|
Activity Number:
|
111
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #305526 |
|
Title:
|
Identifying an Optimal Risk Window Length for Self-Controlled Case Series Studies
|
|
Author(s):
|
Lijing Zhang and Stanley Xu*+ and Chan Zeng and Jennifer Nelson and John Mullooly and David McClure and Jason Glanz
|
|
Companies:
|
Kaiser Permanente Colorado and Kaiser Permanente Colorado and Kaiser Permanente Colorado and Group Health Center for Health Studies and Kaiser Permanente Northwest and Kaiser Permanente Colorado and Kaiser Permanente Colorado
|
|
Address:
|
, , 80237,
|
|
Keywords:
|
Vaccine safety study ; Optimal risk window ; Self-controlled case series ; Relative risks ; Changing point
|
|
Abstract:
|
Misspecification of risk window length in a self-controlled case series studies for evaluating vaccine safety introduces bias in relative risk (RR) estimates. We propose a data-based approach for searching the optimal length of risk windows. Conditional Poisson models are used to obtain RRs for a series of pre-specified risk window lengths (L). The relation between L and the corresponding RR is derived and used in a predictive model. Change-point detection involves an iterative process by fitting the model to a subset of the RR-L pairs for larger Ls and predicting the RR for a slightly smaller L. A change point is identified when the RR for the smaller L is not covered by the predicted intervals. Both simulation studies and real data application show that the 'right-side' approach is reliable in identifying the optimal risk window for medium and long risk windows in vaccine safety study.
|
