Background For the evaluation of radiographic progression in RCTs, at least 2 readers should read all films in order to constrain measurement error. If differences in change scores between readers exceed a certain threshold, a third reader (adjudicator) can be used to obtain a more precise result. While the selection of a threshold that triggers this 3rd read is somewhat arbitrary (historically between 7 and 15 points), regulatory authorities suggest a cause for concern if 20% or more of the cases in a given clinical trial results in adjudication.
Objectives To provide data on which percent of sets of radiographs need adjudication for a predetermined cut-off of difference in change score from baseline between two readers in RCTs in patients with early and established RA.
Methods Fifteen datasets from radiographic trials in RA were scored by 13 readers as pairs according to the modified Sharp van der Heijde method (SHS). The reader was blinded to chronological sequence. The number of time points per study was 2 in 2 trials, 3 in 10 trials, and 4 in 3 trials.
We calculated the theoretical adjudication rates if adjudication thresholds from 3 to 20 units were selected per study plus the weighted mean for all studies. We investigated the influence of the number of time points within the same session, the length of the interval, and mean disease duration (early (≤3 years) versus established RA) on the adjudication rates. Statistical analysis was performed using SPSS software version 18.0.
Results 20.098 time points from 7.643 patients were included in the analysis. Median (range) sample size of the studies was 517 (103-901) patients, and the number of time points within one reading session was 2 for 1172 patients, 3 for 5296 patients and 4 for 1175 patients. Median (IQR) baseline radiological damage- and progression scores were 32 (18-48) and 1.05 (0.5-2.0) respectively.
The percentage of adjudicated cases was inversely related to the threshold. The higher the number of time points within the same reading session, and the longer the time interval, the higher the adjudication rate. The adjudication rate was also higher in RCTs with early vs established RA, especially at lower thresholds. In all cases, the adjudication rate remained below 19%, even with a very conservative threshold selection of 3 units.
Conclusions With trained and experienced readers, the adjudication rate in RCTs with patients with RA is low even with very conservative adjudication thresholds, which adds to the validity of the SHS scoring method as a precise and reliable outcome measure in RA. The adjudication rate is higher in RCTs with early- vs established RA.
Disclosure of Interest None Declared