The Importance of Reporting Negative Findings in Data Mining
- PDF / 105,995 Bytes
- 5 Pages / 583.276 x 792 pts Page_size
- 97 Downloads / 166 Views
CURRENT OPINION
© 2008 Adis Data Information BV. All rights reserved.
The Importance of Reporting Negative Findings in Data Mining The Example of Exenatide and Pancreatitis Manfred Hauben1,2,3,4 and Alan Hochberg5 1 2 3 4 5
Risk Management Strategy, Pfizer Inc., New York, New York, USA Department of Medicine, New York University School of Medicine, New York, New York, USA Departments of Community and Preventive Medicine and Pharmacology, New York Medical College, Valhalla, New York, USA School of Information Systems, Computing and Mathematics, Brunel University, London, England ProSanos Corporation, Harrisburg, Pennsylvania, USA
Abstract
The US Food and Drug Administration (FDA) recently published a warning regarding pancreatitis in association with the use of exenatide, an incretin mimetic used for the treatment of patients with diabetes mellitus. We note that this safety issue is not associated with a signal of disproportionate reporting (SDR) in the FDA Adverse Event Reporting System (AERS) database or the World Health Organization (Uppsala Monitoring Centre) Vigibase for any of four data-mining algorithms we tested (proportional reporting ratio, the multi-item gamma-Poisson shrinker, an urn model and the Bayesian Confidence Propagation Neural Network). Exenatide and acute pancreatitis may thus represent a ‘false-negative’ result for disproportionality-based data-mining methodology generally. We evaluate the possibility that this lack of an SDR is caused by the phenomenon known as ‘masking’ (or ‘cloaking’) and reject this hypothesis. While positive findings are understandably more exciting, we discuss why publishing negative findings, such as in this example, is important for placing the capabilities and limitations of drug safety data mining into proper perspective.
A number of articles have discussed the capabilities and limitations of disproportionality-based data mining in the detection of drug safety ‘signals’ within databases of spontaneously reported adverse events.[1,2] The emphasis of many discussions pertaining to data mining is the problem of ‘false-positive’ findings that may be generated by data mining. While this problem afflicts all methods of signal detection, data mining, like any other method, is also subject to ‘false negatives’ that may be particularly problematic if data-mining algorithms, especially those with an elaborate mathematical veneer, promote overconfidence in their outputs and desensitise stakeholders to the profound limitations of the data. Both false-negative and false-positive findings may adversely impact public health surveillance systems. Negative findings in data mining may have three sources: 1. The apparent association reflected in the individual case reports is because of happenstance rather than causality. In this case, the negative finding from data mining is a ‘true negative’.
2. The association has been masked[1,2] (also known as ‘cloaking’). This is the phenomenon whereby a very strong statistical reporting relationship between a drug (or drugs) and an advers
Data Loading...
