GraphPad Statistics Guide

Example of survival data from a clinical study

Example of survival data from a clinical study

Previous topic Next topic No expanding text in this topic  

Example of survival data from a clinical study

Previous topic Next topic JavaScript is required for expanding text JavaScript is required for the print function Mail us feedback on this topic!  

Here is a portion of the data collected in a clinical trial:

Enrolled

Final date

What happened

Group

07-Feb-98

02-Mar-02

Died

Treated

19-May-98

30-Nov-98

Died

Treated

14-Nov-98

03-Apr-02

Died

Treated

01-Dec-98

04-Mar-01

Died

Control

04-Mar-99

04-May-01

Died

Control

01-Apr-99

09-Sep-02

Still alive, study ended

Treated

01-Jun-99

03-Jun-01

Moved, off protocol

Control

03-Jul-99

09-Sep-02

Still alive, study ended

Control

03-Jan-00

09-Sep-02

Still alive, study ended

Control

04-Mar-00

05-Feb-02

Died in car crash

Treated

 

And here is how these data looked when entered in Prism.

 

Prism does not allow you to enter beginning and ending dates. You must enter elapsed time. You can calculate the elapsed time in Excel (by simply subtracting one date from the other; Excel automatically presents the results as number of days).

Unlike many programs, you don't enter a code for the treatment (control vs. treated, in this example) into a column in Prism. Instead you use separate columns for each treatment, and enter codes for survival or censored into that column.

There are three different reasons for the censored observations in this study.

Three of the censored observations are subjects still alive at the end of the study. We don't know how long they will live.

Subject 7 moved away from the area and thus left the study protocol. Even if we knew how much longer that subject lived, we couldn't use the information since he was no longer following the study protocol. We know that subject 7 lived 733 days on the protocol and either don't know, or know but can't use the information, after that.

Subject 10 died in a car crash. Different investigators handle this differently. Some define a death to be a death, no matter what the cause. Others would define a death from a clearly unrelated cause (such as a car crash) to be a censored observation. We know the subject lived 703 days on the treatment. We don't know how much longer he would have lived on the treatment, since his life was cut short by a car accident.

Note that the order of the rows is entirely irrelevant to survival analysis. These data are entered in order of enrollment date, but you can enter in any order you want.