select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
A US study published in PLOS Medicine looks at air pollution as a risk factor for Alzheimer’s Disease.
Prof Matt Loxham, Professor of Respiratory Biology & Toxicology, University of Southampton, said:
“The press release is a good reflection of the work. The paper looked at fine particulate matter (PM2.5) concentrations specifically, rather than air pollution more generally through, for example, studying other pollutants as well.
“Because several of these pollutants are often well correlated with each other, it is not possible from this study to know whether any association, if causal, would be driven by PM2.5, or another likely-correlated pollutant such as NO2.
“Nonetheless, the evidence here for (1) an association between PM2.5 concentrations and Alzheimer’s disease incidence, and (2) there being an increased association in people who have experienced a stroke, is good.
“While the increased risk association with PM2.5 exposure was greater for people who had previously suffered stroke, there was nonetheless a significant PM2.5-associated risk in those who had not. Therefore, saying “individuals with a history of stroke may be especially susceptible to the harmful effects of air pollution on brain health” potentially overemphasises the extent of the increased association following stroke, and may also underemphasise the association in individuals without history of stroke. Also, the term “brain health” potentially over-extends the findings of this study, given that Alzheimer’s disease is only one facet of brain health (albeit one of clear importance).
“One of the strengths of a study like this is the size of the cohort – starting with over 27 million individuals gives the researchers the opportunity to delve deeper than they might be able to with a much smaller population. For example, here, not all of the cohort developed Alzheimer’s disease, and still smaller portions had Alzheimer’s disease diagnoses with prior history of hypertension, stroke, or depression. Starting with such a large cohort gives the statistical power to analyse these smaller groups.
“However, as a downside, studies using cohorts of this size often use estimations of pollution concentration which are inexact and do not necessarily represent exposure well, especially at fine spatial scale, although in this case the performance of the method of PM2.5 concentration modelling has previously been demonstrated. Furthermore, while the authors have attempted to account for confounding factors, such as smoking, body mass index, and various measures of socioeconomic deprivation, all of which may affect the relationship between air pollution exposure and health outcomes, data on several of these confounders were not available on an individual level, but had to be taken as an average over larger geographical areas, potentially resulting in some generalisation. However, these are not uncommon trade-offs when cohorts of this size are used.
“The study is not able to say what the direct pathways are which might allow PM2.5 to cause Alzheimer’s disease. However, given that there is very good evidence for PM2.5 having detrimental effects on the cardiovascular system, it is not unreasonable to hypothesise that there may also be effects on the brain.
“When trying to understand how air pollution affects parts of the body away from the lungs, one of the key questions is how such an effect might occur. The two main hypothesised mechanisms are that PM2.5 causes inflammation in the lungs, with the inflammatory signals released from lung tissue entering the circulation and driving effects in other parts of the body AND/OR that some (likely rather small) portion of inhaled particles crosses into the blood supply where they may exert effects on the blood and blood vessels, as well as potentially other organs (in reality, there is evidence for both these mechanisms). In addition, it is possible that particles may enter the brain directly through nervous system links from the airways.
“The authors’ finding that prior stroke increased the PM2.5-associated risk of Alzheimer’s disease aligns with this. Following a stroke, damage to the blood vessels of the brain may increase the leakiness of the vessels (termed the blood-brain barrier), which could potentially facilitate entry of particles or inflammatory molecules into the brain. Alternatively, it is possible that the injury caused by the stroke may render the brain more susceptible to other effects of inhaled pollution.
“However, the mechanism through which any effect of PM2.5 on Alzheimer’s disease is causes is likely to be complex and multifactorial, as it is for PM-associated associated effects elsewhere in the body, such as the lungs, where we have a better understanding of how effects can occur.
“Indeed, it is notable in this study here that the majority of the association of PM2.5 with Alzheimer’s disease was seen even in individuals without previous history of stroke. It might be that the mechanisms through which effects occur are in place in otherwise healthy individuals, but just heightened, or acting on already injured tissue, following stroke.
“However, much more research across epidemiology and toxicology is required in order to answer key questions such as (1) how air pollution affects the brain, (2) what it is about air pollution which allows this to happen, (3) whether certain air pollutants are more able to do this, (4) whether there are risk factors which mean some people are more likely to be affected than others, (5) how these risk factors interact with exposure, and (6) the time scale of these effects relative the manifestation of the disease clinically.
“The main implication of this study should be a continued impetus to understand more about how air pollution affects health across the lifecourse. Alzheimer’s disease may manifest in old age but, like other diseases occurring later in life, there is evidence that the diagnosed disease represents the cumulative effect of longer-term exposures. As we continually understand more about the myriad ways in which air pollution may affect our health, and the breadth of the population who may be affected, it is becoming clearer that poor air quality exerts a great burden on the health of the world’s population, and that more research is needed to mitigate these effects.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, Open University, said:
“It’s worth noting that this study isn’t about potential effects of all types of air pollution. The researchers only considered levels of very fine particulates in the air (so-called PM2.5, or particles smaller than 2.5 micrometres in diameter), and not other pollutants like larger particles or ozone or nitrogen oxides. And they couldn’t take into account particulate pollution in the home or at work because they did not have the necessary data. So it can only be a partial picture of associations between dirty air and dementia – but one has to start somewhere.
“The press release begins by asserting that this study found that “People with greater exposure to air pollution face a higher risk of developing Alzheimer’s disease.” This needs some unpacking.
“The study did find that people with greater exposure to very fine particulate air pollution (PM2.5) from the air outside their homes were more likely to be diagnosed with Alzheimer’s disease (AD) – but there remain some unavoidable doubts about the extent to which the increased AD risk is caused by the pollution exposure.
“Briefly, the doubts arise because this is observational data. People living in places with different levels of PM2.5 also differ in terms of many other factors, and these other factors could make up all or part of what causes the higher AD risk in more polluted places. The researchers on the new study did make statistical adjustments for an impressively wide range of factors, to avoid them getting in the way of the patterns of cause and effect. But that can only be done for factors on which one has data, and one can never be sure that everything relevant has been measured and fully included.
“The researchers point out that some of the factors that they did take into account weren’t measured in an ideal way. For instance, data for smoking levels were not available for individuals, or even as averages for ZIP codes (US postal areas), but only at the level of whole counties. That’s clearly not all the relevant information about smoking.
“Further, even if the increased AD risk is caused to some extent by air pollution, it’s unclear how that cause and effect might work in detail.
“It’s a little odd that the press release points out this correlation between fine particulate pollution levels and AD risk so prominently. In broad terms it’s not a new finding, and I don’t think it’s the most important finding in the research anyway. I think the important findings are those about how some other chronic health conditions (hypertension or high blood pressure, stroke, and depression) might come into the pattern of cause and effect. I’ll come back to that.
“This isn’t the first study that has reported a correlation between levels of PM2.5 and AD risk. The researchers mention some of them in the discussion section of their paper. One of them (their reference 25) is to a study published in 2021 using the same data source as in the new research, which found a correlation between PM2.5 and AD risk of a similar size to the finding in the new study.
“They also mention two review papers, that include studies going back more than 10 years. In one of those review papers (their reference 40) that the new paper refers to as demonstrating a correlation between levels of PM2.5 and AD risk, the evidence for that correlation doesn’t reach the usual statistical level for accepting there really is an association. But that doesn’t mean that there can’t be any real correlation.
“The fact that some previous studies didn’t find clear evidence for the correlation is a reason for the new study to revisit this question, and anyway the researchers needed to check the position on a correlation between levels of PM2.5 and AD risk in their data to justify the rest of their data analysis.
“Putting all those previous findings together with other research and with the findings in the new paper, there is certainly a question to be investigated about levels of PM2.5 and AD risk.
“The difference in AD risk that was found between areas with lower and higher fine particulate pollution in this study is arguably not particularly large. But because poor air quality exists in many places, and can’t be avoided if you happen to live in such a place, and because AD and other dementias are unfortunately rather common, it’s still important to keep researching this association.
“The main focus of the paper, and of the new results it presents, is about a specific question on how fine particulate matter might cause differences in AD risk, if indeed this pattern of cause and effect is real.
“One possibility is that fine particulates could cause increases in risk of some other chronic diseases than dementia, and that higher prevalence of these other chronic diseases in areas with high air pollution causes, in turn, higher levels of AD in those areas.
“The researchers looked at this question for three different chronic diseases: hypertension (high blood pressure), stroke, and depression. All of these have been linked to AD risk in previous studies.
“First, they report on their investigation of whether the numerical measure of statistical association between levels of PM2.5 and AD risk was different depending on whether individuals had had one of these chronic diseases or not. For stroke, they did find a difference.
“For people who had had a stroke, the difference in AD risk between areas with high and low levels of PM2.5 was greater than it was for people who had not had a stroke. (Only a tiny amount greater, but it’s possible to estimate very tiny amounts when you are using data on nearly 30 million people.) This indicates that stroke may come into a pattern of cause and effect somehow. They did not find statistically firm evidence of a difference of this kind for hypertension or depression.
“This finding about stroke is the reason why the researchers concluded that “Individuals with a history of stroke may be especially susceptible to the harmful effects of air pollution on brain health.”
“In turn, though, this raises the question of how having had a stroke might work in this way. One possibility could be that living in a polluted area increases the risk of having a stroke, and having had a stroke then in turn increases the chance of having AD later (whatever the level of air pollution). If a lot of the pattern of cause and effect works like this, then that would throw some strong light on exactly how high PM2.5 levels might cause increased AD risk.
“But the findings I’ve just described can’t tell us that, because they are based on everyone who had a stroke while they were part of the dataset being used for the study, even though they might have had the stroke after entering the study but before they lived in a place with high PM2.5 levels. In such cases the high pollution level can’t have caused the stroke. So further data analysis is needed.
“For stroke, the researchers worked with a slightly different data set and different statistical methods. They used data from people who had a measure of air pollution where they lived when they joined the study, and had not had a stroke by the start of the study. Some of them had a stroke after that, and some (whether or not they had had a stroke) had a diagnosis of AD later still. They also excluded people who had moved to a different ZIP code, so possibly far enough to change the air pollution measure where they lived. But because the original data set was so big, there were still over 26 million people providing data.
“I’m not going to go into details of exactly how the researchers used this data, because it’s pretty complicated. Briefly, they used a statistical method called mediation analysis to investigate to what extent the association between PM2.5 and AD risk had gone, in terms of cause and effect, via first having a stroke (or not) and then later having AD (or not).
“What they found was that, for the possibility that the pattern of cause and effect goes from particulate levels to stroke and then to AD, only about 4% of AD cases were caused in that way rather than being caused in some other way that starts from particulate levels and then doesn’t involve stroke. That is, assuming for now that higher particulate pollution does cause increased AD risk somehow, some of the cases of AD in high-pollution areas do appear to arise because the person had a stroke because of the air pollution, but only a very few of them.
“The researchers also looked at hypertension and depression in a similar way, and found even lower measures of how much of the patterns of cause and effect might go through these other diseases than for strokes.
“I need to mention that the statistical models that produce the mediation analysis results can only tell us for certain what’s going on in terms of cause and effect if there are no factors, left out of the statistical models, that influence the patterns of cause and effect for three different correlations: between pollution levels and AD risk, between pollution levels and stroke risk, and between having a stroke and AD risk. But we can’t know for sure whether there are any such factors on which we don’t have data, or where they are not measured precisely enough.
“That’s the same issue as for the overall association between PM2.5 levels and AD risk, but it applies to more than one kind of correlation. Despite the care the researchers took to consider only the most relevant people, this raises considerable uncertainty about the detailed findings.
“All this really means is that the findings so far must have a degree of uncertainty, but that depends on data we don’t have, so really nothing definite can be said about it. I think the researchers on this new study did a thorough job of this statistical modelling, but it’s not possible to remove all the doubts.
“Overall, subject to these remaining doubts, these findings about hypertension, stroke and depression imply that the great majority of the extra cases of AD in highly polluted areas arise, not because the pollution caused these chronic diseases, but for some other reason that isn’t accounted for by the many other factors which the researchers included in their statistical models. They refer to this as a ‘direct effect’ of PM2.5, because it does not go indirectly through each of the three chronic diseases.
“But how the cause and effect might work instead is not something that was part of this research study. Maybe it involves other chronic diseases than the three the researchers examined. Maybe it involves the way that very fine particulate material can enter the body and affect the brain without resulting in a chronic disease before AD. The new research paper does mention some other possible routes whereby fine particulate pollution could cause AD, if indeed the pollution is involved in cause and effect, but it does not directly provide new data on these alternatives.
“In a way that’s rather a negative result, but negative results are just as useful as positive ones because they stop one going down routes that are likely to lead nowhere. It does, however, emphasise that there’s still a lot to learn about air pollution and dementias.”
Dr Mark Dallas, Associate Professor in Cellular Neuroscience, University of Reading, said:
“This long‑term study, which tracked 28 million adults, suggests that breathing tiny particles of air pollution may slightly increase the risk of Alzheimer’s disease. People who have had a stroke might be somewhat more sensitive to these effects, although the increase in risk remains small. Because the research relies on broad pollution estimates and medical records, there are important limitations, but the findings align with growing evidence that air pollution is a modifiable risk factor for dementia. Overall, the study reinforces a simple idea: what we breathe over many years can shape how our brains age.”
Dr Sheona Scales, Director of Research at Alzheimer’s Research UK said:
“This study strengthens the case that air pollution poses a genuine threat to brain health. Now researchers are trying to understand the interplay between air pollution and other known risk factors for dementia.
“Researchers explored whether existing health conditions—like stroke, high blood pressure, or depression—might add to the effects of air pollution. They found that people who had previously had a stroke seemed slightly more vulnerable to the effects of air pollution, while high blood pressure and depression did not account for most of the increased dementia risk. This suggests that polluted air may be affecting the brain more directly than we once thought.
“The implications for this US study are global. Air pollution is a challenge we need to tackle together. Setting stronger, health-based air quality targets and reducing exposure would help protect our brains as well as our lungs and could lower the overall risk of dementia.
“Here in the UK, the NHS 10-Year Health Plan recognises the harm polluted air can cause. However, current policy is still aiming to meet the World Health Organisation’s 2040 air quality targets, despite evidence suggesting these could be achieved by 2030. We should act sooner to protect the health of current and future generations.”
Prof Eef Hogervorst, Professor of Biological Psychology, and Director of Dementia Research, National Centre for Sports and Exercise Medicine, Loughborough University, said:
“A nationwide US study investigated air pollution exposure to fine particulate matter (PM2.5) due to burning of fossil fuels (but also other sources including fine dust) between 2000 and 2018. These fine particulates have been associated with cardiovascular disease, and pulmonary conditions, including cancer and asthma.
“In this study the outcome was Alzheimer’s disease (AD), the most common form of dementia. People who were included had not been diagnosed with AD for 5-years before the study started, so this was incident AD developed in 11% of participants during a median follow-up of 6 years.
“The study included 27,763,593 participants aged around 76. The majority were White (89%), female (58%), but most were ineligible for Medicaid (84%). The 5-year average PM2.5 concentration was 10 μg/m3. The WHO suggests this should be half that to avoid health-risks. Of these study population, 87% had hypertension, 23% stroke, and 35% were diagnosed with depression during the study. These co-morbidities are all risk factors for dementia and were also affected by pollution exposure in this study.
“The main finding was that associations of pollution were largely independent of having these co-morbidities, although people with stroke seemed slightly more vulnerable. This suggests that the association of this type of pollution could act via other dementia pathways, like promotion of misfolding of proteins, deposition of the toxic beta-amyloid plaques, oxidative stress, neuroinflammation or other unknown.
“It is not the first study to report this association, with several papers reporting an overall more than 3 times higher risk for AD with PM2-5 exposure. Swedish data showed similar (partly mediating) associations of having had a stroke, but also found increased vulnerability in people with co-morbid heart failure and ischemic heart disease.
“Regardless of the pathways that mediate this risk, as this environmental exposure has been linked to brain changes in children and young adults – which may be precursors for risk of AD – it is a serious reminder that this modifiable factor needs urgent attention to avoid future human and economic costs. This type of pollution is especially high in Africa, India and China, where dementia risk is also increasing at alarming rates. Indonesia with very high urban pollution was found to possibly have double (or even triple) the percentage of people over 65 suspected of having dementia, compared to European areas.
“There are a number of aspects in this study that require closer scrutiny and which are acknowledged by the authors. Firstly, AD diagnoses data came from the Medicare data base based on insurance claims. This was shown to have only moderate sensitivity (64%, meaning many AD cases are missed) but good specificity (95%, so when someone is included as having AD that is usually correct). The data also used postcodes rather than actual addresses of people which could have impacted both on actual outdoor pollution (which could vary from street to street, due to proximity of highways, hills, forests etc). Other risk factors, such as smoking and body mass index were also derived from areas (e.g. as % ever smokers at county level), rather than from individuals. Individual data pertained to age, sex, race, and Medicaid eligibility. This is crucial data to reflect poverty as a major confound.
“Previous studies found that being poor and having low Socioeconomic Status (SES) is often associated with having had less education and not (being able to) engage in many protective lifestyle factors (such as good sleep, nutrition, holidays, gym memberships etc), as well as living in poor and often polluted environments with high population density. This can cause chronic stress, but also a higher risk of spread of infectious disease, such as was seen during COVID-19 with disproportional death rates in low SES areas in people with dementia. In this study, a number of postcode-level average SES indicators were included. Again, there were no individual data on education or employment, but being Medicaid eligible which was used indicates having a low income after age 65 and/or having a substantial disability. Area-level SES variables included healthcare capacity indicators (number of hospitals in that area), population density, median household income, the percentage of Black individuals, people with less than a high school education, who are below the poverty line, and lived in renting accommodation. These covariates were selected as confounds for the association.
“Regardless of these methodological issues, pollution is a significant modifiable risk factor that should not be ignored for not only dementia, but also other high-cost morbidities, which can impact on disability, early death and independence.”
Prof Brian Castellani, Professor of Sociology and Co-Director, Wolfson Research Institute for Health and Wellbeing, Durham University, said:
“There is already a substantial and growing body of international evidence linking long-term exposure to fine particulate air pollution with increased risk of dementia. This very large national study does not establish the link for the first time, but it adds important statistical precision and reinforces what the field has increasingly been showing over the past decade.
“This finding fits with what UK and international research has already been showing: long-term exposure to fine particle air pollution is associated with higher dementia risk, including Alzheimer’s disease. That makes it relevant here. But this is US Medicare data, so we shouldn’t import the exact risk estimates into the UK — the direction of the effect likely translates, the size of it may not.
“There are already multiple UK reports identifying air pollution as a modifiable dementia risk factor, and UK Alzheimer’s and dementia organisations recognise this already. The implication is that brain health needs to be built directly into air-quality policy. That means reducing exposure now for older adults and for people already living with dementia, where pollution can worsen cognitive decline, while also protecting early cognitive development to strengthen long-term resilience. It requires embedding brain health into transport, planning and air-quality standards, strengthening action on traffic exposure and engine idling, and making the cognitive impacts of pollution part of routine public health messaging. This study adds further weight to that direction.”
‘The role of comorbidities in the associations between air pollution and Alzheimer’s disease: A national cohort study in the American Medicare population’ by Yanling Deng et al. was published in PLOS Medicine at 19:00 UK time UK time on Tuesday 17th February.
DOI: https://doi.org/10.1371/journal.pmed.1004912
Declared interests
Prof Matt Loxham: “No industry funding for either my own research or scientific meeting attendance – all funding for my current and previous work has been from either research councils, research charities, or local government. A post-doc previously in my group was funded by AXA Research Fund, although this work is now finished and the post-doc is no longer employed in my group. For scientific meetings, the same, although sometimes (if I’ve been invited to give a talk) travel costs, accommodation, and food will be covered by the conference organiser (which is normally a learned society e.g. British Thoracic Society, European Respiratory Society). Again, nothing from industry.
“I am a member of the Committee on the Medical Effects of Air Pollutants, which advises the government on all matters concerning the health effects of air pollutants (although I am commenting here independently, and these views are purely my own). This is voluntary and does not pay, although I am reimbursed (within limits) for travel costs and food. I have also been a member of a working group of the National Allergy Strategy in 2025. Again, entirely unpaid, and all meetings were online so no reimbursement of any kind was needed.
“There is no relevant prior employment for any company. I have worked for University of Southampton since my PhD in 2013.”
Prof Kevin McConway: “I have no conflicts of interest to declare.”
Dr Mark Dallas: Dr Dallas receives research funding from the MRC and the Carbon Monoxide Research Trust and is a member of the Alzheimers Research UK Thames Valley Network.
Dr Sheona Scales: “No declaration of interest”
Prof Eef Hogervorst: “I acted as a consultant for Proctor to investigate nutrients and memory.”
Prof Brian Castellani: “I do not have any conflicts of interest.”