The Clinical Data Repository

1
The Clinical Data Repository

• Ken Scully, MS
• LSP Conference
• March 13, 2003

2
Clinical Data Repository

• An information system consisting of
• a database containing information from UVa
  legacy systems, as well as additional clinical
  details
• a custom-built web interface for creating
  standard and ad-hoc reports and for downloading
  data to your desktop

3
Purpose

• Enable detailed, flexible and rapid retrospective
  queries of patient data
• Supply data sets for research, education and
  medical management applications
• Provide insights into opportunities for improving
  outcomes and containing costs

4
Has Been Used For

• Clinical research for publication
• Outcomes evaluation and assessment
• Analysis of utilization
• Treatment patterns and protocols
• Cost containment studies
• Education

5
Some examples

• Is there an association between inflammatory
  bowel disease and lymphoma?
• What happens to patients who have gastric bypass
  surgery for obesity?
• What are the causes of death for patients who die
  within 30 days of surgery?
• What factors are associated with readmission
  attributable to influenza?

6
Some examples

• Does care of patients with diabetes improve with
  implementation of a case management program?
• Is there a July effect, i.e. does house staff
  experience affect patient outcomes? Likewise, is
  there a weekend effect?
• What are the factors associated with bleeding
  complications of anticoagulation?
• How many women deliver vaginally after prior
  c-section?

7
Data Sources

• SMS Inpatient and OutpatientJuly 1993 Dec
  2002
• HSF Physician Encounters (IDX) July 1992 - Dec
  2002
• Lab Results (In/Out Patient)July 1996 - Dec
  2002
• Micro-biology Results (In/Out Patient)
  Sep 2001 - Dec 2002
• TCV-STS Summit (Heart Center)July 1994 Dec
  2002
• Virginia Department of Health (VDH)July 1992
  Dec 2000

8
System architecture

• Dell PowerEdge 4600 (dual 2.2GHz Intel
  processors)
• RH 7.3 Linux
• Sybase RDBMS (version 12.5)
• 4GB main memory
• 8 - 30GB (15,000 rpm) RAID Disk Array
• Perl, CGI, C, JavaScript
• Housed in ITC-ACHS data center

9
Volume

• 60GB data
• 650,000 patients and 8.3 million visits
• For year 2002
• 1 million visits
• 8.4 million lab results
• 8.3 million billing transactions (includes
  medications)
• 3.3 million diagnoses
• 1.8 million procedures

10
Project history and organization

• Began as the Clinical Research Database in 1996
• Formerly housed in ITC-ACHS
• Moved to Health Evaluation Sciences July 2001
• Currently 3 FTE 1 open position

11
Security and Confidentiality

• Disguised patient and MD Identifiers
• Patient identifiers kept separate from data
• Uses https for web access
• Uses are tracked in log files
• Authorization required for access
• Individual database login/passwords required
• 2nd level authorization required to obtain real
  identifiers

12
Privacy and the CDR

• Human Investigation Committee approval may be
  required for any research study, especially
• Charts will be reviewed
• Patients will be contacted
• Study will be published

13
HIPAA

• User access and display will be HIPAA compliant
  by mid-April
• Requesting HIPAA waiver for receiving patient
  data
• Users will be given different menus depending on
  usage type
• With HIC number or where HIPAA regs allow access
  to patient specific info users will have access
  to near de-identified info (dates, county of
  residence).
• Other users will only have access to totally
  de-identified info.

14
De-identification Requirements

• Names
• All geographic subdivisions smaller than a State
  and their equivalent geocodes. 1st 3 digits of
  zip code can remain for geographic unitsgt20,000.
• Dates birth, admission, discharge and death
  dates. Nothing smaller than a year.
• Telephone numbers
• Fax numbers
• Email addresses
• Social Security numbers
• Medical Record Numbers
• Health plan beneficiary numbers

• Account numbers
• Certificate/license numbers
• Device identifiers and serial numbers
• Vehicle identifiers and serial numbers
• Web URLs
• IP address numbers
• Biometric identifiers, including finger and voice
  prints
• Full face photographic images and any comparable
  images
• Any other unique identifying number,
  characteristic, or code including Clinical Trial
  Number

15
CDR Intended Use

• HIC Approved Research
• Review preparation for research
• Education
• Quality Assessment
• Other

16
HIPAA The Silver Lining

• Currently, all users must request patient
  translation to real MRNs and patient names with
  division head or department chair signature.
• HIPAA - HIC approved research will be able to
  access patient names, MRNs without the need for
  the extra form.

17
CDR Architecture
Load Programs
18
CDR Demonstration