In a randomized controlled clinical trial (RCT), participant responses to interventions may be measured on an ordinal, rather than interval, scale. The study may also have a stratified design, such as randomization within centers, and collect information at multiple visits to evaluate treatment efficacy. While random baseline imbalances between the treatment groups are expected to be minimal due to randomization, baseline characteristics may be strongly associated with the outcome, and adjustment for them can improve power. The win ratio (ignores ties) and the win odds (accounts for ties) can be useful when analyzing these types of RCT data. This work presents randomization based methodology for covariance and stratified adjustment of the win ratio and the win odds for ordinal outcomes from a multi-visit clinical trial with stratified randomization. A large and small sample methodology is made available through a SAS macro and an R package. The software is illustrated for a multi-visit clinical trial with an ordinal outcome for a respiratory disorder. For this example, the presented methods are noted as more sensitive to the treatment differences than unadjusted counterparts.
