OBJECTIVES: Often patient-reported outcomes are used to understand a patient’s change from baseline at specific time points throughout trial follow-up. Group-based trajectories give an informative perspective of patient worsening and improvement patterns over time. This exercise evaluates the findings of Jensen-Battaglia et al.¹ to inform the validity of conducting group-based trajectory modelling (GBTM) analyses across different patient-reported outcome (PRO) measures collected in acute myeloid leukaemia (AML) studies. Furthermore, this study aims to assess how patient trajectories may inform clinical decision-making.

METHODS: The Jensen-Battaglia et al. publication was peer reviewed to assess the parameter values chosen for the model, and their associated justifications. An evaluation of the assumptions and limitations of each parameter choice was performed. Subsequent implications on demographic and clinical characteristics of each trajectory group were summarized to understand how patient trajectories may inform clinical decision making and intervention.

RESULTS: Unique patient trajectories may be impacted by pre-specified model parameters such as polynomial degree and censoring distributions, choice of covariate adjustment, and probabilistic model selection with Akaike/Bayesian information criterion. Clinical feedback was used to support model selection, ensuring that the number of groups reflected the clinical experience observed in patients with AML. Model limitations such as data missingness were discussed to improve the accuracy and strengthen the validity of using GBTM for PRO data.

CONCLUSIONS: The results of this appraisal allow researchers to leverage these learnings in new applications of GBTM using PRO data. Parameter choice may influence resulting trajectories; therefore, clinical expertise is critical to ensure group patterns are reflective of the patient experience. This publication by Jensen-Battaglia et al. highlights key learnings that may be extended across varied PRO measures and indications.