Compliance with a 12-month assessment in an elderly cohort participating in a preventive intervention study: the San Diego Medicare Preventive Health Project.

Non-compliance in long-term cohort studies contributes to bias in the estimation of study parameters and loss of power in hypothesis testing. This paper identifies baseline factors associated with non-compliance in a 12-month health assessment among elderly participants in the San Diego Medicare Preventive Health Project, a randomized preventive intervention study. A non-complier was defined as a study subject contacted by telephone by study personnel to schedule an appointment for the 12-month health assessment but who would not or could not attend. Of the almost 1600 contacted subjects, 14.1% did not participate in the evaluation. Using logistic regression, non-compliance was found to be associated with older age, intervention group assignment, nonsingle family residence, no alcohol use, and, to a lesser extent, depression. No interactions between group assignment and other baseline factors were detected. These results may be useful in designing cohort maintenance strategies where greater resources may be devoted to participants at higher risk of non-compliance and attrition.