May 3, 2021

expert reaction to study looking at short-term air pollution, cognitive performance and nonsteroidal anti-inflammatory drug use

The Veterans Affairs Normative Aging Study, published in Nature Aging, looks at short-term air pollution, cognitive performance and use of nonsteroidal anti-inflammatory drugs (NSAIDs).

Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This new piece of research raises some interesting questions, but some aspects of the way it was conducted mean that there’s a lot it can’t answer.  Many studies of possible harmful effects of air pollution have looked at long-term measures of air pollutants at the places where people live, and investigated how they might be related to aspects of health.  But there have also been concerns about shorter-term effects of peaks in air pollution.  There have been previous studies that found associations between high long-term pollutant levels, particularly of very fine particles in the air, and problems with cognitive function – that is, how well thinking and memory are working.  But there seems not to have been much study of associations between shorter-term particulate pollution and cognitive functioning, and that’s what this new research looks at.

“The researchers used data on a group of around 1,000 elderly white men living in the area around Boston, USA.  This group was original recruited to study things other than air pollution, which presumably explains why no women were included, and this study looked only at white men because the group happened to contain very few who were not white.  But that all does make it impossible to know whether similar results would be found in other parts of the world (or even of the USA), in women, or in people of different ethnicity.  The study participants provided data on several quantities, including two measures of their cognitive function – the Mini-Mental State Examination (MMSE), which is often used in screening for dementia, and an overall measure of cognitive function that they called the Global Cognitive Function score (GCF).  They also reported whether they had been taking nonsteroidal anti-inflammatory drugs (NSAIDs), a class of drugs including aspirin and ibuprofen.  These have various effects, including reducing pain and (as the name indicates) inflammation, but they can also help prevent blood clots.  Most of the participants in this study who took NSAIDs in fact took aspirin, and aspirin is quite commonly prescribed or recommended in older people to help prevent heart disease and strokes rather than for pain.  As well as all that, the researchers recorded the level of particulate air pollution (PM2.5, and also a measure of the concentration of black carbon in the atmosphere).  These levels were recorded for the day the participants attended to do the cognitive tests, and also at weekly intervals before that day up to four weeks before.  Then, the researchers analysed the data statistically to see if there were associations between the pollutant levels and the measures of cognitive functioning. It turned out that there were associations – the higher the level of particulates in the air in the weeks before testing, the worse the participants did on the cognitive tests.  Those associations weren’t very strong on the GCF, but they were clearer on the MMSE.  But what I found particularly interesting is that these associations between air pollution and cognition were different, depending on whether the participants were taking NSAIDs (that is, in most cases, aspirin).  The associations largely disappeared in people taking NSAIDs.

“One interpretation of these findings would be that higher levels of particulate pollution, a few weeks before testing, cause elderly men to do worse on cognitive tests, but that, if they take aspirin or perhaps other NSAIDs, this causes that effect of pollution to be partly or entirely avoided – but there’s no way that we can know from this study that that is how it works.  The main issue is that the study is observational – the researchers did not deliberately expose people to different pollution levels, they just recorded the pollution levels in the days and weeks before the participants were tested.  There are likely to be many differences between men tested on different days when the pollution levels were different, and indeed there will also be differences between the days other than the particulate pollution levels, for instance in aspects of the weather, or in levels of other air pollutants.  So the observed associations between the cognition measures and the particulate pollution might be cause, in whole or in part, by any of these things, and not by the PM2.5 pollution levels at all.  The researchers made statistical adjustments to allow for some of these other differences between participants and between days, but it’s simply not possible to be sure that everything has been dealt with fully in this sort of study, so we just can’t be sure of what is causing what.  The same points apply about the NSAIDs – the researchers knew who was taking NSAIDs, and indeed which NSAIDs they were taking, but there will be differences between participants who take or don’t take NSAIDs – if nothing else, because the men would have been prescribed NSAIDs for a reason, and that reason wouldn’t be so likely to apply to people not taking them.  Maybe that reason for prescription would be the real cause of any difference between those taking and no taking NSAIDs, rather than an effect of the NSAIDs themselves.  The researchers did carry out an extra statistical analysis to check for this possibility, which led to the same kind of results as their original analysis, but some doubt has to remain.

“One thing that I’d like to know from this study, and which can’t be answered from what was measured, is how long any short-term effects of pollutants on cognitive function actually last.  The study found practically no association between the pollution levels on the day a participant was tested, and their cognitive results on that day.  But there was an association with the pollution levels the week before the test, and the week before that, right up to four weeks before they were tested.  So how long before the cognitive tests do we have to go before the association goes away?  Another extra analysis done by the researchers, where they adjusted for the average particulate pollution level in the year before the cognitive tests were done, indicates that the association possibly does have a limited timespan, but how long?  The researchers did not look specifically at associations over periods longer than four weeks.  That’s another indication that there are possible important things that this study simply can’t tell us.”

‘Short-term air pollution, cognitive performance and nonsteroidal anti-inflammatory drug use in the Veterans Affairs Normative Aging Study’ by Xu Gao et al. was published in Nature Aging at 16:00 UK time on Monday 3 May 2021.

DOI: 10.1038/s43587-021-00060-4

Declared interests

Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee.  However, my quote above is in my capacity as an independent professional statistician.”