What are double blind trials?

 Clinical research to test a treatment, may use a series of double blind trials, ideally conducted independently in different places. An active treatment is compared with an indistinguishable dummy treatment, without anyone knowing which is being used on a particular person. Only when the trial is over, do the participants find out who got what.

There is often an independent group who monitor the clinical trial as it proceeds, to ensure that it is ethically satisfactory to continue. Medical studies have been stopped because it has become likely that no firm answer could come of them, or because it was likely people were being harmed.

In some conditions, up to two thirds of people will benefit from a dummy treatment. Without double blind trials, any old treatment may then appear to be pretty good. The full title includes "placebo controlled" and a lot of medicine involves supposedly useful treatments which really have only this effect - encouraging the person to expect to improve, so that their natural healing is facilitated by the "power of the mind." There is absolutely nothing wrong with this, as long as the placebo (but supposedly active) treatment is safe.

Some people reading this will immediately respond there is something wrong, as this is encouraging dependency of the patient on the (God) doctor rather than helping them to own their own power. My response is yes, true, but it's horses for courses. I hope that a person will move towards taking at least an equal share in decisions and believing in their body's healing capacity. In the meantime, I'll play the part of the "healer."

However, read on...

Quoting from a lecture published in the British Medical Journal1, Dr. M. Weatherall wrote...
"Sometimes double blind trials produce extraordinary results, such as the notable success of the dummy preparation with complete failure of the reputed remedy under trial - for example, Baker and Thorpe (1957)...it transpired that the theoretical conditions of a double blind trial - namely, that neither the patient nor physician knew which tablets contained active remedy - did not apply....the placebo was sweet and the tranquilizer was bitter, and, on the authors' interpretation, the sweetness of the placebo was responsible for the improved behaviour of the (schizophrenic) patients. ...It is unusual to find a double-blind trial in which the authors report either an experiment to test whether the supposedly indistinguishable preparations were in fact indistinguishable by all the procedures open to patients (including chewing the tablets and inspecting the crushed fragments or emptied capsules) or estimations on body fluids to determine whether the drugs administered had actually been consumed. in the absence of evidence on these points it is always legitimate to question whether a supposedly double-blind trial really was so."

These days his second condition, of checking that treatment is actually taken, is done often (by pill counts or blood tests.)

On random allocation

It would be easy to bias a trial by having the treatment group in better shape to start with. To avoid this people are allocated at random to treatment or placebo. Most reports detail how this worked out, tabulating comparisons of the two groups to show they were very similar.

The easy way to check is to compare numbers in the two groups, which should be approximately but not exactly the same.

The following is from the title and abstract of a paper published in 2012.

"Randomized, Double-Blind, and Placebo-Controlled Clinic Report of... In this paper, 90 patients with coronary heart disease or cerebral infarction were randomly divided into two groups, 60 in the treatment group and 30 in the control group, and..."

Pretty dodgy randomization.

Multiple determination and synergy

Life is complicated. Medical trials have to be simple. The people in the trial have to be selected so that the result can be extrapolated to other similar groups of people, and really can only be assumed applicable to similar groups - not everyone.

Multiple factors are involved in any outcome, and often work in cahoots. If two keys are needed to open a casket, testing one will be fruitless. Medical research trials use single treatments for comparisons, as testing multiple combinations would be too expensive and require too many participants.

Mendelian Randomization

Say we wish to study the effect of low vitamin D levels. Lots of things can alter our vitamin D level and also alter the other things one is interested in as possibly caused by the low vitamin D.

Smoking, physical activity, weight and one's income are all associated with lower vitamin D. Any or all of them could be responsible for some other illness being more if vitamin D were low. It might have absolutely nothing to do with the low vitamin D.

It just happens that two of our genes have inherited variants (alleles) causing low vitamin D whatever the lifestyle we lead. They are DHCR7 and CYP2R1. That is all the polymorphisms of these genes do, so one can measure them and get a pure effect of low vitamin D quite apart from all the other confounding things mentioned above.

This has been done and showed low vitamin D (measured by the genetic polymorphisms) is likely to contribute to causing cancer but not heart disease.

It's as though nature had done a randomized trial of low vitamin D, a bit like the experiments conducted with peas by Gregor Mendel.

Genetically low vitamin D concentrations and increased mortality: mendelian randomisation analysis in three large cohorts BMJ 2014

On safety

Trials run over limited times and involve limited numbers of individuals. Serious adverse effects of treatments may only show up with widespread community use, as was the case with Thalidomide and Vioxx. There is one school of thought which recommends using the "seven year rule" - waiting this time before prescribing new drugs.

It has just been reported (December 2014) that tramadol can occasionally cause hypoglycaemia, severe enough to require hospital admission. This analgesic drug was apparently marketed first in 1977.
This wasn't seen in the clinical trials as they didn't include enough people for such a rare event to show.

And efficacy over long periods

Medications tested for maybe 5 or 10 years are then prescribed for a lifetime, with no direct trial evidence for benefit continuing beyond the duration of the trial. Statistical analysis of results is often presented to show whether benefits are changing with time over the course of the trial, but one is still flying blind later.


1. Editorial on tranquilizers in BMJ Saturday May 5th 1962

Click to return from double blind trials to main index home page.


Is there something else you would like to read about?
This search button will bring up anywhere on this site your words are mentioned...


Custom Search


The contactpage.

If you are interested in a home study course on examination of the spine, please send me your e-mail address by the contact form.