CRIF_CRDB_and_TPS

Clinical Research Database

(CRDB) Overview









Dawn Caron-Fabio

Manager, Data Management Resource

Office of Clinical Research

CRDB Milestones

 Pilot Version November 1991



 Release 1.0 July 1992



 Release 2.0 April 1994



 Release 3.0 November 1996



 Release 3.3 October 1999



 Conversion to Web Summer 2004

CRDB Objectives

 Institutional accountability for research data

 Flexibility to accommodate a large variation of

studies, goals and data types

 User friendliness

 Prospective registration and randomization

 Easy access for global reporting of

institutional clinical research

 Minimize disruptions of frequent staffing

changes

CRDB Goals and Challenges

 Data coding

 Standardized data codes



 Security

 Ensuring protocol-specific access

 Controlling read/write access



 Flexibility and Ease of use

 Accommodate a large number of protocols

 Easy to enter and retrieve data

CRDB Current Status



 366,776 distinct patients in CRDB as of 3/2/2007

 MSKCC Patients 329,068 (89.7%)

 Non-MSKCC Patients 37,708 (10.3%)



 >148,000* patient records in >3,700 IRB protocols



 >483,000* patient records in >290 prospective databases

* Patients may be registered to multiple protocols and/or databases

CRDB Design and Structure

Hardware



 Database Server Cluster

(2) Compaq AlphaServer GS80 machines

(4) EV68/1224 CPUs and (6) GB memory, each



 Cluster is connected to a shared EVA storage array

containing 2.5TB raw disk space in either mirrored or

RAID5 configuration



 (26) Slot tape change with (2) DLT4 tape drives

CRDB Design and Structure

Software



 HP Alpha Tru64 Operating System, v5.1A





 HP Alpha TruCluster 1.5



 Oracle 9i RAC database software

CRDB Network Diagram

Storage Area Network (SAN) connections

10101









0 1 2 3 4 5 6 7

IP

StorageWorks SANSwitch 2/8-EL









SAN switches

HP AlphaServer GS80 HP AlphaServer GS80 10101









StorageWorks SANSwitch 2/8-EL

0 1 2 3 4 5 6 7

IP









Database Server crdbds1 Database Server crdbds2



High-speed cluster interconnect (memory channel)





Gigabit Network Switch

hp procurve

switch 5304XL

J4850A

Hewlett-Packard

StorageWorks

J4820A 24 port 10/100TX J4820A 24 port 10/100TX MSL 6030









HP RP5470 Application

DLT backup device

Server crdbas1

Run Attn. Fault Remote Power









hp A l p h a S e r v e r E S 8 0 hp A l p h a S e r v e r E S 8 0









HP RP5470 Application

HP9000 rp54x0









Server crdbas2

Run Attn. Fault Remote Power









HP Proliant DL380

HP9000 rp54x0





Application Server

crdbas3

hp proliant DL380g3 Int

Ext

Lan1









Tape

Lan2

UID









5 3 1









4 2 0









EVA disk array



Local Area Network (LAN) connections

MSKCC 1000BaseT LAN WAN



Standard Non-standard

Remote users

workstations workstations

Internet

hp workstation xw4000 hp workstation xw4000

Printers

Protocol Set-up

 PI specifies the data requirements of the protocol



• DMR staff define protocols using on-screen form:

– Eligibility criteria

– Randomization criteria (if protocol is a

randomized trial)

– Applications (protocol-specific subset of data

entry forms, reports, etc.)

– Users and their privileges

– Protocol-specific subsets of data codes

Protocol Structure

 Users can work on only one protocol at once

– In data entry, a user can enter data on only one

patient on one protocol at a given time



 Hierarchical

– You can establish a “parent” or “umbrella”

protocol covering multiple IRB protocols

– This enables you to analyze multiple protocols

across the same disease, service, department,

etc.

Hierarchical Protocol Structure

Example – Department of Medicine

MP-DOM





MP-BRST MP-GI MP-HEM MP-LEUK









00-093 01-011 01-129

Patient Registration

 When a patient is registered to a protocol:

– Demographic data retrieved from SMS, last 30 days lab data

retrieved from Laboratory Computer System

– User is prompted with protocol eligibility questions

– If an eligibility question can be answered from retrieved lab data (e.g.,

“Is the patient’s WBC above N?”) the system enters it automatically



 CRDB automatically determines eligibility

– (user can register the patient even if ineligible, but it is recorded)



 User prompted with randomization strata questions



 MSKCC pharmacy later consults CRDB

– to confirm registration, treatment assignment before dispensation

Prospective Databases

 POEs (points of entry):

– “private” (usually service-specific) non-IRB prospective

databases

– Collect data on (e.g.,) all patients entering a particular

service or department, regardless of whether they will

ever go on an IRB protocol

 Set up on request from investigator, who decides

what data should get collected

– DMR staff set up the database in much the same way as

an IRB protocol

– Data feeds from other hospital systems will happen

automatically (flow starts at patient “registration”)

CRDB Data Sharing

Outside MSKCC In MSKCC

SMS

Pharmacy

Clinical Labs

OR System

Surgical Complications



PIMS CRDB

Disease Mgm



IDB (ARC)



Pathlite

Industrial Sponsors



CDUS (CTEP)

Additional CRDB Collaborations

 Information Systems

 System Tracking and Reporting (STAR) System

 Pancreatic Cyst Tracking Application

 Computational Biology

 Study Tracker

 Pathlite

 NCI caBIG

 Clinical Trials Management System Contract

 Tissue Banks and Pathology Tools Contract

 Design and implement a public interface using caTissue (along

with the Inter-Prostate SPORE Biomarker Study )

CRDB Specialized Programs

 AICT (T Cell) Laboratory Database

 Cell Marker Laboratory Database

 Mouse Tracking Database

 Patient Survivorship Form

 Family History Pedigrees

 Health, Habits, and History Questionnaire

 Scannable Forms (Teleforms)

 Coding of Lab Values to CTC Toxicity Grades

 Analysis Tables

CRDB Specialized Programs

 Pet Scan Forms

 Biostatistics Collaboration Form

 Departmental Conference Sheets (eg, BMT

and Sarcoma)

 Comprehensive Phase Based Reports

 Specimen Protocol Registration Form

 Pharmacy Patient Query Form

 Specimen Tracking System

CRDB/ TPS Specimen Tracking

CRDB - Specimen Tracking Banks



 Tissue Procurement Service

 Sarcoma Bank

 Myelodysplastic Syndrome Bank

 Clinical Chemistry Lab

Specimen Tracking Bank Data 2006





Clinical Chemistry Lab Specimen

Accession 2006



July August Sept Oct Nov Dec



# Patients with New Specimens 4199 5328 6298 6410 6280 5964



# Specimens Stored 5532 7468 8792 8967 8809 8098

Specimen Tracking Bank Data

Tissue Procurement Specimen

Accession and Distribution 2004-2006



2004 2005 2006



# Patients with New Specimens 3821 4895 5033





# Specimens Stored 14238 23391 24031



# Specimens Distributed 3042 3847 4091

Single Patient Request 13 17 14



Clinical Review 615 683 698



Pilot Research Project 343 582 489



Clinical Trial 2071 2565 2890

TPS Specimen Tracking Design Team:



 Chair of the Dept. of Biostatistics

 Director of the Office of Clinical Research

 Clinical Research Database Administrator

 Database Developers (2)

 Data Management Resource Staff (2)

 Tissue Procurement Service Staff

(TPS Director and Coordinator)

1. 2.

Standard

PI PI

Patient Clinical

PI

Tests



Patient Consent

Required





Tissue

OR Pathology Procurement

4.

Request







Request

3.

Specimen Banking NT

Tumor

Process Bank

T

A

B C1

(Specimen Flow from Surgery to

C C2

Distribution from Tumor Bank) D C3

E

Goals for the System:

1. Relational Database Design

2. Coherent Management of Tissue Requests



3. Information Management of Resource Depletion



4. Linkage of Distributed Specimens with Clinical

Outcome Data



5. Ability to Distribute Specimens and Related Data

Anonymously



6. Improved Accession of Tissue

Database Relationships for Specimen

Tracking and Research Results:





Specimen Patient CRDB

Tracking MRN



System Specimen (Clinical Outcome

ID# Data)

Specimen ID#

Only





Pathological

Research Results

Decision

Support Tables



CRDB = Clinical Research Database

Specimen Tracking Table Structure

Data Entry Forms/Tables:

 Specimen Accession Form

 Specimen Accession

 Specimen Part

 Specimen Bank

 Specimen Bank Detail



 Specimen Request Form

 Specimen Request

 Specimen Request Detail



 Specimen Distribution Form

 Specimen Distribution

Specimen Accession Tables:

 Specimen Accession

 Information about the accession of individual

specimens and clinical details about the patient from

whom the specimen was obtained

 Specimen Part(s)

 Details about the individual part(s) that are created

from a single surgical procedure

 Specimen Bank

 Catalogues all specimens received by the bank

 Specimen Bank Detail

 Details about all specimens and their distribution

Specimen Accession: Data Elements

 Patient Identifiers & Diagnosis Information

 Pathology Date & Accession #

 Procurement Date

 Service Code

 Specimen Part #

 Specimen Identifiers (Site and Cell Type)

 Bank #

 Specimen Type (e.g., Tissue, Blood)

 Amount of Specimen Available

 Distribution ID

 Allocation Time

 Distribution Time

Specimen Accession Form (Page 1)

Specimen Accession Form (Page 2)

Specimen Request Tables:



 Specimen Request

 Contains information about the individual

request for specimens





 Specimen Request Detail

 Information about the types of specimen that

are requested (including the specific site(s)

and cell type(s) needed to fill the request)

Specimen Request: Data Elements

 Request ID

 Requester Information

 Request Date

 Distribution Type (e.g., Anonymous)

 Type of Request

 IRB Protocol

 Pilot Study

 Clinical Review

 Single Patient Request

 Specimen Information (Site and Cell Type)

 Billing Criteria

Specimen Request Form (Page 1)

Specimen Request Form (Page 2)

Specimen Distribution Table:



 Specimen Distribution

 Information about specimens distributed from

the bank. This process includes:

- Selection of Current Requests

- Ability to Query Available Specimens in Bank

- Distribution of Specimens Based on

Patient’s Consent Status

- Anonymization of Information

Specimen Distribution:Data Elements



 Distribution ID



 Request ID



 Additional Data From Previous Tables by

Way of Hidden Patient Identifiers

Specimen Distribution Form (Search)

Specimen Distribution Form (Allocation)

Specimen Distribution Form (Validation)

Specimen Distribution Form (Summary)

Non-Anonymized:

 Non-Anonymized (patient identity known)

 Patients give full consent for the specific

molecular or genetic tests that will be

conducted



 No restrictions are placed on the coding

variables (each specimen may be identified

by patient identifiers, facilitating a direct link

with clinical outcome data)

Investigator Anonymized:

 Investigator Anonymized (samples coded

but linked to patient identity )



 Investigators will be unable to identify the

patient



 Specimens will be distributed with a coded

identification number that will identify the

specimen



 Clinical information may also be supplied via

the Specimen Bank #

Anonymized:

 Anonymized (unlinked to patient identity)

 Once a specimen is distributed it can never

be re-linked to the patient



 All clinical data must be available at the time

of distribution

Additional Specimen Forms

Liquid Accession Form

Clinical Chemistry Lab:

The Next Steps…

 CRDB

 Availability of CRDB on the Web

 Initial Projects

- Specimen Protocol Registration (06-107)

- Minimal Data Set Data Collection







 Specimen Tracking

 Incorporation of the new 06-107 Consent

Process