The limits of evidence-based medicine

Advancements in medical, social, and economic fields over the past century have led to increase in longevity and better living conditions. Statistical methods pioneered by Robert A Fisher, based on mathematics of probability, have contributed the lot to these developments.[1] With faster computers, and statistical packages handling large data, our dependency on these methods have increased over the years. We are optimistic that statistical methods will continue to yield breakthroughs in the future. However, there are misgivings.

Papers contradict each other. This together with flip-flops by experts on many crucial issues, is eroding public trust in science. About two decades ago, Stanford researcher, John Ioannides expressed concerns that many scientific papers, based on principles of evidence-based medicine (EBM), are often not replicable.[2] Even top scientific journals are not immune to errors in the results.[3] Are we moving backwards from certainty toward uncertainty?

For example, the role of low dose aspirin that was accepted as a preventive for coronary heart disease is now being debated.[4,5] The benefits do not offset the risks of gastrointestinal bleeding,[6] which perhaps got underestimated earlier. Similar was the fate of hormone replacement therapy for menopausal women that was launched with a bang but fell from grace with fresh evidence.[7,8]

These are not infrequent phenomena. Ideally newer research should build upon previous research leading to stronger evidence and trust in the scientific method. Regrettably this is not happening. Each research paper stands in isolation contributing to millions of papers in journals with “statistically significant” results, which often cannot be replicated adding little to the literature. Clinicians depending on hard evidence from these studies for their practice are none the wiser. Researchers striving for zero error certainty, unachievable by methods depending solely on probability theory, have driven medical research toward sterility sanitizing it from the nuances of the art of medicine.

This has resulted in polarization of views among the scientific community. Much of the blame lies with the ascension of EBM and modern statistical methods to the exclusion of all alternative means that could enrich the practice of medicine.

While EBM is an important research tool grounded in modern probability-based statistical methods, it has its limitations. The quantitative methods of EBM need to be combined with an approach that is based on qualitative methods. While critics dismiss the latter as being subjective with inherent biases, they overlook the potential of this approach to capture the nuances of the art of medicine.

Mukherjee in his slim volume, “The Laws of Medicine,”[9] has captured the nuances of this art perfectly. The three laws, with illustrative case studies, have been enumerated as follows: firstly—A strong intuition is much more powerful than a week test; second—“Normal” teaches us rules while “Outliers” teach us laws; and lastly—for every perfect medical experiment, there is a perfect human bias.

Among all medical disciplines, the fields of psychiatry and psychology pose the greatest challenges for developing valid diagnostic tools, defining normality, and conducting experiments, which form the foundation of EBM. This is because of the complex determinants of psychiatric illnesses and human behavior depending on cultural, social, economic, family, and genetic factors, besides many unknowns.

A combined approach of EBM supplemented with qualitative research methods would be most appropriate for research in psychiatry and human behavior.

Author contribution

Concept, and drafting of the article: AB; Editing and revising article: AB, SC, JP, KS. All authors approved the final manuscript