The matching quality of experimental and control interventions in blinded pharmacological randomised clinical trials: a methodological systematic review

Table 3 Main and supplementary results

Paper	Main results: proportion of trials with inadequate matching based on the conclusion of the publications	Supplementary results: Proportion of trials with inadequate matching based on high/low thresholds^a
Paper		High threshold	Low threshold
Hill and collegues, 1976	7 of 22 trials	6/22 trials	14/22 trials
Blumenthal and colleagues, 1974	5 of 6 trials	5/6 trials^b	5/6 trials
Walter and colleagues, 2005	1 of 1 trial	0/1 trial	1/1 trial
Fai and colleagues, 2011	0 of 1 trial	0/1 trial	1/1 trial^c
Dupin-Spriet and colleagues, 1993	1 of 3 trials	0/3 trials	3/3 trials
Wen and colleagues, 2004	1 of 1 trial	1/1 trials	1/1 trials
Farr and colleagues, 1987	1 of 2 trials	0/2 trials^d	1/2 trials^e

^aThe high thresholds are in most cases based on at least 75 % of assessors finding clear differences. The low thresholds are in most cases based on 25 % of assessors finding clear differences
^bTwo groups of assessors were involved. In one group (patient-simulated), only in one trial did 75 % or more of assessors find clear difference between experimental and control intervention. In the second group of assessors (experimenter-simulated), over 80 % of assessors found clear differences in five out of six experimental and control drugs
^cTaste was excluded from the study because “taste was not adjusted with bitter agent” though the experimental intervention most likely had a more bitter taste
^dBased on chance alone, 50 % of assessors would be expected to correctly guess intervention types. We defined high threshold to imply that at least 75 % of assessors randomized to the experimental intervention perceived being on the experimental intervention
^eSignificantly more than 50 % of assessors randomized to experimental group perceived to have received the experimental intervention

ISSN: 1471-2288