Social inequalities in health: are there gender differences?
Introduction
The existence of socio-economic status (SES) inequalities in health is well documented in the industrialised world (Fox, 1989; Davey-Smith et al., 1990; Macintyre, 1997, Macintyre, 1998; Marmot et al., 1997). For most ages and causes of death there are improvements in health with each successive increase in the social hierarchy. That is, the general pattern is linear rather than showing a threshold effect, at least among men. The pattern for women is less consistent with some deviation from a linear trend (Macintyre, 1998). Moreover, there is some evidence that the magnitude of SES inequalities differs, with women having shallower gradients than men across a broad spectrum of morbidity measures (Dahl, 1993; Stronks et al., 1995) and mortality (Valkonen, 1989; Blane et al., 1990; Koskinen and Martelin, 1994; Elo and Preston, 1996). Although smaller SES inequalities for women are cited frequently, there are studies in which men and women have similar SES gradients (using both occupational and non occupational measures of social position). For example, similar patterns of SES inequality have been demonstrated for coronary heart disease (Diez-Roux et al., 1995), self perceived health and psychological well being (Marmot et al., 1997), overall health problems (Gijsbers van Wijk et al., 1995) and minor illnesses (Popay et al., 1993). There are other examples in which greater SES inequality has been found for women, specifically for asthma (Eachus et al., 1992) and for myocardial infarction events and case fatality (Morrison et al., 1997). SES gradients can also vary for particular sub-groups, with less pronounced gradients in limiting long-standing illness and self-rated health among women not in paid work (Arber, 1997). Further evidence suggests that gender differences in SES inequalities have changed over time. For example, in several western European countries SES mortality differentials appear to have widened among men but not women (Valkonen, 1989). In the US, SES mortality differentials increased substantially from the 1960s among middle aged and older white men, while they remained similar for women (Feldman et al., 1989). This divergent trend for men and women continued during the 1980s (Elo and Preston, 1996). In Finland, SES inequalities in limiting long-standing illness declined among men but not women between 1986 and 1994, hence the gender difference in SES inequality has diminished in recent years (Lahelma et al., 1997).
Some of the inconsistency in the evidence for greater differences in SES gradients may therefore relate to the period of investigation, but they may also relate to the age group studied. In Finland, SES inequalities in mortality were more pronounced among men than women between ages 35 and 49 (Valkonen et al., 1990), but less pronounced during childhood, early adulthood and among the elderly (Valkonen et al., 1993). Further support is available from studies focusing on particular age groups. Thus, among 15–24-yr old Finns, there were no SES inequalities in limiting long-standing illness for men or women (Rahkonen and Lahelma, 1992). However, in Scotland, a significant gradient in self-rated health was found for 18-yr old men but not women (West, 1997). There may therefore be gender differences in SES inequalities at some ages and not at others, that vary across time and place and across different health measures (Blane et al., 1994; Elo and Preston, 1996). Gender differences in SES inequalities in health are of interest partly because they provide evidence relevant to the underlying causes of SES inequality. Further evidence is therefore needed on the existence of gender differences in SES inequalities. Moreover, where gender differences in inequality occur it is necessary to exclude artefactual explanations, such as those associated with the measurement of social position for men and women. In this regard, concerns arise with the use of occupational based measures, because of the different relationships men and women have to the formal labour market (Macran et al., 1994; Arber, 1997). More women than men are employed part-time or are out of the paid labour market as full-time homecarers, while most men are employed full-time and, if not, are usually unemployed. These major gender differences in labour market experience can affect the extent to which men and women are included in an occupation-based classification of social position (Rose, 1995). Furthermore, the Registrar General's classification of occupations was constructed primarily on the basis of the male occupational structure. Given that women tend to occupy particular sections of the labour market (Rose, 1995), the magnitude of SES inequalities might reflect gender differences in the measurement of social position rather than true differences in the relationship between health and social position for men and women. Nevertheless, several studies demonstrating wider SES inequality in health for men have used alternative SES indices, such as education, that are less subject to these problems (Valkonen, 1989; Koskinen and Martelin, 1994; Elo and Preston, 1996). Other explanations have been suggested as to why the magnitude of SES inequality may differ for men and women (Koskinen and Martelin, 1994; Macintyre and Hunt, 1997; Macintyre, 1998). Specifically, factors associated with SES may vary by gender and hence there may be a different pattern of exposure and experience among men and women. For example, there appears to be less social variation among women than men in those employment status categories (unemployed and long-term disabled) associated with poorer health (Stronks et al., 1995). Moreover there may be gender differences in the impact of the same SES related factors due to different susceptibilities either through cultural or biological differences. Two further explanations for gender differences in the magnitude of SES inequalities have been investigated, namely the influence of other socio-demographic variables and the causes of death structure for men and women (Koskinen and Martelin, 1994). With regard to socio-demographic factors, there is some evidence showing shallower SES inequalities in married women compared with married men, but not in the single, divorced or widowed, among whom SES inequality in mortality is just as steep for women as for men (Koskinen and Martelin, 1994). In relation to cause specific mortality, the social gradients for most major causes of death were relatively similar for men and women, except for a steeper social gradient for accidents and violence among men and a slightly steeper gradient for circulatory diseases among women. The weaker SES gradient for all-cause mortality among women was largely attributable to a reverse gradient (that is, higher mortality in higher SES groups) for breast cancer (Koskinen and Martelin, 1994).
Whilst there are many studies focusing on explaining gender differences in morbidity and mortality, few studies specifically compare explanations for SES inequalities in men and women. One study that has been used to investigate SES inequalities is the 1958 British birth cohort, but so far gender differences in such inequalities have largely been neglected (Power et al., 1991, Power et al., 1996, Power et al., 1997, Power et al., 1998; Manor et al., 1997; Matthews et al., 1998). In previous work on the cohort we have adopted a lifecourse perspective in order to understand how SES inequalities develop (Power et al., 1991, Power et al., 1996, Power et al., 1998; Power and Matthews, 1997). Hence, the longitudinal data available for the cohort have been indispensable. The conceptual framework used in earlier work also integrates different potential explanatory factors, in order to understand their relative importance. This has involved examining the contribution of health-related behaviour, material circumstances, education, working characteristics, family structure and social support towards the development of SES inequalities (Power et al., 1998). These explanatory areas were examined because there is extensive evidence linking these areas both to SES and to health (Power et al., 1991, Power et al., 1998). Although most previous analyses were conducted separately for men and women, no systematic comparisons have been undertaken to establish if SES inequalities are of similar magnitude for men and women, nor whether there are underlying similarities in the explanations. The cohort study provides an opportunity to clarify relationships between gender, social position and health in a relatively young population.
This paper addresses two questions, first does the magnitude of SES inequality in health differ between men and women at ages 23 and 33; second, do explanations for SES inequalities in health vary for men and women? For the purposes of the first aim we use 7 measures, representing a spectrum of physical and psychological health. For the second aim, we focus on one particular measure, self-rated health at age 33, for which we have previously identified factors contributing to social inequalities (Power et al., 1991, Power et al., 1998). Self-rated health was selected for this more detailed analysis because there is now extensive evidence that it predicts mortality, morbidity and has been shown to be associated with fitness (Kaplan and Camacho, 1983; Idler and Angel, 1990; Wannamethee and Shaper, 1991; Cox et al., 1993; Appels et al., 1996; Moller et al., 1996), the latter two being especially useful in view of the cohort's age.
Section snippets
Study sample
The 1958 birth cohort includes all children born in England, Wales and Scotland during one week in March 1958. The study originated in the Perinatal Mortality Study whose aim was to determine the social and obstetric factors associated with stillbirth and death in early infancy. Information was collected on 98% of births totalling 17 414. Five follow up studies have been conducted at ages 7, 11, 16, 23 and most recently in 1991 at age 33, with 11 405 subjects included in this latest sweep (
Results
Table 2 shows 6 health measures at the 2 ages 23 and 33 yr, for men and women separately. In general, prevalence differences are modest despite achieving statistical significance for most health measures. The exceptions are psychological distress and obesity, with substantial excesses among women. No gender differences were evident at age 33 for poor-rated health, limiting long-standing illness and asthma/wheezing. Table 3 shows the odds ratios (classes IV and V versus I and II) for the 7
Discussion
A recent review by Macintyre and Hunt (1997)highlighted the tendency within health inequalities research for investigators to either standardise for gender or to focus solely on male populations. Where separate analyses for men and women have been carried out, systematic comparisons between men and women are rare. Studies either examine gender differences ignoring SES or focus on SES inequality without considering gender. There have, however, been several exceptions in which gender, SES and
Acknowledgements
This research was supported by a Grant from the (UK) Economic and Social Research Council under the Health Variations Programme (L128251021). CP is a Weston Fellow with the Canadian Institute for Advanced Research. Data acknowledgment: City University Social Statistics Research Unit, National Child Development Study Composite File including selected Perinatal Data and sweeps one to five [computer file]. National Birthday Trust Fund, National Children's Bureau, City University Social Statistics
References (70)
- et al.
Self rated health and mortality in a Lithuanian and Dutch population
Soc. Sci. Med.
(1996) Comparing inequalities in women's and men's health: Britain in the 1990s
Soc. Sci. Med.
(1997)- et al.
The shape of the relationship between income and mortality in the United States: evidence from the national longitudinal mortality study
Ann. Epidemiol.
(1996) - et al.
Social patterning of medical mortality in youth and early adulthood
Soc. Sci. Med.
(1994) Social inequality in health: the role of the healthy worker effect
Soc. Sci. Med.
(1993)- et al.
Educational differentials in mortality: United States, 1979–85
Soc. Sci. Med.
(1996) - et al.
Patterns of class inequality in health through the lifespan: class gradients at 15, 35, and 55 years in the west coast of Scotland
Soc. Sci. Med.
(1994) - et al.
Male and female health problems in general practice: the differential impact of social position and social roles
Soc. Sci. Med.
(1995) - et al.
Why are socioeconomic mortality differences smaller among women than among men?
Soc. Sci. Med.
(1994) - et al.
Comparisons of inequalities in health: evidence from national surveys in Finland, Norway and Sweden
Soc. Sci. Med.
(1994)