There are no empiric data to support guidelines for duration of therapy with antidementia drugs. This study examined whether persistent use of antidementia drugs slows clinical progression of Alzheimer disease (AD) assessed by repeated measures on serial tests of cognition and function.
Six hundred forty-one probable AD patients were followed prospectively at an academic center over 20 years. Cumulative drug exposure was expressed as a persistency index (PI) reflecting total years of drug use divided by total years of disease symptoms. Baseline and annual testing consisted of Mini-Mental State Examination (MMSE), Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog), Baylor Profound Mental Status Examination (BPMSE), Clinical Dementia Rating-Sum of Boxes (CDR-SB), Physical Self-Maintenance Scale (PSMS), and Instrumental Activities of Daily Living (IADL). Annual change in slope of neuropsychological and functional tests as predicted by follow-up time, PI, and the interaction of these two variables was evaluated.
PI was associated with significantly slower rates of decline (with, without adjustment for covariates) on MMSE (P < 0.0001), PSMS (P < 0.05), IADL (P < 0.0001), and CDR-SB (P < 0.001). There was an insignificant trend (P = 0.053) for the PI to be associated with slower rate of decline on BPMSE. The association of PI with ADAS-Cog followed a quadratic trend (P < 0.01). Analysis including both linear and quadratic terms suggests that PI slowed ADAS-Cog decline temporarily. The magnitude of the favorable effect of a rate change in PI was: MMSE 1 point per year, PSMS 0.4 points per year, IADL 1.4 points per year, and CDR-SB 0.6 points per year. The change in mean test scores is additive over the follow-up period (3 ± 1.94 years).
Persistent drug treatment had a positive impact on AD progression assessed by multiple cognitive, functional, and global outcome measures. The magnitude of the treatment effect was clinically significant. Positive treatment effects were even found in those with advanced disease.