Evidence-based medicine is broken: why we need data and technology to fix it

A recent paper in the British Medical Journal suggests that evidence-based medicine is in crisis. Evidence-based medicine is based on the practice of employing treatments that have scientific research that backs up their effectiveness. It is usually set against medical practice that is based on anecdotal experience or simply doing things because that is the way they always have been done.

Ways evidence-based medicine is broken

The authors of the paper point out however that there are a number of problems with evidence-based medicine that together, significantly compromise its effectiveness. The first problem is that evidence-based medicine plays right into the hands of companies and organisations with a vested interest in seeing a treatment recommended. Of particular concern are drugs that are recommended for treating pre-diseases. The classic example of this is the use of statins to lower cholesterol in the attempt of reducing cardiovascular disease.

The National Institute for Health and Care Excellence (NICE) has offered evidence-based guidance on prescribing statins that has reduced the required level of risk before treatment is offered, effectively increasing the number of people taking statins significantly. The bar for being put on statins is very low and cholesterol levels can be considered normal and the person could still have a risk profile that puts them on the pathway for treatment.

Drug companies have a vested interest in funding trials and journals are more likely to publish positive results than negative ones. The authors of the BMJ paper cite “one review of industry sponsored trials of antidepressants showed that 37 of 38 with positive findings, but only 14 of 36 with negative findings, were published”.

Another major problem with evidence-based medicine is the fact that statistical significance in a clinical trial does not necessarily mean very much at an individual level. This works to overemphasise benefits and to underestimate risks.

People writing the guidelines to implement evidence-based medicine face a growing challenge of dealing with ever increasing amounts of evidence. At the end of this are the clinicians that have to deal with an explosion in the guidelines themselves which can be complicated and even more so when they have to be applied to a patient with all of the real-life complexity and idiosyncrasies that they bring.

The authors are not suggesting abandoning evidence as the basis for clinical treatment. What they end up recommending are modifications of how the evidence is gathered, reported and implemented.

Technology and data as a solution

What the authors don’t mention though is the role that technology has to play in providing a solution for the problems that they have outlined. The first element of a technology-based solution actually runs counter to the author’s point of crisis. What is needed is more evidence and data, not less. Large data sets providing detailed monitoring of patients combined with genetic information will eventually be able to reveal specific information about what makes a treatment work in one patient and not another. This type of information goes much further than the limited range of measures normally collected as part of current clinical trials. To enable this collection of data, we will need to be monitoring people pervasively and collecting that information on a continuous basis.

Clinical analytics of this so-called “clinical big data” can potentially reveal much more information about the effectiveness of particular drugs and treatments given a specific individual with a set of symptoms and illnesses. What we will then need is sophisticated artificial intelligence as portrayed in the fictional film “Her” that guides clinicians through a range of true evidence-based treatment options that have been personalised for the specific patient at that moment in time.

It is possible that as a society we are not ready to trust the likes of Google CEO Larry Page with the task of collecting and analysing patient data on this massive a scale. He declared our ability to save thousands of lives through this approach. Despite our reticence, it is clear that he is right.

Companies such as PatientsLikeMe have also faced skepticism when arguing their ability to contribute to clinical research through the massive, but self-reported data they collect from patients.

Despite the resistance, sites like PatientsLikeMe, CureTogether and the personalised genetic profiling company 23andMe are part of the future of medicine.

These companies have the ability of providing extensive information about the effectiveness of treatments, symptoms and side effects of people outside of the restrictions, and some would say artificial settings, of a clinical trial.

All of this does not deny the needs for the skills of a doctor, at least until the technology has advanced to the point where it makes those skills redundant.