Introduction

DISCLAIMER: this version has not been fully reviewed by the authoring team and is subject to revision prior to the Second MPI Workshop.

This white paper presents a proposal for the specification of a Master Patient Index (MPI) for use in the healthcare industry. This proposal grew out of work started at the "First MPI Workshop" held in Santa Fe, NM on May 20-22, 1996.

The essential role of a MPI is to improve the quality of patient care. Originally, MPIs were only manual; almost every provider had some sort of manual MPI used primarily to link patient episodes for continuous care. In today's world of automated and a potentially global network of MPIs, improving the quality of patient care is still their most important role. By linking information through a MPI, an authorized provider can not only access a patient's pertinent clinical information for medical decision-making purposes, but can also link to the patient's clinical, financial and administrative information in a longitudinal fashion. These capabilities can dramatically help to improve the quality of patient care delivery.

A MPI can be summarized as being a dynamic, secured directory of uniquely identifiable patients, and "pointers" to where the respective medical and possibly insurance or other medically related information about that patient resides. A MPI is neither a repository for medical records themselves, nor a superset/subset of patient data that is not used strictly and exclusively for coordinated and consistent patient identification. Such a directory can exist within a single healthcare provider or enterprise, at a local community level, at a state level, or even, it is envisioned, at a national or international level.

Each patient that is represented in a MPI may be identified in more than one computer system, across multiple organizations in a wide variety of settings over time. Therefore, a MPI must be able to accomodate identification instances possibly originating from a variety of index-producing algorithms producing identifiers with different formats. Even where public policy legislates unique identifiers, a MPI must accomodate legacy identifiers as well as identifiers from other regions beyond the jurisdiction of the policy.

Minimally, a MPI must provide pointers to medical information that is reasonably available and, when automated, accessible using real time, interactive information technology. Optionally, a medical record "pointer" may point to data that does not reside in electronic form, such as a phone number to call for more information about the record.

Even though a MPI is not a repository for administrative information or medical records, a MPI may provide pointers to special categories of medically related data. For example a MPI may have pointers to an Essential Emergency Data Set (EEDS), or other such patient-supplied information. A MPI may also have pointers to contact information for healthcare "providers" (such as hospitals and doctor's offices) and "payers" (such as insurance companies or the patient's employer).

This document discusses:

• What is the goal of a MPI standard?
• What are the requirements to be met?
• How are patient privacy and confidentiality concerns addressed?
• The conceptual or reference object model for the proposed MPI
• The services provided by the MPI
• Possible business opportunities related to the MPI
• The Interface Definition Language (IDL) for the MPI

References

The following references were used in creating this proposal:

1. Proceedings from the "First Master Patient Index Workshop", at World Wide Web site http://www.acl.lanl.gov/cpr/proceedings
2. "NCHICA Specifications Workgroup Document S1: Specifications of the MDS and MPI-ERN Standard SUMMARY", Version 11/12/95 (Note: this is not a public domain document, but has been made available for MPI Workshop participants).
3. OMG CORBAmed Request for Proposal, "Master Patient Index & Patient Record Indexing Services", draft 1, July 31, 1996.

Goal

The goal of this work is to define an interface specification for the interoperability of MPI systems. Having such an interface will allow the various internal (within a single enterprise) proprietary MPI systems in existence today to exchange information with each other. Having a network of interacting MPIs will establish a foundation upon which to build large-scale external inter-enterprise (beyond a single enterprise) MPIs.

It is increasingly perceived that having a MPI standard is critical to the future of the healthcare industry. Electronic medical information cannot be shared or exchanged unless it can be located! The following items reflect the importance of MPI standards:

• The Object Management Group (OMG) recently formed a healthcare vertical market focus group, CORBAmed, which identified MPI as the most important item to standardize.
• The National Information Infrastructure, Health Information Network Program (NII-HIH), a public-private effort sponsored by the Defense Advanced Research Projects Agency (DARPA) of the United States government is creating a national architecture for the exchange of critical medical information, of which MPI is an important component.
• The Health Insurance Portability and Accountability Act, signed into law by the President in August 1996, requires the use of " a standard unique health identifier for each individual, employer, health plan, and healthcare provider for use in the healthcare system." It would seem such use of unique identifiers in the health care system, complemented with the need to locate medical information about a patient will require the use of MPI technology.

To reach this goal, a conceptual object model for an MPI is defined, from which a CORBA IDL can be derived. CORBA is the "Common Object Request Broker Architecture" from the OMG. IDL is the "Interface Definition Language" used to specify CORBA interfaces. CORBA is specified as the target architecture as it is becoming supported on a wide variety of distributed computing environments, including the Internet and World Wide Web (WWW), and in healthcare the Andover Group for Healthcare Interoperability; and in a number of programming languages (including C, C++, Smalltalk, ADA, Cobol, Java, and Visual Basic).