On 2013 Jul 01, lu tian commented:

The authors have made the timely call for moving beyond the traditional statistical analysis plan that “within the tried and true comfort zone for clinical acceptance and regulatory approval” in the setting of survival analysis with re-current events. In this case, the “comfort zone” is the so-called time to the first event analysis which is not only often underpowered but also completely fails to characterize the treatment effect beyond the first event. The true difficulty to go beyond the “comfort” zone is in summarizing the whole patient experiences during or after the treatment in an objective clinically meaningful way. For example, how to weigh the relative importance of non-fatal events compared with death? This is a difficult task and probably we will never be able to come up with a perfect consensus.

However, the authors convincingly conveyed the message that albeit those difficulties, clinicians and statisticians need to be bold to go beyond the simple “time to the first event analysis” to address the clinical meaningful question which is to understand the treatment effects on entire disease burden characterized by both non-fatal and fatal events in this case. The initial solution may not be perfect but at least we should answer the right question.

On 2013 Jul 01, lu tian commented:

The authors have made the timely call for moving beyond the traditional statistical analysis plan that “within the tried and true comfort zone for clinical acceptance and regulatory approval” in the setting of survival analysis with re-current events. In this case, the “comfort zone” is the so-called time to the first event analysis which is not only often underpowered but also completely fails to characterize the treatment effect beyond the first event. The true difficulty to go beyond the “comfort” zone is in summarizing the whole patient experiences during or after the treatment in an objective clinically meaningful way. For example, how to weigh the relative importance of non-fatal events compared with death? This is a difficult task and probably we will never be able to come up with a perfect consensus.

However, the authors convincingly conveyed the message that albeit those difficulties, clinicians and statisticians need to be bold to go beyond the simple “time to the first event analysis” to address the clinical meaningful question which is to understand the treatment effects on entire disease burden characterized by both non-fatal and fatal events in this case. The initial solution may not be perfect but at least we should answer the right question.