A New Framework for Interpreting the Outcomes of Imperfectly Blinded Controlled Clinical Trials

It is well known that the outcome of an intervention is affected both by the inherent effects of the intervention and the patient's expectations. For this reason in comparative clinical trials an effort is made to conceal the nature of the administered intervention from the participants in the trial i.e. to blind the trial. Yet, in practice perfect blinding is impossible to ensure or even verify post hoc. The current clinical standard is to follow up the trial with an auxiliary questionnaire, which allows trial participants to express in closed form their belief concerning the intervention, i.e. trial group assignment (treatment or control). Auxiliary questionnaire responses are then used to compute the extent of blinding in the trial in the form of a blinding index. If the estimated extent of blinding exceeds a particular threshold the trial is deemed sufficiently blinded; otherwise, the strength of evidence of the trial is brought into question. This may necessitate that the trial is repeated. In this paper we make several contributions. Firstly, we identify a series of problems of the aforesaid clinical practice and discuss them in context of the most commonly used blinding indexes. Secondly, we formulate a novel approach for handling imperfectly blinded trials. We adopt a feedback questionnaire of the same form as that which is currently in use, but interpret the collected data using a novel statistical method, significantly different from that proposed in the previous work. Unlike the previously proposed approaches, our method is void of any ad hoc free parameters and robust to small changes in the participants' feedback responses. Our method also does not discard any data and is not predicated on any strong assumptions used to interpret participants' feedback. The key idea behind the present method is that it is meaningful to compare only the corresponding treatment and control participant sub-groups, that is, sub-groups matched by their auxiliary responses. A series of experiments on simulated trials is used to demonstrate the effectiveness of the proposed approach and its superiority over those currently in use.