Using CDC Edits Metafile in the Registry to Support Clinical Trials Recruitment

1
Using CDC Edits Metafile in the Registry to
Support Clinical Trials Recruitment

• Alan R. Houser, MA, MPH
• C/NET Solutions
• Dennis Deapen, DrPH
• Los Angeles Cancer Surveillance Program

2
Finding a Clinical Trial

• NCI web site, ClinicalTrials.gov, has 5933 open
  trials for cancer patients.
• Focused search tool to filter on disease, age,
  location, and treatment.
• Checked on 2/21/2007

3
ClinicalTrials.gov

• Optimal for searching for available trials for a
  single patient or patients with similar
  characteristics.
• Not suited for screening for trials for a large
  number of patients with dissimilar
  characteristics at one time.

4
Another Approach CDC Edits Tools (1)

• Descriptive cancer terminology built in
• Excellent at complex pattern matching
• Readily customizable each trial is translated
  into a single edit edits can be removed from
  Metafile when closed to recruitment

5
Another Approach CDC Edits Tools (2)

• Match multiple patients against multiple trials
• Match hundreds of trials against large numbers of
  cases at one time
• Portable can distribute metafile widely

6
ClinicalTrials.gov

• For each listed trial, three elements of
  eligibility criteria
• Disease characteristics
• Patient characteristics
• Prior concurrent therapy
• EDITS language can test each of these elements

7
How to Write an Edit to Select a Trial

• Identify inclusion and exclusion requirements for
  trial
• Match requirements to data fields in registry
  data set
• Failing an edit means matching a trials
  requirements require failure to display message
• Missing data write edits to exclude cases
  that dont match requirements, leaving cases that
  are still potential matches

8
Demonstration Project (1)

• Write metafile edits for a selection of actual
  clinical trials selected from ClinicalTrials.gov
• Select trials that require a diagnosis of cancer
  no prophylactic studies
• Select trials (five for each site) that use data
  fields available in registry data

9
Demonstration Project (2)

• Match metafile edits against a sample of real
  cancer case reports from central registry
  (California Cancer Registry)
• Select cases from 2004 forward to take advantage
  of Collaborative Staging

10
Methodology

• Datafile selected from California Cancer Registry
  Eureka database NAACCR 11.1 format
• Clinical Trials Metafile created with EditWriter
  3.0 five trials for each site
• Edits Metafile run against Eureka datafile with
  GenEdits Plus (beta)

11
Selecting Trials fromClinicalTrials.Gov (1)

• Select by Primary Site and Location (California)
• Breast Cancer
• Total Trials 789
• California Trials 131
• Translated into Metafile Edits 5
• Checked on 2/21/2007

12
Selecting Trials fromClinicalTrials.Gov (2)

• Select by Primary Site and Location (California)
• Prostate Cancer
• Total Trials 366
• California Trials 70
• Translated into Metafile Edits 5
• Checked on 2/21/2007

13
Selecting Trials fromClinicalTrials.Gov (3)

• Select by Primary Site and Location (California)
• Lung Cancer
• Total Trials 575
• California Trials 107
• Translated into Metafile Edits 5
• Checked on 2/21/2007

14
Case Data File (1)

• Extract test file from California Cancer
  Registrys Eureka database
• All patients diagnosed 2004-2006 (about 2.5
  years)
• Three sites breast, prostate, lung
• Vital status alive
• Los Angeles County residents at diagnosis
• 31,007 cases identified

15
Case Data File (2)

• Selected from California Cancer Registrys Eureka
  database
• 3 sites, patients alive at last contact
• Breast (C500-C509), 15,708 cases
• Prostate (C619), 11,197 cases
• Lung (C340-349), 4102 cases

16
Tracing the Filtering Process (1)

• Breast Clinical Trial NCT00382070
• Start 31,007 cases
• Exclude if not female, not alive
• Step 1 19,588 cases
• Exclude if not breast or if bilateral
• Step 2 15,613 cases
• Exclude if not invasive, not microscopically
  confirmed
• Step 3 12,688 cases

17
Tracing the Filtering Process (2)

• Breast Clinical Trial NCT00382070
• Exclude if not Stage I, II, IIIA
• Step 4 10,870 cases
• Exclude if ERA, PRA are within normal limits
• Step 5 8430 cases
• Exclude if hormone therapy not given
• Step 6 867 cases
• Exclude if not lumpectomy or simple mastectomy
  with lymph node staging
• Final 756 cases

18
Results Five Breast Cancer Trials

• Total breast cases 15,708
• NCT00074152 11,302 (72)
• NCT00127205 573 ( 3.6)
• NCT00382070 756 ( 4.8)
• NCT00388726 341 ( 2.2)
• NCT00390455 1075 ( 6.8)

19
Results Five Prostate Cancer Trials

• Total prostate cases 11,197
• NCT00004124 97 ( 0.9)
• NCT00063882 9 ( 0.1)
• NCT00110214 247 ( 2.2)
• NCT00123838 2814 (25)
• NCT00402285 12 ( 0.1)

20
Results Five Lung Cancer Trials

• Total lung cases 4102
• NCT00008385 222 ( 5.4)
• NCT00268489 948 (23)
• NCT00293332 1682 (41)
• NCT00368992 183 ( 4.5)
• NCT00409188 154 ( 3.8)

21
Summary

• Metafile technology can be used to screen large
  data sets for potential clinical participants
• Matching criteria is limited by registry data set
• Additional criteria not available to registry may
  exclude patients identified by metafile matching

22
Limitations of Metafile Scanningof Registry Data
(1)

• Data not collected
• Her2/neu (except in California)
• Date treatment ended
• Clinical factors (lab tests, fitness)

23
Limitations of Metafile Scanningof Registry Data
(2)

• Incomplete treatment data
• Cases may be reported before treatment is
  completed
• No identification of specific agents
• No identification of multiple courses

24
Limitations of Metafile Scanningof Registry Data
(3)

• Incomplete recurrence data
• May not be available to central registry unless
  reported from hospital registries

25
Extending the TechnologyHospital Registry

• More timely identification of eligible cases
• Monitor changes in patient status that could
  trigger eligibility (e.g., recurrence, additional
  treatment)
• Notification of managing physician when patient
  become eligible
• Interactive or batch processing
• Data set not limited to state requirements

26
Extending the TechnologyPhysician Reporting

• Immediate notification of potential eligibility
• Passive, not active, screening by physician or
  staff

27
Thank You

• Dennis Deapen (LA CSP), for guiding the direction
  of this work with his helpful suggestions
• Mark Allen (CCR), for providing the data extract
  from the Eureka database
• Winny Roshala (CCR), for help in translating
  clinical trial requirements into ICD-O-3 codes
• Tom Rawson (CDC), for making available GenEdit
  Plus (beta) for running the edits

28
For more information

• Alan R. Houser
• C/NET Solutions
• 1936 University Ave, Suite 112
• Berkeley CA 94704-1024
• [phone]
• alanh_at_askcnet.org