Send to

Choose Destination
Drug Saf. 2014 Jan;37(1):53-64. doi: 10.1007/s40264-013-0124-9.

Comparing time to adverse drug reaction signals in a spontaneous reporting database and a claims database: a case study of rofecoxib-induced myocardial infarction and rosiglitazone-induced heart failure signals in Australia.

Author information

School of Pharmacy and Medical Sciences, Quality Use of Medicines and Pharmacy Research Centre, Sansom Institute, University of South Australia, GPO Box 2471, Adelaide, SA, 5001, Australia,



The objective of post-marketing surveillance of medicines is to rapidly detect adverse drug reactions (ADRs). Early ADR detection will enable policy makers and health professionals to recognise adverse events that may not have been identified in pre-marketing clinical trials. Multiple methods exist for ADR signal detection. Traditional quantitative methods employed in spontaneous reports data have include reporting odds ratio (ROR), proportional reporting ratio (PRR) and Bayesian techniques. With the development of administrative health claims databases, additional methods such as sequence symmetry analysis (SSA) may be able to be employed routinely to confirm ADR signals.


We tested the time to signal detection of quantitative ADR signalling methods in a health claims database (SSA) and in a spontaneous reporting database (ROR, PRR, Bayesian confidence propagation neural network) for rofecoxib-induced myocardial infarction and rosiglitazone-induced heart failure.


This study demonstrated that all four signalling methods detected safety signals within 1-3 years of market entry or subsidisation of the medicines, and for both cases the signals were detected before post-marketing clinical trial results. By contrast, the trial results and subsequent warning or withdrawal were published 5-7 years after first marketing of these medicines.


This case study highlights that a post-marketing quantitative method utilising administrative claims data can be a complementary tool to traditional quantitative methods employed in spontaneous reports that may help to verify safety signals detected in spontaneous reporting data.

[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Springer
Loading ...
Support Center