EBM+ is a consortium taking part in a 3-year, AHRC-funded research project called ‘Evaluating Evidence in Medicine’. Our aim is improve Evidence Based Medicine (EBM) by developing innovative new ways of finding and evaluating different types of clinical evidence, in order to better inform medical decisions.
What is EBM?
Causal claims are crucial in medicine. Bugs, injuries and environmental factors cause disease and other symptoms; medicines, other treatments and public health policies alleviate or prevent such problems. Evidence-based medicine is a collection of methods for evaluating the evidence for and against causal claims like these. It provides grading systems and hierarchies of evidence, to help weigh up the evidence and to help decide whether there is sufficient evidence to establish a causal claim.
What is EBM good at?
EBM is good at weighing statistical evidence of associations. Statistical trials are used to test whether there is an association between the putative cause and effect. These trials vary in size and methodology, and EBM has developed ways of ranking these statistical studies.
What is EBM bad at?
EBM is bad at a few things, such as detecting reference classes within populations. For example, blood pressure trials are typically administered on clean populations, whereas the treatment is usually given to people with multiple morbidities. EBM is also affected by ‘pragmatic’ clinical trials – i.e. hardly any trials are done on pregnant women or for drugs with little financial promise. More generally, EBM does not account for the fallibility of randomised controlled trails (RCTs), most notably in their lack of mechanisms to support correlative evidence.
How can EBM be improved?
Evidence of mechanisms also plays an important role in establishing causal claims. Such evidence is often crucial when it comes to devising a statistical study; interpreting its results; deciding whether an association is causal, due to some other sort of connection, or a statistical blip; or applying the results of a study to a new population or a particular individual. High quality evidence of mechanisms can be produced by statistical studies, but it can also come from literature searches, one-off experiments, imaging, simulations etc. At the moment, EBM does not take explicit account of the role of non-statistical evidence of mechanisms. Arguably, though, we need to take all relevant evidence into account, not just statistical evidence of associations.
Visit our website: http://www.ebmplus.org
The EBM+ research project is jointly funded by