Confronting Common Challenges in Managing Laboratory Data

1
Confronting Common Challenges in Managing
Laboratory Data

May 16, 2008
2

Confronting Common Challenges in Managing
Laboratory Data

• Laurie Callen
• Senior Technical Manager
• Synta Pharmaceuticals, Corp

Cory Donovan Manager, Database
Programming Prometrika
Ajay Sadhwani Executive Director Strategic
Product Development Premiere Research Group
Boston
3
Deal with Lab Data? Id rather.
4
With just a little planning..
5
This.
6
Not this.
7
Confronting Common Challenges in Managing
Laboratory Data

• What did we have to work with?
• One Clinical Program
• (One Therapeutic Agent)
• Multiple Studies, Multiple Indications

8

Summary of in-house studies
9

• Confronting Common Challenges in
• Managing Laboratory Data

• Plus
• Many legacy studies
• Multiple vendors
• Multiple database systems
• With varied data structures

10
Tools to Collect the Local Lab Data for
in-house studies

• 12 pt NIH sponsored study
• No CRF Lab Reports straight from the lab
• 22 pt RA study
• Local Lab CRF

11
Example Local Lab CRF
12
Central Lab Data from Vendor

• Over 50 variables
• Study Information Id, Protocol Code
• Patient Info Id, Number, Demography Info
• Lab Test Name, Lab Test Value, Date, Time,
  Ranges, Normals, Converted Values, Alert Flags,
  etc. etc.

13

• Confronting Common Challenges in
• Managing Laboratory Data

14
Confronting Common Challenges in Managing
Laboratory Data

• What did we have to do?
• Utilize information gathered from past studies
• Manage, maintain data for the new studies using
  Oracle Clinical and Oracle Clinical NormLab2
  Module

15

Confronting Common Challenges in Managing
Laboratory Data

• What did we have to do? (cont)
• Employ CDISC/SDTM naming conventions and
  structure within native Oracle Clinical.. as
  much as possible
• Minimize pre and post-processing of all lab data

16
Confronting Common Challenges in Managing
Laboratory Data

• What did we have to do? (cont)
• Needed to implement uniform Safety Analysis and
  Reporting requirements
• Ultimately create a uniform, consistent dataset
  for all studies to be used by all team members

17

Confronting Common Challenges in Managing
Laboratory Data

• How were we going to do it?
• 1. We needed to Prioritize needs
• Statistical Reporting, CDISC/SDTM, Data Cleaning,
  Lab Loading
• 2. We needed to Identify the required or
  mandatory fields for all of our needs
• Analysis, CDISC/SDTM, OC, NormLab2

18
Confronting Common Challenges in Managing
Laboratory Data

• How were we going to do it (cont)?
• 3. Determine the easy wins
• Example all labs datasets need a lab test name
  and lab unit field
• 4. Identify the redundancies and overlaps
• Remove unnecessary data fields

19
Confronting Common Challenges in Managing
Laboratory Data

• How were we going to do it (cont)?
• 5. Compromise when necessary

20
As a result.. A few major themes

• I. Develop a laboratory units/ normals management
  strategy
• II. Develop a uniform data structure that would
  be used for analysis, reporting and working with
  Oracle Clinical/NormLab2 and CDISC/SDTM

21
I. Develop a Laboratory Unit/Normals Maintenance
and Management Strategy

• Conversions

• Local Lab Units

• Lab Test Names

• Local Lab Ranges

• Date of Birth

• Standard Units

• Gender

22
Key Components of Lab Unit/Normals Maintenance
and Management Strategy

• Step 1 Create Library of Lab Test Names
• Step 2 Plan for Units/Conversions
• Step 3 Define Standard Units
• Step 4 Create Library of Local Lab Ranges
• Step 5 Identify Key Subject Characteristics

23
Step 1 Create Library of Lab Test Names

• Names will be used across all studies
• Document naming convention
• Full name/Abbreviation/Other
• Source documentation
• Important to have library
• Critical to standardization
• Most efficient for maintenance

24
Step 1 Create Library of Lab Test Names (cont.)
25
Step 2 Plan for Units/Conversions

• Create repository of all likely/possible units
  for each test

26
Step 2 Plan for Units/Conversions (cont.)

• Create conversions

27
Step 3 Define Standard Units

• Final results will always be reported in these
  units
• Standard units
• Company wide
• Program/Project/Study/Client specific
• Requires input and approval of all affected
  parties

28
Step 3 Define Standard Units (cont.)

• Company X Standard Units

29
Step 4 Create Library of Local Lab Ranges

• Lab X

Lab Y
30
Step 4 Create Library of Local Lab Ranges (cont)

• Lab units for Glucose will always be evaluated as
• mg/dL for Lab X
• mmol/L for Lab Y
• With a comprehensive lab management approach, CRF
  printed/recorded units dont really matter

31
Why Printed/Entered CRF Units Wont Matter

• Lab Y

x
x
32
Step 5 Identify Key Subject Characteristics

• Key Data Points
• Sex
• Date of Birth
• Used for
• Calculating age at time of exam
• Applying correct lab range

Need lab test date in order to calculate age at
time of exam
33
Ready to Go
1.Get Sex and DOB 2. Calculate age at time of
test
LAB Y RANGES
34
Ready to Go (cont.)
Confirm standard unit
Perform conversion to Standard Units (in this
case from mmol/L to mg/dL)
4.1 X 17.9 73.4 mg/dL
35
Ready to Go (cont.)
36
Importance of This Approach
Standard Units Conversions Local Lab Ranges
Test names
Possible units
37
Special Considerations Character Results

• Text Responses
• RBC Morphology, Pregnancy
• Pos/Neg, Normal/Abnormal, Clear/Cloudy
• Assign numeric value to results
• Positive 1, Negative -1
• Normal1, Abnormal -1
• Assign numeric range

38
Character Results (cont.)
1. Create equivalent numeric results
2. Create numeric ranges
39
II. Develop a uniform data structure that would
be used for analysis
40
Confronting Common Challenges in Managing
Laboratory Data
41
Confronting Common Challenges in Managing
Laboratory Data
42
Confronting Common Challenges in Managing
Laboratory Data
43
Confronting Common Challenges in Managing
Laboratory Data
44
Confronting Common Challenges in Managing
Laboratory Data
45
Confronting Common Challenges in Managing
Laboratory Data
Raw Data Structure used in OC/NormLab2 Addit
ional Post-processing necessary to make fully
CDISC/SDTM compliant
46
Confronting Common Challenges in Managing
Laboratory Data

• Some special considerations
• Unpredictability of local lab data reports
• (tests performed, test names, units reported,
  etc.)
• Lab CRF utilizing codelist values, or free text
  Other lab test performed
• Investigators Clinical Significant vs.
  Programmed Alert Flags

47
Confronting Common Challenges in Managing
Laboratory Data

• Some special considerations, cont
• Local lab units vs. Central lab units
• Not Done/ Missing Data
• Comment fields

48
Confronting Common Challenges in Managing
Laboratory Data
Some special considerations, cont
49
Confronting Common Challenges in Managing
Laboratory Data

• In Summary
• Identify Understand the needs/requirements
  (Early and Often)
• Plan Ahead develop a project plan
• Identify the tools/resources
• Key goal from outset should be standardization
• Units/Normals Management
• CDISC/Data Structure
• Develop Expertise

50
Confronting Common Challenges in Managing
Laboratory Data

• Thank you!
• lcallen_at_syntapharma.com
• cdonovan_at_prometrika.com