Purpose: Statistical shrinkage is a potential statistical method to improve the accuracy of signal detection results and avoid spurious associations detected by disproportionality analyses. In this study, we introduced statistical shrinkage influence on disproportionality methods in spontaneous reporting system in China.
Methods: We added the shrinkage parameters in the numerator and denominator, denoted as in the formula of disproportionality analysis. The shrinkage parameters were subjectively set to between 0 and 5, with an interval of 0.1. Adverse drug reaction product label database was deemed as a proxy of golden standard to evaluate the effect of statistical shrinkage. Reports in the years of 2010-2011 were extracted from the national spontaneous reporting system database as the data source for analysis in this study.
Results: When α was around 0.5, the Youden index reached the maximum for each disproportionality methods in this study. The value of 0.6 was suggested as the most appropriate statistical shrinkage parameter for reporting odds ratio and proportional reporting ratio and 0.2 for information component based on the spontaneous reporting system of China.
Keywords: adverse drug reaction; false positive signal; pharmacoepidemiology; pharmacovigilance; signal detection; statistical shrinkage.
Copyright © 2015 John Wiley & Sons, Ltd.