Semiparametric Grouped Backward Recurrence Cox Model for the Analysis of Current Duration Data with Preferential Reporting

• Activity Number: 680
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Alexander C. McLain*+ and Marie Thoma and Rajeshwari Sundaram and Germaine M. Buck Louis
• Companies: Department of Epidemiology and Biostatistics, University of South Carolina and Center for Disease Control and National Institute of Child Health and Human Development and National Institute of Child Health and Human Development
• Keywords: Backwards recurrence times ; Current duration ; Digit preference ; Grenander estimator ; Grouped survival data ; Proportional hazards model

Abstract

Current duration data arise in cross-sectional studies from questions on the length of time from an initiating event to the time of interview. For example in the National Survey on Family Growth (NSFG), women who were considered at risk of pregnancy at the time of interview were asked (a) "Are you currently attempting pregnancy,'' and (b) "If yes, how long have you been attempting to get pregnant.'' It is of interest to make inference on the distribution of the unobserved total length of pregnancy attempt based on responses to (b). Current durations are obtained only from women attempting pregnancy at the time of interview, and thus the observations are length-biased. In this article, we propose a semi-parametric grouped backward recurrence Cox model, and a piecewise constant baseline model draw inference on current duration data. Further, we discuss nonparametric methods to estimate the distribution of current duration data. We discuss, and investigate through simulation studies, the robustness properties of our grouped methods when the underlying distribution is discrete and continuous, and when digit preference is present. Lastly, we present an analysis of the NSFG data.