OBJECTIVES: In oncology trials, it is common practice to offer patients the possibility to switch from their randomized treatment to another active treatment upon disease progression. When patients switch to, and benefit from, active post-progression therapies which do not form part of the standard treatment pathway, a standard intention-to-treat (ITT) analysis may inaccurately estimate the ‘‘true’’ overall survival (OS) benefit associated with the investigational product [Latimer et al, 2014]. Using published data from the interim analysis of the PREVAIL study [Beer et al, 2014], we estimated the effects of enzalutamide compared to placebo on OS, adjusting for the potential confounding effects of switching to antineoplastic therapies that do not form part of the standard mCRPC treatment pathway in the UK. METHODS: Published treatment switching adjustment methods were applied including the Inverse Probability of Censoring Weights (IPCW) and a two-stage accelerated failure time model. The suitability of each method was evaluated by examining the study characteristics, the treatment switching mechanism, and by assessing the plausibility of the underlying assumptions of the models in the context of the PREVAIL study. RESULTS: Overall, 180/845 (21%) patients in the placebo arm and 129/872 (15%) patients in the enzalutamide arm of PREVAIL switched to non-standard antineoplastic therapies. The unadjusted ITT analysis for OS resulted in a hazard ratio (HR) of 0.706 (95% CI: 0.595 - 0.837) for enzalutamide compared to placebo, which was reduced to 0.662 (95% CI: 0.568 -0.809) using the two-stage method with no recensoring and 0.625 (95% CI: 0.517 - 0.752) using the IPCW method. CONCLUSIONS: Adjusting for non-standard subsequent therapies given in PREVAIL resulted in a larger estimated treatment effect on OS associated with enzalutamide. This suggests that receipt of non-standard subsequent antineoplastic therapies caused the ITT analysis to underestimate the true treatment effect of enzalutamide on OS.