Statistics from Altmetric.com
Burns et al 1 report a significant excess of deaths due to amyotrophic lateral sclerosis (ALS) in a cohort of Dow employees potentially exposed to the herbicide 2,4-dichlorophenoxyacetic (2,4-D), but then argue against the plausibility of a causal association, concluding that the association “is not consistent with previous human or animal studies”.
This conclusion and the authors' characterisation of the relevant epidemiological studies seem to rely entirely upon the significance of the statistics, which downplays the importance of their finding. Firstly, the authors state that “cohort studies of people with exposure to 2,4-D (have not) reported increased rates of ALS,” citing two studies,2 3 both of which have limited power to detect the risk of ALS. One of the two studies assessed risk in a cohort that was quite young with a relatively short follow up,2 and would therefore be unlikely to detect an increased risk for a disease such as ALS, which has a much older median age at onset. Burns et al then go on to state that “exposure to pesticides and agricultural chemicals have shown no significant association in several studies” (emphasis added).1
In each of the three case-control studies cited, however, ALS was positively associated with pesticides or agricultural chemicals, with reported ORs of 1.4,4 2.0,5 and 3.0,6 although the associations do not reach significance. Finally, Burns et al refer to a case-control study,7 which found a significant association between ALS and pesticides, but, they emphasise, “did not find a significant association of exposure to herbicides”.1 The association between ALS and exposure to herbicide was increased, however, and the lack of significance reflected, at least in part, small numbers.
None of this is meant to say that the finding of a significant association between ALS and 2,4-D is conclusive. The finding is, however, consistent with several previous studies, and instead of being played down, warrants serious attention in future studies.
We appreciate the interest taken in our study by Freedman. At the heart of the discussion are the interpretation of the significance of the statistics in our study,1-1 and the lack of significance in others. A critical point in valuing causation is the weight of the evidence to be placed upon the non-significant increase of non-specific exposures found in human studies of amyotrophic lateral sclerosis (ALS) compared with the weight placed upon controlled animal studies specific to the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D).
I agree with Freedman that undue reliance upon significance is ill advised. He is correct that the case-control studies cited in our paper showed increased odds ratios,1-2-1-5 but there is no evidence that any subjects were actually exposed to 2,4-D as the exposures were limited to pesticides, agricultural chemicals, and herbicides. The cohort studies examined workers who were definitely exposed to 2,4-D and thus provide a more valid assessment of risk even though they are less powerful than the case-control studies.1-6 1-7 The cohort studies of 2,4-D do not consistently show increased risk of ALS.
The associations found in the case-control studies are clearly unsupported by the experimental studies that have been conducted on 2,4-D. Environmental causes of ALS remain unknown. If future epidemiological studies investigate the neurotoxicity of herbicides such as 2,4-D, the researchers must improve upon the status quo of surrogate exposure information used in case-control studies or perform further studies of the 2,4-D workers. Epidemiologists must make a commitment to quality exposure assessment of individual pesticides, perhaps coupled with biomonitoring, to assess the putative health concerns associated with pesticides.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.