Reply to letter to editor: “RespOnse Shift ALgorithm in item response theory (ROSALI) for response shift detection with missing data in longitudinal patient-reported outcome studies”

Excerpt

In an insightful Letter to Editor, Gunn HJ [1] critiques our algorithm ROSALI for response shift (RS) analysis using Item Response Theory (IRT) [2]. Four points are raised highlighting some of the imperfections and pitfalls of this version of ROSALI on which progress has been made since its first development leading to a new version of ROSALI that has recently been assessed [3]. To put things into perspective, we can first recall that when ROSALI was initially developed, the Structural Equation Modeling (SEM)-based method proposed by Oort [4] was the only available method based on latent variable models to detect RS (at the domain-level). ROSALI was, to our knowledge, the first IRT-based method to explore RS at item-level, leaving room for improvement, as we highlight below by following the four points raised. …