Tuesday, February 18, 2014

Noisy trials

A recent study in the European Heart Journal that shows that the common wisdom that high HDL has a protective positive effect on cardiovascular health could be wrong, is a warning signal to traditional medicine, handicapped with normality statistics and decades long practice of small randomized trials. As many other disciplines prove routinely, including the mathematically pure Physics, once stated, a hypothesis is easy to prove. And, as everybody knows, economists can prove pretty much anything including the Princeton hypothesis that 100 Trillion in US debt will cure all the world’s ills. Medicine, has been lagging for decades – driven by the collection and use of noise from limited observations to satisfy regulators blinded by ignorance and institutionalism.

Mendelian randomization (MR) that combines genetic information with clinical observations is in the right direction. But till the industry realizes that normality statistics have lured them into conclusions that are utterly wrong, it will continue on the well-trodden and disastrous path. With little market feedback, the scientists are unlikely to see the emergence of black swans or the assumption that white swans walk in a straight line is wrong – and this will continue to propel them into bringing incremental therapies to market with little patient benefit. In the financial market, incompetence is rewarded by bankruptcy but in science, generally, incompetence is rewarded with prizes as the lack of feedback let those adept at data manipulation rise to the top.

Statistics – may have led humans astray in most fields with complex non-linear effects – such as Physics, Medicine and Economics. The engineers, who are happy with this conventional tool, may ultimately end up leading everybody else into dead-end alleys.