Sunday 8 December 2019

Can we attribute 36% of premature deaths to "socioeconomic inequality"?

The last issue of the Lancet before the election contains a study looking at the association between deprivation and health. It concluded that...

One in three premature deaths are attributable to socioeconomic inequality, making this our most important public health challenge. 

On Twitter, the study's lead author went further, saying that a third of premature deaths could be avoided if society was 'fairer'...

The counterfactual is, apparently, a 'fair country'...

The study used the Index of Multiple Deprivation which splits England into 32,844 areas and gives them scores based on income, employment, crime, education, health, housing and the environment, with income and employment having the heaviest weightings (between them they make up 45 per cent of the final score). The authors stripped out the health bit and calculated the number of premature deaths in each area.

In line with common sense and virtually all previous research, they found that the most 'socially deprived' areas had the highest mortality while the least deprived areas had the lowest. (For the sake of simplicity, I'll use the words 'poorest' and 'richest' in this post to describe these areas. It's a bit more complicated than that, but not enough the affect my argument.)

There's nothing surprising about the correlation between health and wealth. The problem is in how the authors have presented their findings.

Firstly, they make causal claims based on crude correlations. When challenged about this on Twitter, the authors denied it, but the study is unambiguous. It is titled 'Premature mortality attributable to socioeconomic inequality in England between 2003 and 2018' for a start, and the authors use the words 'attributable' and 'attributed' time and time again. For example...

35·6% (95% CI 35·3–35·9) of premature deaths were attributable to socioeconomic inequality

Inequalities in premature mortality are persistent despite reducing total mortality, and we estimate that one in three premature deaths can be attributed to socioeconomic position.  

Their evident belief that social deprivation is the cause of the excess mortality is underlined by their assertion that the deaths would be avoided if there was no social deprivation...

Every year tens of thousands of premature deaths in England alone could be avoided by closing socioeconomic inequalities in mortality.

The large proportion of premature deaths that could be avoided by eliminating socioeconomic differences in mortality is likely to exist across high-income countries, because inequalities in premature mortality are ubiquitous.  

The second problem is that they do not settle for merely claiming causation between excess mortality and social deprivation. They make the stronger claim that the excess mortality is caused by (sorry, 'attributed to') socioeconomic inequality. For example...

Years of life lost to inequality was calculated as the difference between the years lost due to death before age 75 years in each cohort and the corresponding least deprived cohort. 
This is a different proposition. There are some who believe that inequality per se lowers life expectancy (most notably, the authors of The Spirit Level). Under this hypothesis, the stress of living in an unequal society manifests itself in unhealthy behaviour and bad health outcomes. The evidence for this is weak, however, and the authors of the present study do not make that argument, nor does the study contain any data that could test that hypothesis.

Perhaps it is just sloppy language, but the difference between poverty causing premature deaths and inequality causing them is enormous. In the former case, premature deaths could be reduced by making people richer. In the latter case, premature deaths could be reduced by reducing the gap between rich and poor even if the poor do not get any richer (ie. by simply making the rich poorer). 

Thirdly, they use the the richest decile as their baseline, as shown below.

All the lower deciles have a higher rate of premature mortality than the richest decile, and the authors describe every death above this baseline as an 'excess death'.

That would be just about acceptable if this were an academic exercise to imagine what would happen if everybody in Britain lived in the richest parts of Britain, behaved like rich people and had the income of a rich person.

But it isn't. According to the lead author, the study tells us what would happen in a 'fairer society'. The clear implication, underlined in the study, is that most or all of these deaths could be avoided in practice.

The presumption is that there would no socioeconomic inequality in a 'fairer society'. The idea that fairness implies total equality of outcome is debatable, to put it mildly, but let's go with it for the sake of argument.

It is possible, in theory at least, to reduce most of the criteria in the Index of Multiple Deprivation to close to zero. We could - again, in theory - ensure that everybody has exactly the same income.

But there are simply not the resources available to ensure that everybody has a top income. The median income in the UK is £28,400. The mean income - which is what everybody would get (in theory) if we had full equality of outcome - is £34,200. Incomes in the top decile are around twice that. The country would need to be at least twice as rich for everybody to have the kind of income that the authors think is 'fair'.

This is obviously unrealistic, even if we accept the other wild assumptions in the study (such as people in rich areas having better health for no other reason than that they live in rich areas, and radical egalitarianism having no negative effect on the amount of wealth created). The authors' counterfactual is not a 'fair country' but a fantasy country. It is not a lack of political will that prevents every area of the country from being as wealthy as the richest areas. The problem is economic. With limited resources, you cannot raise nearly everybody up without bringing some people down.

It makes no sense to view the happy situation of the least 'socially deprived' as a state of nature which is only denied to other people because of inequality. If you used the poorer areas as the baseline, you could cite the lower mortality rates in the higher deciles as evidence that inequality saves thousands of lives (a point conceded by the lead author).

That would be provocative, fatuous and fairly pointless, but so is the conclusion of this study.

No comments: