Title: 2001 NIC 011 Integrating Systems: The Challenges of Developing a Master Child Index
1Integrating Systems The Challenges of Developing
a Master Child Index to Integrate the NYC
Immunization and LeadQuest Registries
Robert Brackbill Deborah Walker Amy Metroka Paul
Schaeffer Alex Ternier Nicolas Gaglioti John
Somma Carla Rodriguez Noam Arzt (HLN Consulting,
LLC) New York City Department of Health e-mail
rbrackbi_at_health.nyc.gov
35th National Immunization Conference May 29 -
June 1, 2001 Atlanta, GA
2Objectives of Talk
- To describe the data processing issues related to
child health registries - To provide conceptual basis for integrating two
child health registries - To give an overview of challenges faced by
project developers and how they were overcome -
3Size and scope of City Immunization Registry
(CIR)
- Mandated by Amendment to NYC Health Code on
January 1, 1997 required to have immunizations
reported for children up to 7 years old - Contains demographics and immunizations of
children up to 18 years of age - 91 of private providers are currently reporting
- Contains nearly 2 million children with over
11.6 million immunization events
4Size and scope of LeadQuest (Lead Poisoning
Registry)
- Originated in 1982
- Contains demographic data, blood lead test
results, and follow-up of children up to 18 years
of age - 60 laboratories report blood lead tests for NYC
residents 49 report electronically - Contains approximately 3.9 million blood tests
from 1.7 million children - Contains 21,800 cases with elevated blood lead
levels - Approximately 49 of NYC children (6 months - lt 6
years) were tested in 2000
5Data Base Issues of CIR
- Matching and de-duplication of records
- Having consistent and standardized address
information - Finding answers to analytic questions
- Reducing back-end processing
6Data Base Issues of LeadQuest
- Improving information on children without high
test results - Having access to better provider data
- Finding answers to analytic questions
7Commonalities of Two Registries
- Each registry covers all children in NYC
- Each registry matches new information with
existing records - Each registry seeks being able to provide
complete child specific information
8Other Considerations
- Increasing accuracy of registry matching
processing - Creating system links for knowing as much as
possible about individual children - Obtaining the capacity to guide comprehensive
prevention and intervention programs
9Concept of Master Child Index (MCI)
- Create a core demographic data base into which
demographic data on each child would be loaded
from participating registries - Provide front end registry matching for new
immunizations and lead test data - Generate a common ID linking participating data
systems
10Basic Development Steps in Creating MCI
- Select data elements required by MCI
- Design logical relational database
- Training and testing of matching algorithm
(program) MEDD (Maximum Entropy De-duplication - Formulate business rules governing the
loading/translation of data from registries to
MCI - Initial data-load (order of loading is important)
from CIR, LQ, and vital records - Build application primary interfaces (APIs)
between components in system
11Basic Development Steps in Creating MCI
- Build communication services for linking
components of system including MCI, LQ, CIR, MEDD
and geocoding service - Design and develop MCI administration and query
tools - Synchronize all the databases and generate MCI ID
for linking two registries
12Steps 1 and 2 Select Data Elements and Design
Logical Relational Database
Table structure
Entity Diagram
Child ID
Birth Date
Master Child Index (MCI)
Address
13Step 3 Training And Testing of MEDD
SEARCHING AND BLOCKING
MEDD
CIR/LQ
Manual Review
14Steps 4 and 5 - Initial Loading and Translation
of Data From Registries to MCI and MEDD
MCI
Error File
MEDD
MCI Services
Geo-coding (City Planning)
Error File
Load using Business Rules
CIR
LQ
Vital Records
15Steps 6 and 7 Build APIs and Communication
Services
MCI Database (Oracle)
MCI Database Server (Unix)
Geo-coding
MCI Services (NT)
MCI Client
MEDD
CIR Database Server (UNIX)
LQ Client
LQ Database Server (NT)
CIR Clients Power builder Provider client Batch
processing
CIR Database (Oracle)
LQ (Microsoft SQL)
16Steps 6 and 7 (Continued) Build APIs and
Communication Services
MCI Database (Oracle)
Communication Service (CORBA)
LQ Powerbuilder Application
CIR Powerbuilder Application Provider
Client Batch processing
MCI Visual Basic Application
17Organizational Challenges
- Working with multi-vendors and DOH programs
- Coordinating feedback from DOH programs with
building of data system - Maintaining momentum on project and avoiding
de-synchronized development steps - Making decisions on communication technology with
insufficient information
18Technological Challenges
- CIR and LQ registries contain different data
elements - CIR and LQ registries apply different processes
to registry matching/updating records - Conflicting CDC standards
- Lack of over-arching DOH technical architecture
- CIR resides on a UNIX system and LQ on NT platform
19Meeting These Challenges
- Bi weekly DOH and vendor meetings and weekly DOH
management meetings - Formal, detailed requirements and specifications
documents - Separate workgroup meetings
- Providing vendors explicit criteria for each
deliverable - Achieving consensus from all parties on major
decisions - Maintaining a focus on the conceptual goals of
the project
20Conclusions
- Developing strategies to overcome organizational
and technological challenges is central to
project management - Maintaining consistent leadership encourages
programmatic buy-in - Fact finding and consensus building maximizes
ownership by separate program areas