The Nature of Scientific Evidence

Roy Walford in the opening chapter of his book entitled “The 120 Year Diet” (Publishers Simon and Schuster, New York, 1986) outlines the four categories of evidence that medical researchers use for making recommendations for any new medical treatments (pages 33-37). These categories are: testimonial, argumentative, correlational and experimental.

The third category, namely correlational evidence, implies that if two events occur together all or most of the time, this demonstrates a cause and effect relationship. The current dilemma hitting the newspapers - multiple sclerosis and narrowed neck veins - or to use the scientific terminology "MS and CCSVI (chronic cerebrospinal venous insufficiency)” is an example.

According to a recent article in our local newspaper "Narrowed veins’ link to MS questioned" (Winnipeg Free Press, April 15, 2011), data from a University of Buffalo study suggests little correlation between venous obstruction and MS. Because investigators have found patients with MS who do not have the anatomical abnormality of CCSVI, and others without MS who do have it, they are questioning the validity of the recommended surgical treatment (a vein-opening procedure known as the "liberation treatment").

There has been a similar "anatomical correlation" drawn between heart attacks and narrowed coronary arteries. Surgical procedures like coronary bypass, stents, ballooning etc. have been globally and routinely performed on patients who have suffered a heart attack or are considered at risk. The heart attack correlation is more complex because of the cholesterol correlation - or shall I say lack of it - when we find patients with elevated cholesterol who don’t get heart attacks, and conversely others with low cholesterol who do. This biochemical discrepancy does not deter the medical profession from writing millions of prescriptions for cholesterol lowering drugs.

As Walford mentions in his first chapter: “Correlation can be seductive, but it's not actual proof.” I still vividly recall attending a seminar in the 1980’s given by a visiting British cardiologist at the Manitoba Medical College. He projected a graph on the screen which showed a direct correlation between the incidence of death from heart attacks in England with the sale of colored televisions. Both showed the same increase in incidence (slope) over the years, as well as the same dip during one specific year! Does this correlation necessarily imply causation? It seems to me that many treatments in medicine are still based on correlational evidence.

Walford’s fourth category, namely experimental evidence, is the scientific "gold standard". As he points out on page 36, the experimental approach establishes "... the causal nature of a relationship (cause and effect ) between two phenomena." However, because of moral, ethical and legal restrictions, designing repeatable experiments with the necessary blind controls can make the experimental model difficult to achieve.