Randomised Clinical Trials in Medicine
Western medical science is based on the use of randomized clinical trials (RCT) to determine whether a particular medical procedure is effective or not. This approach was introduced in the 1950s and has since become the dominant means, the so-called “gold standard,” such that it is the very foundation for the more recent move to “evidence-based” medicine. Prior to this, medicine was based essentially on observation by the skilled clinician (the art of medicine).
At the heart of the rather meteoric rise of the RCT in medicine is the issue of causality. Causality as an issue implies lack of knowledge or doubt regarding what one has observed. This problem of the split between what has been observed and to what extent it represents what actually occurred goes back to the dawn of rational thought in human history, the Ancient Greeks. This dawn brought with it a reciprocal darkness or veil regarding the essence of reality. This was not a problem experienced before. Even within Greek philosophy, the problem did not arise as such until the time of Aristotle, who turned from the world of Platonic Ideas, to be discovered and understood through an internal thought process, to the world of nature, that is, the external world. Here, one is thrown back on observation of events which must then be interpreted.
Aristotle identified levels of cause which meant that causation was a complex, and not simply reducible to “A causes B.” This he did because for him, reality existed beyond the material, external plane. Thus, while he included the more modern definition of cause (efficient cause), as that external entity from which the change commences, he also included a formal cause, which is the plan or Idea of the thing, such that we can understand the context of the emergence of B. He also included the final cause, which is the purpose or end of the thing, what today we only find in the humanities in the form of motive. Thus, for Aristotle, A is the efficient cause of B, but not the formal cause, nor the final cause. Aristotle also included the various parts of an event or thing that were necessary for it to happen. We can understand these levels of causation by considering that a movie is produced by an author, working from an original Idea (final cause), put into effect by a producer (formal cause) working from a script, then effected by a director (efficient cause) who needs actors (material cause), locations, sets, etc to make it all come together.
Because of the reduction of natural science to the material plane, only the second two causes were retained. Out of the one came the view that there must be a link between A and B as per the following (as regards the efficient cause):
1. That laws exist in nature whereby the occurrence of B (effect) is dependent on the occurrence of A (cause).
2. That the occurrence of A is prior to, or at least simultaneous, and not after B.
3. That there is a continguity, or chain of events that connect A to B, or B to A. (Wikipedia – Causality)
and that there could be necessary, sufficient and contributing causes at the level of material cause.
However, the advance of quantum physics since the 1920s has forced a reconsideration of this rather simplistic approach to the issue of causality. What happened is that causality became a matter, not of determination but of probability. Prior to the last half century, the efficient cause was considered to be of greater importance than the material cause, thus seeing causality in deterministic terms. Now, however, the order was reversed. Further, the problem of the non-neutrality of the observer presented by psychology and quantum physics, along with the problems in Western philosophy regarding any clear judgment as to the link between what is observed and what is real, led to a a need for a different approach, one that worked with uncertainty, but provided a degree of certainty, that is, that could manufacture order out of seeming randomness.
The problem was particularly acute in medicine, long considered more of an art dependent on the clinical skills of the physician, which, in the modern era, produced a medical empiricism, in which what was used worked, often simply because what worked was what was used, meaning that the beliefs of the physician often determined the positive judgement regarding a medical procedure.
The change from the observational approach in medicine came about because of the work of one man in particular, A.B. Hill, who had trained as a doctor, but had to cut short his medical career for one involving research. He used a training in economics to devise a means to try to screen or control for various possibly biases. Initially intended simply to allow for a better judgment as to whether a factor was causal in nature or not, his list of criteria became essentially fixed as to what was causal or not. He then devised an experiment involving the random selection of test subjects to try to minimize the problem of operator bias.
Lacking any clear basis any more for asserting causality, Hill decided that causality could be expressed as a probability, namely by the probability of event B (effect) occurring in the presence or absence of event A. He himself admitted the problem of dealing satisfactorily with the issue of causality itself as a concept (“I have no wish, nor the skill to embark upon a philosophical discussion of the meaning of ‘causation.’” – Hill AB: The environment and disease: Association or causation? Proceed Roy Soc Medicine – London 1965, 58:295-300.)
This approach, termed “counterfactual causality” derives from the works of David Hume in the 18th Century. Humes scepticism about knowing anything lent itself to the use of statistics with their probability, which was seen as the closest one could get to determining causality in the modern era. The idea of randomisation was invented by R.A. Fisher and applied to agriculture in the 1920s and 1930s. Fisher was a good friend of Hill, though he later strongly criticised the Hill’s research on the link between smoking and lung cancer.
The counterfactual outcomes have no objective criteria of measurement. Thus, “the best option is usually to estimate population average effects.” (http://www.ete-online.com/content/2/1/11#B6) To overcome the bias effect due to clinical observation, Hill adopted the randomised selection process.
Hill’s approach was increasingly adopted in medicine as a solution for a growing epistmological problem, that of the problem of determining causality in a world of relativity and growing complexity due to rapidly growing biochemical discoveries about the processes of the body as well as the nature of disease (in this latter case, from the rather simple causal model of infectious disease as reflected in Koch’s postulates, to the more complex situation of modern chronic disease conditions where, as Hill himself held, several if not many factors were involved in one condition, or the reverse, one factor involved in several conditions).
However, Hill himself was undecided as to the value of his approach to determine causality: “None of my nine viewpoints can bring indisputable evidence for or against the cause-and-effect hypothesis.” and the cautionary notes he actually made were essentially tossed to the winds (http://www.ete-online.com/content/2/1/11#B6) And as other commentators on the issue of causality have stated, there really is no longer any causal criteria at all in epidemiology: “Causal inference in epidemiology is better viewed as an exercise in measurement of an effect rather than as criterion-guided process for deciding whether an effect is present or not.” (Rothman KJ, Greenland S: Causation and causal inference in epidemiology. Am J Publ Health 2005, 95:S144-150.)
Another careful observer of Hill’s work has argued that “the application of seven of [his] nine considerations…involve comprehensive causal theories…If complexity becomes very large, the uncertainty regarding whether or not to apply a given consideration can be expected to approach a decision made by coin toss. Thus, with increasing complexity, the heuristic value of Hill’s considerations diminishes.” (http://www.ete-online.com/content/2/1/11#B6) Indeed, Hill’s considerations were essentially treated a causal criteria. (Phillips CV, Goodman KJ: The missed lessons of Sir Austin Bradford Hill. Epidemiol Perspect & Innov 1965, 1:3.)
Thus, we have today a situation where a probability analysis is used to stand in for a true causal analysis, but also where that probability analysis is itself subject to increasing complexity in the light of advances in biochemistry, neurology, psychoneuroimmunology, genetics and epigenetics to mention a few areas, which complexity diminishes the validity of any such analysis given the inherent scope for bias due to the nature of the human mind. The very system that was designed initially to reduce operator bias and error is not capable of dealing with increasing complexity. The state of affairs is such that any drug tested is usually barely better than placebo, and there can be as many studies showing a negative outcome as positive. As one medical researcher commented: “So the FDA conducted an analysis of antidepressant drug trials and found that of 38 trials for which the evidence appeared favorable, 37 had been published. Whereas of 36 trials for which the evidence did not appear favorable toward antidepressant drugs, 22 were not published at all, and 11 were published in a way that misleadingly conveyed the outcome as though it was favorable.” (http://articles.mercola.com/sites/articles/archive/2010/06/12/beatrice-golomb-interview.aspx) This certainly reduces the value of conventional medicine’s claim to knowledge based on published studies.
While there are indeed influences from drug companies to bend studies their way, and there are tremendous conflicts-of-interest in that most drug trials are funded, at least in part, by pharmaceutical companies, and the problem of the regulators becoming captured by those they regulate, there is a fundamental problem at the heart of the RCT which makes it a very poor tool for determining the efficacy of drugs or vaccinations. The RCT has severe methodological and epistemological flaws.