Air Pollution Deaths
How deadly is it?
Dear readers!
During my career as an air quality consultant, I have been faced with many forms of the same basic question; 'how deadly is air pollution?'
The answer normally references headline figures such as those given by the WHO, that claim 4.2 million people die each year around the world from outdoor air pollution, millions more from indoor air pollution.
But what do those headlines actually mean? Are there really millions of people dying every year from air pollution?
Can it be as big a killer as say road accidents, or smoking? After all, I don't know anyone who has (at least knowingly) died from 'air pollution', unlike, sadly, these other categories.
In this letter we will dig deeper into these claims.
You might find it helpful to refer back to my earlier letter to get an instant upgrade in air pollution science before reading this.
The industry-standard answer
Some years ago I attended an air quality conference at the Danish Embassy in London. In attendance were professionals from academia, local government, technology and consultancy firms.
During one of the breaks, after listening to a health expert cite alarming statistics about things like 'premature deaths', 'life-years lost' and so on, I asked a fellow attendee:
'but how do they know how many people die of air pollution?'
I was met with the industry bog-standard response, something along the lines of:
'they look at spikes in hospital admissions, and cross reference with days in which there are high levels of air pollution, and then work it out from that'
Nothing about clinical trials, randomized control tests, or the weaknesses of correlational epidemiological studies. Nothing about confidence intervals, probabilities or other statistical references. And I still don’t know anyone who has actually died of air pollution!
Feeling somewhat ashamed that people in the business of ‘air pollution’ -myself included - couldn’t come up with something better, I became very conscious of this central question. With heightened sensitivity, I started to pay particularly close attention when seeing such claims being made.
A strange choice of words
Reading air pollution headlines I would notice that words were often chosen that didn’t quite mean what was being implied. For example:
‘premature deaths..’
‘the equivalent of X deaths’
‘accelerated deaths’
and so on.
It turns out that directly claiming deaths from air pollution isn’t so straight forward, hence the sometimes strange choice of words.
Some statistical detective work
An often cited UK statistic is that air pollution causes 40,000 deaths each year. Where does this figure actually come from?
One level of investigation takes you to an often cited 2016 Report published by the Royal College of Physicians called 'Every Breath We Take'.
This report states that;
‘..around 40,000 deaths are attributable to exposure to outdoor air pollution..’
Does ‘attributable to’ mean ‘caused by’?
The report later uses the following language;
‘The annual mortality burden in the UK from exposure to outdoor air pollution is equivalent to around 40,000 deaths’.
Notice the use of 'annual mortality burden' and 'equivalent to'.
It doesn’t really sound like ‘caused by’ any more. If not ‘caused by’, then what?
I really recommend this article from the University Of Cambridge, which does some excellent detective work in tracking down the source of such claims, going deeper than the Royal College of Physicians paper, referencing work undertaken by the UK’s Committee on the Medical Effects of Air Pollutants (COMEAP).
And yes, you guessed it, the COMEAP reports themselves reference further studies undertaken principally in the UK and the US. It really is like tumbling down an air pollution study rabbit hole!
The University of Cambridge article explains how such death claims are in fact a statistical construct rather than an actual number of people who have ‘air pollution’ written on their death certificates. I will return to this later point below.
I will not attempt to describe all of the findings of the University of Cambridge article, but I will make the following points.
As you seemingly get passed from one study to the next , it is striking to me that none of them are based on actual ‘gold standard’ clinical and randomised control tests. Do they exist? Yes, although for obvious ethical reasons they are not undertaken very often. I will write more about such studies and their findings in a future letter.
The limitations and assumptions upon which these correlational studies are based tend to get diluted (or left out entirely) with each subsequent reference to it. So in the end you get a statistically constructed number, but with no appreciation of the level of confidence that should be placed on such values.
To illustrate this second point, one of the original studies upon which these claims are based analysed mortality across the US over a given time period and compared those that died with air pollution levels in the areas in which they lived. There are a few potential problems to be aware of with this type of approach, but perhaps most notable is that this study did not correct for the socio-economic status of the individuals concerned.
In other words, can you say for certain that air pollution was causing higher death rates observed, as opposed to say a lack of lifestyle choices associated with lower standards of living (diet, exercise, time with family, etc.)?
If lower socio-economic status also correlates with living in higher air pollution areas (which doesn’t seem unreasonable - lower cost housing might tend to be closer to highways, for example), then all subsequent bets are off as far as I’m concerned.
The problem with correlational studies are well illustrated by my favorite ‘ice cream sales vs shark attack’ analogy. There is of course a strong correlational relationship between the two, but we really shouldn't be holding ice cream salespeople responsible when too many people venture out for a bathe in shark infested waters on a hot day!
Does Air Pollution appear on death certificates?
Until quite recently, the answer was ‘no’. Perhaps never. This did change in December 2020, however, when, after a lengthy investigation, a coroners court in London ruled that air pollution was a cause in the tragic death of nine-year-old Ella Adoo-Kissi-Debrah.
Ella had suffered from severe asthma, and had been repeatedly admitted to hospital after suffering attacks. A pattern emerged whereby the asthma attacks tended to occur during the winter months, when air pollution levels were typically higher in the heavily congested part of London where she lived.
In his verdict, the coroner said the cause was "multi-factorial. It was down to both genes, and the environment".
It has also been acknowledged that despite this landmark ruling, it is still ‘unlikely’ that air pollution would start to appear on death certificates without similarly lengthy inquests.
Conclusions
Headline air pollution death figures are often statistical constructs. In such cases, you could not actually identify the claimed number of people who were killed by air pollution. In fact you probably couldn’t find any.
These statistical constructs are themselves based largely on correlational studies. With such studies, one must be careful of the ‘correlation vs causation’ trap, and it is important to contextualize such claims with confidence levels.
Clinical trials looking more directly at the impact of air pollution on human health are rare, but have been undertaken - more on this topic to follow.
There is only one case that I am aware of where ‘air pollution’ has been officially recorded as a (not ‘the’) cause of death, following years of investigation and inquiry.
A large number of health professionals, vindicated in some ways by the sad case of Ella Adoo-Kissi-Debrah, believe that air pollution can ‘play a role’ in the deaths of vulnerable people.
I hope you have found this interesting. As an air quality professional of almost two decades it is certainly a topic that, while often skimmed over by many, has always interested me!
-Tristan