Discussion: View Thread

Over-adjusting on an age-related exposure; case-control

  • 1.  Over-adjusting on an age-related exposure; case-control

    Posted 05-17-2012 17:22

    Hello,

    In a study, having the outcome before age 55 defined cases & controls were at least 78 and never had the outcome; no other matching occured.  Several exposure variables of interest may be age-related (eg cataract).  It is of interest to determine the odds ratio (OR) of each exposure variable to the outcome.   

    Are age-adjustments in this situation redundant; circular; over adjustment on the outcome? If no adjustments for age are made and the exposure is very control-heavy (i.e. age-related), how is a higher odds ratio interpreted (assuming the exposed is top row and cases are 1st column in 2x2)?   

    What other issues have I not thought of? 

    (I understand there may issues in the defintion of cases/controls, a secondary concern of mine, but sample size issues drove the defintion & hopefully this won't become the main focus of this thread.) 


    Kind regards, 
    -------------------------------------------
    Emmeline Sangeorzan
    Biostatistician
    Arthritis Research Institute of America
    Clearwater, FL
    -------------------------------------------