Aims: The purpose of this study was to estimate the extent to which symptomatic progression mediates the association of treatment with PRO symptom scores in oncology.
Methods: Using simulated data modeled after results from a phase 3 clinical trial in prostate cancer, repeated measures mixed effects models of the FACT-P Symptom Index (FAPSI) were used to estimate the total effect (TE), the controlled direct effect (CDE), the controlled indirect effect (CIE), and the average marginal controlled direct effect (AMCDE) of treatment mediated by symptomatic progression.
Results: Progression was significantly lower in the treatment group compared to placebo. The TE of treatment on FAPSI score change was near null and non-significant. The CDE was also near null, but the CIE through progression showed a significant treatment effect. The AMCDE showed a modest, but non-significant treatment effect.
Conclusions: These results demonstrate an attenuation of the treatment effect in the mediation-naïve model (TE) compared to the mediation model (AMCDE), which highlights the importance of modeling causal pathways in determining an overall treatment effect on health-related quality of life.