1. The Registration of Clinical Trials Deborah A. Zarin, M.D. Director, ClinicalTrials.gov May 2007

2. 2 Public Policy Challenges Use of Human Volunteers Concerns about the conduct of Trials Need to ensure access to (valid) Results

3. 3 Proposed Role of Trial Registries and Results Databases

4. 4 Definitions Prospective Registry –Captures summary protocol information before or during the trial –Provides “denominator” data for monitoring “publication bias” Results Database –Captures the results of a completed trial ClinicalTrials.gov –A prospective registry –Can link to publications

5. 5 Trial Registration Policies US Federal legislation CMS policy US State legislation ICMJE WHO Other (Ottawa, AAMC, etc.)

6. 6 The Rationale for Trial Registration Use of Human Volunteers –Participation is informed by prior research –Participation contributes to public knowledge Conduct of Clinical Trial –Provide public record of protocol items Ensuring Access to (Valid) Results –Provides “denominator” –Provides pre-specified outcome measures

8. 8 Boldfaced Print: Items required by ClinicalTrials.gov (incorporates FDAMA 113) WHO: Items additionally required for ICMJE (WHO minimal registration data set) * Additional information required in some instances Italicized print: Items not displayed at the public site

9. 9 ClinicalTrials.gov Statistics NumberPercent Total40,343 Interventional Drug Trials25,900100% Non-IND 17,232 66.5% IND 8,68833.5% –Phase I 1,759 6.8% –Phase II 3,19012.3% –Phase III 2,89211.1% –Phase IV 588 2.3% –Missing 239 1% Source –NIH 9,19635.5% –Industry 9,23135.6% –University, other 7,47328.9% 05/09/2007

10. 10 ICMJE

11. 11 Compliance Update Intervention Name –Only two companies with any “missing” data –Total of 14 records (out of 39,000+) Primary Outcome Measures, since 1/1/06 –Percentage with entries: NIH - 74% Industry - 88% Other - 89% –Continued heterogeneity in quality

12. 12 Necessary Features of a Trials Registry Provide non-duplicated list of all trials that meet users’ search criteria –Necessary to determine options for potential subjects –Necessary to determine “denominator” for scientific purposes This is difficult

13. 13 Key Challenges: Drug Names Search engines depend on known names, lists of synonyms, and hierarchies “Code” names, without “de-coders”, lead to “hidden” trials

14. 14 WHO Search Portal

15. 15 WHO Search Portal - Example

16. 16 ClinicalTrials.gov - Search

17. 17 ClinicalTrials.gov - Example

18. 18 ClinicalTrials.gov - Synonymy

19. 19 Drug Serial Numbers: “Hidden” Trials Gardasil® was approved on June 8, 2006

20. 20 Pubmed Gardasil Search One month after approval (and promotion)

22. 22 Use of Serial Numbers 11% of all drug studies in registry 20% of all industry drug studies in registry –43% Phase 1 –35% Phase 2 –11% Phase 3 –1% Phase 4

23. 23 Potential Impact of the Maine Law

24. 24 Unmet Goals of Registration Policies: How can States Help? Produce a comprehensive list of clinical trials –Only some are required by law –Unknown numbers are not registered despite array of policies States can broaden legal mandates re: scope –Use clear, objective language –Harmonize with Federal and other policies –Avoid subjective terms (eg, serious, exploratory)

25. 25 Unmet Goals of Registration Policies: How can States Help? Ensure that entries are complete and informative –Federal law leaves room for vague entries –Some providers are reluctant to disclose all information States can create clear mandates to provide complete information –Use WHO minimum data set –Harmonize requirements

26. 26 Unmet Goals of Registration Policies: How can States Help? Ensure that users can find all trials that meet their search criteria –Multiple registries make searches incomplete –“Hidden” trials will result if search engines do not use synonyms, hierarchies, etc States can mandate use of ClinicalTrials.gov States can require drug names and synonyms

27. 27 Unmet Goals of Registration Policies: How can States Hurt? Creating multiple new registries without adequate search functions will lead to confusion Non-harmonized data requirements will pose large burden and make registration very complicated Use of vague terms in mandate will create confusion; e.g.: –Chemopreventative, exploratory, serious condition,

28. 28 Reporting Trial Results Current –Link to PubMed/MEDLINE citation –Link to Drugs@FDA database –Link to unpublished results on clearly marked sponsor site

29. 29

30. 30 Sources of Independently Reviewed Trial Results Published literature FDA reports on website –Sparsely populated –Difficult to navigate –Trials not identified

31. 31 Many Trials do not have Publicly Available Results Not all trials have a published article Not all articles include all key outcome measures

32. 32 JAMA 2007:297;117:1917-1920. FDA Site Not Complete

33. 33 Trial Design Details of Methods Statistical Plan Participant Flow Baseline Data Outcomes in Each Arm, Absolute Numbers Statistical Analyses Discussion/ Conclusions General Public Clinicians Systematic Reviewers Policy makers Results Databases: Overview of Key Components and Intended Users

34. 34 Validation of Data on Results Data are more complex, and stakes are higher No proposals provide access to protocol or source data Implications of an error

35. 35 Possible Error in Registry that Would be Difficult to Detect

36. 36 Possible Error in Results Database that Would be Difficult to Detect

37. 37 How to “Validate” Results? Proposals call on FDA to “ensure accuracy and completeness” of structured data May be between 40 and 200+ trials per week What about “conclusions” or “risk-benefit” summary statements?

38. 38 PAGE 3: More results PAGE 2: Results GSK Results Entry: COPERNICUS COREG (Carvedilol) in preventing death of patients with severe heart failure Primary outcome was all- cause mortality Stopped early by DMC for benefit at the 4th interim analysis Results are published on GSK website –1 page of protocol and analysis methods –1.5 pages of results in tabular format PAGE 1: Protocol summary

39. 39 FDA CDER Review of COPERNICUS

40. 40 Results Databases—Key Questions Who are the intended users? What will they use it for? Are non-reviewed sponsor submissions adequate? Is independent review needed? If so, how can it be done?

41. 41 Implications of Pending Legislation May facilitate linkages with improved FDA site May mandate registration with all 20 WHO data elements May mandate a results feasibility study May mandate a results database May pre-empt State laws?

42. 42 Contact Us dzarin@mail.nih.gov www.ClinicalTrials.gov