Following some discussion earlier on about the pos
Post# of 154560
(Total Views: 719)
Posted On: 04/11/2025 10:25:59 AM
Following some discussion earlier on about the possible trial methodology:
What will be very likely used is the Log-rank test (also called Mantel-Haenszel Test) to compare the survival distributions of two or more groups. Basically one starts with two arms, for example: Leronlimab + Trodelvy vs. Trodelvy. Let’s call them LT vs T.
The patients start to be treated and the time-to-events are recorded. These events obviously happen at different times and don’t have to be parametric, meaning it doesn't assume a specific distribution for the survival times. If, for example, one defines an event as a death of a person, every time there is a dismissal it is recorded and plotted in what are called Kaplan-Meier curves. Below an example from Wikipedia
https://en.wikipedia.org/wiki/Kaplan%E2%80%93Meier_estimator
This sounds complicated but it is just a plot against time of the surviving (or event-free) participants. If, for example, the LT arm is superior the curve will go down slower (les steeply) than the corresponding to the T group (and, hopefully won't go to zero, that is, surviving participants at the end of the trial).
Then a statistical package compares the two curves and produces the famous P number. If it is less than 0.05 we are golden.
In regard to the number of participants, this is calculated to ensure the trial has enough statistical power to detect meaningful differences between the groups and the practical constraints, such as budget and patient availability . The larger the groups the better the statistical power of the trial.
What will be very likely used is the Log-rank test (also called Mantel-Haenszel Test) to compare the survival distributions of two or more groups. Basically one starts with two arms, for example: Leronlimab + Trodelvy vs. Trodelvy. Let’s call them LT vs T.
The patients start to be treated and the time-to-events are recorded. These events obviously happen at different times and don’t have to be parametric, meaning it doesn't assume a specific distribution for the survival times. If, for example, one defines an event as a death of a person, every time there is a dismissal it is recorded and plotted in what are called Kaplan-Meier curves. Below an example from Wikipedia
https://en.wikipedia.org/wiki/Kaplan%E2%80%93Meier_estimator
This sounds complicated but it is just a plot against time of the surviving (or event-free) participants. If, for example, the LT arm is superior the curve will go down slower (les steeply) than the corresponding to the T group (and, hopefully won't go to zero, that is, surviving participants at the end of the trial).
Then a statistical package compares the two curves and produces the famous P number. If it is less than 0.05 we are golden.
In regard to the number of participants, this is calculated to ensure the trial has enough statistical power to detect meaningful differences between the groups and the practical constraints, such as budget and patient availability . The larger the groups the better the statistical power of the trial.
(17)
(0)