Keywords: Patient-Reported Outcomes, Rasch Measurement Theory Models, Response Shift, Longitudinal
The growing interest in the analysis and interpretation of longitudinal Patient-Reported Outcomes (PRO) measures logically appears in the context of chronic diseases where patients have to regularly adapt to their illness. Thus, patients might give different answers on PRO measures over time, not only because their health has changed, but also because their perception of what health or quality of life means to them has changed. This phenomenon referred to as response shift (RS) needs to be accounted for to provide more insight on patients’ adaptation to illness and reliable measurements of RS effects and PRO changes. The "RespOnse Shift ALgorithm in Item response theory” (ROSALI) allows for RS analyses in the individual items that comprise a PRO using longitudinal Item Response Theory Models including RS detection and adjustment assuming that patients experience RS in the same way and magnitude on average which can be questioned. It is likely that RS and its effects on PRO changes can be diversely impacted by covariates (medical/psychological). ROSALI is therefore extended to take into account the effects of covariates on RS and PRO changes.