A Natural B-Spline Varying-Coefficient Method for Longitudinal Data with Nonignorable Dropout

• Activity Number: 360
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Jeri E.F. Harwood*+ and Samantha MaWhinney and Erika L. Hernandez
• Companies: University of Colorado, Denver and Colorado School of Public Health, University of Colorado Denver and Brigham Young University
• Keywords: Nonignorable dropout ; B-spline ; Varying-coefficient ; HIV

Abstract

Dropout is common in longitudinal studies and occurs for reasons that may be related to the outcome of interest. When the probability of dropout depends on the unobserved outcomes, even after conditioning on observable data, the missing data are missing not at random and therefore nonignorable. We present a natural B-spline varying-coefficient (NSV) method to account for dropout. Given that subjects who drop out early provide limited data for slope estimation, a natural B-spline left boundary offset is introduced, providing a linear trajectory prior to the left boundary point of the dropout mechanism. We present simulation results comparing the NSV with the classic conditional linear model. The method will be applied to HIV/AIDS clinical trial data.