Accounting for Dropout in Longitudinal Cohort Studies

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

Abstract

Prospective longitudinal cohort studies are a valuable resource for biomedical and public health research. Dropout due to loss to follow-up or death is common in these studies and may be nonignorable. To account for dropout, we consider a natural spline varying-coefficient method for continuous outcomes and an extension for binary outcomes. We present an application of these methods to publicly available cohort data where we examine the impact of illicit drug use on longitudinal CD4+ T cell count and HIV-1 viral suppression in untreated and treated HIV-infected subjects, respectively.