Dr Fiona Reddington

1
Data Sharing

• Dr Fiona Reddington
• Informatics Coordination Unit

2
Today ..

• The current situation
• The role of the NCRI
• Progress to date
• Resources currently available

3
(No Transcript)
4

• SCIENCE TECHNOLOGYClinical trials The
  sounds of silenceSep 9th 2004 From The
  Economist print edition
• Negative clinical-trial results are
  underreported. But this may soon change
• SOMETIMES, no news is news. In clinical trials,
  where new medical treatments are tried out on
  human subjects, no newsan inconclusive result,
  indicating that the treatment is useless, or a
  negative one, indicating that it is harmfulcan
  be as scientifically important as a positive
  result. Unfortunately, such a result is much less
  likely to be reported.
• That is particularly true for trials sponsored by
  the pharmaceutical industry which, according to
  the American Medical Association (AMA), accounted
  for over 70 of the funding for such trials in
  America in 2002, the most recent year for which
  figures are available. The lack of reporting of
  null or negative findings is pernicious because
  it skews the results of so-called
  meta-analyses, which compile data from previous
  studies of a treatment. If only positive results
  are reported, then a meta-analysis risks being
  too laudatory.
• The medical profession has been aware of this
  problem for a long time. However, pharmaceutical
  companies have a vested interest in keeping
  negative results quiet, so change has been slow
  in coming. But the proper balance between
  commercial confidentiality and public disclosure
  in the case of drugs, where ignorance can cost
  lives through misprescription, is different from
  that for, say, computer chips. The widespread
  government funding of basic drug research also
  gives the public a moral claim on the results.
  And a confluence of forces in the past few weeks
  may well succeed in pushing drug companies
  towards greater openness.

5
The current situation

• The US National Institutes of Health now require
  data sharing
• American Medical Association lobbying for a
  federal database, open to clinicians and
  patients, which would contain information on all
  clinical trials
• World Health Organization (WHO) has formally
  launched its International Clinical Trials
  Registry Platform project dataset and unique id
• The International Committee of Medical Journal
  Editors will no longer publish the results of
  trials that have not been registered in advance
  in an independent database open to the scrutiny
  of all
• PLOS Medicine proposes
• First tier registration in a publicly available
  database.
• Second tier the publication of a peer-reviewed
  summary of every trial, regardless of its
  outcome, in a traditional journal format
• Third tier deposition of detailed trial data in
  a structured, computable format that allows
  sophisticated searching and analyses across
  trials.

6
The NCRI Partnership
7
NCRI - work to date

• July 2003 - Strategic Framework approved
• Oct 2003 Data Sharing group established
• March 2004 Draft data sharing policy published
  online
• Summer 2004 pilots commence in collection of
  data sharing strategies
• July 2005 NCRI Board formally adopts Data
  Sharing Policy

8
Data Sharing Guiding Principles

• Joint approach across NCRI
• Consistent messages regarding policy
• Build on existing resources and standards
• Recognise this will be easier in some areas than
  others
• Phased implementation
• Empowering the research community
• Recognise resource requirements
• Flexibility

9
Focus areas

• Acquire good quality data
• Store data properly
• Share data appropriately
• Mine data and generate new knowledge

10
Progress

• Summer 2004 data sharing pilots
• CR-UK
• Tenovus
• Health Technology Assessment
• Scottish Executive
• Now at implementation phase
• MRC
• BBSRC
• Imaging and Pathology demonstrator project

11
NCRI Imaging and Pathology Demonstrator project

• Data sharing in action!
• Based on the MERCURY Trial
• http//www.virtualpathology.leeds.ac.uk/

12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
Resources currently available

• NCRI Data Sharing documentation
• NCRI Planning Matrix
• Documentation from Partner organisations

16
(No Transcript)
17
(No Transcript)
18
Other resources available

• caBIG workspaces
• Clinical Trial Management Systems
• Data Sharing and Intellectual Capital
• NCRI Informatics Unit happy to facilitate
  participation
• https//cabig.nci.nih.gov/workspaces

19
Workshop Groupings

• Group 1 North Hall
• (Chair Doug Fridsma, Scribe Fiona)
• Group 2 Garden Room
• (Chair Monica Jones, Scribe Peter)
• Group 3 Committee Room
• (Chair Peter MacCallum, Scribe Max)

20
What are the burning issues?
21
Discussion Session

• Transatlantic modelling efforts
• Variability in semantics is dangerous
• Variability in implementation is good
• How do we ensure semantic interoperability?
• NCRI Unit to facilitate UK interaction in
  relevant caBIG working groups, participate in
  Listservs, face to face meetings, etc. Interact
  at many levels

22
Discussion (2)

• How do we build national service infrastructure
  to support data sharing in clinical trials?
• Need to reach consensus on use of agreed data
  elements, datasets, etc
• Need dialogue with those who work in these areas
  (domain experts)
• Coordination from NCRI Unit, a funding stream to
  be made available
• Refer to Business case on use of standards
  (CDISC)
• NCRI to facilitate blanket UK membership of HL7

23
Discussion (3)

• Discussion groups on informatics website
• Matrix to include things beyond resources (e.g.
  listservs both UK and international)
• Community to help by linking to NCRI Informatics
  website and spreading the word

24
Discussion (4)

• Need long-term data curation strategies
• Software systems will change, need to plan for
  this
• Need to recognise resource requirements
• Industry not required to keep data indefinitely
  by Regulatory Authorities
• Partial centralisation may help

25
Discussion (5)

• Linking trial data to other data types is
  problematic need stewardship
• Trial Registers need to avoid duplication of
  effort for researchers
• Need to be able to map between registers
• Need economic argument to engage industry (7 a
  second can data stds interoperabilty, etc reduce
  this. Also, patients are dying while we wait.