WHO WILL DROP OUT NEXT? CHARACTERIZATION OF DROPOUTS IN A LARGE LONGITUDINAL STUDY

Abstract

One of the biggest problems of large-scale longitudinal studies is the phenomenon of attrition of participants. Even with the best effort a significant number of participants decides to discontinue – which increases with study duration. For aged study samples with a high number of diseases characteristics of dropouts and the prediction of future dropout are of particular interest. 1201 participants of the TREND study (Tübinger evaluation of Risk factors for Early detection of NeuroDegeneration; www.TREND-studie.de) underwent a detailed general, neurological and psychiatric examination at four time points (baseline and three two-year follow-ups). To assess cognitive functions, the neuropsychological CERADplus test battery was used, and the participants were asked for subjective memory impairment and their respective concerns. Participants with objective cognitive impairment had a higher risk for dropping out (16.5 % of the dropouts after baseline had an amnestic MCI vs. 4.1 % of the non-dropouts, p<.001), whereas subjective memory impairment was no risk factor (p=.257); counterintuitively, dropouts had significantly less memory concerns (p=.015). The completers of all four visits (506 participants, participants with less than four visits so far were excluded from the analyses) showed a significant relation between subjective and objective impairment in cognitive tests for the baseline assessment and similar for all follow-up assessments. Dropouts (225 dropouts after one of the four visits, pooling based on their last conducted visit) did not show any significant relations to any of the mentioned objective and subjective domains in the last visit prior to dropout. Surprisingly, subjective memory impairment was not predictive for dropout. There was no concordance of subjective and objective memory impairment in the dropout group, but in the completers’ group. These results indicate that dropouts display an inaccurate subjective memory evaluation, particularly in the last visit before dropout. The loss of the ability to estimate one's own cognitive performance adequately may be a useful predictor for future dropout. Therefore, these participants need greater attention regarding their study adherence.