Comparison of two clinical severity scoring systems in two multi-center, developing country rotavirus vaccine trials in Africa and Asia

Abstract

Clinical severity scoring systems are used in rotavirus vaccine efficacy and effectiveness studies to define the primary endpoint, severe rotavirus gastroenteritis (RVGE). Understanding how scoring systems perform in diverse settings is critical for proper design and interpretation. This investigation aims to understand how the Vesikari scoring system (VSS) and Clark scoring system (CSS) categorize severe disease among children under 2 years of age using data from two Phase III efficacy trials conducted in five developing countries in Africa and Asia. Signs and symptoms were collected on trial participants who presented to a medical facility with study-defined gastroenteritis. Severity scores were calculated using pre-established VSS and CSS criteria and compared to identify differences in the proportions of severe RVGE within regions and sites, and by gender and age. In Africa and Asia, 40.6% and 56.0% of rotavirus-positive episodes were severe according to the VSS, while 9.5% and 6.3% of episodes were severe according to the CSS (Fisher's Exact, p ≤ 0.001). Using the mean scores in these trials (VSS: ≥ 10 Africa, ≥ 11 Asia; CSS: Africa and Asia ≥ 10) as the severity thresholds, agreement between scoring system severity classifications improved substantially within each region (Africa: kappa = 0.67; Asia: kappa = 0.78) as compared to the original severity classification (Africa: kappa = 0.27; Asia: kappa = 0.10). Using the mean score, 17.1% and 9.5% of severe VSS cases in Africa and Asia, respectively, were classified as not severe according to the CSS and 14.7% and 9.5% of severe CSS cases in Africa and Asia were classified as not severe according to the VSS. The two scoring systems performed differently among developing country populations in Africa and Asia, with the VSS classifying more cases as severe in both regions. One accurate and reliable scoring system should be developed and implemented for all trials so that results may be more comparable.