EBM and the Era of Big Data
The mission of a doctor is to heal the sick and save lives. The reality is that life itself is a sexually transmitted ‘disease’ that carries 100% mortality. On the other hand, it is true that many great medical discoveries came from close collaboration between the academia and manufacturers. However, it seems that the mission of the pharmaceutical manufacturer is financially driven and suspicious of intentionally labelling healthy people sick. In addition, it is difficult to find a specialist who has no tie with the biomedical or pharmaceutical industry. A recent American survey of 50 medical schools that own hospitals has revealed that every researcher has received on average $33 417 capital assistance from an industry. US Food and Drug Administration also heavily relies on the funding from biomedical industry and in 2010, such funding has reached a total of $1.25 billion, nearly 46% of its drug research budget.1 To great extent, billions of sales profit has created trouble in clinical research. Scientific research has been ‘stained by the stinkiness of money’, filled with false information, wrong diagnosis and confused standards. Therefore, it has generated untrustworthy survey, statistically significant but clinically meaningless research results, misleading data, and hard to differentiate false data but somehow all have been easily published.6 ,7 One example is the recent announcement by a major pharmaceutical company that manufactures a drug for Alzheimer's disease. The company announced that it will continue on its clinical trials by eliminating patient's daily functional ability as an outcome measure. Although the reason behind this change was that patient with early Alzheimer's disease would have mild functional impairment and 18 months of trial could not detect it, one would suspect that such change could possibly affect the fate of this new drug, which has been closely watched by the medical profession.8
Another example that can illustrate this phenomenon was the Woman's Health Study, which enrolled 160 000 postmenopausal healthy women. The trial has found that women on hormonal replacement therapy had more breast cancer, heart disease, stroke and thrombosis than those in the control group. The price paid to treat these complications was far higher than the benefit of preventing colon cancer and hip fracture. This project was stopped 6 years into its originally planned 15 years of research. Another shocking example was the use of β-blocker in the Guidelines of European Heart Association. This recommendation was based on the fabricated research results from Holland researchers, which was possibly associated with nearly 800 000 deaths. These researchers initially found that perioperative use of β-blockers was cardioprotective in two clinical trials. This finding was incorporated into the European 2009 guideline. Nevertheless, a recent meta-analysis of 11 clinical trials of the same topic has shown that patients treated with perioperative β-blockers had 27% more deaths than those in the control group, which was equivalent to 800 000 deaths in Europe.
It is undeniable that medical profession is influenced by potentially biased information from some publicised scientific publication. Although mandated by the top medical journals that all clinical trials should be registered online first, a recent Journal of the American Medical Association (JAMA) article has found inconsistency of information registered on clinicaltrials.gov and its final publication.9 The researchers selected 96 clinical trials published in 19 journals with high impact factor (>10). They have found that 70 trials were sponsored by the industries. The most popular areas of research interest included cardiovascular disease, diabetes and hyperlipidaemia (23%), cancer (21%) and infectious disease (20%). The results of these trials were published in the New England Journal of Medicine (NEJM) (24%), Lancet (19%) and JAMA (12%). Nearly 93–100% of these clinical trials published the information of cohort analysis, interventions and outcome. However, 93 of 96 trials contained at least one outcome that did not match the information registered. The inconsistency between the cohorts and intervention was about 2–23%. Ninety-one clinical trials generated 156 positive outcomes. Among them 132 (85%) described the findings in clinicaltrials.gov and journals, while only 14 were published in the website and only 10 in the journals.
We know well that the essence of EBM is to combine the best external evidence, physician's personal experience and patient's wishes together. All three are needed to help a physician make the most appropriate clinical decision when treating an individual patient. Randomised and controlled studies and meta-analysis are not equivalent to the EBM. They are the reflection of external evidence. When the external evidence is lacking, the experience of a treating physician becomes very important. In this era of big data, biomedical science will have a major role in the world and the use of internet can support transparency and honesty. The explosion of large data has made the traditional research methodology obsolete. Randomisation of samples could be replaced with a complete data set. Statistics has been in use for over 100 years and perhaps one day it will be outdated. The best statistical methodology is probably the exhaustive attack method, which is to have the entire data points at once: samples=entirety. In this era of big data, we could have digitised human body and data on an individual but not a cohort. Everyone can be defined at the individual level as a single entity. The force that has the impact on this change is the internet.
For the physicians today, the technology is ready but new concepts and ways of thinking are still lacking. The physicians of the future will not play the role of knowledge storage but knowledge administrator. They should interact with the patients better, provide compassionate care, consult the patient, assist in decision-making process and be a partner with the intelligent patients. By focusing on solving clinical problems, they will apply the knowledge learnt from a complete set of data to their daily clinical practice and serve the patients even better.