Minimizing Attrition

• Welcome to the Outcome Assessment forum by John Grohol, 9/18/96
  • (...)
    
    • General comments and attrition by Cindy Gruman, 10/2/96
      
      

If you are still collecting the data, you might be able to prevent some of the attrition by investing time and money into the process of tracking. For example, you can use both the telephone as well as mailed questionnaires to find participants. At baseline you can ask for a proxy respondant who will always know how to locate the participant. It is so hard to retain people for follow-up, but I often think attrition is a function of both the level of persistance and the data manager's ability to keep information up to date (i.e. address changes etc.)

Nevertheless, it is inevitable that participants will be lost to follow-up. It sounds like you are describing an "intent to treat" design where you need to acknowledge that participants of a moderate level of severity of illness will not be represented well in terms of your outcome data. It is hard to talk about how to deal with this statistically, without knowing more about the study. Do you want to talk about the study further and see whether we can get some opinions from statistical experts?

Ellen

• What to do about attrition by E. H. Fischer, 10/8/96