Extended PDQ Detailed Profile Proposal
1. Proposed Profile: Extended PIX and PDQ
- Proposed Editors: Alean Kirnak (email@example.com), Sandy Thames (firstname.lastname@example.org), David Shields (email@example.com)
- Date: N/A (Wiki keeps history)
- Version: N/A (Wiki keeps history)
- Domain: ITI Infrastructure (Public Health)
Patient Demographics Suppliers are only required to support matching against six specific demographic traits in the Patient Demographic Query transaction.
Experience with cancer and immunization registries has shown matching problems with poor data quality or diffentiating twins.
An extension requiring the Supplier to support matching against any or all of the demographic traits (fields) supplied by Patient Identity Feed would improve the situation.
Also, the Patient Identity Feed (based on HL7 V2.3.1) should optionally include fields supported by HL7 V2.5.
2. The Problem
The IHE IT Infrastructure Technical Framework Volume 2, Section 3.21, Patient Demographics Query, starting on line 4137 and ending on line 4145, states that “the Patient Demographic Consumer shall specify, and the Patient Demographic Supplier shall support” a list of six demographic traits for searching. Public Health registries, including immunization and cancer registries, contain population databases with many demographic records collected from a variety of sources. As a result of this, the quality of demographic data may be uneven, as registries often prefer partial or mixed quality data to no data at all. Effective searches on such demographic databases are thus more difficult and require a larger number of demographic fields both to locate matches and to avoid falsely matching records for separate patients. The cited section is found to be too restrictive for practical use by public health registries.
3. Key Use Case
Use Case #1 – Problematic Case
Two Patient Identity Feeds are sent to SDIR for twins Casey and Carey Smith. No multiple birth indicator or birth order is supplied. Upon receiving the feeds, the San Diego Immunization Registry (SDIR) matching software first adds Casey, then Carey. Upon adding Carey, the records are improperly determined to belong to the same patient and linked. Mrs. Smith brings twins Casey and Carey into North County Health Services clinic for childhood immunizations. SDIR is searched for Casey’s record using Name, Date of Birth, Sex and Address. The combined record is returned and there is no indication that this record represents an incorrect combination of both twins’ records.
Use Case #1 – Solution Case
SDIR is instead searched for Casey’s record using Name, Date of Birth, Sex, Address, Multiple Birth Indicator and Birth Order. The combined record is returned with an alert that this could represent an incorrect combination of both twins’ records, and a recommendation to unlink.
Use Case #2 – Problematic Case
Naomi Gonzalez brings son Jose into North County Health Service clinic for childhood immunizations. Jose does not have an account number at North County Health Services, and Mrs. Gonzalez does not have an immunization record with her for Jose. SDIR is searched for Jose’s record using Name, Date of Birth, Sex and Address. However, Date of Birth was incorrectly recorded in SDIR as 11/24/04 and not 1/24/04, and his last name is misspelled. A large number of low-probability matches are returned.
Use Case #2 – Solution Case
In addition to Name, Date of Birth, Sex and Address, Naomi’s first name and her husband’s first name are also passed as search criteria. The set of records returned is substantially smaller (possibly a single candidate match) in spite of the Date of Birth and Last Name errors in the SDIR data.
4. Standards & Systems
- Immunization Information Systems (IISs)
- Central Cancer Registries (CCRs)
- Other Public Health Registries
- Master Person Indices (MPI)
- Other systems implementing PDQ
- IHE PIX and PDQ profile and transactions Version 4.0.
- HL7 Version 2.5 Chapter 3 Sections 3.3.8 and 3.3.57.
- Implementation Guide for Immunization Data Transactions Using V 2.3.1 of the Health Level Seven (HL7) Standard Protocol.
- HSSP Entity Identification Service
- Service Functional Model (SFM), balloted HL7 Draft Standard for Trial Use (DSTU), references PIX and PDQ. See: http://hssp-eis.wikispaces.com/
- Initial submission to OMG includes a profile that demonstrates PIX/PDQ compatibility, with the above stated extensions. See:
5. Technical Approach
In addition to the demographic traits currently required by PDQ, the following traits have been found to be very useful in the practical experience of immunization registries:
- Patient Home Telephone
- Patient Birth State
- Patient Multiple Birth Indicator
- Patient Birth Order
- Last Update Date/Time
- Last Update Facility
- Mother’s Name
- Mother’s SSN
- Father’s Name
- Father’s SSN
However, our recommended solution is to enhance the text of the Patient Demographics Query section to optionally allow the Patient Demographics Consumer to specify, and the Patient Demographics Supplier to support, any or all of the demographic traits (fields) supplied by Patient Identity Feed. In addition, we recommend that Patient Identity Feed be enhanced to optionally include additional fields supported by HL7 Version 2.5 (Patient Identity Feed is based upon HL7 Version 2.3.1). This would include but not be limited to the Last Update Date/Time and Last Update Facility fields mentioned above. This approach would be more consistent with existing HL7 2.5 implementations and would allow more flexibility in developing finely-tuned matching algorithms suited to any particular population database.
Patient Demographics Supplier, Patient Demographics Consumer, Patient Identity Source, Patient Identity Cross-Reference Manager
No new actors
Under the Extended option:
- Allow the Patient Demographics Consumer to specify, any or all of the demographic traits (fields) supplied by Patient Identity Feed
- Require the Patient Demographics Supplier to accpet any or all of the demographic traits (fields) supplied by Patient Identity Feed
- Include additional fields supported by HL7 Version 2.5 in Patient Identity Feed (Patient Identity Feed is based upon HL7 Version 2.3.1).
New transactions (standards used)
No new transactions are needed
Impact on existing integration profiles
Existing implementations of PIX and PDQ would not be affected; however optional additional demographic fields are added for future implementations.
New integration profiles needed
No new profiles are needed.
Breakdown of tasks that need to be accomplished
The resulting changes to the text of PIX and PDQ is expected to be small and to mostly affect Volume 2 Transactions.
6. Support & Resources
The Public Health Data Standards Consortium has led the effort to formulate this proposal. Many organizations including the CDC Cancer Registries, the American Immunization Registry Association, representatives of HL7 and HSSP, and numerous individuals associated with public health organizations, integrators and vendors have taken part in its formulation. The primary PHDSC contact is Anna Orlova (firstname.lastname@example.org).
- It is possible that the addition of new search fields may create questions about the semantics of the PDQ profile from implementers or users wishing to know specifically how the added demographic traits are to be used.
- It is possible that the same reasons for which HL7 Version 2.3.1 was used by Patient Identity Feed in the first place, instead of Version 2.5, still remain in effect now.
8. Open Issues
This proposal does leave open the question of defining the specific use of demographic traits in search algorithms that might be used by PDQ implementers. We feel there are other possible solutions to this issue besides limiting or constraining the possible search traits. For example, HSSP (See “Systems and Standards”) has taken the approach of defining service semantics separately from interface definitions (in a “Service Functional Model”). It is also possible that this issue can simply be treated as remaining out of scope of the profile.
9. Tech Cmte Evaluation
<The technical committee will use this area to record details of the effort estimation, etc.>
Effort Evaluation (as a % of Tech Cmte Bandwidth):
- 35% for ...
Responses to Issues:
- See italics in Risk and Open Issue sections