“Number of chemicals linked to problems such as autism doubles in just seven years,” reports the Mail Online. This headline uncritically repeated the conclusions of a new literature review by two researchers.
They argue that exposure to certain industrial chemicals, which have become a ubiquitous feature of modern life, being found in everything from solvents to smartphones, may be disrupting brain development during pregnancy. This in turn may be increasing the prevalence of neurodevelopment disorders such as autism spectrum disorder, attention-deficit hyperactivity disorder and dyslexia.
The “doubling” quote stemmed from news that a 2006 review by the same study group reportedly found five chemicals they considered were linked to neurodevelopment disorders, and now in their current review they report that research has found six more.
However, the studies quoted in this review found associations rather than solid proof of cause and effect.
Also, the literature review did not appear to be systematic and did not seek out any unpublished findings to assess publication bias. This means the evidence they found and used in the review might not represent the full range and balance of research available on the topic; this could bias the reviews’ conclusions.
This review may stimulate debate but does not add much in the way of proof. It is unclear whether low levels of industrial chemicals are causing harm to children and adults on a wide scale, and if, or how, they should be regulated differently than they are today.
Where did the story come from?
The study was carried out by researchers from Denmark and the US, and was funded by the US National Institutes of Health, National Institute for Environmental Health Sciences.
The study was published in the peer-reviewed medical journal Lancet Neurology.
The Mail’s reporting was broadly accurate in the sense that it repeated the conclusions of the research and included many quotes from the main author. However, it did so in an uncritical fashion, providing no counter comment from other experts in the field or discussion of the inherent limitations of relying on observational studies to link chemicals to development disorders.
The claim that exposure to chemicals is responsible for conditions such as autism spectrum disorder (ASD) is certainly not a consensus opinion. Most experts hold the opinion that ASD and other neurodevelopment disorders probably arise due to a complex mix of both environmental and genetic factors.
What kind of research was this?
This was a literature review that attempted to identify new literature on the potential damaging effects of environmental toxins on health.
The researchers said “neurodevelopmental disabilities, including autism, attention-deficit hyperactivity disorder, dyslexia, and other cognitive impairments, affect millions of children worldwide, and some diagnoses seem to be increasing in frequency”. In 2006 they reported conducting a systematic review that they say identified five industrial chemicals as developmental neurotoxins – that is chemicals causing problems in the development of the brain and nervous system.
The current review was an update of the original. However, the current review contains limited information on methods, although that indicates a search of only one literature database, and provides little information on how studies were reviewed and selected for inclusion. And possibly more importantly, what studies were not included and why.
With such limited methods provided it is not possible to call this a systematic review.
Systematic reviews generally provide more robust conclusions than literature reviews. As the name suggests, one has a more systematic nature that seeks to identify all the literature on a particular topic. Ideally this includes unpublished evidence, as this is a good way to assess whether any publication bias has clouded the picture.
In contrast, a literature review typically only searches relatively few sources for relevant publications. So it may miss a proportion of the relevant published or unpublished evidence, potentially biasing the conclusions.
What did the research involve?
The evidence for this review came from searching one electronic medical database (PubMed) for relevant published articles from 2006 to the end of 2012. The authors also mention they retrieved additional papers using the reference list of the publications initially retrieved. The search was limited to children (0 to 18 years old).
The number of articles identified as relevant in the PubMed search was not reported in the main article. Nor were any further sifting methods, inclusion or exclusion criteria, to arrive at the final studies that formed the basis of the review.
What were the basic results?
The search identified cross-sectional and cohort studies of relevance. In the published narrative summary it was not always clear what was simply the authors’ considered opinion and what was backed by evidence. The authors’ prose tended toward a line of persuasive, and often passionate rhetoric, rather than a balanced discussion of the pros and cons of the underlying research.
The article was organised around discussion the following issues:
- the unique vulnerability of the developing brain
- new findings about known hazards
- newly recognised developmental neurotoxicants
- developmental neurotoxicity and clinical neurology
- the expanding complement of neurotoxicants
- consequences of developmental neurotoxicity
How did the researchers interpret the results?
The researchers concluded that “since 2006, epidemiological studies [that examine the possible causes for conditions] have documented six additional developmental neurotoxicants – manganese, dichlorodiphenyltrichloroethane, fluoride, chlorpyrifos, tetrachloroethylene, and the polybrominated diphenyl ethers.”
Similarly, they “postulate that even more neurotoxicants remain undiscovered.”
On this basis they concluded that there was a widespread pandemic sized threat posed by industrial chemicals on developmental neurotoxicity and there should be “a global prevention strategy”.
Their central conclusion was that “untested chemicals should not be presumed to be safe to brain development, and chemicals in existing use and all new chemicals must therefore be tested for developmental neurotoxicity. To coordinate these efforts and to accelerate translation of science into prevention, we propose the urgent formation of a new international clearing house.”
This literature review provides stimulation for debate, but does not add much in the way of proof, around the issue of whether low levels of industrial chemicals are causing harm to people on a wide scale, and if or how they should be regulated differently than they are today.
The study highlights a number of valid issues for debate (see below), but provides only half of the debate in the publication. The review may benefit from a more balanced account or critique of the underlying studies it drew upon. For instance, some were cross sectional studies that provide little evidence on causality. And even those that were cohort studies may still be subject to significant confounding from other factors.
These limitations weren’t discussed in the published article. Hence, whether or not there is a firm evidence base on which to assert that these chemicals are causing damage is not clear from this publication. They may be present in the original 2006 review, which was not assessed as part of this critique.
Currently chemicals are presumed safe until negative health consequences are proven. An example of this was the use of lead piping that contaminated water and resulted in lead poisoning, or the use of asbestos fibres in buildings that caused lung cancer.
There is legitimate argument around whether this is the correct approach given the often large time delay between these chemicals being routinely used and any health effects being detected. An alternative, proposed alongside other measures by the study authors, would be to prove they aren’t harmful first, before they are able to be used wholesale across the world.
As well as the conceptual shift, there would also be practical challenges in this approach, for example, interactions between chemicals would presumably need to be tested, non-industrial chemicals would need testing, and different countries might impose different regulations.
A systematic review with explicit methods making it clear how it identified all relevant published and unpublished literature on this topic would have been more useful. This type of research could have potentially provided more robust evidence to inform the debate; ideally this would seek to include cohort and toxicology studies.
A final point to consider is that the fact that more children are being diagnosed with conditions such as autism spectrum disorder does not necessarily mean that these conditions are becoming more common. It could be the case that health professionals are more aware of the condition, and are getting better at diagnosing it in children.
Children who previously would have been labelled as “painfully shy” or a “problem child” are now rightly diagnosed as being on the autism spectrum.
Analysis by Bazian
Edited by NHS Website
Links to the headlines
Mail Online, 15 February 2014
Links to the science
Lancet Neurology. Published online February 15 2014