Subscribe Bookmark RSS Feed

Kaplan-Meier Analysis

chittalr

Community Trekker

Joined:

Jul 30, 2014

I am very new to JMP. I have some data from SEER - which is a huge sample size and they are survival data. I was trying to create a Kaplan-Meier with that. But I don't know how to code/ prepare the data table for this. I am attaching the table I have. I will appreciate any help.

      

Time to eventTotal patients                Proportion Alive
0 mo42,9061
12 mo42,9060.938
24 mo42,9060.894
36 mo42,9060.867
48 mo42,9060.85
60 mo42,9060.839
72 mo42,9060.83
84 mo42,9060.822
96 mo42,9060.811

Thank You.

1 ACCEPTED SOLUTION

Accepted Solutions
Solution

You will need to prepare the data in three columns: Time, number died at a given time, and a censoring column.  The censoring column is needed to take into account the 81.1% of the patients who are still alive at the end of the trial.

With a little work, I was able to get your data into this form:

months    died    censor

0      .          0

12    2660    0

24    1888    0

36    1158    0

48    729    0

60    472    0

72    386    0

84    343    0

96    472    0

96    34797    1

I rounded the product of the proportion alive and total patients to get number died.  I think due to rounding error, the total number of died in this data doesn't exactly equal 42906.  Also, notice that I added a final row to account for the 34797 censored observations (patients still alive at 96 months).

From here, you can use the Survival platform (Y=months, Censor=censor, and Freq=died) or the Life Distribution platform (same roles).  If you use the Life Distribution platform, you will probably want to select the "Show Survival Curve" option from the red triangle menu in the Life Distribution report.

Both the Survival and Life Distribution platforms can be found in the Reliability and Survival submenu of the Analyze menu.

Hope this helps!

Michael

Michael Crotty
Sr Statistical Writer
JMP Development
2 REPLIES
Solution

You will need to prepare the data in three columns: Time, number died at a given time, and a censoring column.  The censoring column is needed to take into account the 81.1% of the patients who are still alive at the end of the trial.

With a little work, I was able to get your data into this form:

months    died    censor

0      .          0

12    2660    0

24    1888    0

36    1158    0

48    729    0

60    472    0

72    386    0

84    343    0

96    472    0

96    34797    1

I rounded the product of the proportion alive and total patients to get number died.  I think due to rounding error, the total number of died in this data doesn't exactly equal 42906.  Also, notice that I added a final row to account for the 34797 censored observations (patients still alive at 96 months).

From here, you can use the Survival platform (Y=months, Censor=censor, and Freq=died) or the Life Distribution platform (same roles).  If you use the Life Distribution platform, you will probably want to select the "Show Survival Curve" option from the red triangle menu in the Life Distribution report.

Both the Survival and Life Distribution platforms can be found in the Reliability and Survival submenu of the Analyze menu.

Hope this helps!

Michael

Michael Crotty
Sr Statistical Writer
JMP Development
chittalr

Community Trekker

Joined:

Jul 30, 2014

YearsDeath FrequencyColumn 3Column 4
000
16.20SEER
24.40SEER
32.70SEER
41.70SEER
51.10SEER
60.90SEER
70.80SEER
81.10SEER
881.11SEER

Thanks Michael. Your answer solved it for me.  The only change I made in the table is instead of numbers I converted them to percentages. Which gave me percentages on the Y axis.

Thank you.