Retest effects in a diverse sample: sociodemographic predictors and possible correction approaches.

Abstract

Repeated cognitive assessment in longitudinal studies favors the occurrence of retest effects, usually increasing the scores obtained at the follow-up assessments when compared to baseline. Therefore, retest effects can compromise the evaluation of cognitive decline in older adults.

Objectives: We aimed to verify the occurrence of the retest effect and the impact of sociodemographic characteristics on the follow-up scores in a sample of 5,592 participants with a diverse sociodemographic profile, who were assessed twice during 4 years of follow-up.

Methods: We tested two possible approaches to correct the retest effect and calculated the Reliable Change Index.

Results: We observed increased scores at the follow-up assessment after 4 years, but the results indicate a modest occurrence of retest effects. The regression difference correction successfully generated follow-up corrected scores, while the mean difference did not provide effective corrections. Sociodemographic characteristics had a minor impact on the retest.

Conclusions: We recommend the regression difference correction for retest effects. The absence of this methodological approach might lead to biased results using longitudinal cognitive scores.