The rise of a pseudo-scientific links lobby
Every day there seems to be a new study making a link between food, chemicals or lifestyle and ill-health. None of them has any link with reality.
Dr Michael Fitzpatrick
Scarcely a week passes without a spate of banner headlines proclaiming a ‘link’ between some familiar activity or environmental factor on the one hand and some disease, disorder or social problem on the other.
‘Fast food linked to surge in child asthma and allergies’, declared the Guardian on 15 January. ‘UK scientist backs link between lead pollution and crime’, The Times (London) told us on 9 January. ‘Autism: traffic pollution linked, study suggests’, reported BBC News on 27 November.
These headlines usually arise from the publication of an epidemiological study and are generally warmly welcomed by other scientists (1). The claims advanced in these reports that diverse environmental hazards may have seriously adverse social consequences pass smoothly into the realm of conventional wisdom without being troubled by sceptical or critical commentary.
The key factor that appears to justify the exemption of such claims of ‘links’ from scientific scrutiny is that the activity or exposure in question is already the target of social disapproval. Powerful forces in the worlds of politics, the media, medicine and public health are firmly convinced that certain sorts of activity (or inactivity), food and drink, atmospheric pollution and chemicals, have damaging effects on the health of individuals and society. Industries associated with stigmatised products – alcohol and tobacco, chemicals, cars, fast food, pharmaceuticals – have become the focus of such moral opprobrium that any study or report that raises concerns about possible adverse health effects of these products is guaranteed a positive reception, irrespective of the quality of the science on which it is based.
The prevailing climate in relation to claims of environmental ‘links’ was well expressed in a comment from Sophia Xiang Sun, a scientist at the prestigious University of Cambridge Autism Research Centre, on the claims of a link between traffic-related air pollution and autism. Brushing aside the question of the scientific basis of the claim, Dr Xiang Sun ‘argued that cutting pollution would be a good idea anyway: “However, whether or not the potential association between autism and traffic-related air pollution exists, reduction of traffic-related air pollution would be good for public health.”‘ (2)
In other words, ‘it doesn’t really matter whether the scientific claim is valid, we are all opposed to traffic pollution, so we needn’t bother examining the research too closely’. But it does matter. If people attribute autism to traffic fumes, they are likely to blame themselves – or others – for this exposure. They may well embark on a quest for compensation, which, in a court more demanding in relation to scientific evidence than Dr Xiang Sun, is likely to prove futile and costly. They are also likely to be offered the services of charlatans promising quack remedies to ‘detoxify’ them or their children of these pollutants. Bad science thus has damaging consequences for the health and welfare of those who believe it.
In a celebrated lecture nearly half a century ago, Austin Bradford Hill, the statistician who is credited, together with Richard Doll, with establishing the causal link between cigarette smoking and lung cancer, laid out ‘nine different viewpoints from all of which we should study association before we cry causation’ (3). These criteria are, of course, well known in the worlds of epidemiology and public health, indeed to every first-year medical student. It is clear that, if we put environmentalist prejudice and fashionable dogmas aside, and stringently apply these criteria to current ‘links’ claims, scarcely any of the standards for causation are met:
|Hill criterion||environment/disease link|
|cigarettes / lung cancer||lead /
|fast food / asthma||pollution / autism|
It is not possible here to examine each of these links in detail in relation to all nine criteria, but three examples will serve to illustrate the gulf between current claims and the pioneering work of Doll and Hill.
Firstly, the strength of the association. Whereas Hill observed that moderate smokers were 10 times more likely to get lung cancer and heavy smokers 20 to 30 times more vulnerable, in the recent studies the risk of autism was said to be raised by a factor of three and that of asthma by 40 per cent.
Secondly, the specificity of the association. Hill emphasised the importance of a link between a definite exposure (smoking) and a definite outcome (lung cancer). What is the cause of lead poisoning? Is it atmospheric lead from tetraethyl lead in petrol or lead from pipes, paint and other industrial processes? Is the damage caused by inhaling or ingesting lead? And what exactly is the adverse effect? Different studies focus on IQ levels, rates of diagnosis of attention-deficit hyperactivity disorder (ADHD), behavioural disorders, delinquency, and different sorts of crime, from burglary to murder (all of which are socially constructed categories, subject to diverse cultural and political influences). Or, turning to the food/asthma link, what exactly constitutes fast food? A burger? A cheese sandwich? Asthma and allergy are both diagnostic categories that have expanded dramatically in recent decades and rates of diagnosis vary widely according to cultural influences.
Thirdly, the biological plausibility of the link. Lead has long been known to be a neurotoxin, but that does not explain claims for its role in diminishing criminal responsibility. In one of the few critical responses to any of the recent claims, the eminent autism researcher Uta Frith observed that the traffic pollution study did not ‘get us any further since it does not present a convincing mechanism by which pollutants could affect the developing brain to result in autism’, though the authors claimed plausibility for a highly speculative hypothesis (somewhat reminiscent of the ill-fated theory that the MMR vaccine caused autism).
In the concluding sentence of the paper linking fast food and asthma, everything hangs on the opening conjunction: ‘If the association between fast foods and the symptom prevalence of asthma, rhinoconjunctivitis and eczema is causal, then the findings have major public-health significance owing to the rising consumption of fast foods globally.’ If, on the other hand, as the results of applying the Hill criteria suggest, the association suggests a mere correlation rather than a causative relationship, then the findings have no particular interest or significance, other than to confirm the popular prejudice against fast food. A similar conclusion is justified in each of the links examined here – indeed in relation to almost all the environmental factors that have been ‘linked’ to diseases in the wake of the historic work of Doll and Hill.
Since this article was written, another ‘link’ has appeared: ‘Childhood asthma “admissions down” after smoking ban’, reports BBC News on Monday. Before his death in 2005, aged 92, Professor Richard Doll was critical of the extension of the causative link he and Hill had established between cigarette smoking and lung cancer to claims for diverse adverse effects of passive smoking – mainly on the grounds that the association was so weak. Yet, since the bans on passive smoking, statistically feeble associations have been mobilised to claim the benefits of reduced rates of passive smoking on various conditions, including heart disease – and now asthma. But we all hate smoking and smokers, so once again science is casually sacrificed to propaganda. See Christopher Snowdon’s critique of this study.
Another day, yet another ‘link’: ‘Cleaning products “linked to adult asthma”‘, reports BBC News, Tuesday. A quick look at the study soon demonstrates that this ‘link’ also fails to fulfil a single one of Austin Bradford Hill’s criteria for causation.