caBIG CTMS Workspace F2F

1
caBIG CTMS Workspace F2F San Francisco, August
29-30, 2005 Lab Interface SIG John Speakman
2
Agenda

• Introduction (15 minutes)
• Three concurrent breakouts (40 minutes)
• Units of Measure
• ebXML hub solution
• Lab-to-toxicity mapping
• Possible report back on Units of Measure (5
  minutes)

3
Lab SIG background

• A lab interface to clinical trials databases was
  identified as one of the top four cancer center
  clinical trials needs at the caBIG kick-off
  meeting (February 2004)
• Center need simplify the process of building a
  data pipeline from clinical lab systems to
  clinical trials systems
• Lab data constitutes the majority of clinical
  trial data
• Rekeying especially time-consuming and
  error-prone
• caBIG need make the data presentable in a
  standard, shareable, interoperable form

4
caBIG Need a Data Standard

• Identified, evaluated and selected
• HL7v3 Clinical Trials Laboratory (CT Lab) message
• derived from the Clinical Data Interchange
  Standards Consortium (CDISC) LAB teams
  laboratory data transmission standard
• developed to facilitate the transmission of
  clinical trial laboratory data from local and
  central laboratories to the pharmaceutical
  company and/or Contract Research Organization
  (CRO) that is managing a drug development
  clinical trial

5
Centers Need Clinical Lab Feed

• Institutional political/cultural issues
• Security, privacy concerns (real or perceived)
• Especially acute where feeds from affiliate
  institutions are desired
• Clinical lab systems are heterogeneous
• To get a data feed from clinical lab systems, a
  custom interface has to be developed at both ends
• Development and maintenance are non-trivial
• Not viable at institutions that cannot corral
  significant software engineering resources

6
Breakout 1 Units of Measure

• Request received from caBIG caDSR Context
  Administrators
• HL7v3 implies the use of Logical Observation
  Identifier Names and Codes (LOINC)
• most labs identify tests by internal non-standard
  code values LOINC is a standard code set
• gt40,000 universal identifiers for laboratory and
  other clinical observations
• Maintained by Regenstrief Institute (Indiana
  University)

7
Breakout 1 Our Input Required

• Focus of Units of Measure
• Lab, Pharmacy, Radiology, Other Diagnostics
• Granularity of Units of Measure
• Bulk (one general list)
• Categories of Units
• per Use Case

8
Breakout 1 Whats Next

• Supply your input today
• Bear in mind that CTMS has to interoperate beyond
  cancer center/NCI community
• Pharma, regulatory bodies
• A proposal will be brought to next Lab SIG call
• Solicit CTMS Workspace opinion
• Pass to VCDE Workspace
• Will become a caBIG standard

9
Breakout 1 Audience, Significance

• Those interested in standards, ontologies and the
  CDE curation process
• Significance beyond the lab SIG
• this will introduce the CTMS workspace to the CDE
  curation process
• CTMS input on other areas will surely be
  solicited in the future
• I would encourage you to comment on the process
  as well as the issue at hand

10
Breakout 2 Integration Hub Solution

• Clinical lab systems are heterogeneous
• Misys, Cerner are most common in cancer centers,
  many others
• Most (incl. Misys, Cerner) talk HL7 v2.x
• High degree of optionality - there is no
  guarantee that any two lab systems will be
  interoperable
• To get a data feed from clinical lab systems, a
  custom interface has to be developed at both ends
• Development and maintenance are non-trivial

11
Breakout 2 The Problem Space

• Clinical lab systems will not go HL7v3 for
  several years
• Most HL7v3 lab messages are not yet ready
• Without HL7v3, a standard end-to-end solution is
  impossible
• Can we take centers half way and lower the
  burden?
• Provide a lab system integration hub solution
• ebXML suggested as a possible technology

12
Breakout 2 ebXML Questions

• What is it about ebXML that can lower the burden
  for cancer centers seeking to hook up
  heterogeneous lab systems?
• What do we need to know in order to prepare an
  SoW/RfP for the development of a hub for the
  transfer of lab data to clinical trials systems?
• How can we write the spec so it can be reusable
  in other areas of caBIG and beyond?
• Can we leverage the efforts of other groups that
  have implemented, or are implementing, ebXML
  solutions? In other words, how are our needs
  different?

13
Breakout 2 Adopter Characteristics

• What would be the characteristics and/or
  responsibilities for an early adopter of a pilot
  system? For instance
• Minimal cultural/political issues around access
  to lab data by clinical trials, e.g., access to
  data by affiliates
• Should be able to instantiate a clinical trials
  lab database around the HL7v3 standard
• Doesnt preclude centers using commercial vendors

14
Breakout 2 Audience, Significance

• Those with an interest in software engineering
  and XML
• The solution can provide a model for other areas
  within caBIG and beyond
• Transmission of data from any clinical system to
  a clinical trials system
• EMR, Billing, Order Entry, Decision Support, etc.
• CRIX is looking at ebXML also

15
Breakout 3 Lab-to-toxicity

• NCI Common Terminology Criteria for Adverse
  Events (CTCAE) v3.0, formerly Common Toxicity
  Criteria (CTC) v2.0, is NCIs system for ensuring
  uniform grading of toxicities
• Some of the grading criteria use lab values as
  their basis
• absolute (e.g., above a value of N)
• relative (e.g., double the maximum normal range)
• We can take lab values and use CTC/CTCAE as
  algorithms to map them to toxicities
  automatically
• Note a toxicity does not necessarily mean an
  adverse event someone has to determine if an AE
  has occurred

16
Breakout 3 Issues

• Define toxicities, adverse events and serious
  adverse events
• Who has a system we can leverage?
• What proportion of CTC criteria map to a lab
  value?
• Are there other scales than CTC?
• Do we do this only for lab values downloaded
  automatically?
• What if there are multiple values for the same
  patient on the same day? What if a lab value is
  deleted or updated?
• Are there opportunities to provide further
  automation?
• Can we use baseline values to automatically
  determine when an adverse event is expected?

17
Breakout 3 Audience, Significance

• Clinical Trial Domain Experts
• caAERS looks like being the first caBIG
  application out of the gate
• This is a chance to add further functionality
  quickly
• There are existing systems to leverage
• City of Hope, MSKCC, others?

18
Thats it!

• Three breakouts (upstairs)
• Units of Measure Room 202
• Integration hub solution Room 261
• Lab-to-toxicity mapping Room 271
• Questions?