Socioeconomic inequalities in health

Socioeconomic differences in health have been described since the 16th and 17th centuries,1 2 but only recently has reducing them been central to public health policy in many Western countries.3 Over the past three decades, epidemiological studies have confirmed the existence of socioeconomic inequalities in a range of health outcomes, including premature mortality, cardiovascular disease, obesity, diabetes, self reported ill health, and smoking related cancers, and have explored potential mechanisms linking lower socioeconomic position to poorer health.4 The Whitehall cohort studies have made important contributions to this literature.5

Several studies,6 7 8 9 including a publication from Whitehall II,10 have found that poorer socioeconomic position is associated with worse morbidity, mortality, and self reported health in older people. In this week's BMJ, a new analysis of data from Whitehall II by Chandola and colleagues examines the extent to which socioeconomic inequalities in self reported physical and mental health continue into older age.11 The paper adds to the literature by using repeated measures of socioeconomic position and self reported health, both of which may change with age. The paper demonstrates one of the strengths of prospective cohort studies—the ability to examine changing relations between health related characteristics over time.

Three key messages emerge: firstly, self reported physical health declines with age in all groups (women and men, people who are retired and those who continue work, and people in all employment grades); secondly, in contrast, self reported mental health increases in all groups; and thirdly the rate of decline in physical health with age is greater in those from lower employment grades than those from higher employment grades, which results in a widening of health inequalities with age.11

The authors focus specifically on socioeconomic inequalities. But their repeated measurements and detailed analyses allow other inequalities to be explored. Figure 2 in their paper shows the trajectories of health change with age by occupational grade for the final phase (2002-4) of the study. However, the authors do not highlight that the interactions of age with time period included in their statistical model suggest that these trajectories changed over time. We calculated the trajectories of physical and mental health for each time period that the study covered using data from the full results of model I, presented in the appendix to the paper (figs 1⇓ and 2⇓).11 We found that in the first period (1991-3) physical health did not decline with increasing age, and during the rest of the 1990s the decline in self reported physical health with age was much less pronounced than that seen since 2000 (fig 1⇓). With respect to self reported mental health, in the early 1990s the increase with age was more noticeably linear—continuing to increase into later older age—than in more recent years, where at older age the improvement in mental health flattens off (fig 2⇓). The differing impressions given by trajectories in the different periods are a reminder of how difficult it can be to summarise the results of complex statistical models in a transparent way.

Fig 1 Trajectories of age related short form 36 (SF-36) physical health in Whitehall II cohort, predicted from model I of Chandola and colleagues11

• Download figure
• Open in new tab
• Download powerpoint

Fig 2 Trajectories of age related mental health component in Whitehall II cohort, predicted from model I of Chandola and colleagues11

• Download figure
• Open in new tab
• Download powerpoint

These findings suggest that people in recent years perceive a greater decline in their physical health and a smaller improvement in their mental health as they age than people did a decade ago. Reasons underlying this cannot be determined from the data presented, but continued reporting in the media of the “burden” of an older population, together with changing roles of the family and society, and changing attitudes in society towards care for older people might be important.

The results of the statistical model also show that sex is the strongest predictor of physical health; the physical scores of the women in the reference group were, on average, 2.65 points lower than those for the men in that group. This compares to a difference of 1.60 points between the lowest and highest employment grades in this group. Women also reported worse mental health (difference of 1.96 points on the mental health score). As the authors report no evidence of statistical interaction between sex and age, the results suggest that the sex differences found in the reference age persist as people get older.

In summary, the full model results suggest that socioeconomic inequalities in self reported health persist and possibly widen with age, that the relation between age and self reported health changes over time, and that women have worse self reported health than men at all ages and time points.

The implications of the findings for public health are uncertain because the meaning of differences of this size in self reported physical or mental health is unclear. A difference of 1 in the short form 36 (SF-36) score probably corresponds to 0.05-0.07 of a standard deviation: in previous UK based studies the standard deviation has ranged from 15-20, with similar means to those published in table 1 of the paper.11 Quantifying similar trajectories for objective health outcomes (such as blood pressure, fasting and postload glucose, lipid values, incident diabetes, and cardiovascular disease) that have a clearer meaning to clinicians, public health practitioners, and the public, and exploring how these change with socioeconomic position, age, and sex over different time periods, is something that Whitehall II can do and that we look forward to seeing.