]. In order to improve the cost-effectiveness of treatment and prevention, there is a need for studies that can improve our understanding of who will benefit the most. Our results suggest that single men who have a manual job should be one group. However, it has not yet been proven in controlled trials whether the cardiovascular risk associated with obesity can be lowered by weight reduction. With the exception of surgical intervention there seems to be no method available to achieve a permanent weight loss [35]. Needless to say the occurrence of CVD amongst those with obesity can only to some extent be accounted for by the excessive body weight. In this urban cohort, 47% of the obese had hypertension, yet only 10% reported use of blood pressure lowering medication. Sixty per cent of them had been regular smokers; of those less than one-third had quit. One of 12 had diabetes. Hence in order to reduce the cardiovascular morbidity and mortality associated with obesity there is a need for preventive programmes that take into account the complicated interaction between modifiable risk factors.

Some methodological issues need to be considered. Local and national registers were used for case-retrieval. A validation study from the National Hospital Discharge register showed that the diagnosis ‘myocardial infarction’ was false only in 5% of the cases [23]. Thus, based on the experience from the Swedish National Board of Welfare [23], the Malmö Heart Infarction register [25] and several other follow-up studies of this cohort there should be no reason to believe that observed associations could have been confounded by biased retrieval and validation of end-points [17]. Because of the small number in each group, no distinction was made between those who never had been married and those who were divorced or widowers. It is quite possible that amongst men who are living alone one can define subgroups with a lower risk of CVD and death [36–40]. Thus, a limitation of our study is the lack of data on other lifestyle patterns and psychosocial factors, associated with socio-economic circumstances and cohabiting status [36, 37, 41–43]. It is known that increased levels of catecholamines, such as those resulting from environmental stress, could contribute to coronary disease through a variety of mechanisms [44]. Furthermore, in a population sample of 51-year-old Swedish men it was recently demonstrated that lower socio-economic status was associated with visceral obesity and higher cortisol values in relation to perceived stress [45]. It is however, noteworthy that the RRs changed little after adjustment for lifestyle factors.

Change in exposure is an inherent problem in long-term cohort studies. Some of those who were single may have married and vice versa, some have lost their jobs, others have changed occupation. Single men who married during follow-up have probably reduced their risk. Conversely, married men who divorced or became widower would have been exposed to an increased risk. Thus, it is reasonable to assume that changes in civil status during follow-up, if anything, tend to dilute existing associations. All who were diagnosed with hypertension, lipid disorders or diabetes were referred for further evaluation and treatment, smokers were advised to quit, but received no help to achieve that [15]. As the prevalence of these risk were more common amongst men with obesity and men having a manual job, they should benefit most from the interventions and this would if anything tends to dilute existing associations. Intervention against overweight and obesity was not part of the preventive programme. To what extent change in body weight in groups defined in terms of civil status and occupational level have confounded the results remains to be evaluated.

Adjusted for life-style and biological risk factors, the increased risk of coronary events and death for obese men with manual jobs was in this cohort study applicable only on those who were single. It is concluded that being single significantly increase the cardiovascular risk associated with obesity.

Acknowledgements

The Swedish Medical Association, the Swedish Council for Social Research and the Malmö City Council financially supported this study.

Ancillary