Infant mortality and income inequality

Yesterday, the Lancet published infant mortality rates for every major country in the world. In line with previous estimates, they showed that Britain did much better than the global average but slightly worse than most EU countries. The death rate for under-5s in the UK is 4.9 per 1,000 live births. Globally, the average is 44.0 per 1,000 live births and the Western European average is 3.9 per 1,000 live births.

This was initially reported in a matter-of-fact way by the BBC, but when I looked at the report again in the afternoon it had changed substantially. I am quite sure of this because when I first saw the headline I vaguely wondered if the Beeb would do its usual thing of trying to blame everything on smoking, drinking and/or obesity, and I noticed that it hadn't. By the afternoon, however, it had. It had even put up a photo of a pregnant woman holding a pack of cigarettes (a plain pack, naturally). Furthermore, it had also introduced the fringe theory that inequality 'causes' infant mortality.

All of these additions arrived in the form of quotes from Dr Ingrid Wolfe of the Royal College of Paediatrics and Child Health, and the BBC has now given her an entire op-ed in which she expands on her opinions ('Why the UK has a high child death rate').

In reality, it's a combination of factors; but just because the causes are complex that shouldn't stop us looking for solutions.

This week the Royal College of Paediatrics and Child Health launched a report - Why Children Die - that dug deeper into what it is children die from and which of these deaths can be prevented.

Our overarching conclusion was that risk of child death disproportionally affects poorer families.

When you compare Britain to the rest of Europe, we are one of the most unequal societies, with a growing gap between rich and poor.

It is no coincidence that our child mortality rates are also the worst.

Our analysis of the causes of child deaths shows many deaths in under-fives are due to risky behaviours (such as smoking) during pregnancy, which is more common among women who are socially disadvantaged.

Drinking during pregnancy is another risk factor, as is children being around second-hand smoke and unsafe sleeping - all of which can contribute to premature death.

So what do we want to see?

The focus has to be on reducing the growing gap between rich and poor - put simply, countries that spend more on social protection have lower child mortality rates.

... The messages are stark; living in an unequal society raises the risk of children dying.

You will notice that Wolfe, having correctly noted that the "causes are complex", provides some very simple and predictable solutions: don't smoke, don't drink and let's reduce inequality.

On the first point, it seems clear that smoking and excessive drinking during pregnancy increase the risk of infant mortality. However, Wolfe offers no evidence to indicate that either behaviour is especially common in the UK, and smoking in general is less common in the UK than in most EU countries. So whilst these are two (of many) risk factors, they apply to all countries and are not decisive in answering the question of why infant mortality in this country exceeds the Western European average.

On the second point, Wolfe makes a basic, but very common, error in conflating income inequality with material deprivation. It is undeniable that rates of infant mortality are highest amongst low income groups. This, again, is true in all countries. But the level of inequality (which, contrary to her claim, is falling, not rising, in the UK) tells us nothing about how many people are materially deprived, nor does it tells us about their level of material deprivation. Nor, indeed, does the level of inequality tell us anything about how much money is spent on 'social protection'.

She puts it like this...

Our overarching conclusion was that risk of child death disproportionally affects poorer families.

When you compare Britain to the rest of Europe, we are one of the most unequal societies, with a growing gap [sic] between rich and poor.

It is no coincidence that our child mortality rates are also the worst.

This is a non sequitur. The prevalence and income of 'poorer families' is not related to the level of inequality. Income inequality is very low in a country like Namibia and very high in a country like Luxembourg. Neither statistic tells anything about poverty in those countries.

Wolfe may have made an honest category error in confusing the two measures. Lots of people do. Or she may genuinely believe that the economic variable of income inequality is causally linked to rates of infant mortality; that certainly seems to be what the reader is supposed to conclude. If so, she may have been influenced by The Spirit Level, which blames income inequality for almost everything. The authors of that book do not provide a single reference to support their claim that inequality 'causes' infant mortality and I have yet to meet anyone who can offer a plausible explanation for how it could cause a rare biological problem. Robert Waldmann suggested there might be a link in 1992, but his research has since concluded that the statistical association is actually the result of a third variable—healthcare spending (he explains it here).

When I give talks about The Spirit Level, I often say that the inequality hypothesis offers policy-makers a panacea. The Spirit Level suggests that almost every problem can be alleviated by reducing the income gap. This seems to good to be true and, like all panaceas, it is. I suggest that one of the dangers of swallowing The Spirit Level theory is that the real causes of health and social problems could be overlooked in the rush to adjust a single economic variable.

This is particularly true in the case of issues like infant mortality. Its causes are indeed complex (there is a whole chapter about it in The Spirit Level Delusion). Rates of infant mortality tend to be higher in the English speaking nations, lower in Europe and even lower in the richer Asian countries (including Singapore, which has a very unequal income distribution). Within the English speaking countries, rates are particularly high amongst certain ethnic groups (eg. Maoris, Aboriginals, African Americans, Pakistanis). And, as Wolfe correctly notes, rates are consistently highest amongst poorer groups.

I am convinced that much of the success of The Spirit Level results from people making the simple category error described above. If one wrongly assumes that rates of material deprivation are fixed to rates of income inequality, it is easy to conclude that reducing inequality will materially benefit the poor. This is simply incorrect. The two measures can easily go in different directions. Indeed, they have been going in different directions in Britain since 2008. Conversely, the living standards of the poor typically improve when income inequality grows. There is no correlation between income inequality and infant mortality over time.

Finally, it should be noted that the Lancet study itself does not mention smoking, drinking or inequality at any point. The points raised above are pure editorialising on the part of Wolfe and the BBC to make a political point from a non-political piece of research.