the duke integrated subject cohort and enrollment research
TRANSCRIPT
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
1
H D W A 2 0 1 1 S A N D I E G O
Business Perspective
Use Cases
Technical Overview
Tips & Lessons Learned
Future Goals
Duke University Health System
2
H D W A 2 0 1 1 S A N D I E G O
2008 Enterprise Data Warehouse
Duke University Health System
3
H D W A 2 0 1 1 S A N D I E G O
Data goes back as far as 1996
4.09 Million Patients Represented
30 Million Encounters
17.55 Million CPT coded Procedures
39.1 Million CPOE Orders
271 Million Lab Records
250 Million Technical Charges
109 Million Measurements evaluated on 74,412 patients in the CDM disease registry
DSR Facts and Figures
H D W A 2 0 1 1 S A N D I E G O
2008 Enterprise Data Warehouse
2009 DISCERN Poster
Duke University Health System
5
H D W A 2 0 1 1 S A N D I E G O
An Open Source Integrated Subject Recruitment System
Bill Gilbert, Howard Shang, Jeffrey Ferranti, M.D., M.S.,
Duke Health Technology Solutions - Duke University
Health System, Durham, North Carolina
INTRODUCTION HL7 MESSAGE PROCESSING
• Failures or delays in meeting patient enrollment goals are widely
considered the most prevalent bottleneck to efficient clinical trials.
The secondary use of healthcare data coupled with open source
technologies promises to greatly improve the cost and efficacy of
research recruitment and enrollment. In recent years, the wide
spread implementation of Hospital Information Systems (HIS) have
led to the accretion of staggering amounts of healthcare data, but
few institutions are leveraging these data in support of research
enrollment. While several commercial and home grown systems
have attempted to ameliorate this problem, an open source
methodology for data-driven clinical research recruitment has yet to
be devised .
• The Duke University Health System has designed a
comprehensive open source solution that leverages both
retrospective data warehouse information and real-time Health
Level Seven (HL7) streams to aid in research recruitment efforts.
The Duke Integrated Subject Cohort and Enrollment Research
Network (DISCERN) uses these data sources to identify a cohort of
potential study recruits, cross reference this group with real-time
scheduling and lab data, and alert study personnel of potential
research candidates . The open source Mirth integration engine
(Mirth Project, supported by WebReach, Inc.) is being used to
ensure generalizability to other research institutions at minimal
cost. DISCERN can relay potential subject information to study
staff through clinic lists, email, and real time text pages. It can also
empower patients with context specific study information via the
Duke University Health System patient portal.
DESIGN METHODS
Improving the efficiency of subject enrollment will decrease the cost of clinical trials and ultimately accelerate lifesaving treatments into clinical
practice. The clinical trial timeline is amenable to technological solutions designed to shorten the recruitment process. Such solutions fall into
four broad categories: information, communication, workflow, and data management. (Marks 2002).
1) Information: Secondary Use of Retrospective and Prospective Data in Support of Research.
The design of the Duke Integrated Subject Cohort Enrollment Research Network (DISCERN) addresses the information category by utilizing both
retrospective and prospective data. The retrospective data is available in the data warehouse, and is used to build cohorts of patients that are potential
subjects for a clinical trial. In the case of rare conditions, retrospective data may provide what is needed to begin recruitment. More often, the inclusion or
exclusion criteria for subjects require current information about the patient, such as the last result for a particular laboratory test. Access to real-time Health
Level 7 (HL7) (HL7, Inc., Ann Arbor, MI) messages allows us to reason over real time data while identifying potential subjects. This comprehensive
approach leverages both retrospective and prospective data in an effort to define the best candidate list possible.
2) Communication: Leveraging Technology to Efficiently Communicate with Study Staff
Automatic communication with research staff is a large part of DISCERN. The design includes the ability to page study personnel, send secure email, post
advertisements on the Duke Health View Patient Portal, and provide "on demand" reports. Each clinical trial may choose one or more communication style
to receive information. For time critical studies, investigators may choose to be paged immediately once DISCERN identifies a potential subject.
The identification and recruitment of eligible patients during the course of busy clinical practice is difficult. This problem is compounded when physicians not
directly involved in a trial must maintain a working knowledge of trial timelines, inclusion and exclusion criteria, and consent details. (Tollman 2001)
DISCERN removes some of this burden on clinicians by e-mail alerting them of potential trial candidates on their clinic panel.
3) Workflow: Enhancing Workflow through the Thoughtful Application of Technology
Providing specific data on when and where subjects can be contacted may also achieve workflow enhancement. Rather than employing recruitment
methods such as cold calling or direct mailings, the study staff can meet with the patient in person, which is the most effective method of enrollment
(Verheggen 1998).
4) Data Management: Utilizing Economies of Scale for Data Storage and Security
Data management is also addressed by DISCERN. The retrospective data collected as a by-product of care will be stored in our secure data warehouse.
Strategies to integrate this data and provide the most accurate information possible are handled by our centralized data warehouse program. Prospective
data is managed by the rigorous standards of HL7 messaging. The data unique to the utilization of DISCERN is also stored in the Data Warehouse and
managed by the Information Management team. This centralized management of recruitment and enrollment data allows for economy of scale, and
removes the burden of data management from the individual investigator.
DISCERN is a comprehensive solution to the challenge of improving recruitment through the secondary use of health care data. Because DISCERN is
based upon open-source standards it is generalizable and easily adopted by other institutions for a relatively low cost.
OPEN SOURCE
DISCERN IMPLEMENTATION
A new patient is admitted to a clinic and a notification event message is
transmitted over TCP/IP.
MSH|^~\&|ClinicXYZ|W|PatientRegistry|UHN|2001052319
27||ADT^A01|22139243|P|2.4
EVN|A01|200105231927|
PID||123^^|2216506^||John^Doe^^^MR.^MR.||19720227|M
|||123 Any Street^^TORONTO^ON^M6G 3E6^CA^H^~123 Any
Street^^TORONTO^ON^M6G 3E6^CA^M^|1811|(123) 456-
7890||E^ ENGLISH|S| PATIENT DID NOT
INDICATE|211004554^||||||||||||
The message is received by the Mirth gateway listening over TCP/IP.
Mirth Administrator Dashboard
Mirth HL7 Message Flow Start Process
Mirth Engine
Receive HL7
message From IB
Successful
ACK
Process Message
For DISCERN
Incomming
Message
Process
Outgoing
Message
ACK
Store Message
For
Processing
(DISCERN.MESSAGE_QUEUE)
Outgoing
Process
Process Message
For Storage
Mirth HL7 Channel
Process
Mirth HL7 Message Transformation
Mirth Server Manager
Mirth Channel Writer
Mirth Alert Error Handler
Web Based Future Appointment Report On
Demand
HL7 FeedReal Time Data
HL7 FeedReal Time Data
HL7 FeedReal Time Data
ADT
Orders
Labs
DEDUCECohort Generation
STABILITY
COOLCAP
ROCKET
STABILITY
DEDUCECohort Generation
STABILITY
COOLCAP
ROCKET
DEDUCECohort Generation
DEDUCECohort Generation
STABILITY
COOLCAP
ROCKET
STABILITY
DISCERN EngineDISCERN Engine
DEDUCECohort Generation
STABILITY
COOLCAP
ROCKET
Age ≥ 60HDL < 40GFR 29 > 60ICD-9 = “Diabetes”(AND) “Periph Art Dis“(AND) “CerebroVascular Dis”
DEDUCECohort Generation
DEDUCECohort Generation
STABILITY
COOLCAP
ROCKET
Age ≥ 60HDL < 40GFR 29 > 60ICD-9 = “Diabetes”(AND) “Periph Art Dis“(AND) “CerebroVascular Dis”
Eligible Patient Arrives at ClinicEligible Patient Arrives at Clinic
01: Patient : Doe, John | AB1234Has arrived : Duke South Clinic 2JMeets criteria for: STABILITY Trial
01: Patient : Doe, John | AB1234Has arrived : Duke South Clinic 2JMeets criteria for: STABILITY Trial
01: Patient : Doe, John | AB1234Has arrived : Duke South Clinic 2JMeets criteria for: STABILITY Trial
Eligible PatientsEligible Patients
1. Study cohort defined
by researcher.
2. Data warehouse
queried for eligible
subjects.
3. Future appointment
schedule report
available on demand
for eligible subjects.
4. Eligible subject
arrives in outpatient
clinic.
5. Notification of
principle investigator
via email and/or
page.
6. Subject is contacted
by investigator for
recruitment into
study.
ACKNOWLEDGEMENTS
The project described was supported by Grant Number UL1RR024128 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH) and NIH Roadmap for
Medical Research, and its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH. Information on NCRR is available at
http://www.ncrr.nih.gov/. Information on Re-engineering the Clinical Research Enterprise can be obtained from http://nihroadmap.nih.gov/clinicalresearch/overview-translational.asp. The authors thank
Dr. Jeff Ferranti MD and the DUHS data warehouse team for their support and contribution.
Mirth
MSH|^~\&|PCM|A01|PHR|A01|20080903142352-0400||OMP^O09|1154941|D|2.3|||AL|NE PID|1||AR6280^^^NONE^MR^A||PLECZKOWSKI^PAUL^PED||20040714000000-0400|M||||||||||
Channel
Channel
Transforms Message to XML from a template
Transforms XML to Database calls Insert /Updates
Stores entire
message with
received ACK field
Insert into Message Queue
Where Message =
„MSH|^~\&|PCM|A01|PHR|A01|2009952-0400|
|OMP^O09|1000000|D|2.3|||AL|NE
PID|1||^^^NONE^MR^A|
|FJJEJ^PddL^PED||2004000000-0400|M|
|||||||||‟
Insert into Message Queue
Where Message =
„MSH|^~\&|PCM|A01|PHR|A01|2009952-0400|
|OMP^O09|1000000|D|2.3|||AL|NE
PID|1||^^^NONE^MR^A|
|FJJEJ^PddL^PED||2004000000-0400|M|
|||||||||‟
MSH|...
EVN|...
PID|John^Doe
MSH|...
EVN|...
PID|John^Doe
<MSH>...
<EVN>...
<PID>
<PID.1>John</PID.1>
<PID.2>Doe</PID.2>
</PID>
<MSH>...
<EVN>...
<PID>
<PID.1>John</PID.1>
<PID.2>Doe</PID.2>
</PID>
INSERT INTO Patients
VALUES(„John‟, „Doe‟);
Patient patient = new Patient();
patient.setFirstName(“John”);
patient.setLastName(“Doe”);
SQL
Java
<patient>
<name>
<first>John</first>
<last>Doe</last>
</name>
</patient>
XMLHL7 v2.X XML
Encoding
Transformation
(mapping, script, HL7
generator, XSLT) TO: [email protected]
SUBJECT: Visit
Dear John Doe,
Your last visit …
Message transformation
H D W A 2 0 1 1 S A N D I E G O
2008 Enterprise Data Warehouse
2009 DISCERN Poster
2010 DISCERN in production
2011 DISCERN how-to
Duke University Health System
7
H D W A 2 0 1 1 S A N D I E G O
WHY? There is a great need for
effective adoption of
integrated technology and
information solutions that
enable high quality clinical
care, research, and education.
Duke University Health System
8
HOW IT WORKS? By leveraging and integrating
the existing capabilities of:
• EDW Retrospective Data
• HL7 Information Broker
• Open Source MIRTH Engine
H D W A 2 0 1 1 S A N D I E G O
QUALITY OF CARE: Improve real-time notification
of patient-related events to interested parties
RESEARCH Improve capacity to identify,
recruit and enroll subjects to clinical trials within a highly constrained timeframe
Duke University Health System
9
DISCERN ALLOWS ADVANCES IN TWO
SIGNIFICANT AREAS OF HEALTH CARE:
15%
85%
QI RESEARCH
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
10
• Adolescent HPV Study Recruitment
USE CASE 1
• Cord Blood Study
USE CASE 2
• General Medicine Readmission within the last 30 days
USE CASE 3
Research
Research
Clinical QI
Report
Text Page
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
12
WHY?
Aid in the recruitment effort for a clinical study called “Alternate Dosing Study for HPV Vaccine “ (in collaboration with the CDC Control & Prevention) to determine the efficacy of longer dosing intervals.
HOW?
PROCEDURE: HAVE REC’D 2
DOSES OF HPV VACCINE
FUTURE APPT DATE
Generate a Cohort & Send Alert
GIRLS
AGED 9-18
PRIMARY CARE
PHYSICIAN
USE CASE 1: Adolescent HPV Study
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
13
USE CASE 1: Adolescent HPV Study
H D W A 2 0 1 1 S A N D I E G O
USE CASE 2: CORD BLOOD TRIAL
Duke University Health System
14
The early administration of umbilical cord blood cells may act as an important adjunct to total body cooling in cases of neonatal hypoxic ischemic encephalopathy (HIE)
The protocol mandates that cord blood be procured and prepared immediately after birth.
Cells must be administered within 24 hours of birth
Multiple missed opportunities at recruitment because of this aggressive timeline
H D W A 2 0 1 1 S A N D I E G O
USE CASE 2: CORD BLOOD TRIAL
Duke University Health System
15
INITIAL DATA Pre DISCERN (7/1/2009 – 1/1/2010)
7 infants underwent cooling at cord blood collection sites
The study team was not notified of study candidates for 6-24 hours
Patients With Cord Blood Successfully Procured : 0
Patients Successfully Enrolled in Trial : 0
Post DISCERN (1/1/2010 – 11/1/2010)
9 infants underwent cooling at cord blood collection sites
Study team was paged immediately for all cooling patients (Day or Night)
Patients With Cord Blood Successfully Procured: 9
Patients Successfully Enrolled in Trial : 8
H D W A 2 0 1 1 S A N D I E G O
USE CASE 3: PATIENT READMISSION ALERT IF DISCHARGED WITHIN 30 DAYS
Duke University Health System
16
WHY? To alert the discharging
physician if a General Medicine patient is admitted to a hospital or seen in the ED within 30 days of discharge.
Goals: › Feedback to Provider
› Identify trends in readmissions
› Identify targets for QI
HOW?
DISCHARGE DATE < 30
DAYS
INPATIENT
LOCATION
DISCHARGE
PHYSICIAN
Join Cohort with real-time alert
H D W A 2 0 1 1 S A N D I E G O
USE CASE 3: PATIENT READMISSION ALERT IF DISCHARGED WITHIN 30 DAYS
Duke University Health System
17
IMPACT
Provide real time information & feedback to discharging physicians for assessment
Provide leadership with data for hospital-wide improvement of readmission statistics
H D W A 2 0 1 1 S A N D I E G O
We
ek o
f 3/2
1
We
ek o
f 3/2
8
We
ek o
f 4/4
We
ek o
f 4/1
1
We
ek o
f 4/1
8
We
ek o
f 4/2
5
We
ek o
f 5/2
We
ek o
f 5/9
We
ek o
f 5/1
6
We
ek o
f 5/2
3
We
ek o
f 5/3
0
We
ek o
f 6/6
We
ek o
f 6/1
3
We
ek o
f 6/2
0
We
ek o
f 6/2
7
We
ek o
f 7/4
We
ek o
f 7/1
1
We
ek o
f 7/1
8
We
ek o
f 7/2
5
We
ek o
f 8/1
We
ek o
f 8/8
We
ek o
f 8/1
5
We
ek o
f 8/2
2
No
. o
f N
otific
atio
ns
Week
General Medicine Notifications Weekly Totals
Total Notifications/Week
Total ER Returns/Week
Total Readmits/Week
Notifications Trends
USE CASE 3: PATIENT READMISSION ALERT IF DISCHARGED WITHIN 30 DAYS
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
19
USE CASE 3: PATIENT READMISSION ALERT IF DISCHARGED WITHIN 30 DAYS
Patients Survey
Physicians Survey
H D W A 2 0 1 1 S A N D I E G O
USE CASE 3: PATIENT READMISSION ALERT IF DISCHARGED WITHIN 30 DAYS
Duke University Health System
20
Appropriate
Yes
No
Not Answered
Preventable
Yes
No
Not Answered
Prescriptions Filled
Yes
No
Not Answered
Follow-up with Primary
Yes
No
Not Answered
H D W A 2 0 1 1 S A N D I E G O
“The DISCERN alerts have been very useful … and I believe will lead to a better understanding of the targets (and limits) for our readmission and care transitions work.”
Thomas A. Owens, MD
Chief Medical Officer
Duke Hospital Medicine Programs
Duke University Health System
22
“We have found the DISCERN readmission notifications a very powerful tool. It has helped us track with a great deal of granularity … surveys regarding the use of the notifications suggests that they have influenced provider practice related to transitions of care… [and] has helped generate project ideas for general medicine.”
Jonathan G. Bae, MD
Duke University Medical Center
“...it was extremely reassuring to study staff to have DISCERN on board… We expect creatinine stopping rules will be in place for this multicenter NICHD funded trial, and we believe the DISCERN safety monitoring is crucial for subject safety… C. Michael Cotten MD MHS Medical Director Neonatology Clinical Research Duke University Medical Center
H D W A 2 0 1 1 S A N D I E G O
Receive Real-Time
HL7 messages from Duke’s Institutional IB
Provide Retrospective Data
Run ETL Processes
Generate Cohorts
Use DEDUCE
Duke University Health System
23
INFORMATION BROKER
MIRTH INFORMATION MANAGEMENT
Listen
Filter
Transform
Send
HL7
HL7
HL7
H D W A 2 0 1 1 S A N D I E G O
Mirth is an HL7 gateway facilitates the routing, filtering, and transformation of messages between health information systems over a variety of protocols.
All configuration is done through a Java web client which can connect remotely
Duke University Health System
24
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
25
Cohort Surveillance
Real Time Business
Rules
Notification Engine
.
2. The receiving channel then feeds eleven customized channels
INFORMATION
BROKER
eGate
1. Mirth receives & sends > 500,000 HL7 msgs each day
3. Which further filters, joins, transforms and sends only relevant data.
H D W A 2 0 1 1 S A N D I E G O
The MIRTH engine is easily configured:
1. Specify source connector
LLP, HTTP, FTP, POP3, files, web services, and more…
2. Create or use built-in filters and validation
profiles (i.e. Accept only administrative,
financial, or lab events)
Duke University Health System
26
H D W A 2 0 1 1 S A N D I E G O
3. Create or use built-in transformers
Mapping transformer: map data from incoming message to variables
Script transformer: execute custom script on message (Ex. JavaScript)
HL7 message generator: construct HL7 messages from data source
XSLT transformer: run XLS Transformations on incoming HL7 v3 or XML encoded messages
Duke University Health System
27
H D W A 2 0 1 1 S A N D I E G O
4.Specify destination connector(s)
LLP, ODBC/JDBC, J2EE, JMS, XML, SMTP, files, web services, and more…
› All messages and transactions are logged to an internal or external database (Ex. Oracle)
› Can be configured to auto-generate HL7 acknowledgement response (ACK)
Duke University Health System
28
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
29
Source Connector
Destination Connectors
Filters and Validation
Transformers (mapping)
HL7 IB
Application
Ch
an
nel
• Message received from Broker HL7 Message
• HL7 Encoded into XML Source
• Accepted or discarded based on filter rules or validation
Filter
• Message segments and fields can be transformed using JavaScript
Transformation
• The binding map is sent to the connector
Destination
• Sent to multiple destinations if needed
Data
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
30
HL7 Feed Real Time Data
ADT
Orders
Labs Cohort
Surveillance
Real Time Business
Rules
Notification Engine
DEDUCE Cohort Generation
STABILITY
Gen Med Patients
Adolescent HPV
Age ≥ 60 HDL < 40 GFR 29 > 60 ICD-9 = “Diabetes” (AND) “Periph Art Dis“ (AND) “CerebroVascular Dis” Eligible Patient Arrives at Clinic
01: Patient : Doe, John | AB1234 Has arrived : Duke South Clinic 2J Meets criteria for: STABILITY Trial
Eligible Patients
H D W A 2 0 1 1 S A N D I E G O
Main Objectives:
› LISTEN
› FILTER
› TRANSFORM
› SEND
Duke University Health System
31
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
32
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
33
MSH|...
EVN|...
PID|John^Doe
<MSH>...
<EVN>...
<PID>
<PID.1>John</PID.1>
<PID.2>Doe</PID.2>
</PID>
INSERT INTO Patients
VALUES(„John‟, „Doe‟);
Patient patient = new Patient();
patient.setFirstName(“John”);
patient.setLastName(“Doe”);
SQL
Java
<patient>
<name>
<first>John</first>
<last>Doe</last>
</name>
</patient>
XML HL7 v2.X XML
Encoding
Transformation
(mapping, script, HL7
generator, XSLT)
SUBJECT: Visit
Dear John Doe,
Your last visit …
Message Encoding and Transformation
Mirth Corporation 18831 Von Karman Ave
Suite 300 Irvine, CA 92612
H D W A 2 0 1 1 S A N D I E G O
MSH|^~\&|DMCADT|DUH|DWHIS1||20103343223||ADT^A01|DMCADT10034410|P|2.2|||NE
EVN|A01|20101100200200|||SIMS
PID|||ABC123||DOE^JANE||19001117|F||W|WHITE LANE ^^DURHAM^NC^277705^US^M|091|(919)555-1212|(919)555-1212|EN|S|LUT|XXX|012034442|||O
PV1||I|N79^7907^01|2|||1111111^SMITH^JOHN^A|1111103^CLARK^JANE^A||CAR||||H||||A||027|||||||||||||||||||A01||10||| 20103343223
ZV1|P|I|N||||||||||NADM
DG1||||CHF||A
Z99||6||| 20103343223
Duke University Health System
34
H D W A 2 0 1 1 S A N D I E G O
1. LISTEN & ENCODE HL7MESSAGES
2. FILTER THE DEFINED DATA
Duke University Health System
35
H D W A 2 0 1 1 S A N D I E G O
MSH|^~\&|DMCADT|DUH|DWHIS1||20103343223||ADT^A01|DMCADT10034410|P|2.2|||NE
EVN|A01|20101100200200|||SIMS
PID|||ABC123||DOE^JANE||19001117|F||W|WHITE LANE ^^DURHAM^NC^277705^US^M|091|(919)555-1212|(919)555-1212|EN|S|LUT|XXX|012034442|||O
PV1||I|N79^7907^01|2|||1111111^SMITH^JOHN^A|||CAR||||H||||A||027|||||||||||||||||||A01||10||| 20103343223
ZV1|P|I|N||||||||||NADM
DG1||||CHF||A
Z99||6||| 20103343223
Duke University Health System
36
IT’S A MATCH!
H D W A 2 0 1 1 S A N D I E G O
1. LISTEN
2. FILTER
Duke University Health System
37
3. TRANSFORM IT!
• Join Filtered HL7 & EDW Data
• Create contents for ALERT, EMAIL
or REPORT
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
38
SELECT * FROM
EDW.PATIENT_DISCHARGE
WHERE
PATIENT_ID = „ABC123‟
SQL <MSH>...
<EVN>...
<PID>
<PID.1> DOE^JANE </PID.1>
<PID.2>ABC123</PID.2>
</PID>
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
39
FROM: Data Warehouse Group
Duke Health Technology Solutions This is an Automated Email from DISCERN. DISCERN is a patient cohort alerting service provided by the Data Warehouse Group (DWG) of Duke Health Technology Solutions (DHTS). Please direct questions to Jane Smith [email protected] The visit type is: Inpatient Admission. The event took place on: 07/01/2011 00:05. The patient was discharged on 06/23/2011 17:13 by Dr. JANE DOE, M.D. 7 days ago with a diagnosis of CHF. The attending physician is: DR. GREGORY HOUSE The diagnosis for Inpatient Admission is: HYPOTENSION. PATIENT: SMITH, JOSEPH MRN: XXX123 TIME MESSAGE RECEIVED: 07/01/2011 00:05
H D W A 2 0 1 1 S A N D I E G O
Duke University Health System
40
DISCERN ALERTS & EMAILS
RESULTS SUPPORTING 22
PROJECTS
H D W A 2 0 1 1 S A N D I E G O
DISCERN framework is not a stand-alone tool but a service provided by a technical team.
The hybrid model requires both an enterprise feed of HL7 messages and an enterprise data warehouse.
All relevant staff must be alerted to the implementation.
It is IMPERATIVE that you follow up with the customer.
Modularize your MIRTH code for reusability.
Duke University Health System
41
H D W A 2 0 1 1 S A N D I E G O
Competition for patient enrollment may be exacerbated
Disruptive Technology
Requires effective communication and roll-out to all users
Duke University Health System
42
H D W A 2 0 1 1 S A N D I E G O
Extensibility of DISCERN tool directly to the research community
Extensibility of DISCERN services to other projects
› Stoplight Project (STAT Alerting System)
› Infectious Diseases
› Real Time DW
Extensibility to other patient health portal accounts and sources
Advertising research to patients
Duke University Health System
43
H D W A 2 0 1 1 S A N D I E G O
Combining open-source MIRTH HL7 messaging services with the data warehouse proves to be a highly effective model in addressing recruitment needs.
The flexibility of the model has proven powerful in addressing Quality Improvement health management initiatives.
The DISCERN model may help others seeking to deploy similar systems to achieve success.
Duke University Health System
44
H D W A 2 0 1 1 S A N D I E G O
Journal Publication pending in the Journal of American Medical Informatics Association (JAMIA)
“The design and implementation of an open-source, data-driven cohort recruitment system: The Duke Integrated Subject Cohort and Enrollment Research Network (DISCERN)”, In Press, Acceptance date 08/06/2011
Includes: Detailed Technical discussion of the Cardiac Readmit Project
(DUH Heart Center)
Detailed ‘How to’ Appendix
Duke University Health System
45
H D W A 2 0 1 1 S A N D I E G O
Howard Shang [email protected]
ACKNOWLEDGEMENTS:
Dr. Jeffrey Ferranti, M.D., M.S.
Chief Medical Information Officer
Duke Medicine
Monica Horvath, Ph.D
Senior Research Analyst
Health Analytics & Information Management
Duke Health Technology Solutions
Bill Gilbert [email protected]
Ilona Stashko, Senior Data Analyst
Duke Health Technology Solutions
Mirth Corporation
18831 Von Karman Ave Suite 300 Irvine, CA 92612
And thanks to all the staff that has been involved in the many successes of this project!
Duke University Health System
46
The project described was supported by Grant Number UL1RR024128 from the National Research Resources (NCRR), a component of the National Institutes of Health (NIH) and NIH Roadmap for Medical Research, and its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH. Information on NCRR is available at http://www.ncrr.nih.gov / Information on Re-engineering the Clinical Research Enterprise can be obtained from http://nihroadmap.nih.giv/clinicalresearch/overview-translational.asp The authors thank Dr. Jeff Ferranti MD and the DUHS data warehouse team for their support and contribution.