Not in our stars but in ourselves

Why are poorer people unhealthier? A commonsense response would be because they behave more unhealthily. Their reduced life expectancy is the result of choices made, or forced on them, by their relative poverty.

But conventional wisdom, expressed most recently in the Marmot Review Fair Society, Healthy Lives, does not accept this explanation. “Serious health inequalities do not arise by chance, and cannot be attributed simply to genetic make-up, ‘bad’ unhealthy behaviour, or difficulties in access to medical care, important as those factors may be”, it says.
 
A principal reason why Professor Michael Marmot believes this is his own Whitehall Study, which compared the health of civil servants high and low. Consistently, those at the top of the tree had better health and longer lives than those at the bottom – and he could not account for these differences solely by differences in diet or lifestyle.
 
There seemed to be some other ingredient that linked social status with health status, independent of behaviour – possibly (one theory) the stress felt by those who are not in control of their own lives. Hence the view has grown that the only way to reduce health inequalities is to reduce social inequalities, a huge task. The Marmot Review is merely the latest exposition of this view, which has captured almost all policymakers in this field.
 
It was surprising, then, that a recent paper in the Journal of the American Medical Association, by a team that included some of Professor Marmot’s colleagues at the Department of Epidemiology and Public Health at University College London, did not attract more attention. This is essentially a re-analysis of the Whitehall Study, but one in which health behaviours were assessed not just at the start of the study but as it continued. The data covers 24 years in which many changes of behaviour may have taken place, so assessing those behaviours only at the start may underestimate their effect.
 
So it turns out, and the results are quite startling. Over the 24 years, the raw data show that those on the lowest rungs of the civil service ladder were 60 per cent more likely to have died than those at the top. When behaviours such as smoking, drinking, diet and physical activity measured at baseline were factored in, the effect of social position was weakened, but not eliminated. Poorer men and women were still 31 per cent more likely to die over the following 24 years.
 
But measurements of these factors were made not once, but four times during the course of the study. When all these sets of measurements were factored in, the gap in all-cause mortality fell to just 14 per cent. So smoking, diet, drinking and exercise can explain almost all the mortality difference between rich and poor.
 
The weakest effects were found for cancer, the commonest cause of death, where social class seems to have little direct influence. For heart disease, the poorer participants were 3.05 times more likely to have died than the richer ones, but factoring in smoking, drinking, diet and exercise measured throughout the study reduced this to 1.85.
 
The biggest effect was in causes of death other than heart disease or cancer, which accounted for almost a quarter of deaths. Here, correcting for the lifestyle factors virtually eliminated the difference between rich and poor.
 
Summing up, the team says that the four lifestyle factors chosen, when measured throughout the course of the study, account for 77 per cent of the difference in mortality between richest and poorest participants. The effect is greatest for deaths from causes other than cancer and heart disease, where it accounts for 94 per cent of the difference, and lowest for heart disease, where it accounts for 59 per cent of the difference.
 
So it appears that "bad, unhealthy behaviour" can indeed account for the bulk of the difference in mortality between rich and poor, in contrast to claims made in the Marmot Review. The authors do not quite say this, however. They say that the confidence intervals around their estimates are wide, and that even if their results are true, it remains possible that poverty, even if not itself the cause of health inequality, may be the cause of the unhealthy behaviours that are responsible.

Is it possible to eliminate those differences in behaviour without eliminating social differences altogether? Professor Marmot would say no, but others may disagree. Whatever the conclusion, the new study has rendered less plausible some of the more fanciful explanations for health inequalities.