Eliminating the Impact of Reporting Delay on Survival Analysis with Right-Truncated Data

• Activity Number: 403
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Ruiguang Song*+
• Companies: CDC
• Keywords: survival analysis ; right truncated data ; reporting delay

Abstract

We consider the analysis of survival time with right truncated data: the initial event is observed only if the ending event both occurs and is observed. If there is no lag between the occurrence and the observation of the second, or ending event, then standard survival analysis for right truncated data can be used. However, the analysis is not so straightforward when there is a delay in reporting the second event, which is a common phenomenon in disease case surveillance systems. The estimate for the survival time distribution will be biased if the reporting delay is ignored. Kalbfleisch and Lawless (1989) proposed a stratification method based on a discrete time scale to address this issue. In this paper, we introduce a simple approach to eliminating the impact of reporting delay on the estimation of the survival time distribution. There are two steps in this approach: (1) redefine the dates of the two events based on each subject's reporting lag (2) apply the estimation procedure for right truncated data to the modified data. This approach generates the same estimation results as the stratification approach does. It's easy to use and does not require the lag time to be discrete.