Mr. Chairman, members of the Committee, honored guests. It is my pleasure to be with you today representing EDS in this testimony to the National Committee on Vital and Health Statistics regarding the Draft Criteria for Selection of Patient Medical Record Information (PMRI) Standards.
My primary healthcare experience has been wit h the Veterans Health Administration (VHA) Enterprise Architecture Office, where I serve as their Application Architect and representative to the Object Management Group and Health Level Seven standards organizations. In addition, I have been a part of the reference modeling activities under the Government Computer-based Patient Record (GCPR) Project.
EDS is a large-scale systems integrator that leverages standards, technologies, and products to deliver integrated solutions that add value for our clients. This testimony will center on this perspective, providing a view from the community intending to apply PMRI standards, and their impacts both to healthcare information systems and to the tools with which those systems are built.
PMRI standards will offer healthcare organizations and integrators the opportunity to purchase, integrate, and develop systems that minimize risks associated with single-vendor, single-product solutions. I see PMRI affecting several business lines (clinical, administrative, research, etc.) by providing consistent, meaningful, data of quality and integrity.
I commend the committee on the PMRI initiative and encourage you to remain steadfast in recognizing that there will never be consensus on technology, platform, product, or business practices. Initiatives such as open interface specification provide the loose-coupling needed to allow disparate organizations to continue to interoperate while doing business autonomously. As you proceed, it is important to recognize that integration challenges do not end at organizational boundaries, but extend within organizations as well. This work can significantly benefit healthcare institutions in addressing internal interoperability issues as well as enabling interchange across organizations.
In its consideration of standards, I encourage the committee to recognize the issues that extend beyond our national boundaries. Be it to support a multilingual environment, epidemiological interests across borders, corporations with global presence, co-operation in a multinational peace-keeping force, or simply citizens traveling abroad, the PMRI must be positioned to support interchange needs that are affected by non-US parties.
For maximum PMRI utility, organizations must be capable of extending the standard to address these "external" interests without sacrificing the fundamental interoperability offered by PRMI interchange.
Though much attention is being paid to "customization," this capability must be differentiated from "extensibility." In my view, sacrificing the core standard to bend to local interests (e.g., optionality) at the expense of interoperability is "customization." Conversely, the addition of information to support interests beyond the scope of the core standard would be extensibility. Without an extensibility mechanism, organizations are shackled to a standard and often forced to misuse its constructs, encouraging localities to bend the rules. An extensibility mechanism benefits the long-term utility of the PRMI.
I urge the committee to broaden its view of integration technologies to extend beyond message-based integration, such as middleware solutions. Though messages and message-based technologies have a clear value and several advantages to information interchange, they alone are not the only technology capable of doing so. The SDO Questionnaire may benefit from being broadened to allow NCVHS to consider other alternatives.
Acknowledging a separation of concerns between the content of the Patient Medical Record and the transport of that record is of tantamount importance. Without separating these concerns, the PMRI will not be able to take advantage of evolving and maturing technology solutions.
The following are my directed comments on the PMRI Criteria and Questionnaire:
Section 1: Draft Criteria for Selection of PMRI Message Format Standards
In my view, the criteria identified are well suited for selecting the PRMI standards. I feel strongly, however, that use of weighted criteria to place emphasis on those areas that are particularly important would add significant value. My recommendation is to place a heightened importance on:
I propose altering one of these criteria slightly. The current criterion --consistency with other standards-- implies that there must be alignment across all relevant standards activity. In practice I feel this is both undesirable and unrealistic. Each standards organization brings a differing perspective and thus a different value. Instead, if the emphasis is on the relationships among standards, the emphasis shifts from alignment to synergy, where the touch-points between standards can be explored and understood, allowing each standard to be used to its advantage. This is a stronger value proposition to the industry than uniformity.
"Market acceptance" is an important criterion, so long as it is viewed in the context of a changing landscape. Standards and technologies wax and wane. The technologies that are emerging should not be discounted solely because of minimal healthcare penetration, particularly if they show strength in solving similar issues for other market sectors. Conversely, standards with significant market share should not be blindly embraced at the risk of prolonging a potentially aging standard.
With regard to data quality and accountability, there is a need to transmit relevant metadata in conjunction with the data itself. For instance, this metadata will provide insight into the steward and source system of the transmitted information, and facilitates a shared meaning and understanding.
As is, the SDO Questionnaire provides a strong basis for understanding much of what the Standards Organizations have to offer to the PMRI initiative. There are a few objectives identified in the Uniform Data Standards report that may not be addressed by responses to the questionnaire as-is, however. The following suggestions are intended to provide insight as to how SDO's and their standards can address some of the specific objectives from the report.
Interoperability Categories. The report introduces three categories of interoperability: basic, functional, and semantic. The SDO Questionnaire, however, does not expressly differentiate among these categories. Instead, the interoperability level is left to inference based upon items such as the relationship between terminology and message structure.
Categorization Vehicle. In my experience, I have found that some objective measure, such as a categorization matrix, assists in differentiating standards and relevant touch-points between them. I would urge the committee to identify or define some objective metric to serve this role. Existing public work such as the viewpoints from the ISO Reference Model for Open Distributed Processing -- RM-ODP(1) may provide good seed material for such an effort. In doing so, I believe that NCVHS can provide a great service to the industry--government, private, and standards groups alike.
Cost Metrics. Though completion of the questionnaire would provide some value in understanding the costs incurred to purchase a standard, it would not sufficiently identify the reasonable expected costs to use the standard. The cost of implementing standards within organizations extends well beyond licensing of the standard itself. In fact, the initial integration costs, off-the-shelf product costs, development costs, and recurring maintenance (such as keeping current with the evolution of the standard) are likely to be the costs of primary interest.
I recommend soliciting from SDO's some notional cost model for organizations of a pre-determined, specified size with respect to organizational experience with the standard. For instance, one model might include expected costs for an inpatient hospital with 200 beds using HL7 2.x to upgrade to HL7 version 3.0, with relevant assumptions being documented. This figure should be provided both for initial development and integration as well as for subsequent maintenance stages.
I felt that the questionnaire would benefit by expanding the scope of questions in Subsection VII (discussing timely standards development), and Subsection VIII (addressing flexibility concerns). Below are some thoughts for consideration:
Subsection VII:
As I am primarily representing a technical community, I will defer comments in this area to the relevant clinical community. While discussing the PRMI with colleagues, however, concern was expressed in that identified transaction set did not appear to address the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) ambulatory care data requirements, gin particular, " a summary list of all significant diagnoses, procedures, drug allergies, and medications."[JCAHO STANDARD IM 7.4].
Based on my experience, I believe the following areas are worthy of consideration for inclusion in the PMRI identification and selection process:
When considering the interoperability of patient medical record information, it is important to recognize that two concerns are being addressed. The first of these concerns is the content itself: identifying the information that is of interest, and appropriate authoritative sources for describing that information. The second concern is more technically oriented, addressing the means by which that content is transferred.
While messaging and message set identification is a proven, effective vehicle for interchanging information, several alternatives (such as middleware, service-based architectures, etc.) may be as well or better suited to address these needs. My concern is that given the current framework for SDO responses and the apparent emphasis of the questions, these alternatives may be deemed out-of-scope though they merit consideration. I would urge the committee to broaden the questionnaire to ensure that these alternatives can be considered.
As previously mentioned, I believe that the NCVHS is in a pivotal position to identify an objective framework that can be used industry-wide to compare and understand standards. I would not view such a tool as a measure of quality, completeness, or judgement of the standards themselves, instead serving as a roadmap for the industry to illustrate how the standards fit together. For instance, this could be instrumental in identifying the touchpoints between terminologies, message formats, and middleware and, as such, provide guidance both to understand the landscape of standards and to leverage each to its strengths.
Given the importance of allowing organizations the flexibility to carry out their business practices within the general framework of the PMRI standards, a public forum is necessary to allow for the open registration of these extensions. This forum would allow localities and organizations to relate their extensions to the core standard and to define relationships to other extensions.
This activity would likely be a natural extension of the ANSI HISB Metadata Registry work. There is precedence for this activity both within and outside of healthcare, with initiatives such as RosettaNet (http://www.rosettanet.org) and Good Electronic Health Record (GEHR) Archetypes (http://www.gehr.org).
Regardless of the forum, it is essential that the PMRI standard allow for interchange at the semantic (or knowledge) level while still affording organizations the freedom to build on that standard to address their local or specific needs. If the core standard is not capable of addressing these concerns, organizations will be faced with implementing multiple standards or relying on a semi-standard, semi-proprietary solution approach that ultimately has an impact on interoperability.
I would like to thank the Committee for the opportunity to present my views in this area. Though ultimately this testimony reflects my views, it also reflects input from several individuals who were kind enough to offer me their thoughts, comments, and support. In particular I would like to recognize Tom Beale, Jim Demetriades, Doug Felton, David VonRinteln, Steve Wagner, and Sherri Walter for their substantial contribution to this work.
Respectfully Submitted,
Kenneth S. Rubin
Senior Systems Architect
EDS
Ken.Rubin@acm.org
703-845-3827
http://www.eds.com
(1) ISO/IEC 10746-3 Part 3: Architecture is the volume containing the viewpoint specifications. RM-ODP identifies five viewpoints which have proven useful in separating concerns in complex distributed systems.