Unpublished work presented at American Association of Pediatrics conference has reported an association between a certain type of pesticide and autism and developmental delays in children.
Dr Anna Hansell, Reader in Environmental Epidemiology, MRC-PHE Centre for Environment and Health, said:
“The possible role of environmental factors in autism is being actively discussed and investigated by environmental scientists, but as yet there are few other studies available exploring pesticides as a possible factor in ASD development –so this remains a hypothesis under investigation.
“The type of study presented can highlight associations – while interesting, we would need more evidence than currently available to conclude this is causal. Further, this report is a conference presentation –the paper has not yet been published in a peer reviewed journal so the full details of the study are not available, including details such as number of cases . Full details are important to be able to assess limitations of the study and to consider possible alternative explanations. These might include chance and other factors that might lead to observed higher rates in those areas e.g. better diagnosis and recognition of ASD, higher exposures to other agents also linked with ASD.
“A study was published in 2014 also finding associations with pesticides in California, using a similar type of study design (see http://ehp.niehs.nih.gov/122-A280/ for report). Journalists might want to contact authors of this study. However, both these types of study show associations. While they add to the evidence available, they do not prove causation – this would require an accumulation of evidence from various studies of differing study designs.
“There are differences between the case and control areas that may make detection of ASD more likely in the case areas – e.g. the control areas have a five times higher neonatal death rate and statistically significant higher referral rates for the control disorders. An alternative explanation is that the control areas are more deprived and this may lead to ascertainment bias of ASD in case areas. If reviewing this I would ask authors to provide other assessments of deprivation than just the provided % below poverty level (which is higher than in the case areas, although not statistically significant).
“As the authors state, ‘The influence of mosquito-borne illness on regional prevalence of ASD/DD is a potential confounder’ – areas with more pesticide use have higher mosquitoes in the first place, therefore higher risk of mosquito borne illness – exposure to this in utero may affect ASD/DD risk.”
Prof. Alastair Sutcliffe, Professor of Paediatrics at UCL, said:
“There is no justification for this scientific hypothesis which would seem implausible to me, and there appears to be no evidence for the allowance of multiple confounders.
“Childhood autism has been in the public eye following spurious claims of increased risk after the MMR vaccine. One of the areas that is known is that genetic factors have a powerful contribution to its origins and there have been genes implicated at the last count on all 23 pairs of human chromosomes.
“Furthermore its origins affect structures such as the amygdala, and these parts of the brain develop from about 5 weeks post-conception. Autism therefore cannot be ‘acquired’ after birth, which is why the Wakefield conspiracy was eventually dismissed as best it could be.
“Rates of autism appear to be rising in the Western World so anything that increases incidence can seem alarming. But anybody can claim one thing is associated with another. This new unpublished, unreviewed study seems likely to be merely observational and therefore not cause and effect.”
Prof. Jean Golding FMedSci, Emeritus Professor of Paediatric & Perinatal Epidemiology at the University of Bristol, said:
“It is impossible to give this study any credence on the basis of this poster. The comparison between the area where there had been aerial spraying with the rest was different in a number of ways, including that it was a swamp. The authors took no account of differences between the areas that may have been explained by the age of the parents or other circumstances known to increase the risk of autism. It would be especially helpful to have information on the prevalence of autism in these areas before 2003, which was when the airborne spraying started, so that that can be compared with the subsequent prevalence figures.
“Indeed these must be considered as very preliminary findings – they have not been peer-reviewed, and should be ignored until more details are provided, and the data analysed more coherently.”
Dr Christopher Connolly, Reader in Neurobiology, School of Medicine, University of Dundee, said:
“Given the target site for pyrethroids and its likely impact on the developing nervous system such studies may provide some insight into potential risks to human brain health from chemical spraying. However, good evidence is required and must be published in the peer-reviewed scientific literature and available for expert scrutiny. It is premature and inappropriate to discuss the significance of this study before this time. A particular caveat (from the limited information available) is that the study appears to be done at a single site (with local technical replicates that may not be independent).
“Even if the claims of this study are substantiated, caution is required in its interpretation as the study can indicate, at best, a correlation between the spraying of pyrethroid(s) and autism in this area. It does not demonstrate a causative role for pyrethroid(s) and other correlations cannot be excluded. So the study needs to expand to multiple sites and the correlation studied at a mechanistic level in carefully controlled laboratory studies to establish cause and effect.”
Dr Rosa Hoekstra, Institute of Psychiatry, Psychology and Neuroscience (IoPPN), King’s College London, said:
“The information provided doesn’t provide me with confidence that the study is robust and the conclusions drawn are warranted.
“Based on the limited information available I have the following concerns and queries:
– The poster suggests the researchers considered all children <20 years living in each area. It seems they studied a single time point. This suggests they have not corrected for the length of exposure nor when that exposure occurred (Prenatally? Postnatally? In the newborn period? When the child was 18?). Some of the children in the studied areas may only just have moved there, while others may have lived there for the past 20 years (i.e. were also exposed prenatally). Some may have moved from the exposed to the unexposed area over the past 20 years. To me this inexact method makes little sense.
– In both the exposed and unexposed regions the rates of ASD/DD are suspiciously low. Recent USA autism estimates (that is, autism only, not also including other DD) suggest a prevalence rate of 1/68. The large difference with the ‘prevalence’ reported here (1/120 and 1/172) raises questions about the reliability of the diagnostic information used in this study.
– Why were the referral rates for these 4 control diagnoses lower in the pesticide-exposed regions? Could this be due to referral substitution? That is, the children in the ‘exposed’ area may have been referred for autism/DD instead of the control diagnoses such as ADHD, leading to inflated rates of autism/DD and deflated rates of ADHD.
– The ‘unexposed’ area has a 3.5 -fold higher neonatal death rate compared to the ‘exposed’ area. Why is this?
– The exposed area also has a nominally higher percentage of whites, fewer poor people and a higher percentage of men. In rigorous epidemiological studies high SES, being male and being white are all associated with higher ASD rates. Based on the limited information presented here I can’t evaluate whether these covariates have been taken into account appropriately.
“In short, based on the limited information provided in this poster I am as yet unconvinced of the study’s methods and findings. I continue to strongly urge journalists to wait publicising this work until it has been published in full, after scrutiny by peer reviewers.”
Dr Max Davie, mental health lead for the Royal College of Paediatrics and Child Health (RCPCH), said:
“This is the latest of a series of abstracts to emerge from the US about alleged environmental triggers for Autism Spectrum Disorder (ASD). Like all these possible links, they should be carefully explored and we must not dismiss them.
“In relation to this paper, the effect is small and similar to the usual variance in districts for ASD referral, and although the investigators made some effort to control for confounders, there are many factors that might influence referral and diagnosis rates that are very hard to correct for – the most obvious being differing policies across districts regarding the necessity of receiving a diagnosis for additional support in school. This has been shown to be a powerful influence in other studies, and even subtle changes in policy interpretation can have an impact. And finally, it is not clear why delivering a molecule that is not linked convincingly with ASD in a different way will suddenly increase the risk of ASD.”
Dr Daniel Hawcutt, Senior Lecturer in Paediatric Clinical Pharmacology at the University of Liverpool, said:
“I am a paediatric pharmacologist – not a neurologist – so am looking at this from that perspective. My understanding of the current evidence for development of autistic spectrum disorders is that it is a complex interplay of genetic and environmental factors (which are not fully understood).
“There is some background evidence in this area – for example in the CHARGE study (a case/control study from California (2014)) it was shown that “Children of mothers residing near pyrethroid insecticide applications just before conception or during third trimester were at greater risk for both ASD and DD, with ORs ranging from 1.7 to 2.3” (http://www.ncbi.nlm.nih.gov/pubmed/24954055 ). However, this study did not show that the pregnant women were actually EXPOSED to the drug – and there are plenty of other reasons why people living in one area may be more at risk than others living on other areas (different rates of affluence, toxicity in the soil/water, proximity to a school for children with special needs, etc, etc), so a criticism of this previous paper is that causality was not established.
“This abstract only contains limited information, and has not been ‘peer reviewed’ as an article in a medical journal would be, and so needs to be treated with some caution. The information presented in the abstract builds on the previous work mentioned above. It attempts to control some of the other factors by looking at affluence, referral rates to hospital, etc. However, from what I can see, It still does not show that the pregnant women were actually exposed to the pesticide. In addition, the development of the brain is a very intricate process. If this is a ‘true’ finding then it is surprising that there was no correlation between exposure of women at a particular stage of pregnancy and the effects. The data was collected retrospectively (i.e. from existing medical records), and this is more prone to error than data collected prospectively (deciding on the important information and collecting as you go along).
“Overall, I think this is interesting, and I would suggest in conjunction with the previous paper, it builds the case to examine pesticides and neurodevelopment more closely – but it is NOT proof that pyrethroid pesticides are causative. It is also important to remember that the mosquitos being treated spread disease, so while I have no idea what the optimal methods of mosquito control are, not adequately reducing mosquito numbers carries a risk from other diseases.”
‘Aerial Pesticide Exposure and Increased Risk of Developmental Delay and Autism Spectrum Disorder’ by Steven Hicks et al is a poster at the American Association of Pediatrics conference.