Academia.eduAcademia.edu

Figure from:Socio-economic status, health and lifestyle

See full PDFdownloadDownload figure
Figure 1: Social Class Gradients for alternative models of SAH effects of education are almost identical to those obtained when excluding the lifestyle variables. Epidemiological studies have also examined the impact of lifestyle on the socia class gradient in SAH and other health indicators, including mortality and specific disease indicators such as heart disease (Borg and Kristensen(2000), Power et al.(1998), Marmot et al.(1996), Lynch et al.(1996)). Using Danish data from a random sample of individual: interviewed in 1990 and 1995, Borg and Kristensen used simple logit models to estimate the odds ratios for reporting a reduction in SAH in 1995 conditional on having good o1 very good SAH in 1990. They found that 17% of the odds ratio for the highest status occupational group relative to the lowest was explained by tobacco consumption and ar indicator for obesity, after controlling for age, gender and an indicator of illness. However. tobacco consumption led to only a 4% reduction.*4 Lynch et al.(1996) using the Kuopic Ischaemic Heart Disease Risk Factor Study data, found that 35% of the relative risk of all. cause mortality for the lowest income quintile relative to the highest could be attributec (conditioning only on age) to behavioural factors (smoking, alcohol consumption, anc physical activity).

Figure 1 Social Class Gradients for alternative models of SAH effects of education are almost identical to those obtained when excluding the lifestyle variables. Epidemiological studies have also examined the impact of lifestyle on the socia class gradient in SAH and other health indicators, including mortality and specific disease indicators such as heart disease (Borg and Kristensen(2000), Power et al.(1998), Marmot et al.(1996), Lynch et al.(1996)). Using Danish data from a random sample of individual: interviewed in 1990 and 1995, Borg and Kristensen used simple logit models to estimate the odds ratios for reporting a reduction in SAH in 1995 conditional on having good o1 very good SAH in 1990. They found that 17% of the odds ratio for the highest status occupational group relative to the lowest was explained by tobacco consumption and ar indicator for obesity, after controlling for age, gender and an indicator of illness. However. tobacco consumption led to only a 4% reduction.*4 Lynch et al.(1996) using the Kuopic Ischaemic Heart Disease Risk Factor Study data, found that 35% of the relative risk of all. cause mortality for the lowest income quintile relative to the highest could be attributec (conditioning only on age) to behavioural factors (smoking, alcohol consumption, anc physical activity).

Related Figures (8)

Table 2: Correlations for full recursive model estimated by MSL(R=50)
Table 2: Correlations for full recursive model estimated by MSL(R=50)
Table 1: Selected marginal effects for reduced form lifestyle models
Table 1: Selected marginal effects for reduced form lifestyle models
Table 3: Selected marginal effects for alternative models of SAH 6.2 SAH Equations We begin by considering the results of the model for SAH at HALS2 where lifestyle factors measured at HALS1 are excluded.”* The ‘excluded’ column of Table 3 presents marginal effects for the social class variables using a simple univariate probit model without the lifestyle variables. It is immediately apparent that a significant social class gradient in SAH exists conditional on education, marital status, ethnic status, type of area, gender, height, and a flexible function of age. For each social class, the probability of reporting excellent or good SAH is greater than for the class immediately below it. These differentials are such that those in the highest social class are significantly more likely to report excellent or good SAH relative to the baseline category, while those in the two lowest social classes are significantly more likely than the baseline category to report fair or worse health. This gradient is emphasized by the graphical illustration in Figure 1. A gradient is not clear for education, with only those with a degree or higher degree having a significantly different probability of reporting excellent or good health relative to those in the baseline category. While these results are indicative of a direct impact of socio-economic characteristics on health, this model omits the lifestyle variables. As reported in the previous section, there are significant effects of socio-economic status and other variables on lifestyle choices at HALS1. By including the lifestyle variables we are able to obtain estimates of the effects of HALS1 lifestyle choices on HALS2 SAH and examine how including lifestyle affects the estimates of the impact of social class and education. We begin by considering the results of the model for SAH at HALS2 where lifestyle factors
Table 3: Selected marginal effects for alternative models of SAH 6.2 SAH Equations We begin by considering the results of the model for SAH at HALS2 where lifestyle factors measured at HALS1 are excluded.”* The ‘excluded’ column of Table 3 presents marginal effects for the social class variables using a simple univariate probit model without the lifestyle variables. It is immediately apparent that a significant social class gradient in SAH exists conditional on education, marital status, ethnic status, type of area, gender, height, and a flexible function of age. For each social class, the probability of reporting excellent or good SAH is greater than for the class immediately below it. These differentials are such that those in the highest social class are significantly more likely to report excellent or good SAH relative to the baseline category, while those in the two lowest social classes are significantly more likely than the baseline category to report fair or worse health. This gradient is emphasized by the graphical illustration in Figure 1. A gradient is not clear for education, with only those with a degree or higher degree having a significantly different probability of reporting excellent or good health relative to those in the baseline category. While these results are indicative of a direct impact of socio-economic characteristics on health, this model omits the lifestyle variables. As reported in the previous section, there are significant effects of socio-economic status and other variables on lifestyle choices at HALS1. By including the lifestyle variables we are able to obtain estimates of the effects of HALS1 lifestyle choices on HALS2 SAH and examine how including lifestyle affects the estimates of the impact of social class and education. We begin by considering the results of the model for SAH at HALS2 where lifestyle factors
Table 4: Lifestyle marginal effects for alternative models of SAH by computing the percentage change in the difference between the marginal effects for the highest and lowest social classes and education groups. This measure is the percent- age change in the difference in the probabilities of reporting an excellent or good SAH comparing the highest and lowest social classes.?°
Table 4: Lifestyle marginal effects for alternative models of SAH by computing the percentage change in the difference between the marginal effects for the highest and lowest social classes and education groups. This measure is the percent- age change in the difference in the probabilities of reporting an excellent or good SAH comparing the highest and lowest social classes.?°
Table 5: Variable Definitions and Sample Statistics
Table 5: Variable Definitions and Sample Statistics
Table 6: Selected HALS1 Variable means by sub-sample defined by number of a priori ‘healthy’ behaviours
Table 6: Selected HALS1 Variable means by sub-sample defined by number of a priori ‘healthy’ behaviours
Table 8: Coefficient estimates for alternative estimators of SAH model(Standard errors in
Table 8: Coefficient estimates for alternative estimators of SAH model(Standard errors in