Discussion: View Thread

  • 1.  SAS procedure for segmented Poisson regression (interrupted time series)

    Posted 07-16-2014 16:29
    Could anyone suggest a sas procedure that will allow to account for correlated errors (AR(1)) in a Poisson regression model for interrupted time series?

    Outcome data are yearly counts (population at risk is used as an offset), study intervention is a binary variable (implemented mid-study), the model includes a few covariates. We need to specify  first order autoregressive covariance structure for this model.

    Any help will be greatly appreciated!

    -------------------------------------------
    Katarzyna Wyka
    Graduate Center, CUNY
    -------------------------------------------


  • 2.  RE:SAS procedure for segmented Poisson regression (interrupted time series)

    Posted 07-16-2014 18:45
    Proc glimmix feel free to contact me if you need assistance.

    -------------------------------------------
    Jason T. Machan
    Director, Lifespan Biostatistics Core,
    Lifespan Hospital System
    Research Scientist, Biostatistics, Research
    Rhode Island Hospital
    Assistant Professor, Departments of Orthopaedics and Surgery
    The Warren Alpert Medical School, Brown University
    Director Biostatistics Externship, Adjunct Assistant Professor, Department of Psychology
    University of Rhode Island
    -------------------------------------------





  • 3.  RE: SAS procedure for segmented Poisson regression (interrupted time series)

    Posted 07-22-2014 16:38

    Try "Proc Glimmix"

    -------------------------------------------
    Shahidul Islam, MPH
    Biostatistician
    Winthrop University Hospital
    Mineola, NY 11590
    -------------------------------------------




  • 4.  RE: SAS procedure for segmented Poisson regression (interrupted time series)

    Posted 07-22-2014 16:48
    Hi,

    You might also take a look at PROC GENMOD with the REPEATED statement, which performs generalized estimating equations (GEE) analysis.

    HTH,

    Tor

    -------------------------------------------
    Tor Neilands
    Professor of Medicine
    UCSF Center for AIDS Prevention Studies
    -------------------------------------------


    Original Message


  • 5.  RE: SAS procedure for segmented Poisson regression (interrupted time series)

    Posted 07-31-2014 11:01
    I am going back to this thread from another angle.   I am working on a project similar to this with a mid-study intervention.   This is not a run-in or a wash-out so the pattern before the intervention is important.   An AR(1) correlation structure makes sense before the intervention and after the intervention but how can we assume that the time points after the intervention have the same correlation with cases before the intervention based on the distance e.g. timepoints  +1 and +3 compared with +1 and -1 where 0 is the start of the intervention?   I think economists would call the intervention a 'shock' in time series but I am not sure how this idea is applied in multi-subject longitudinal studies.

    -------------------------------------------
    Georgette Asherman
    -------------------------------------------