Analysis of Onset of Dementia Data with Interval Censoring

• Activity Number: 438
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Linbo Wang*+ and Xiao-Hua Andrew Zhou
• Companies: University of Washington and University of Washington
• Keywords: mixture model ; immune rate ; survival model ; onset of dementia

Abstract

Modeling time to onset of dementia is important for prevention and treatment of dementia. Many current approaches adapted survival analysis techniques that implicitly assume all subjects are at risk for dementia. We examined this assumption with a logistic-Weibull mixture model, which belongs to the family of cure-rate models. The likelihood is modified to account for potential interval-censoring in the data. Simulation result shows that in presence of immune subgroup, mixture model excels in survival curve estimation. We apply this method to Uniform Data Set from National Alzheimer's Coordinating Center. Demographic variables, comorbid conditions and family history are adjusted in the model. Estimation result suggests the existence of such an immune group, with a proportion of 7.97% (p-value=?10?^(-10)) within our study population. Our study shows that some people may be immune to dementia, and this subgroup of immune individuals should be adjusted for in any survival analysis to produce correct estimates for age of onset of dementia.