+ Reply to Thread
Results 1 to 7 of 7

Thread: Survival Analysis - Time to utilization of medical services

  1. #1
    Points: 1,283, Level: 19
    Level completed: 83%, Points required for next Level: 17

    Posts
    3
    Thanks
    2
    Thanked 0 Times in 0 Posts

    Survival Analysis - Time to utilization of medical services




    Hi all,

    I work in the medical field and Iím looking into using survival analysis calculate time to utilization of medical services, that is, once we have a member enrolled in our medical insurance product, how long does it take until this member sees a doctor for the first time. One detail about this is that these members can join the insurance at any time in the year.

    Iím not sure which approach to use, Iím thinking about two, and would love your assistance on how to proceed.

    Approach 1 (fixed time frame for all members):
    -Have a fixed block of time for the study period, letís say full year 2015 (January through December). Members can enroll at any point within this time period.
    -Let Length of Follow Up (LOF) be the difference in days between the date the member first used medical services and member enrollment date.
    -Observations would be considered censored if members havenít used services after 12/31/2015.

    Approach 2 (dynamic time frame for all members):
    -Have a fixed block of time for the enrollment period, letís say full year 2015 (January through December). Members enrolled after 12/31/2015 would not be included in the study.
    -Let Length of Follow Up (LOF) be the difference in days between the date the member first used medical services and member enrollment date.
    -Now here is the difference in approaches. Letís say a member enrolls in 03/05/2015, then we would follow up with this member for 1 year (that is, until 03/05/2016). This observation would be censored if heíd not used any medical services after 03/05/2016. Another example, letís say a member enrolls on 09/24/2015, we would follow up with this member for 1 year (that is, until 09/24/2016). This observation would be censored if heíd not used medical services after 09/24/2016.
    -Hence, the time frame for enrollment is 01/01/2015 through 12/31/2015, but the time frame to be considered for each member would be of 1 year based on member date of enrollment, which varies per member.

    **************

    Whatís the best approach to do this? Are any of these approaches appropriate at all? Does it matter? Is there a better approach? Also consider that Iíd like to use covariates such as age, race, gender, marital status and so on.

    I will be doing this in SAS, so which of the survival analysis procedures would be the most appropriate to use? Would it be the PHREG procedure (i.e. Coxís Partial Likelihood)?

    Your help would be greatly appreciated. Please let me know what you think.

    Thanks,

    Alex

  2. #2
    TS Contributor
    Points: 14,811, Level: 78
    Level completed: 91%, Points required for next Level: 39
    Miner's Avatar
    Location
    Greater Milwaukee area
    Posts
    1,171
    Thanks
    34
    Thanked 405 Times in 363 Posts

    Re: Survival Analysis - Time to utilization of medical services

    From the analytical perspective, either approach is acceptable. It may change the type of censoring and consequently the type of analysis, so it may impact whether it can be done in SAS (sorry, not a SAS user). In Approach 2, will you have the actual date that service was used, or will it be within an interval of time? Interval censoring can still be analyzed, but different methods must be used and the errors are greater.

  3. The Following User Says Thank You to Miner For This Useful Post:

    avbraga (03-13-2017)

  4. #3
    Points: 1,283, Level: 19
    Level completed: 83%, Points required for next Level: 17

    Posts
    3
    Thanks
    2
    Thanked 0 Times in 0 Posts

    Re: Survival Analysis - Time to utilization of medical services

    Hi Miner, Thank you so much for the reply. Yes, in Approach 2, I would have both dates (enrollment date and date when medical service was first used), this way I am able to calculate Length of Follow Up (LOF). The observation period for each individual member would be 1 year from his enrollment date. If the member didn’t use medical services after his 1 year follow up, then that would be considered a censored observation.

    Thanks for shedding light on this question. I wanted to confirm I was approaching this correctly (though Approach 1 is simpler and I may stick to that). I have a SAS book on survival analysis, and will be able to get things done by following its instructions.

    Again, thank you for the help!

  5. #4
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: Survival Analysis - Time to utilization of medical services

    If you are doing survival analysis you might want to look at "Survival Analysis Using SAS: A practical Guide" 2nd ED by Paul Allison. Its very detailed in both the stats and the code.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  6. The Following User Says Thank You to noetsi For This Useful Post:

    ondansetron (03-13-2017)

  7. #5
    Points: 1,283, Level: 19
    Level completed: 83%, Points required for next Level: 17

    Posts
    3
    Thanks
    2
    Thanked 0 Times in 0 Posts

    Re: Survival Analysis - Time to utilization of medical services

    Hi Noetsi, that's the book I have.

  8. #6
    TS Contributor
    Points: 14,811, Level: 78
    Level completed: 91%, Points required for next Level: 39
    Miner's Avatar
    Location
    Greater Milwaukee area
    Posts
    1,171
    Thanks
    34
    Thanked 405 Times in 363 Posts

    Re: Survival Analysis - Time to utilization of medical services

    Your scenario is very similar to what I deal with when analyzing field return data. I initially organize the data in what is called the Nevada format. To see how this pertains to your situation, refer to the attached image.

    I don't know how SAS handles this. In Minitab, this is referred to as arbitrary censoring. This means that you do not have all subjects starting at the same time, but on a rolling start. While you can analyze the data assuming the same start time and ignoring the rolling start, incorporating the rolling start allows you to make predictions on future events (i.e., enrollments) at specific times. This is obviously very useful when predicting future warranty returns.

    p.s. In the linked article, Suspension = Censored
    Attached Images  

  9. The Following User Says Thank You to Miner For This Useful Post:

    avbraga (03-13-2017)

  10. #7
    Points: 1,741, Level: 24
    Level completed: 41%, Points required for next Level: 59

    Posts
    230
    Thanks
    37
    Thanked 68 Times in 59 Posts

    Re: Survival Analysis - Time to utilization of medical services


    Excellent book! I used/am using it myself to get an intro to survival analysis and also have the Hosmer and Lemeshow Applied survival analysis book for when I graduate from Allison's book.

    Quote Originally Posted by noetsi View Post
    If you are doing survival analysis you might want to look at "Survival Analysis Using SAS: A practical Guide" 2nd ED by Paul Allison. Its very detailed in both the stats and the code.

+ 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