Social class differentials in mortality and morbidity persist in Europe, even though health care allocation in general has become more equitable in recent decades. 1 Such health differences may, in part, be due to behavioral and lifestyle aspects of socioeconomic status rather than a lack of universal health coverage. 2,3 The aim of this study is to examine whether a discrepancy between mortality in lower and higher socioeconomic groups continues to exist in Sweden, where health care has been made accessible to all. Much has been written about socioeconomic status and mortality in men, but relatively less is known about women, among whom these associations may be even more complex. 4–7 Therefore, we measured estimates of social class inequality using both conventional and composite indicators, to assess optimally socioeconomic status in 1968. We will describe the association between these social risk factors and women’s health while also taking into account factors known to vary with socioeconomic status and survival, for example, smoking and obesity.
Subjects and Methods
A prospective population-based study of women from Gothenburg (approximately 450,000 inhabitants) in western Sweden began in 1968–1969, when a representative group of 1,622 women were randomly selected within the strata 38, 46, 50, 54, and 60 years of age; 1,462 of these women attended the primary health examination, resulting in a participation rate of 90%. This investigation consisted of detailed questionnaires describing medical history and socioeconomic variables together with physical examinations including anthropometric measures taken by a single investigator. 8 Three follow-up studies have been subsequently carried out in 1974/1975, 1980/1981, and 1992/1993. After 24 years, 265 (18%) of the original participants were deceased and fewer than 1% have been lost to follow-up. 9
Measures of Socioeconomic Status
Socioeconomic status was characterized in this analysis by two methods: husband’s occupational category and a composite variable reflecting household income and women’s education. Women’s own occupational category could not be used as a socioeconomic indicator, because 39% described themselves as housewives, and of those who worked outside the home, many worked part-time.
Husband’s occupational category is a conventional socioeconomic indicator reflecting community status and financial earnings of the husband in each household, for married women. This variable comprised three levels of socioeconomic status: high (large-scale employers and officials of high or intermediate rank, including 14% of the 1,156 married women), medium (small-scale employers, officials of lower rank, foremen, including 43% of the sample), and low (skilled and unskilled workers, including 43% of the sample). 10,11 Seventy-nine per cent of the women were married and could be characterized in this way.
In addition, we examined the variables household income and education so that we could include all women in the analysis. Self-reported household income at the time of the baseline study was calculated from a woman’s own income plus that of her husband, if married; we used a cutpoint of 35,800 Swedish Crowns per year (median) to discriminate lower vs higher income groups. Educational group also included two categories that were based on a natural cutoff level; the majority of women had attended primary school grades 1 through 7 (70%), whereas only 30% had gone beyond this level, and fewer than 2% had attended university or college in 1968.
Combining information on household income and education, a three-level composite indicator was created. High socioeconomic status consisted of high income with high education and included 20% of all 1,462 participants. Medium referred to high income with low education, comprising 29% of the sample. The remaining 51% were classified as low (low income with either high or low education). This type of composite approach, which has been used previously, 12 combines information concerning resources available to the household with individual social class information as reflected by educational level for each woman. The categorization scheme used here gives greater weight to income than education in the assessment of health risk in women, in accordance with previous research suggesting that income has a stronger relation to morbidity than education in women. 6,13,14
Mortality and Morbidity Endpoints
For each socioeconomic variable, we created statistical models for total mortality, cause-specific mortality, and selected morbidity endpoints. The total mortality model included 265 deaths. Cause-specific mortality included cardiovascular disease (104 deaths) and cancer of all sites (88 deaths). Specifically, mortality due to cardiovascular disease consisted of all deaths from myocardial infarction (40%), other heart diseases (34%), or stroke (26%). Death due to cancer at all sites included the following cancers: breast (21%), uterine (5%), ovarian (12%), lung (9%), blood/lymph (14%), gastrointestinal (22%), skin (1%), multiple site (1%), and other (15%). Two general types of cancer morbidity were studied separately, out of a total of 221 cases: breast cancer (22%) and non-breast cancer (78%). Morbidity from myocardial infarction (92), stroke (79), and diabetes mellitus (82) was also evaluated. Information on mortality was obtained from death certificates, and morbidity was ascertained from cancer registries, hospital registries, and medical examinations.
The data were analyzed using the Cox proportional-hazards model to estimate time to mortality or morbidity. 15 Each socioeconomic factor was studied in an age-adjusted model and also with confounders potentially relevant to cardiovascular disease or cancer, in disease-specific regression models. These covariates (age, waist-hip ratio, body mass index, smoking, age at first birth, and parity) were selected because they had known associations with mortality, morbidity, and socioeconomic status in women. We used reduced models when the number of events was low. Results from the proportional-hazards models were presented with slope estimates, standard errors, hazard ratios [relative risk (RRs)], and 95% confidence intervals (95% CIs).
Selected baseline characteristics are shown in Table 1. These descriptive data are stratified according to the husband’s occupational category, making it possible to document the relation between the husband’s socioeconomic status and other variables used in this analysis.
Husband’s occupational category was analyzed in relation to mortality and morbidity for married women in whom the low group was used as the reference. The following cumulative mortality proportions were observed for each category of the husband’s occupation: high 20% (32 deaths), medium 16% (77 deaths), and low 17% (86 deaths). We found an almost flat association between husband’s occupational group and total mortality, in age-adjusted and multivariate models (Table 2). Nevertheless, lower occupational category was associated with excess cardiovascular disease mortality. We saw a protective effect with respect to cardiovascular disease mortality when high occupational category was compared with low (RR = 0.22; 95% CI = 0.04–0.84). In contrast, higher husband’s occupational category indicated an increased risk in all-site cancer mortality relative to low (RR = 2.29; 95% CI = 1.21–4.31). Both of these associations were independent of all covariates included in the respective multivariate models.
Considering the morbidity endpoints, high occupational category was associated with elevated breast cancer morbidity compared with low (RR = 2.92; 95% CI = 1.30–6.53), although differences were not observed between the medium and low occupational categories. Non-breast cancer morbidity also increased as a husband’s occupational level increased, but only marginally. In contrast, we found inverse trends for stroke and diabetes mellitus morbidity, with the greatest contrast between the middle and low categories [RR = 0.53 (95% CI = 0.29–0.94), and RR = 0.39 (95% CI = 0.20–0.75), respectively]. These findings of excess stroke and diabetes mellitus risk in the lowest socioeconomic group persisted even after covariate adjustment. Compared to stroke and diabetes mellitus, the relation between spouse’s occupational category and myocardial infarction was much weaker.
To include all women in the analysis, regardless of marital status, we conducted separately analyses using household income or educational level. Although the magnitude and direction of these associations followed the same pattern as a husband’s occupational category, the confidence intervals tended to be much wider (results available upon request). Therefore, in addition to the husband’s occupational category, we are reporting results based on a composite socioeconomic indicator derived from income plus education, as described in Subjects and Methods.
As in all previous analyses, high and medium socioeconomic status levels were compared with the reference category, low. The crude 24-year cumulative mortality risks in the three composite socioeconomic status levels were 16% (45 deaths), 16% (67 deaths), and 20% (153 deaths) for high, medium, and low categories, respectively. Consistent with results seen using the husband’s occupation, total mortality in the lower composite socioeconomic group did not differ much from that in any of the other groups. Again, high socioeconomic status was associated with a decreased risk for cardiovascular disease [RR = 0.49; 95% CI = 0.24–0.99 (Table 3)]. Although the composite variable was not strongly related to mortality from cancer, it did present a positive trend similar to that seen using the husband’s occupational category.
The composite measure of socioeconomic status was also related to morbidity from breast cancer. Breast cancer morbidity increased as socioeconomic status increased, in a strong dose-response manner, and marked differences were seen when high socioeconomic status was compared with low (RR = 3.31; 95% CI = 1.74–6.31). Finally, using this measure of socioeconomic status, stroke and diabetes mellitus incidences were lower in the medium compared with low socioeconomic level [RR = 0.56 (95% CI = 0.32–0.99), and RR = 0.71 (95% CI = 0.53–0.96), respectively]. Stroke also showed an inverse trend with socioeconomic level, which became attenuated in the multivariate model. Consistent with the results in married women, we saw only a weak and uncertain association between socioeconomic status and incidence of myocardial infarction using the composite index.
This population-based study showed that low socioeconomic status as measured by two alternate socioeconomic indicators was related to increased cardiovascular mortality, in agreement with a short-term follow-up of the same women. 16 Although socioeconomic differentials in cardiovascular mortality among women, observed here, were similar to those previously reported in men, myocardial infarction morbidity in this investigation did not clearly follow the same pattern using either socioeconomic indicator. 2,17,18 Incidence of stroke was weakly associated with low socioeconomic status using both socioeconomic variables. Diabetes mellitus was also related to both indices of low socioeconomic status, but the association was much attenuated after covariate adjustment, suggesting a likely explanatory role for obesity.
In direct contrast to the cardiovascular disease and diabetes mellitus findings, higher occupational status of the husband was associated with high mortality from cancer, and the cancer morbidity data suggested that a large part of the association in women may be related to breast cancer. In contrast to the other endpoints, in the case of breast cancer morbidity, the composite socioeconomic indicator was stronger than the husband’s occupational category. A number of factors, however, may limit the generalizability of the cancer findings. First, owing to the small numbers, it was not possible to study cancer mortality in a site-specific way. Regarding breast cancer morbidity, the low number of cases limited the number of simultaneously controlled covariates for this endpoint, increasing the likelihood of confounding by variables not included in the model. Individual adjustments for other covariates did not change our results, however. Finally, although the combination of high education and high income was strongly associated with breast cancer risk, there were insufficient data available to analyze the combination of high education and low income.
There are multiple potential pathways by which socioeconomic status may relate to mortality and morbidity in women. 19–21 In this study, we controlled for risk factors measured at baseline (age, waist-hip ratio, body mass index, smoking, age at first birth, and parity) while testing two measurements of socioeconomic status. As expected, 22–24 waist-hip ratio and smoking were independent risk factors for cardiovascular disease in the multivariate model, in contrast to body mass index, which was not as strongly related (results not shown). Because these lifestyle-related variables often covary with socioeconomic status as well as mortality and morbidity, 25–28 it is important to highlight the fact that controlling for them did not alter the main findings of this study. Specifically, high socioeconomic status remained a strong correlate of less cardiovascular mortality but more cancer mortality, and these relations could not be attributed to class differences, for instance, in obesity, smoking behaviors, or reproductive history.
Our knowledge of socioeconomic risk factors for mortality and morbidity in women has been limited for numerous reasons, including assumptions regarding equitable health care allocation, together with a greater historical focus on chronic disease epidemiology in men. 3,29–32 Furthermore, the interpretation of observed inequities in health is complicated by the fact that social differences may be related to early biological influences as well as biobehavioral and environmental exposures in later life. Theories involving the impact of reduced “volitional capability” on health status among women of different social classes also merit consideration. 5,20,33 In this study, the strongest socioeconomic correlate of health outcome was a husband’s occupational category, which may reflect a woman’s dependency on her husband’s income, even if she was employed in 1968. Marital status per se did not alter the associations reported for the entire sample. Specifically, associations between the composite indicator and all endpoints studied were not affected when unmarried women were removed nor when marital status was considered as a confounder in the model.
To summarize, in this representative cohort of Swedish women traced over 24 years, opposing monotonic trends were seen for mortality from cancer and cardiovascular disease in relation to socioeconomic status. In particular, socioeconomic status correlated positively with cancer and negatively with cardiovascular disease. There was, however, some indication of U- or J-shaped associations for diabetes mellitus and stroke morbidity. Our findings confirm previous studies’ indications that associations between socioeconomic status and health still exist in Swedish women and that these associations vary profoundly for different diseases.
1. Feinstein JS. The relationship between socio-economic status and health: a review of the literature. Milbank Q 1993; 71: 383–411.
2. Rosengren A, Wedel H, Wilhelmsen L. Coronary heart disease and mortality in middle-aged men from different occupational classes in Sweden. BMJ 1988; 297: 1497–1500.
3. Vågerö D. Inequality in health: some theoretical and empirical problems. Soc Sci Med 1991; 32: 367–371.
4. Marmot MG, Shipley MJ, Rose G. Inequalities in death: specific explanations of a general pattern? Lancet 1984; 1 (8384): 1003–1006.
5. Kawachi I, Marmot MG. Commentary: what can we learn from studies of occupational class and cardiovascular disease? Am J Epidemiol 1998; 148: 160–163.
6. Davey Smith G, Hart C, Hole D, MacKinnon P, Gillis C, Watt G, Blane D, Hawthorne V. Education and occupational social class: which is the more important indicator of mortality risk? J Epidemiol Community Health 1998; 52: 153–160.
7. Abbot P, Sapsford R. Women and Social Class. London: Tavistock, 1987.
8. Bengtsson C, Blohmé G, Hallberg L, Hällström T, Isaksson B, Korsan-Bengtsen K, Rybo G, Tibblin E, Tibblin G, Westerberg H. The study of women in Gothenburg 1968–1969: a population study: general design, purpose, and sampling results. Acta Med Scand 1973; 193: 311–318.
9. Bengtsson C, Ahlqwist M, Andersson K, Björkelund C, Lissner L, Söderström M. The Prospective Population Study of Women in Gothenburg Sweden, 1968–69 to 1992–93: a 24-year follow-up study with special reference to participation, representativeness, and mortality. Scand J Prim Health Care 1997; 15: 214–219.
10. Carlsson G. Socialgruppering: Social Mobility and Class Structure. Lund, Sweden: University of Lund, GWK Gleerup, 1958.
11. Halling A, Bengtsson C. Number of teeth and proximal periodontal bone height in relation to social factors. Swed Dent J 1984; 8: 183–191.
12. Krieger N, Chen JT, Selby JV. Comparing individual-based and household-based measures of social class to assess class inequalities in women’s health: a methodological study of 684 US women. J Epidemiol Community Health 1999; 53: 612–623.
13. Arber S. Comparing inequalities in women’s and men’s health: Britain in the 1990s. Soc Sci Med 1997; 44: 773–787.
14. Schrijvers CT, Stronks K, van de Mheen HD, Mackenbach JP. Explaining educational differences in mortality: the role of behavioral and material factors. Am J Public Health 1999; 89: 535–540.
15. Parmar M, Machin D. Survival Analysis: A Practical Approach. Chichester, United Kingdom: John Wiley and Sons, 1995.
16. Lapidus L, Bengtsson C. Socio-economic factors and physical activity in relation to cardiovascular disease and death: a 12 year follow up of participants in a population study of women in Gothenburg, Sweden. Br Heart J 1986; 55: 295–301.
17. Marmot MG, Rose G, Shipley M, Hamilton PJS. Employment grade and coronary heart disease in British civil servants. J Epidemiol Commun Health 1978; 3: 244–249.
18. Kaplan GA, Keil J. Socio-economic factors and cardiovascular disease: a review of the literature. Circulation 1993; 88 (4 pt 1):1973–1998.
19. Gravelle H. How much of the relation between population mortality and unequal distribution of income is a statistical artefact? BMJ 1998; 316: 382–385.
20. Marshall G, Roberts S, Burgoyne C, Swift A, Routh D. Class, gender, and the asymmetry hypothesis. Eur Sociol Rev 1995; 11: 1–15.
21. Marmot MG, Kogevinas M, Elston MA. Social/economic status and disease. Annu Rev Public Health 1987; 8: 111–135.
22. Björntorp P. The association between obesity adipose tissue distribution and disease. Acta Med Scand 1988; 723: 121–134.
23. Brunner E. Inequalities in diet and health. In: Shetty PS, McPherson K, eds. Diet Nutrition and Chronic Disease: Lessons from Contrasting Worlds. New York: John Wiley and Sons, 1997.
24. Lapidus L, Bengtsson C, Larsson B, Pennert K, Rybo E, Sjöström L. Distribution of adipose tissue and risk of cardiovascular disease and death: 12 year follow-up of participants in the population study of women in Gothenburg, Sweden. BMJ 1984; 289: 1257–1261.
25. Goldberg RJ, Gore JM, Gurwitz JH, Alpert JS, Brady P, Strohsnitter W, Chen ZY, Dalen JE. The impact of age on the incidence and prognosis of initial acute myocardial infarction: the Worcester Heart Attack Study. Am Heart J 1989; 117: 543–549.
26. Thur L. Risk of breast cancer. J Lymphol 1989; 13: 74–82.
27. Franceschi S. Reproductive factors and cancers of the breast, ovary, and endometrium. Eur J Cancer Clin Oncol 1989; 25: 1933–1943.
28. Trichopoulos D, Hsieh CC, MacMahon B, Lin TM, Lowe CR, Mirra AP, Ravnihar B, Salber EJ, Valaoras VG, Yuasa S. Age at any birth and breast cancer risk. Int J Cancer 1983; 31: 701–704.
29. Rosengren A, Wilhelmsen L. Are there socio-economic differences in survival after acute myocardial infarction? Eur Heart J 1996; 17: 1619–1623.
30. Peltonen M, Rosén M, Lundberg V, Asplund K. Social patterning of myocardial infarction and stroke in Sweden: incidence and survival. Am J Epidemiol 2000; 151: 283–292.
31. Davis DL, Dinse GE, Hoel DG. Decreasing cardiovascular disease and increasing cancer among whites in the United States from 1973 through 1987. J Am Med Assoc 1994; 271: 431–437.
32. Anderson RT, Sorlie P, Backlund E, Johnson N, Kaplan GA. Mortality effects of community socio-economic status. Epidemiology 1997; 8: 42–47.
33. Wilkinson RG. Socio-economic determinants of health, health inequalities: relative or absolute material standards? BMJ 1997; 314: 591–595.