Prediction models for assessing long-term outcome in Alzheimer's disease: a review

Abstract

In Alzheimer's disease (AD), placebo-controlled long-term studies of cholinesterase inhibitors (ChEIs) are not permitted for ethical reasons. Therefore, in these studies, patients' outcomes on cognitive and functional assessment scales must be compared with mathematical models or historical data from untreated cohorts. PubMed and previously published long-term extensions of clinical trials and naturalistic studies of ChEIs were examined to identify empirical statistical models and other approaches, such as use of data from historical cohorts or extrapolated changes from extension studies, that were used to draw comparisons between ChEI-treated and untreated patients. The models and methods were described. It is essential to be aware of the limitations of comparisons made with these approaches. Prediction models based on ChEI-treated patients can be used in the studies of new treatments when those treatments are added to ChEIs. More sophisticated models that also accommodate patient-specific characteristics should be developed for comparisons in future long-term AD studies.

Keywords: Alzheimer’s disease; activities of daily living; cognition; disease progression; longitudinal studies; statistical models.