+ Reply to Thread
Results 1 to 5 of 5

Thread: Survival Analysis: Unequal Follow-up Time

  1. #1
    Points: 20,006, Level: 89
    Level completed: 32%, Points required for next Level: 344

    Posts
    568
    Thanks
    50
    Thanked 20 Times in 19 Posts

    Survival Analysis: Unequal Follow-up Time




    Hi!

    I wanted to get some feed back for a study idea that's in the works. I haven't done much applied work in survival analysis, so hopefully you guys can chime in.

    Breakdown:
    We're interested in a particular kind of cancer. We have data from 2000-2005 of newly diagnosed cases. The main exposure is whether or not they were on a particular drug within the last 5 years prior to diagnosis. The primary outcome would be mortality. We have accurate mortality data up until Mar 31 2013.

    My question concerns unequal follow-up time or chance for follow up time. For example, if a patient is diagnosed in the earlier year, say 2000, they will have more time to be observed for death than someone whois diagnosed in 2005. What kind of issues or biases does this present into a kaplan-meier / cox model analysis? Are there work arounds? One idea that comes to mind would be to stratify by year of diagnosis? Or should I only follow each patient for a fixed amount of time?

    Any thoughts / feedback are welcomed!
    Thank you in advance!

  2. #2
    Human
    Points: 12,676, Level: 73
    Level completed: 57%, Points required for next Level: 174
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,362
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Survival Analysis: Unequal Follow-up Time

    Was the event of being sampled statistically independent of mortality? Then I believe that it is ok.

    Even if there is oversampling in risk groups, but if the response variable given the explanatory variables, is still independence of the sampling event I would guess that it is ok.

    That's one way of starting a discussion about this. I guess that this is a common problem in many areas.

  3. #3
    Points: 20,006, Level: 89
    Level completed: 32%, Points required for next Level: 344

    Posts
    568
    Thanks
    50
    Thanked 20 Times in 19 Posts

    Re: Survival Analysis: Unequal Follow-up Time

    Quote Originally Posted by GretaGarbo View Post
    Was the event of being sampled statistically independent of mortality? Then I believe that it is ok.

    Even if there is oversampling in risk groups, but if the response variable given the explanatory variables, is still independence of the sampling event I would guess that it is ok.

    That's one way of starting a discussion about this. I guess that this is a common problem in many areas.
    Thanks for the reply Greta. I'm a bit confused though, so maybe I'll give some additional detail about our study.


    I have about N=3,000 patients that were diagnosed between the years of 2000 and 2005. I'll be using their diagnosis date as point of entry into the study. The main exposure variable is whether or not they had used a particular drug within 5 years prior to their diagnosis. The outcome is death, for which I have the exact date if it did occurr. If they did not die, I censor at the end of the study window, Mar 31 2013.

    Thanks!

  4. #4
    Point Mass at Zero
    Points: 13,828, Level: 76
    Level completed: 45%, Points required for next Level: 222
    ledzep's Avatar
    Location
    Berks,UK
    Posts
    684
    Thanks
    188
    Thanked 143 Times in 139 Posts

    Re: Survival Analysis: Unequal Follow-up Time

    Such problems are common in medical field. One suggestion would be (as you also said) to compare the survival estimates at the least possible common time point, i.e. the time point when the data is available on all the patients (for eg, after 3 years of follow-up).

    I know its not easy. There are confounders such as time (improved treatment/facilities/surgical procedures greatly reducing the risk of mortality over time) and patient age (natural reasons) has a great effect on mortality. So, you can certainly examine mortality stratified by time.
    Oh Thou Perelman! Poincare's was for you and Riemann's is for me.

  5. The Following User Says Thank You to ledzep For This Useful Post:

    jamesmartinn (06-18-2014)

  6. #5
    Points: 20,006, Level: 89
    Level completed: 32%, Points required for next Level: 344

    Posts
    568
    Thanks
    50
    Thanked 20 Times in 19 Posts

    Re: Survival Analysis: Unequal Follow-up Time


    Quote Originally Posted by ledzep View Post
    Such problems are common in medical field. One suggestion would be (as you also said) to compare the survival estimates at the least possible common time point, i.e. the time point when the data is available on all the patients (for eg, after 3 years of follow-up).

    I know its not easy. There are confounders such as time (improved treatment/facilities/surgical procedures greatly reducing the risk of mortality over time) and patient age (natural reasons) has a great effect on mortality. So, you can certainly examine mortality stratified by time.
    Thanks for the input Ledzep! I ended up stratifying by year of entry into study.

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats