Predicting Long-Term Cognitive Outcome with New Regression Models in Donepezil-Treated Alzheimer Patients in a Naturalistic Setting.

Background/Aims: To build and analyze regression models predicting (1) the long-term cognitive outcome in donepezil-treated patients with Alzheimer's disease, and (2) the short-term (6 months) cognitive impact of treatment depending on cognitive severity at baseline. Methods: The Swedish Alzheimer Treatment Study (SATS) is an open-label, non-randomized, 3-year, multicentre study in a routine clinical setting. A total of 435 patients, mostly in the mild and moderate stages of Alzheimer's disease, received the cholinesterase inhibitor donepezil. They were assessed with the Mini-Mental State Examination (MMSE) and Alzheimer's Disease Assessment Scale-cognitive subscale (ADAS-cog) at baseline and every 6 months for a total period of 3 years. Regression models were fitted from the actual scores at different intervals for the prediction of the cognitive outcome. Results: The ADAS-cog and MMSE scores during the 3-year treatment period could be predicted with a high degree of explanation using regression models (p < 0.001). Moreover, there was a significant relation between the mean cognitive change after 6 months of treatment and the baseline scores on MMSE (p < 0.01) and ADAS-cog (p < 0.001), respectively. Conclusion: Statistical models can be used to predict cognitive outcome in donepezil-treated cohorts of AD patients. These models can be clinically valuable, for example when assessing the efficacy of new therapies when added to cholinesterase inhibitor treatment.