OBJECTIVES: The baseline risk, reflecting pre-treatment conditions or characteristics, is crucial in determining treatment outcomes. The main goal of this research is to study how baseline risk connects to treatment benefits, giving insights into the complicated relationship between baseline risk and treatment effects in clinical trial analyses.

METHODS: In the methodology, we employed the framework proposed by Thompson et al. to integrate baseline risk (μi) as a covariate within each trial. Within this model, μi is treated as a random variable, allowing for its variability in each iteration of the Markov Chain Monte Carlo (MCMC) simulation. This approach adeptly addresses the inherent correlation between the intercept and slope of the model. A distinctive feature of this Bayesian formulation is its incorporation of the "true" baseline, estimated by the model, as a covariate, thereby automatically considering the uncertainty associated with each baseline risk. The implemented code resembles a meta-regression analysis, examining various factors. In this instance, instead of utilizing a variable named x[i] for each study, we employed μi to represent the baseline risk for that study. A minor adjustment was made to the calculation method to ensure precision in computer calculations, with no substantial impact on the overall results.

RESULTS: The examination of baseline risk revealed considerable overlap in the confidence intervals pertaining to the effectiveness of various therapies in preventing relapses. This implies a lack of robust statistical evidence indicating the significant superiority of one therapy over others.

CONCLUSIONS: A thorough examination of patient and study particulars, our findings propose that baseline risk serves as a surrogate for latent patient characteristics that may influence treatment efficacy. The employed methods, with a specific focus on patients with RRMS, revealed a noteworthy and significant decline in the ARR over recent years.