Explaining temporal trends in annualized relapse rates in placebo groups of randomized controlled trials in relapsing multiple sclerosis

Abstract

Recent studies have shown a decrease in annualized relapse rates (ARRs) in placebo groups of randomized controlled trials (RCTs) in relapsing multiple sclerosis (RMS). This dissertation aimed to describe this trend in a different set of RCTs and to investigate whether patient baseline characteristics, eligibility criteria and other study design features could explain this phenomenon. A literature search of randomized, placebo-controlled trials in RMS offering data on relapses in placebo groups identified 56 suitable trials. Data on eligibility criteria and baseline characteristics were extracted and tested for significant trends over time. Several temporal trends were identified: The number of years considered for the calculation of pre-trial ARR as well as pre-trial ARRs themselves decreased, as did the duration of placebocontrolled follow-up. Pre-trial ARRs of the first and second preceding year before baseline showed major inconsistencies. Limits of eligible age as well as mean age increased, as did the mean disease duration, the number of eligibility criteria and the number of words and characters describing them, scores on the OQS, the number of treatment arms and the average number of patients per treatment arm. A meta-regression was conducted to estimate the contribution of these temporal trends to the decrease of trial ARRs over time. In the final meta-regression modeling of the trial placebo ARR, the date of publication was found to be insignificant in explaining the variation in trial placebo ARR, whereas pre-trial ARR, the number of years used to calculate pre-trial ARR, disease duration and the duration of follow-up became major contributors. In conclusion, the decline in trial ARRs most likely results from decreasing pre-trial ARRs and a shorter time period over which these were calculated. Increasing duration of illness may also contribute to the phenomenon.