Socioeconomic differences in health indices, social networks and mortality among Swedish men. A study of men born in 1933.

Background: In previous survey we found large socioeconomic differences in mortality among urban Swedish men which remained unexplained after controlling for smoking and standard coronary risk factors. The present analysis was undertaken in order to investigate a broader set of possible explanatory factors in another cohort of Swedish men.

Study population and methods: Occupation was coded into five occupational classes for 717 of 776 participant men from a random population sample of 1016 men who were born in 1933. All were living in Göteborg and were 50 years old at the baseline examination. After 12 years' follow-up, 68 of the 717 men had died (9.5%).

Results: Low occupational class was associated with a higher prevalence of smoking at baseline, but no association was found with systolic blood pressure, body mass index, waist to hip ratio, serum triglycerides or serum cholesterol. Subjects from higher socioeconomic strata were taller, had higher maximum peak respiratory flow, lower plasma fibrinogen and lower body temperature. Low occupational class was associated with low social integration, low home activity levels, low levels of activity outside home and low social activity levels (p = 0.001 for all) and with low emotional support (p = 0.018). There were also associations between low occupational class and poor self-perceived health, as well as with several cardiovascular symptoms. During 12 years' follow-up, there was a graded and inverse relationship between occupational class and mortality from all causes. The highest mortality was found among the men who could not be classified (23 per 1,000 person years) Of the men in the lowest occupational class, 12 per 1,000 died, compared to 3 per 1,000 in the highest class (relative risk 3.7 (1.4-9.8)). After controlling for smoking, the relative risk decreased to 3.2 (1.2-8.6) and after further adjustment for emotional support, self-perceived health, activity level at home, and peak expiratory flow, the relative risk was still twofold but not significantly so (RR 2.1 (0.8-5.8)).

Conclusion: We were able to confirm earlier results as to the wide mortality differentials in urban middle-aged men in Sweden. There were also large differences in several other factors, including constitutional factors, health variables, lifestyle and social support indices, which explained important parts of the social mortality gradient, the most prominent being smoking, respiratory function, social network factors and subjective health.