1. SAE data entry: Clinical versus Pharmacovigilance standards Daniel Becker Solvay Pharmaceuticals Hannover, Germany Daniel.Becker@solvay.com T: +49 511 857 3708 © Daniel Becker PSDM, Weesp, 22 May 2008

2. 2 Who am I?  Global Safety Physician in Solvay Global Clinical Safety (part of Global Clinical Development)  Chair of Solvay-Quintiles Joint Clinical Safety and Coding Team  Solvay SAE reporting coordinator

3. 3 Disclaimer These are my current views and not necessarily those of every concerned Solvay colleague

4. 4 SAE data discrepancies between Pharmacovigilance and Clinical databases: Origins I.Different philosophies II.Different industry standards III.Multiple datasets IV.Technical aspects

5. 5 I. Philosophies Pharmacovigilance Safety first! Clinical CRF data authenticity first!

6. 6 Consequences (1) Pharmacovigilance data entry 1.interpretation of supplied data is entered 2.if inconsistent data is supplied (e.g. SAE report vs. discharge summary) the medically worse data is entered 3.selected non-serious AEs are updated to medically serious Clinical data entry 1.data is entered/accepted (in case of eCRF) as is 2.only data reported in the CRF/as query response is accepted 3.no change in seriousness

7. 7 Consequences (2) Pharmacovigilance data entry  potential AEs mentioned e.g. in the narrative which are not reported as AEs are extracted and entered  worst-case assumption is entered, and confirmation is requested thereafter  Pharmacovigilance controls entries Clinical data entry  no data transfer to other variable  questionable entry is kept and query is issued  Reporter controls entries

8. 8 II. Industry Standards Pharmacovigilance ICH-E2B Clinical SDTM, HL7,...

9. 9 Consequences  different variables and formats  not all related variables may match in definition, formats and/or rules  particularly, if non-current versions of the industry standards are used or when not defined in them

10. 10 III. Multiple datasets Pharmacovigilance 1.SAE form entry, as well as formless submissions (e.g., hospital discharge summary, query replies) 2.Pharmacovigilance database Clinical 3.CRF entries in several modules (demographics, concomitant medication, adverse event) 4.Clinical database

11. 11 Consequences  investigators regularly report inconsistently to the clinical versus the pharmacovigilance databases or provide corrections / relevant follow-up information like a final hospital discharge diagnosis only to one database  unless the same person codes for both databases, codes will not always match  if a standard or its guide is modified the other three may have to be updated, too

12. 12 IV. Technical aspects (1) Pharmacovigilance  database never locks  medical history occurs before the SAE  SAE start date is date of first sign / symptom / diagnosis. Clinical  database locks at study end  medical history occurs before the study. Medical events between study start and SAE are AEs.  SAE start date is date AE became serious.

13. 13 IV. Technical aspects (2) Pharmacovigilance  If an SAE started during core study but extended into extension study it is recorded under the core study only.  Concomitant medication is only that administed at the time of the SAE and which is non- suspect to have caused the SAE. Clinical  If an SAE started during core study but extended into extension study it is recorded under both studies.  Concomitant medication is any medication administered during the study, whether suspect to have caused the SAE or not.

14. 14 IV. Technical aspects (3)  unless data entry in both databases is done by the same person, discrepancies are inevitable, e.g. inconsistent ➤ splitting of verbatims (e.g., „pneumonia, malaise“ in „pneumonia“ and „malaise“) ➤ combination of related, concomitant events to code to inclusive term (e.g., „shortness of breath“ and „asthma“ to „asthma“) ➤ duplication of verbatims to code to two terms (e.g., „toxic psychosis“ duplicated and coded to „toxic reaction“ respectively „psychosis“)

15. 15 My thoughts  I don‘t see how you can get around two databases  reconcile only medically most relevant items, e.g.: 1. Study Number 2. Subject Number 3. Randomization Number (if available) 4. Gender 5. Date of Birth 6. SAE verbatim 7. Stop Date 8. Ongoing 9. Outcome 10. Study Drug-Event relationship as judged by the investigator 11. Action taken with study drug

16. 16 Questions? Comments?