select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
A study published in Thorax uses UK Biobank data to look at lifetime occupational exposures and risk of chronic obstructive pulmonary disease risk (COPD).
Dr Richard Parsons, Senior Lecturer in Biochemical Toxicology, King’s College London, said:
“This is a robust study that clearly shows an increased prevalence of COPD amongst those exposed to pesticides. It also shows that, the longer your exposure, the more at risk you are of developing COPD. A link between pesticides and disease has been hypothesised for other diseases, for example farmers exposed to organophosphates in sheep dip has long been linked to the development of Parkinson’s disease. Therefore, it is not a surprise that this paper reports this link between pesticide exposure and increased risk of COPD. What is concerning is that patients who you would consider as having “healthy lungs”, for example those who don’t smoke or don’t have asthma, show this trend. The concern I have is whether the risk higher for those who do smoke and who do have asthma? Do environmental pollution such as diesel particulates, NO from exhaust fumes, etc. increase the risk posed by pesticides? We have been seeing a decade-upon-decade rise in the number of cases of asthma, especially amongst children, which is linked to increased NO in the atmosphere (remember, the UK historically were fined for breaches of EU limits on atmospheric NO). It is fair to say that, as these patients already have damage to their lungs, they may be even more at risk from pesticide exposure. Unfortunately, this study doesn’t appear to test whether such pollutants increase the risk of COPD from pesticide exposure.
“What can the study tell us about non-occupational exposure to ordinary members of the public? The study shows a correlation between cumulative exposure and increased COPD risk. This is important, as pesticides are prevalent throughout our environment, even in the fly sprays we can readily purchase for use in the home. Therefore, we are all being exposed to low-level pesticides. Although we cannot make any definitive conclusions, I think it raises the possibility that non-occupational cumulative exposure to pesticides may increase our risk of developing COPD – although this study itself doesn’t provide any evidence on that.
“Ethnic minorities are underrepresented in this study. This needs to be addressed, as we need to ensure that our ethnic communities are not being disproportionately affected by exposure to pesticides, which is something we do see in other diseases and syndromes.”
Prof Andy Smith, Retired MRC Toxicology Unit and Honorary Professor at the University of Leicester, said:
“This is a study with Biobank UK data of individuals clinically assessed for COPD to find possible association with occupation history, particularly with pesticide usage. The findings are based on sound and detailed epidemiological methods with appropriate consideration for occupation category, co-exposures and confounders. A positive correlation with workers using pesticides was observed. Although the proportion of pesticide workers of all the occupations is small the large population of Biobank individuals is a strength of the study.
“As the authors agree, particular types of pesticides implicated were not available and so related quantitative exposure data to dissect out causal mechanisms was not possible.
“Present exposure of workers to agricultural and amenity pesticides is now greatly improved by both regulatory requirements and advisory information to users. Many pesticides that might be implicated from the findings have subsequently been withdrawn.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“This is a careful study, but it has some inevitable limitations that mean that care needs to be taken in interpreting the results. It follows up previous research1 from the same team, based on the same cohort of people and generally the same data, which investigated associations between people’s occupation and the chance that they would get COPD.
“One important point to understand is that, though the new research looks at associations between occupational exposures to various pollutants and COPD risk, the researchers had no data on how much of the pollutants each individual was exposed to at work. Instead they used a “job exposure matrix” which gives a measure of the typical level of exposure to each of the pollutants the researchers looked at, for workers in a particular occupation. In practice, of course, not all workers in a given occupation will have the same exposure to a given type of pollutant, but the researchers couldn’t take that directly into account. That might not matter much – but it could matter, particularly if the people in this study weren’t typical of the workers in the occupational groups involved. The UK Biobank cohort, which was used for this research, it an important and widely used data source on many aspects of public health, but the people in it were volunteers and in many ways not typical of the UK population. They are likely to be more interested in their own health than would be typical of the population; one indicator of that is that only about one in 18 of the people in this study were current smokers when they were recruited to the study, while in the whole British population of the same ages, at the same time, about one in five were current smokers.
“This is an observational study, and as the press release makes clear, that means that you can’t use the results to conclude that exposure to pesticides at work causes an increase in COPD. That’s pretty well always the position on observational studies, but I think it’s particularly clear here. The whole basis for being able to do this study at all was that different occupations have different typical levels of exposure to pollutants. But people in different occupations differ in a lot of other ways, apart from just their exposure to pollutants. Differences in COPD risk between people with different pollutant exposures could be caused, in part or entirely, by these other differences between occupational groups. (Remember that the study used no direct measure of pollution exposures, only measures via people’s occupations.) I’m not saying that exposures to pesticides can’t cause an increase in COPD risk – it’s certainly plausible that they might. I am saying only that a study like this can’t show that that’s how cause and effect work here.
“In looking at correlations and associations in an observational study, researchers can make statistical adjustments to the differences in risk that they find, to allow for some of the differences between the participants other than the ones they are interested in. That can’t get rid of the doubts about cause and effect, because you can never be sure that you have adjusted for everything relevant (and you can’t make statistical adjustments for factors that you don’t have data on). But it can clarify things to some extent. In this study the researchers made statistical adjustments for the centre where people were recruited for the study (which could take care of some geographical differences), their sex, their age, and a measure of how much they had smoked. In their main calculations for each pollutant group, they also adjusted for the levels of exposure to the other pollutants in the study, which makes sense in order to try and pick out the effects of each pollutant group separately. But, for example, they didn’t adjust for other job characteristics that might have been related to COPD.
“To allow for the possibility that smoking or asthma were complicating or obscuring their findings, the researchers repeated their statistical analyses using only the participants who had never smoked, and (separately) those who had never had asthma. In broad terms, the findings were the same as their analyses involving all the participants. Again that can’t make the issue of cause and effect go away, but it does give some indication that smoking and asthma can’t be interfering with the interpretation too much.
“The key finding is that the risk of having COPD was higher in people who had been exposed to pesticides in their work. That pattern is there in the main analysis, and also in the findings based just on people who had never smoked, and just on people who had never had asthma (though the numerical estimates of the increased risk are a little different in those two smaller groups). What’s more, they appear pretty consistently in three different ways of measuring exposure to pesticides: whether people had ever been exposed in their work, a numerical measure of how much exposure they had had in their working lives so far, and a measure of the length of time people had been in jobs where they were exposed to pesticides (though the pattern is probably not so clear, statistically, in that time-based measure).
“Arguably, though, the increase in risk is not immense. In the comparison of people who were never exposed to pesticides at work with those were ever exposed, the press release rightly says that those who were ever exposed had a 13% higher risk than those never exposed (in the fully statistically adjusted analysis). What’s that mean in numbers? Well, in those never exposed to pesticides, 8% had COPD on the definition used in this study – that’s about 1 in every 12 or 13. So in a thousand people, who are like those people in this study who were never in jobs that involved being exposed to pesticides at work, about 80 would get COPD. If they had all worked in a job, at some point in their life, that typically involved exposure to high pesticide levels, about an extra 10 of the 1,000 would get COPD, so about 90 in all – though the extra cases aren’t necessarily caused by the pesticides, because the study can’t tell us about cause and effect. Because of statistical uncertainty in the estimation, that extra number could be smaller or larger than 10. The data are compatible with the extra number being somewhere between 1 and 22, though it’s more likely to be somewhere in the middle of the range, nearer 10.
“The difference between in COPD numbers between those who were never exposed and those who at some point, were exposed to a high level of pesticides is larger. Again consider our 1,000 people, who are like those in this study who were never in jobs where they would have been exposed to pesticides. Then about 80 would have COPD. If they were instead in a job, at some point in their lives, that typically involved a high level of pesticide exposure, then an estimated extra 26 or so would get COPD – though again we can’t say that these extra cases are actually caused by the pesticide exposure. And the statistical uncertainty means that that extra number could possibly have been somewhere between about 6 and about 48 – not huge numbers out of 1,000 people, but not good for those extra people affected.
“The researchers did look at other potentially hazardous exposures at work, and generally found no clear statistical evidence of an association with COPD levels. That doesn’t mean that all the other pollutant groups that they considered definitely have no association with COPD risk. It means what I said, that the researchers didn’t find evidence for such an association. That could be because there really is no such association, but it could also mean that there is a real association for some groups of pollutants, but despite having data from almost 95,000 people, that didn’t provide enough evidence to find some associations that really do exist.
“It’s also important to realise that the study couldn’t give information on which specific pesticides were importantly associated with COPD risk. That’s for several reasons. As I’ve explained, the findings are all based on typical exposures related to people’s jobs, and while that is probably reasonable for broad comparisons of risk, it’s not going to tell us (for example) whether some specific chemical used in agriculture has an association with COPD, while some other chemicals did not. Also, as I’ve said, the study can’t tell us clearly about cause and effect. There’s a statistical issue too. The researchers did look at separate measures of exposure to three different broad types of pesticide: herbicides, insecticides, and fungicides. But they could present findings only for all three of these combined. That was because the type of jobs that make use of one of these groups would also tend to make use of the others as well, so that it’s impossible to pick out their separate effects statistically from a study like this. So it might be the case that the harmful exposure, in relation to COPD, is exposure just to herbicides, or even just to a few specific types of herbicides, but the type of data available and the nature of the study just couldn’t tell us anything about this.
Further information:
“The researchers didn’t adjust statistically for whether people had asthma, in their main analysis. Though I’m no respiratory doctor, that seems strange to me. Asthma levels might vary according to pollution exposure, if people with asthma choose to avoid jobs where pollution could worsen their asthma. Asthma levels could also be related to COPD, in terms of how COPD was defined in this study. The researchers explain in their report that the measure of COPD that they used could be affected by asthma, not because asthma could cause COPD, but because having asthma could affect the breathing measures that they used to define COPD. The non-adjustment for asthma could mean that some of the numerical measures of risk from the main analysis could be inaccurate in some way – but the researchers also repeated their statistical analysis using only the participants who had never had asthma, and found similar results to their analysis of data from all participants, so probably I am nit-picking here.
“The press release says that the association between pesticide level and COPD risk is positive for all three measures of pesticide exposure (intensity, cumulative, duration) – that is, the more pesticide, the higher the COPD risk. That’s true. But the release also says that the association is linear for all three. That’s misleading, and incorrect in one case. It is linear for the cumulative measure of quantity of exposure and for the duration of exposure, though it’s only linear if you take logarithms of the exposure measure. If you don’t take logarithms, the COPD risk would rise relatively more slowly at high exposure levels than at low ones. For the measure of intensity of exposure, the association isn’t linear. In the main analysis, there’s effectively no difference in COPD risk between people who were never exposed to pesticides at work, and people who were exposed but never at a high level, but there is an increased COPD risk in people who were exposed at a high level in their work, at some point, compared to those who were never exposed. There’s perhaps some evidence in the two subgroup analyses (people who never smoked, and people who never had asthma) that COPD risk is higher in those who only ever had a low level of exposure to pesticides compared to those with no exposure – but that evidence isn’t statistically significant, that is, the apparent increased risk is consistent with nothing going on except the effects of chance.”
1 https://erj.ersjournals.com/content/54/1/1900186.long
Prof Sir Colin Berry, Emeritus Professor of Pathology, Queen Mary University of London (QMUL), said:
“In this study of pesticides we have a widely differing group of compounds with variable mechanisms of action and in differing formulations. The actual exposure of any user might be: by mixing (high chance of contamination, but usually well regulated) and spraying (which might be from a ventilated cab in a tractor), or by other means including a simpler system such as a rucksack sprayer – all of which need different PPE and where regulations and observance of those regulations vary. Also exposure varies, contract sprayers do it all the seasons, vs. those who do it perhaps once per compound on a single crop. The confidence limits in the paper (though good in the sense of being narrow) look unconvincing.
“There are too many variables for a tiny change in relative risk to be of value as an observation when global terms such as ‘exposure’ and ‘pesticides’ are used. The authors do not suggest causation – what do they suggest? The investigation needs to be broken down to categories, for example the Ph of formulations, organosphosphates, neonics, thiazoles and so on before one could even suggest a sensible explanation.”
‘Lifetime occupational exposures and chronic obstructive pulmonary disease risk in the UK Biobank cohort’ by Sara De Matteis et al. was published in Thorax at 23:30 UK time on Tuesday 25 January 2022.
DOI: 10.1136/thoraxjnl-2020-216523
Declared interests
Dr Richard Parsons: “None. Although I am a toxicology expert, my expertise in pesticides is limited solely to neurodegenerative disease.”
Prof Andy Smith: “I am not aware of any conflict of interest.”
Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee. My quote above is in my capacity as an independent professional statistician.”
Prof Sir Colin Berry: “I have no conflicts of interest.”