Discussion: View Thread

  • 1.  Greenwood's estimate of standard deviation of survival function

    Posted 05-23-2012 17:53
    To whom it may concern:

    Does anyone know about "Greenwood's estimate of standard deviation of survival function " ?

    Any good reference would be helpful !

    THank you !

    Cindy Weng

    -------------------------------------------
    Statistician
    -------------------------------------------


  • 2.  RE:Greenwood's estimate of standard deviation of survival function

    Posted 05-23-2012 18:07
    Yes, the original publication dates back to 1926.   Some useful reference texts that discuss it include:

    Collett, D.   Modelling Survival Data in Medical Research, Chapman & Hall / CRC, 1994.

    Lawless, JF.   Statistical Models and Methods for Lifetime Data, 2nd Edition.   Wiley, 2003.

    Hosmer DW, Lemeshow S, May S.   Applied Survival Analysis, 2nd Edition.   Wiley, 2008.

      I hope that will be helpful to you.


    -------------------------------------------
    Lance Heilbrun
    Karmanos Cancer Institute
    -------------------------------------------








  • 3.  RE:Greenwood's estimate of standard deviation of survival function

    Posted 05-23-2012 18:25
    This is the value of this group. I used that estimator many years ago, but I forgot what it is. Nice to be reminded of it. 

    Best wishes,

    Nayak



    -------------------------------------------
    Nayak Polissar
    Principal Statistician
    The Mountain-Whisper-Light Statistics
    -------------------------------------------






    Original Message


  • 4.  RE:Greenwood's estimate of standard deviation of survival function

    Posted 05-23-2012 18:55

    Just to comment on this.  In my text "The Essentials of Biostatistics for Physicians, Nurses, and Clinicians on page 165 I provide the Greenwood estimate of standard error of the KM curve and show how it is used to get confidence intervals for the survival estimate at a givn point in time.  There is also a simpler formula due to Peto that can be used for an alternative method for an approximate standard error and confidence interval.  I mention a paper by Korn and Dorey that explain the pros and cons of Peto's method vs Greenwood.  In my experience with the FDA they have a slight preference for Peto's interval because of its conservative tendency.
    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------






    Original Message