"Statistically Significant" Statistical Nonsense
In chiropractic college I was taught a test that can predict who will have stroke or heart disease. Out of 100 people tested 95.5% of people who perform poorly on the test do not have heart disease over a ten year period, but 4.5% do have a heart event. Of those who perform well on the test, 99.9% do not have heart disease, but over .64% do have a heart event.
It is obvious that your risk of having a heart event is 7 times as high if you test poorly. The question: Should we treat 100% of the population with poor test scores to attempt to help the 4% who will have heart disease because of the risk factor?
The same test will also predict stroke. 98.6% of those who score poorly will not have stroke compared to 99.9% of those who score well. The relative risk is 15 times as great. In other words you are 15 times more likely to have a stroke if you score poorly on the test than if you score well. Should we treat 100% of the population who score poorly to keep the 1.4% from having a stroke?
I think maybe NO! The test is blood pressure and the research is cited in my book on the science of medicine. Hypertension is certainly statistically significant. Is treating hypertension statistical nonsense?
This topic is more thoroughly explained in
Exposed The "Science" of Medicine and the Dollars It Generates