Pages

2015-10-04

Archie Cochrane and the challenge of evidence-based medicine: intelligent system design and back

Already in his lifetime, Scottish physician Archibald Leman Cochrane earned a reputation as the ‘father of evidence-based medicine.’ Hailed as the gold standard in life sciences, the notion of ‘evidence-based medicine’ leads us to the rightful question typical of ideas whose time has come whether and how there could ever actually be such a thing as non-evidence based medicine. And if we look around, it becomes quickly and depressingly clear that there is. And plenty of it, too. Whether one is to call it conjectural or speculative or intuitive medicine, it may have merits in respect of matters where the evidence is just not in, or studies have yet to be completed (say, typically: to be financed…) and may explain why medicine retains the aura of an art as well as of a science, but it falls short of scientific method, reliability, verifiability and duplicability of results.

Evidence based medicine needs to rely on data and on diligent and cautious interpretation thereof. Oxford dedicated an interdisciplinary Center to the concept. Of course there are plenty of areas where that is far from reality. Stanford’s John Ioannidis, a leading scholar of meta-research innovation, has claimed that every second medical study is either wrong or seriously flawed. Part of the problem is that studies yielding negative results are frequently not published at all. That which is published represents the tip of an iceberg, cherries picked carefully by greatly interested parties.  But negative results contain a wealth of information that may well point to different lines of reasoning or research that can be almost as valuable as positive results. Even for those, quite often nothing is disclosed or at least published about potential side effects. That leads to significant distortions of the picture in certain instances.

So long as publication is the primary incentive fueling careers in science research (resulting sometimes in a downright bizarre number of co-authors, far exceeding motion picture credits), passing that hurdle is all that matters in reality. Conflicts of interest abound. Pharmaceutical companies, having invested billions of dollars in clinical studies, have no interest in missing out on a return on their investment. They have decidedly no interest in publishing results that would appear to cast doubt on a product. As it stands, conflicts of interest are crass and would not be tolerated in any other setting: pharmaceutical companies who are paying for clinical studies also assess them and decide on the publication of their results. The egregious bias inherent in this situation becomes clear if we were to substitute car manufacturers and had them control tests about quality and product safety. Think of Volkswagen assessing the emissions of its engines, or Ford assessing the rear-end collision safety of its Ford Pinto tanks.

It would have to be a long-term objective to have clinical studies evaluated by independent institutions far beyond institutional review boards – however unlikely this is to happen in the foreseeable future. But just like in a lot of other areas where disruptive ideas press to the fore because their time has come – think capital markets transparency, public integrity, freedom of information, executive liability and private equity – corporations had better get in front or at least on the bandwagon of a momentum that will not be stopped. Change is seldom desirable for established interests, but where it has gathered the gale force of nature, it is not sensible to resist it, especially where it is actually also good business. In fact, the only way of doing business is the longer term. Of course, here we go again, trying to presuppose that the actors prioritize long-term interests of the institution over short-term interest in boni and performance reviews. That, too, is a principal matter of flawed systems design.

Evidence-based analyses have lately been used in multifunctional nanotechnologies and risk analysis, not least at Columbia University and the University of Michigan.

Evidence-based medicine optimizes choices by using qualitative criteria of epistemological strength for the empirical evidence to be relied upon, that is, meta-analyses, systematic reviews and randomized controlled trials, and by making all data available for future reference and additional analysis. It is itself an offshoot of a greater conceptual category, evidence-based design, one of the key elements in intelligent systems design.  Systems design and regulation provides a powerful methodological momentum for the advancement of quality protocols and the improvement of decisional quality.