Piloting a Comprehensive Knowledge Base for Pharmacovigilance Using Standardized Vocabularies
Download as PPTX, PDF4 likes13,116 views
The document discusses the establishment of a comprehensive knowledge base for pharmacovigilance, focusing on synthesizing adverse drug event evidence within a standardized framework utilizing the Observational Health Data and Informatics Initiative (OHDSI). It highlights the challenges of managing vast amounts of disparate drug safety information and outlines the development of the Laertes pilot project to merge sources into a standard vocabulary. The aim is to improve the quality of clinical research and promote better understanding of adverse events associated with medications.
Piloting a Comprehensive Knowledge Base for Pharmacovigilance Using Standardized Vocabularies
- 1. Piloting a Comprehensive Knowledge Base for Pharmacovigilance Using Standardized Vocabularies Authors and contributors Vojtech Huser, MD, PhD Jeremy Jao Jon Duke, MD, MS Patrick B. Ryan, PHD Scott D. Nelson, PharmD Richard D. Boyce, PhD Erica A. Voss, MPH Michel Dumontier, PhD Nicholas Tatonetti, PhD Lee Evans Majid Rastegar-Mojarad, MS Abraham G. Hartzema, PhD Johan Ellenius, PhD Rave Harpaz, PhD Magnus Wallberg, MSc Christian Reich, MD, PhD AMIA CRI 3/26/2015
- 2. 2 Disclosures • I disclose that neither I nor my wife have relevant financial relationships with commercial interests
- 3. 3 Problem statement • An overwhelming amount of information relevant to drug safety-relevant is being generated – stored in a wide array of disparate information sources – using differing terminologies – at a faster pace than ever before
- 4. The relevant evidence sources Spontaneous adverse event data (FAERS, VigiBase™, ClinicalTrials.gov) Literature (PubMed, SemMed) Product labeling (SPL, SPC) Indications / Contraindications (FDB™) Observational healthcare data (claims + EHR) FAERS – FDA Adverse Event Reporting System; SPL – Structured Produce Labeling; SPC – Summary of Product Characteristics; FDB™ - First DatabankTM EHR – Electronic Health Record;
- 5. 5 Objective • Synthesize adverse drug event evidence within a standard framework for clinical research – The Observational Health Data and Informatics Initiative (OHDSI) • A common data model and standard vocabulary – Easy to adopt and used by numerous sites • A suite of tools that improve the value of observational clinical data – data characterization, population- level estimation, patient- level prediction, – phenotyping, cohort and quality measure design
- 6. A new adverse event evidence base built into OHDSI Largescale Adverse Effects Related to Treatment Evidence Standardization (Laertes)
- 7. 7 The pilot version of Laertes • Merging sources into the OHDSI standard vocabulary • The data schema • Current progress
- 8. Merging the sources Drugs (RxNorm) Conditions (SNOMED) Spontaneous adverse event data (FAERS, VigiBase™, ClinicalTrials.gov) MedDRA -> SNOMED Freetext, ATC -> RxNorm Literature (PubMed, SemMed) MeSH, UMLS -> SNOMED MeSH, UMLS -> RxNorm Product labeling (SPL, SPC) Freetext -> MedDRA® -> SNOMED SPL Set ID -> RxNorm Indications / Contraindications (FDB™) ICD-9-CM -> SNOMED NDC/GenS eqNum -> RxNorm Observational healthcare data (claims + EHR) ICD-9-CM, ICD-10 -> SNOMED NDC/GPI/ATC -> RxNorm Drug classifications (ATC, NDF-RT) Condition classifications (MedDRA®, Ontology of Adverse Events) Source to Drug Mapping Source to HOI Mapping Evidence Sources
- 9. Current progress on evidence sources Spontaneous adverse event data (FAERS, VigiBase™, ClinicalTrials.gov) Literature (PubMed, SemMed) Product labeling (SPL, SPC) Indications / Contraindications (FDB™) Observational healthcare data (claims + EHR) Evidence Sources PubMed (Avillach et al.): • Case reports: 84,181 • Clinical trials: 25,813 • Other: 1,146 SemMed (Kilicoglu et al) • Case reports: 2,372 • Clinical trials: 1,169 Avillach P, Dufour JC, Diallo G, Salvo F, Joubert M, Thiessard F, Mougin F, Trifirò G, Fourrier-Réglat A, Pariente A, Fieschi M. Design and val idation of an automated method to detect known adverse drug reactions in MEDLINE: a contribution from the EU-ADR project. J Am Med Inform Assoc. 2013 May 1;20(3):446-52 Kilicoglu H, Rosemblat G, Fiszman M, Rindflesch TC. Constructing a semantic predication gold standard from the biomedical literature. BMC Bioinformatics. 2011 Dec 20;12:48 Duke, Jon, Jeff Friedlin, and Patrick Ryan. "A quantitative analysis of adverse events and “overwarning” in drug labeling." Archives of internal medicine 171.10 (2011): 941- 954. FAERS : • Subset with counts, EB05 and EBGM: 301,332 ClinicalTrials.gov: In process VigiBase™: In process US SPLs (Duke et al.): • Adverse Drug Reactions: 2,411,943 EU SPCs (PREDICT): • Adverse Drug Reactions: 42,767 In process Can be done on local installations • Public data pending
- 10. The schema supports two use cases Example association: Drug X – Renal Failure Summary Drill down Spontaneous reporting EHR Data Scientific Literature Product Labeling Other evidence EB05 OR Count Count …
- 11. More details on the schema
- 12. 12 Lets look at two example uses of Laertes • Finding and reviewing evidence • Using Laertes and other OHDSI tools to address quality improvement
- 13. 13 Pharmacovigilance example • HOIs associated with Lisinopril – An ACE inhibitor that treats high blood pressure and heart failure (WebMD) – The blood pressure lowering effect might help reduce the risk of diabetes nephropathy • Better understanding adverse events associated with diabetes is a top priority (DHHS 2014) WebMD - http://www.webmd.com/diabetes/tc/diabetic-nephropathy-treatment-overview U.S. Department of Health and Human Services, Office of Disease Prevention and Health Promotion. (2014). National Action Plan for Adverse Drug Event Prevention. Washington, DC: Author.
- 14. The query select s.ingredient, s.hoi, s.ct_count as clintrials, v.medline_mesh_clin_trial_link, s.case_count, v.medline_mesh_case_report_link, s.splicer_count as label_count from laertes_summary s join drug_hoi_evidence_view v on s.hoi_id=v.hoi and v.drug=ingredient_id where ingredient_id = 1308216 and report_name='Stratified by ingredient and HOI' and coalesce(ct_count, case_count, other_count, splicer_count) is not null and case_count is not null order by case_count desc limit 100;
- 16. Clinical trial with evidence on lisinopril- angioedema
- 18. Case reports with evidence on Lisinopril- angioedema
- 20. Case reports with evidence on lisinopril- aplastic anemia
- 22. Structured product label with evidence on lisinopril-aplastic anemia
- 23. 23 Opportunity - Quality improvement in the nursing home setting • The prevalence of anemia in 5 nursing homes is 36% affecting quality of life. • The health system is interested in identifying potential interventions. – Could prescribing be better optimized to reduce this potential adverse event?
- 24. Quality improvement in the nursing home setting What drugs have evidence for an association with anemia? • Laertes Which kinds of anemia? • Standard vocabulary • Cohort definition (Circe) • Phenotyping What is the prevalence of exposure to those drugs in my facilities? • Cohort characterization (Heracles) Are exposed patients at risk? • OHDSI Methods library
- 25. 25 Summary • Laertes is a new adverse event evidence base built into clinical research framework – Enables summary and drill down evidence search – Can be integrated into other clinical research workflows
- 26. 26 Acknowledgements • Funding: The American taxpayers via: – National Library of Medicine (1R01LM011838-01) – National Institute of Aging (K01AG044433-01)
- 27. 27 Discussion
- 28. 28 How to get Involved • Learn about OHDSI: http://www.ohdsi.org/ • Wiki: http://www.ohdsi.org/web/wiki/doku.php?id=pr ojects:workgroups:kb-wg • GitHub: https://github.com/OHDSI/KnowledgeBase