Prognostic regression models of life expectancy after diagnosis of Alzheimer’s disease – a 20-year follow-up

Abstract

Background: To determine characteristics that might affect survival times in Alzheimer’s disease (AD) patients treated with cholinesterase inhibitors (ChEI) in clinical practice, and create statistical models of estimated life-span after AD diagnosis. Method: The Swedish Alzheimer Treatment Study (SATS) is a prospective, observational, multicenter study to evaluate the use of long-term ChEI therapy. The SATS enrolled 1,021 participants with a clinical diagnosis of mild-to-moderate AD (Mini-Mental State Examination (MMSE) score, 10–26) at the start of ChEI treatment (shortly after AD diagnosis). Cognitive ability, instrumental and basic activities of daily living (ADL) were evaluated at baseline and semiannually over 3 years, and the date of death was recorded. Result: After up to 20 years of follow-up, 966 patients (95%) were deceased. Two types of multivariate linear regression models were created, one using only sociodemographic and clinical characteristics at baseline, and another model including baseline and longitudinal measures. Both main models included cognitive ability, and the extended models also included ADL capacities, as well as aspects of ChEI therapy in the longitudinal model. Significant independent predictors of longer life expectancy in all models were female sex, younger age, no use of antihypertensive/cardiac therapy or antidiabetics, (but not the specific apolipoprotein E genotype or solitary living), indicating the stability and validity of our regression models. A higher education level predicted shorter survival time in both baseline models (with and without ADL) and in the longitudinal model including cognition only. In the baseline model including ADL capacities, instrumental ADL (IADL) and basic ADL scores were stronger predictors of mortality than cognition. In the longitudinal model including ADL capacities, IADL and basic ADL scores (but not MMSE score) at baseline, progression rates in cognition and basic ADL predicted survival independently. A longer period of ChEI treatment in the study was also a strong predictor of longer life-span of up to ~2 years. Conclusion: The survival time after a diagnosis of AD could be predicted with a high degree of explanation using multivariate regression models including sociodemographic and clinical factors. These clinically relevant prognostic models can be used to estimate life expectancy for AD patients.