An article recently appeared in Science investigates the correlation between the frequency of tumours (all, indistinctively) and the number of stem cell divisions. The correlation, the authors of the article say, is very strong. This is potentially very significant and informative, except that, unfortunately, only a very small proportion of the variation in cancer risks can be attributed to environmental factors or inherited predisposition. The rest – which is a big chunk to explain – is due to bad luck. I refer the reader to a clear article by Paolo Vineis, where he expresses doubts about the methodology of the paper.
But this is not the point I want to reflect on. I am instead most interested in the explicit use of the term ‘bad luck’, which occurs both in the text of the authors and in the editor’s comment accompanying the article. Needless to say, this is the only aspect that the media picked up (just google ‘bad luck cancer’ and you will realise how much this is resonating).
There is a lot that we don’t know about cancer (and about many other phenomena). But is it ‘bad luck’ if we get ill and we don’t have a cause to blame? Is everything we are unable to explain today due to chance? What kind of message do we convey to the public in this way? The issue is of course delicate, because we find – at the other end – another extreme position: scientism. And surely we don’t want to fall on that side either. While it is intellectually honest to say that we don’t know something, it is misleading to call it bad luck. In the case of cancer, the message can be devastating. If bad luck plays a major role in cancer development, why bothering with prevention and with screening? (This doesn’t imply, however, that all preventive actions are sound or effective, or that screening programmes always lead to correct identification of cases of cancer.)
Also, what the article labels as ‘bad luck’ may be in fact due to factors that have not been thoroughly explored, for instance socio-economic or behavioural factors. I’m not talking about reducing the ‘social’ part of disease to some allegedly objective, bio-chemical element. I’m really talking about understanding the lifeworld of individuals, which includes health and disease.
In sum, ‘bad luck’ is really a bad choice for a scientific article, not just because it shows epistemological clumsiness, but also because it may have quite undesirable consequences on the public understanding of science and on people’s behaviour.
Scienza in Rete is an Italian online magazine that popularises scientific research and also discusses various issues related to science and society, including science policy and ethics. It recently featured an article on possible causes of autism. Autism is a disease affecting the individual’s abilities to interact, verbally and non verbally, with other people. Its aetiology still has grey areas and much research is needed to understand its mechanisms.
I was intrigued by the aforementioned article because the title includes ‘ambiente’ – the environment. As you go through it, though, it becomes clear that the environment has a quite specific, restricted sense: the chemicals to which an individual is exposed, whether in life or even in utero. To be sure, this is precisely what environmental epidemiology investigates. The study mentioned in the ‘Scienza in Rete’ article evaluates, specifically, exposure to methylmercury. Fair enough. It is of utmost importance to understand the total exposome, and there is excellent research in progress in this respect.
But can chemicals exhaust all there is about the environment? What about the social environment?
Biological (or, biochemical) and social (or, socio-economic, psychological, behavioural) causes of disease shouldn’t be studied separately. We should instead strive to understand how the biological and the social realms interact. We should work towards integration of disease aetiologies and try to understand the mixed mechanisms of diseases. It is about time to move beyond the biologisation of disease and return to a more holistic understanding.
At the beginning of November, large part of the philosophy of science community convened in Chicago for the bi-annual meeting of the Philosophy of Science Association. This is usually combined with the annual meeting of the History of Science Society. This year, the programme figured two sessions on interdisciplinarity, one on the history and the other on the philosophy of interdisciplinarity organised by Hanne Andersen.
I contributed to the philosophy of interdisciplinarity with a paper on the social causes of diseases, which I co-authored with Mike Kelly and Rachel Kelly. The paper provides an argument for a mixed aetiology of diseases, especially non-communicable diseases (or, as it was suggested during Q&A, socially communicable diseases). We argue that recent attempts to include socio-economic-behavioural factors do not go far enough, as the explanatory import of these factors is not properly or sufficiently spelled out. If you are interested in the paper, look at the slides, or email me for the latest draft.
But there is another aspect that I wish to focus on in this post. What can ‘interdisciplinarity’ contribute – if anything – to ebm+? A terminological note first. Without too much sophistication*, I take interdisciplinarity to be the collaboration of scholars active in different fields in order to shed light on a shared problem. In this case, I am active in the area of philosophy of science, Mike Kelly has a sociological background and is active in public health, and Rachel Kelly works in molecular epidemiology and has competences in epigenetics. Our shared problem is offering an account of the mixed aetiology of diseases, one that integrates the perspectives coming from sociology (via the concept of ‘lifeworld’) and bio-medicine, and that ‘exploits’ conceptualisations of mechanisms, evidence, or causality developed in philosophy of science.
In this sense, interdisciplinarity is, I think, at the service of ebm+, where the plus precisely indicates what we might add to available eb-approaches in order to improve our understanding of health and disease or the design policies to reduce the burden of disease. The article mentioned in this post is but one example of how to work in an ‘ebm+ perspective’.
* Interdisciplinarity studies is a rich and complex area of research. The kind of interdisciplinarity alluded here is just one aspect of the many possible facets.
Randomized controlled trials (or RCTs, for short) are, according to the paradigm initiated by evidence based medicine (or EBM, for short), the most accredited tool to establish whether drugs or treatments are effective, and to what extent. Surely, this is not stated in such explicit term, but it is quite easy to infer looking at the different versions of evidence hierarchies and of grade systems (for a discussion, see here). The logic behind RCTs is relatively simple, although it is the implementation being problematic in most cases. To know whether a treatment is effective we need to compare two groups of individuals, one that does take and the other that does not take the treatment. A crucial assumption behind RCTs is that individuals (at least in the trials) respond to the treatment in the same way. This is called ‘universal biological response’.
But what if individuals do not respond to treatments in the same way? This is precisely one of the questions behind gender medicine. Gender medicine thus points the finger to an important epistemological issue: to what extent can we safely make the assumption about universal biological response? And even more importantly, should (gender) medicine strive to understand (and accommodate) differences, rather than flatten out diversity?
The question isn’t innocent and in fact sparks a whole battery of new questions. For instance: if relevant difference concerns gender, do they also concern ethnicity? What are the relevant biological, social, psychological factors that play a role in different responses to a same treatment? And mostly, how do we know?
These questions are broadly epistemological in that they concern how we come to know (or fail to have knowledge of) about similar or different responses to treatments. Taking these questions seriously has important repercussions, for instance, on our medical methodology, and on the concept of individual, well being, or efficacy.
A core idea behind EBM+ is evidential pluralism. This was formulated in the context of medicine by Russo and Williamson (2007) and further discussed by several authors (both to support – see for instance here – it and to criticise it – see for instance here). Clarke et al (2014) use the following formulation:
In order to establish that A is a cause of B in medicine one normally needs to establish two things. First, that A and B are suitably correlated—typically, that A and B are probabilistically dependent, conditional on B’s other known causes. Second, that there is some underlying mechanism linking A and B that can account for the difference that A makes to B.
Evidential pluralism of such type has also come to be known as RWT – the ‘Russo-Williamson Thesis.
An important clarification concerns the status of the thesis. RWT does not say what causality is, but how we come to establish causal relations. Philosophers call the first an ontological question and the second an epistemological question.
RWT engages with questions about evidence, evidence evaluation, and use, which have an important epistemological component. This means asking ‘what to look for’ in order to establish causal relations. RWT says that we look for dependencies and for mechanisms. These pieces of evidence are used to make inferences about causal relations, notably about their very existence, their plausibility, their strength, and about how they could be exploited for policy.
From this epistemological thesis about evidential pluralism, ontological pluralism does not follow straightforwardly. It does not follow from RWT that, if we need evidence of correlation and of mechanisms to establish causal relations, causation is itself constituted by correlations and mechanisms. The ontological question is currently unsettled. Russo and Williamson suggested here and here that the epistemic theory of causality offers a possible causal metaphysics. But there are other options worth investigating, such as informational accounts of causal production, as also explored by Illari here and Illari and Russo here.