ACRI

ISSUES OF STATISTICS IN CLINICAL TRIAL-ACRI

ISSUES OF STATISTICS IN CLINICAL TRIAL

ISSUES OF STATISTICS IN CLINICAL TRIAL:

In clinical trials, statistics is important to evaluate treatment effects.

Statistics is at the interface of mathematics and biology. Therefore it gives no certainties, onlychances. Let’s see what chances it gives. E.g. chances that hypotheses are true or untrue. We generally reject the null hypothesis of no effect at p less than 0.05.However,p = 0.05 means 5% chance of a type 1 error of finding a difference where there is none, and 50% chance of type 2 error of finding no difference where there is one. It picturizes very well how limited statistical interferences can be.

In addition to the risks of type1 and type 2 errors, there is the issue of little clinical relevance in spite of statistically significant findings. A sub-analysis of the SOI-VD study found no symptoms of angina pectoris in 85.3% of the patients on enalapril and in 82.5% of the patients on placebo (difference statistically significant at p less than 0.001)in situations like this,one has to wonder about the clinical relevance of the small difference. This is even more so when one considers that an active compound generally causes more side effects than does a placebo. Finally, we have to consider the points of bias. Arguments have been raised that controlled clinical trialsalthough they adjust for placebo effects  and time effects, are still quite vulnerable to the other biases e.g.,psychological biases, and selection biases.In clinical trials, as opposed to regular patient care,patients are generally highly compliant: their high compliance is an important reason for participating in the trials in the first place. They have a positive attitude towards the trial and anticipate personal benefit from it, a mechanism which is known as the Hawthorne effect. Alternatively, patients selected for a trial often refuse to comply with randomization which may render unrepresentative samples. Statisticians have great difficulties in handling such effects and are essentially unable to make sense of unrepresentative samples.

So the statistics has benefits as well as limitations.

February 7, 2018

0 Responses on ISSUES OF STATISTICS IN CLINICAL TRIAL-ACRI"

Leave a Message