Results
Thirty-seven patients of the first 50 randomized patients received two
MADRS scores at each post-randomization study visit, one by a rater at
one of 12 clinical trial sites and by a Sponsor rater. A total of 113
pairs of paired MADRS assessments were conducted. The absolute
difference between the site rater and the Sponsor rater is shown in
Figure 1. Overall concordance between site and Sponsor raters was
93.4%. The intraclass correlation was 0.9837 and an
eta2 = 0.9918 (r = 0.9839, F = 121.91, p <
0.0001). The absolute mean difference in MADRS rating pairs was 1.78
points (95% CI: 1.50-2.06).
Seven of the 113 assessment pairs were discordant, i.e., the MADRS
scores differed by 4 points or more between the site rate and the
Sponsor rater. Of the 106 concordant pairs, 17 pairs had identical
scores, 36 pairs differed by one point, 36 pairs differed by two points,
and 17 pairs differed by three points (range 0 to 9). The seven
discordant pairs occurred across five clinical trial sites. In six of
seven discordant pairs, the site rater assigned a higher MADRS score
than the Sponsor rater. In each case of discordant pairs, the Sponsor
rater contacted the site rater to provide feedback about the MADRS
assessment.