An interesting but very political example of the misuse of statistics. In my school days probability theory was minimally taught. My first meeting with statistics was when my University decided, in their wisdom, that statistics were an essential element of a veterinary degree. Thus in 1953. I was introduced to Prof. Bradford Hill’s book on statistics for medicine and ever since I have been involved with the subject.
My view of statistics is more in relation to its use in medicine and particularly with its use in demonstrating significance of benefits from drugs and procedures and the relevant efficacy frequently by the use of Odds and Hazard ratios (ORs&HRs). While these are acceptable for demonstrating significance between two treatments, the inflated relative rates (RRs) generated are not. They (RRs) do not reflect the true probability of benefit (or its complement, the probability of no benefit) to the individual patient. The statement “If x million treated, y thousands will be saved” is simply irrelevant to me as a patient. The number that is very relevant is y/x * 100 as it defines the efficacy rate of the therapy, the probability of benefit and the probability of no benefit. With these numbers and the true incidence of adverse reactions, I can base a decision on the therapy. To often, and particularly in meta-analyses, the real numbers are left out so that the true rates cannot be calculated.
Unfortunately, the published RRs are far more attractive for advertising purposes for increasing sales. As Ioabnnidis has pointed out (Ioannidis JPA (2005) Why most published research findings are false. PLoS Med 2(8): e124. downloadable in toto) most published research is flawed in some way. Often this is related to hidden “confidential” data.
Rapid Response:
Re: David Oliver: Lies, damned lies, and the NHS
David Oliver - Lies, damned lies, and the NHS
An interesting but very political example of the misuse of statistics. In my school days probability theory was minimally taught. My first meeting with statistics was when my University decided, in their wisdom, that statistics were an essential element of a veterinary degree. Thus in 1953. I was introduced to Prof. Bradford Hill’s book on statistics for medicine and ever since I have been involved with the subject.
My view of statistics is more in relation to its use in medicine and particularly with its use in demonstrating significance of benefits from drugs and procedures and the relevant efficacy frequently by the use of Odds and Hazard ratios (ORs&HRs). While these are acceptable for demonstrating significance between two treatments, the inflated relative rates (RRs) generated are not. They (RRs) do not reflect the true probability of benefit (or its complement, the probability of no benefit) to the individual patient. The statement “If x million treated, y thousands will be saved” is simply irrelevant to me as a patient. The number that is very relevant is y/x * 100 as it defines the efficacy rate of the therapy, the probability of benefit and the probability of no benefit. With these numbers and the true incidence of adverse reactions, I can base a decision on the therapy. To often, and particularly in meta-analyses, the real numbers are left out so that the true rates cannot be calculated.
Unfortunately, the published RRs are far more attractive for advertising purposes for increasing sales. As Ioabnnidis has pointed out (Ioannidis JPA (2005) Why most published research findings are false. PLoS Med 2(8): e124. downloadable in toto) most published research is flawed in some way. Often this is related to hidden “confidential” data.
There is an excellent book by Gigerenzer “Risk Savvy: How to make good decisions” which I would recommend to anyone interested in probabilities and requiring to make decisions relating to self or to patients. Also because of a report by Starbridge on iatrogenic deaths (Starbridge B JAMA, July 26, 2000—Vol 284, No. 4) in the US (possibly similar to the UK per million population) better decisions would lead to reduced unnecessary deaths.
http://www.ncbi.nlm.nih.gov/pubmed/?term=JAMA+2000+284%2C+4+483
http://jama.jamanetwork.com/article.aspx?articleid=1104133
http://jama.jamanetwork.com/article.aspx?articleid=182076
Competing interests: As a patient I depend on local doctors for "right" treatment