2004D-0187 - Draft Guidance for Industry on Premarketing Risk Assessment; Development and Use of Risk Minimization Action Plans; and Good Pharmacovigilance Practices and Pharmacoepidemiologic Assessment; Availability
FDA Comment Number : EC3
Submitter : Dr. Sean Hennessy Date & Time: 07/07/2004 05:07:15
Organization : University of Pennsylvania School of Medicine
Category : Health Professional
Issue Areas/Comments
1. Consider moving section ?VI. Interpreting Safety Signals: From Signal to Potential Safety Risk? so that it appears before section ?V. Beyond Case Review: Investigating a Signal Through Observatioal Studies.? That is, Section VI discusses what to do when a signal has been identified, including the potential for following up with controlled pharmacoepidemiologic studies. Section V discusses those studies. Thus, the document?s flow might be improved by switching the order of these two sections.

2. Consider changing the name of Section III to ?The Role of Pharmacovigilance and Pharmacoepidemiology in Risk Management.?

3. The definition of pharmacovigilance beginning on line 115 seems unconventional, and includes all of pharmacoepidemiology as well. I think that most people use the term pharmacovigilance roughly to mean what one does with spontaneous reports, and pharmacoepidemiology to mean formal studies. I don?t think pharmacoepidemiology as being a subset of pharmacovigilance. Saying in line 136 that good pharmacovigilance practice is ?generally based? on acquiring complete data from spontaneous reports and developing a case series makes sense only if pharmacovigilance is defined in the more traditional sense. In contrast, the definition of pharmacovigilance beginning on line 121 is the more conventional one.

4. Line 168: should this specify that concomitant product therapy details should also include OTC medications, dietary supplements, herbals, and the like?

5. Line 188 seems to imply that all medication errors should be reported to FDA regardless of whether or not there was an ADE. Are manufacturers required or even encouraged to report medication errors to FDA, or should these be reported elsewhere (e.g., ISMP, USP)?

6. Line 216: Consider changing to ?can be used to assess potential cases.?

7. Line 247: ?confounded cases.? Confounding has a particular meaning in epidemiology, and occurs within populations rather than individuals. Consider changing to ?cases with potential alternative causes? or something similar.

8. Line 316: ?Data mining is not the only technique used to make causal attributions between products and adverse events.? Data mining is not even one technique used to make causal attributions. As pointed out in other places in the document, data mining may be useful for generating signals but not for testing hypotheses.

9. Line 465: ?Pharmacoepidemiologic safety studies are nonrandomized observational studies?? Pharmacoepidemiologic safety studies can be either randomized or non-randomized.

10. Line 756: It is unclear what is meant by ?Adverse event collection mechanisms include electronic health information systems??? This seems to imply that such health information systems are useful for generating spontaneous reports. Is that the intent, or is it that they may be useful for pharmacoepidemiologic studies?