There is a philosophical problem within medicine: how to deal with causal complexity and variations. While existing methods are designed for large scale population data and sufficiently homogenous sub-groups, a number of medical conditions are characterised by their heterogenic and complex nature: low back pain (LBP), chronic fatigue syndrome (CFS), fibromyalgia syndrome (FMS), irritable bowel syndrome (IBS), tension-type headache (TTH), post-traumatic stress disorder (PTSD), and many others.
Typical for these conditions is that no common cause or even set of causes can be found. Instead of a clear-cut one-to-one relation between cause and effect, there is a whole range of symptoms and causes: biological, psychological and social factors. Thus, there is no clear psyche-soma division of symptoms. And since the symptoms of these conditions are complex, ambiguous and to a large degree overlapping, classification becomes a problem. But that is not all. In addition, each patient seems to have both a unique combination of symptoms and a unique expression of the condition.
Since no medical cause can be identified for these conditions, they are commonly referred to as medically unexplained symptoms (MUS).
This is not a small problem in medicine. By some estimates, such unexplained conditions amount to 30 percent of all symptoms reported to doctors, and they are linked to a 20-50% increase in outpatient costs and a 30% increase in hospitalisation. The US National Institutes of Health (NIH) identifies medically unexplained symptoms as the most common problem in medicine.
So why do I say that this a philosophical problem, and not, for instance, a methodological one? One reason is that any method designed to discover causes will also bring with it assumptions about what causation is. The methods of evidence based medicine, for instance, includes the Humean orthodox view of causation and probability. In observational studies we search for robust correlation, in RCTs we look for difference-making, and in populations studies we generate probabilistic evidence from statistical frequencies.
Not only do our methods include a notion of causation. They also come with ontological restrictions.
If we want to study genuine complexity, it’s not enough to replace the mono-causal model with a multi-causal one. The bio-psychosocial model was developed as a better alternative to the physiological model. But it still falls short because the methods are designed to study one causal factor at the time. We can perform separate studies for psychological, social and biological factors, and then add the results together. But this is to assume mereological composition: that wholes are sums of distinct parts which don’t interact. To deal with the patient as a unity thus becomes impossible. Instead, one can try to study parts of the problem separately: the irritable bowel, the anxiety, the pain, the fatigue. Genuine complexity cannot be studied in this way.
Heterogeneity and medical uniqueness are also features that require a certain ontology. Singularism about causation is the philosophical idea that causation happens in the concrete. Any statistical claim about what happens in a population will then only be a representation of many such individual causings. On a frequentist theory, in contrast, statistics generate probabilities. This is, basically, what evidence based medicine means: statistical evidence from population studies are applied directly to a patient. This means that each patient is treated as a statistical average, not as a unique individual.
We know that patients respond differently to the same treatments. At the same time, the methods and policies of evidence-based practice are premised on an assumption that the same treatment should be given to all. But what is the rationale for claiming that the same intervention in two different patients is even the same treatment?
My next planned research project, CAUSEHEALTH, is motivated by the idea that medically unexplained symptoms show a limitation of current medical thinking. Their challenging features of multifactorial causation (complexity), heterogeneity (context-sensitivity), medical uniqueness (singularity) and no clear psyche-soma division (holism), are all essential aspects of causation. Rather than being dismissed as marginal, therefore, these unexplained conditions should be taken as exemplary for understanding health and disease in general.