This is the third in a set of six planned essays. Together,
they will document the back-story behind the anti-car policies, which have
plagued the people of the UK, under governments of all parties, for the last 30
years and more.
In the first essay, I looked at the most significant air
pollution episodes, other than leaks of specific chemicals, which have demonstrably
had significant negative health effects on large numbers of people. In the second,
I told the back-story on air pollution politics, from the beginnings of the
green agenda in the early 1970s through to the year 2008.
Today, I will look at a report produced in 2009 by the UK
government’s Committee on the Medical Effects of Air Pollution (COMEAP). It constituted
a watershed, by laying an apparently scientific foundation for the development
of policies to control particulate matter (PM) pollution in the UK.
I do not claim to be an expert on toxicology. I am a
mathematically and scientifically trained generalist, who has a nose for hard
evidence and data, and a strong sense of rightness and wrongness, both
technically and ethically. I am also convinced, by other evidence I have seen, that
on a closely related subject, climate change, groupthink among “scientists” and
experts has fuelled catastrophic scares that, viewed through the lens of hard
evidence, are at the very least grossly exaggerated, and likely totally
illusory.
Moreover, I have found myself, over the years, having to
become something of a philosopher. And the philosophy of science is one of the
areas I have looked at. Indeed, an article I wrote in 2018 describing the
scientific method was re-published at “the world’s most viewed web site on
global warming and climate change.” [[i]].
This essay will be a little more technical than the
others. But I do not apologize for this. The nature of the subject matter means
that a dive into science is unavoidable.
What COMEAP did
COMEAP looked at “risk coefficients” which had been
estimated in air pollution studies, trying to assess which of these were most
appropriate to be used in the UK. The definition of risk coefficient is: “A
quantity expressing the increase in risk per unit of exposure or per unit dose.”
Thus, at a particular level of air pollution exposure, the risk coefficient
gives the relative (or percentage) increase in risk resulting from a one unit
rise in the level of exposure. It can be quoted either as a “relative risk” multiplier,
or as a percentage change. Thus, a risk coefficient of 6 per cent per unit is
equivalent to a relative risk of 1.06 per unit.
It was in 2017 that I first looked at the 2009 COMEAP
report: [[ii]].
Its title is “Long-Term Exposure to Air Pollution: Effect on Mortality.” In
that exercise, my interest was in calculating the social costs of pollution
from cars in London. I did this in order to compare these costs with the fees
to be charged by the then forthcoming (starting in 2019) ULEZ (Ultra Low
Emission Zone). My resulting paper is here: [[iii]].
I did, at the time, comment on a few oddities in the way the
report treated its subject. But I had no notion that there might be anything
amiss, either with the process through which the report was produced, or with
its conclusions. Now, I already knew about the scientific shenanigans, which
had been going on in UK government circles over the climate change issue. Climategate
whitewashes, BBC mis-reporting, interfering with the conclusions of an early
IPCC (UN Intergovernmental Panel on Climate Change) assessment report, blocking
of any possibility of objective cost-benefit analysis on policies, and more. Despite
this, I assumed, naïvely,
that the apparently separate issue of air pollution ought to have been addressed
honestly and without bias.
But since then, I have learned about the conduct of
government scientists and advisors involved in assessment of air pollution,
including a chairman of COMEAP. This has caused me to take another, far more
critical, look at this 2009 COMEAP report. I also examined several later
follow-up reports. I shall recount in the next two essays some of the findings
and incidents, which led me towards the much more cynical view which I now hold
on the matter.
Things that weren’t quite right
Even back in 2017, I had noticed a number of things about
the 2009 report which didn’t look right.
Burden of pollution, or rate of change?
First, the focus was not, as I would have expected, on how
much damage is caused by a particular level of pollution (the burden). After
all, that is what you need in order to work out whether or not the problem is
big enough to need action. Instead, the focus was on how the damage changes
when you reduce (or increase) the level of pollution. That is, the slope, or
rate of change, of the damage function at a particular pollutant concentration.
It seemed, to the mathematician in me, a bit odd to choose
to estimate, not the quantity you ought to be wanting to assess across the
spectrum of pollutant concentrations, but its first derivative. At the time, I
described this as “cart before horse.” Perhaps, I thought, the policy cart
(reducing emissions) might have been pushing the evidence horse (finding the actual
burden of a given level of air pollution)?
To have assessed the burden, rather than the risk
coefficient, may well have been a more difficult set of calculations than the
ones they did in 2009. But it was certainly feasible. The subsequent 2010
report, indeed, did it! Also, unless the burden is near perfectly linear in the
pollutant concentration over a large range, the risk coefficient will vary
depending on what is the concentration at the point of interest.
Expert elicitation
Second, I was flummoxed by the “expert elicitation” exercise,
which COMEAP undertook to try to put error bounds around their central
estimate. Seven experts waved wet fingers in the air, and estimated how likely
they thought it was that the true risk coefficient lay in different ranges.
These guesstimates were then pooled, and the results expressed as a 75%
confidence range. That is, a range within which there are three chances in four
that the value lies. And even this had a factor of 12 between the lower and
upper bounds (1 to 12 per cent)! To those used to 95% confidence limits and
fairly tight error ranges, this was unimpressive, to say the least.
Particularly as evidence given by experts to support the imposition of policies
on millions of people.
Nor was there much correlation between the estimates of
different experts, let alone anything like a consensus. Indeed, one expert gave
every percentage from 0 to 17 almost the same chance. In other words, he was
saying “I have no idea at all what the true risk value may be”. Was he really
an expert, then? Another reckoned there was a half chance of the risk
coefficient being less than 1 per cent; only one-sixth of the pooled best
estimate.
Adjustment for nitrogen oxides
Third, the
attempts to combine separate risk estimates for PM2.5 and nitrogen
oxides into a single figure did not appear to me to be very successful. I found
myself having to use a value for the overlap factor, that was derived from a much
later report, from 2016. And an alarmist one, at that. This subject was to rear
its head again, in a big way, in a 2018 COMEAP report.
More things that aren’t quite right
When I re-read the 2009 report with my new, more critical
attitude, I found some further oddities.
The UN and its WHO
My first observation was a general one. The report took a
lot of account – a lot! – of the recommendations of the United Nations World
Health Organization (WHO). Which had “proposed that the relative-risk
coefficient for all-causes of mortality for the average exposure level reported
in the extension of the ACS study (Pope et al, 2002) should be used.”
And COMEAP ended up using precisely this value for their headline estimate.
“Our recommendation[s] for the individual coefficient[s] that express the
relative risk[s] associated with a
The WHO, being part of the UN, is a
political organization. It is both unelected and unaccountable. As everyone
should know, the UN has been the driving force behind the green agenda since
the early 1970s. Its Rio Earth Summit of 1992 has led to a long series of ever
tightening green regulations, that bit by bit have impoverished us and taken
away our freedoms. And at its Sustainable Development Summit in 2015, our
supposed representatives signed up to the UN’s “Sustainable Development Goals.”
Which I have described as “a blueprint for the destruction of human
civilization as we know it, and for tyranny by a self-appointed global ruling
class over every human being alive.” [[iv]]
The UN had already become activist in the
field of air pollution well before 2009. Notably, in 2005 the WHO updated its
air quality guidelines to specify a guideline concentration for PM2.5
of 10 micrograms per cubic metre. This was an order of magnitude lower than
previous limits. Even than the EU’s limit of 25 µg/m3.
The WHO has also acquired a very negative
reputation in recent years, as a result of its poor handling of the COVID
pandemic, and its subsequent political moves to take unilateral control of
world-wide pandemic responses. So, I became suspicious. Might the decision by
COMEAP back in 2009, to follow the WHO’s recommendation closely, have been a political
one rather than a scientific one? Was the WHO, in actuality, the driver of air
pollution policies in the UK?
Long-term or episodic exposures?
My second observation was, this report is looking only at
long-term exposure to pollution. In the foreword, they said that their earlier
work “has allowed estimates of reducing levels of air pollutants to be included
in the Air Quality Strategy for the UK.” And yet, the factual evidence,
including that from the Great Smog of 1952, tells us that all the most serious
pollution events have involved a large, transient spike in pollutant
concentrations. In other words, a large episodic exposure, rather than a smaller
long-term one.
This raised in my mind the question, is air quality policy
actually being driven by the evidence? Or might a politically desired policy be
pushing the experts towards results that support that policy? Could this be a
case of, in the words of former MP Peter Lilley, “policy-based evidence?” That
would certainly be consistent with the WHO being the policy driver.
PM2.5 alone, or PM2.5 with sulphates?
My third observation was that they chose PM2.5
as “the most satisfactory index of particulate air pollution for quantitative
assessments of the effects of policy interventions.” But all the most serious
pollution events have involved other pollutants beyond PM2.5.
Indeed, virtually all have involved both PM2.5 and sulphates.
Yet COMEAP say: “We have looked in detail at the case for
treating particulate air pollution measured as ‘sulphate’ differently from PM2.5
but have not been convinced that the available evidence warrants and enables
this.” This rather evasive statement made me concerned that COMEAP might be
hiding something.
They even say: “The interpretation of the associations found
with sulphur dioxide is not clear cut.” Even though the Health Effects
Institute (HEI), in an earlier review, had said that adjustment for sulphur
dioxide (SO2) did reduce the relative risks substantially.
The report also mentions studies from Dublin and Hong Kong,
on the effects of sudden reductions in pollutant concentrations on mortality
rates. “In the Dublin study, both particles and SO2 declined; in the
Hong Kong study, the main effect was on sulphur dioxide.” Both “confirmed that noteworthy
and sustained reductions in death rates can occur soon after a major reduction
in air pollution.” This suggests that SO2 is a major contributor to
mortality, and its toxic effects may well be more from short-term episodes than
long-term exposures. This does, indeed, tally with the evidence from historical
“smogs” and “hazes,” which I looked at in the first essay of this set.
When the risk coefficients from the earlier studies are
presented together in a table, they are all over the place. Some, notably James
Enstrom’s 2005 California Cancer Prevention Study, give all-cause risk
percentages an order of magnitude lower than the 6 per cent recommended. Others
give figures almost three times higher. And generally, studies conducted over
smaller geographical areas tend to give higher risk coefficients than larger
ones. These together suggest that, contrary to COMEAP’s focus on PM2.5
alone, other local toxic pollutants may well have more effect on mortality than
the PM2.5 which may, or may not, carry them into the lungs.
The idea that “all PM2.5 is equal” in toxicity
does not seem to me to be anywhere near proven. Indeed, historical evidence, notably
that from the 1952 Great Smog of London, strongly suggests that PM2.5
and sulphates together are far more toxic than either of them alone.
There are several working papers included in the report, the
first four of which treat the subject of sulphates. Working Papers 1 and 2 did
not draw any conclusions. Working Paper 3 reviews much relevant literature, but
the impression it gave me was that the author considered the state of knowledge
on the matter to be simply not good enough to reach any objective conclusion on
the interactions between PM2.5 and sulphur compounds. That is not a good
reflection on the science of toxicology, considering how much importance was now
being given to a problem which had been known about since 1953!
Working Paper 4, however, concluded that: “There is
reasonably consistent evidence that sulphate concentration in particulate
matter is positively associated with adverse health effects in short-term
exposure studies… and long-term exposure studies.” “This suggests that sources
of particles that are related to sulphur-containing fuel combustion may have
adverse health effects.” Of course, we already know that this is so – and the
evidence goes right back to the Great Smog of 1952. Not to mention the ban on
coal with a high sulphur content in Dublin, and the introduction of ultra-low
sulphur diesel in the UK.
So, could COMEAP’s failure to consider adjusting for sulphates,
perhaps, be a cop-out? A political ruse, to make the effects of PM2.5
alone seem worse than they really are?
Confidence intervals
My fourth observation was that the
recommended coefficient value is followed by “with 95% confidence interval (CI)
1.02 to 1.11.” These are exactly the error bounds given by Pope et al,
and thus recommended by the WHO. This is curious, considering that the expert
elicitation exercise had produced only a “75% plausibility interval” from 1.01
to 1.12 (1 to 12 per cent). I am surprised that no-one on the committee caught
this one out. Or perhaps someone did, and was ignored or shouted down.
A
flashback to 2006
My fifth observation relates to a copy of a COMEAP interim
report dating from early 2006, which is included in the 2009 report. It
contains a number of interesting statements. “The identity of the individual
components of the ambient aerosol that affect health remain unknown.” “It is
not possible to distinguish with confidence between the effects of the
different components of the mixture.” “In the absence of clear evidence to the
contrary, therefore, we consider that the coefficient should apply equally to
all components of the mixture, including sulphates.” “PM2.5 is our
chosen index of pollution.” “Calculations should focus on the benefits likely
to be delivered by changes in PM2.5 rather than on estimating the
total impact on health of current PM2.5.”
This tells me that the general direction of this 2009 COMEAP
report had already been set as early as the beginning of 2006. That they had
already made the decision not to look in detail at the effects of combinations
of PM2.5 and sulphates. And that the decision to look at the effects
of reductions, rather than doing a burden calculation in order to work out the
magnitude of the problem, had already been taken. That makes me suspicious that
the entire process might not have been honest.
A possibility of
groupthink?
Now, the majority of COMEAP participants are academics, and
the rest are government employees. In such environments, it is very easy for
“groupthink” to arise among committee members, and among groups of co-workers.
Particularly if some participants, and most of all if they are senior, have
activist views, and may be able to influence others towards their points of
view. It is also likely that, particularly in the early days, the small number
of people working on the subject of air pollution toxicology (most of them in
the USA) would have led them, too, to become prone to groupthink.
Indeed, such groupthink has been very much in evidence among
those working on a subject not so far removed, climate change. And this has led
to activist “scientists,” such as Michael Mann, fudging the scientific evidence
in order to favour their desired policies.
Among the COMEAP experts who produced this 2009 report was
Professor Stephen Holgate of Southampton University. Holgate was later chair of
a working group, which in 2016 produced for the Royal College of Physicians an
extremely alarmist report on air pollution. He was knighted in 2020 for
“services to medical research.” He is also a “Clean Air Champion” for the Clean
Air Programme, which by its own description is “proactively tackling the air
quality challenges facing us all today.” There is no doubt on where Professor
Holgate sits with regard to the politics of air pollution. I cannot judge
whether he influenced other COMEAP members towards his own positions, but would
guess that is not unlikely.
Peer review
Now, there was a peer review, which took place on an earlier
(2007) draft of this report. The reviewers’ comments are in the report. The
peer reviewers made some excellent comments. For example, Reviewer 1 asked: “If
there really is a direct effect of SO2, what will be the effect of
ignoring this when estimating the impact of reducing PM2.5?” And: “The
evidence as presented… certainly supports an effect of PM2.5 on
mortality, but I do not see how it supports translation specifically of the ACS
coefficients.”
Reviewer 2 said: “I would suggest one be very careful using
the guesses of experts as the basis of policy decisions.” “How can one really
think ammonium sulphate or ammonium nitrate will start a catastrophic chain of
events leading to death?” And: “There needs to be more creative thinking as to
the causal factors in particles or the whole aerosol with an emphasis on those
constituents that are likely to drive reactions.” I could not find anything in
the final report that specifically addressed these questions.
Reviewer 3, who was from the WHO, heaped praise on the
report. And said: “The conclusions of the report are fully in line with the
assessments conducted by the expert groups convened by WHO.” If the WHO was
driving the process, that’s what you’d expect.
Reviewer 4, too, was very positive. He had taken part in a
similar review panel in the USA, which included both Dockery and Pope, senior
authors on the ACS studies. And he seemed to be urging, if anything, an even
higher number. Though he did mention studies which “suggest that the more
recent exposures are biologically relevant.” So, perhaps episodic exposure really
is more important than long-term exposure?
Checking against
UK data
The statement, which led to my final observation, raised a
worry. “It would therefore be very presumptuous of us to use UK historical
mortality data to make strong judgements on the plausibility or otherwise of
possible coefficients for the effect of PM2.5 on mortality, based on
informal assessments with no adjustment for confounding factors.”
My immediate reaction was along the lines that the most
“presumptuous” thing any scientist can possibly do is to fail to check the
predictions of his hypothesis against reality. Instead of basing their
calculations on UK data, what COMEAP seem to have done is pick their desired
risk coefficient, then check that it is “plausible” in terms of historical UK
data. This seems to me, not so much cart before horse, as base over apex.
Working Paper 9 concluded that, using the WHO value of the
risk coefficient for PM2.5, the observed past decline in PM2.5,
after adjusting for smoking, accounted for only about a quarter of the observed
decline in mortality. That would suggest, to me at least, that the observed
decline in mortality might well have been more due to reductions in other pollutants
– such as SO2? – than to reductions in PM2.5.
To
sum up
I first read COMEAP’s 2009 report back in 2017, without
trying to judge its rightness or wrongness. On a second reading, far more critical
than before, I now have extremely low confidence, verging on none at all, in the
objectivity or accuracy of its results.
That the focus was on effects of pollution reductions,
rather than gauging the scale of the pollution problem as a whole, worried me
from the start. The expert elicitation exercise to determine confidence
intervals gave me no confidence at all. That the direction of the report, and
what it would study, had been set as early as the beginning of 2006, got me
concerned about the integrity of the scientific process which led to its
results. And that the whole exercise ended up slavishly following the WHO’s
recommendations – even in the confidence intervals! – raised a red flag for me.
In conclusion
All of these pieces of evidence combined to lead me to my
conclusion. This report was not an honest attempt to inform policy assessments
by quantifying the risks arising from PM2.5 pollution in the UK. It
was far more like an exercise in creating “evidence” to suit previously
determined policies. It was simply an excuse – a 180+ page excuse – for falling
into line behind the WHO’s position on the matter. It was not science, but
politics.
On this evidence, I must conclude that some, at least, of those paid out of tax money, that took part in the processes that led to this report, were behaving as enemies of the people they were supposed to be serving.
[[ii]]
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/304667/COMEAP_long_term_exposure_to_air_pollution.pdf
No comments:
Post a Comment