Evaluating blinding in trials of endometrial scratching: a randomised nested survey within two randomised controlled trials.

Abstract

Background: Blinding of participants in randomised controlled trials is considered a robust method to protect against biases associated with departures from the intended intervention and measurement of the outcome. Even when implemented, the success of blinding is not guaranteed, and participants may be able to discern their allocation through the experience of medication side effects or procedural discomfort. To assess the effectiveness of blinding, formal evaluation tools such as the James and Bang blinding indices have been developed.

Aim: To explore whether participants can be effectively blinded in randomised controlled trials involving physical procedures, using endometrial scratching as a case study. Secondary aims were to measure whether symptoms from the procedures (pain, bleeding, and pregnancy outcome) are associated with unblinding, and whether behaviours of trial participants (intercourse patterns and protocol violations) may manifest as a result of unblinding.

Methods: A randomised survey was conducted during two existing trials of endometrial scratching compared to a placebo procedure, in couples attempting to conceive from intercourse. Participants were asked to guess their trial allocation and to provide a reason for their guess. Participants were randomised to guess their allocation either early (within 1 week of study intervention) or late (after their window to conceive had closed). Blinding was assessed using both Bang's and James' blinding indexes, and the association of participant guesses and the measured symptoms and behaviours was evaluated by chi-squared test and the independent Student's t-test.

Results: Of 278 participants, 56% correctly guessed their allocation, 32% guessed incorrectly, and 21% selected 'don't know'. Pain and bleeding were significantly higher in the endometrial scratching arm and were associated with correct guessing. Blinding indices suggested a lack of blinding, particularly in the intervention group (James' blinding index 0.42 (0.36-0.48)). No differences were observed in intercourse timing or protocol violations between groups.

Conclusion: The sham procedure used in these studies does not appear to have achieved complete blinding; however, there was no evidence that lack of blinding led to impacts on important trial behaviour, such as frequency of intercourse.

Trial registration: ACTRN12614000657628, ACTRN12614000656639. Registered on June 24 2014.