I am Associate Vice-Chancellor for Health Affairs, Director of the Informatics Center and Professor of Medicine and Biomedical Informatics at Vanderbilt University Medical Center. I am responsible for the operation and decision support information management infrastructure for the medical center as well as the academic division of biomedical informatics. Outside of Vanderbilt, I am Editor-in-Chief of the Journal of the American Medical Informatics Association and President-Elect of the American College of Medical Informatics.
Vanderbilt Medical Center includes schools of Medicine and Nursing; an integrated delivery system made up of a tertiary hospital, hospital joint ventures, medical group, clinic, and a regional payer neutral provider network; and health plans with products for Medicaid, Medicare, and commercial populations.
The primary purpose of those sections of the Health Insurance Portability and Accountability Act (HIPAA) that we are discussing today, is to improve the "efficiency and effectiveness of the healthcare system." The language of the act directs "the establishment of standards and requirements for the electronic transmission of certain health information" - with an emphasis upon those items needed for efficient administrative management. The language related to patient medical record information is quite different, directing the National Committee on Vital Health Statistics to "study the issues related to the adoption of uniform data standards..."
I think that the Committee should begin that study by asking whether a concentrated effort, by government and the private sector, on the development of uniform data standards for patient medical record information, has a reasonable chance of translating into the desired increases in health care quality, access and affordability, in the near or intermediate term. Only if the answer to that overarching, impact question is YES, does it make sense to focus attention just on the data standard implementation issues raised in the working group's four focus areas. If the answer to that question is No, the Committee should conduct a broader inquiry to identify the set of initiatives, related to electronic patient records, that can be completed on a wide-spread basis and have the desired effect.
For the long haul, I support the vision of an electronic patient record - one that is recorded by providers at the point of care, or by the patient and their family, using uniform data standards. Such standards, together with safeguards for privacy, will facilitate access from different facilities, linkage to knowledge-bases for decision support, and aggregation to document cause and effect of clinical outcomes. Work on these standards should be a high priority, but there are a number of reasons why I feel that this work will not be sufficient to have a significant impact on the health system in the short term.
Most successful standard setting efforts have involved physical systems. Physical systems are well defined and they can be reproduced exactly. Quality can be assured by building one factory that does the same process over and over again. A physical system can often be characterized by an equation, and an instantiation of that system might be specified with a few data points.
Unlike physical systems, biological systems are inherently variable. They are compartmentalized to allow change between high and low energy states with minimal generation or consumption of heat. Each person consists of many factories, each with its own copy of genetic material, material that mutates and evolves randomly. Because of the variability, the number of formulas and data points required to document a biological system increase several fold.
This variability has several implications. First, the number of conditions that would need to be handled by uniform data standards for the patient record are several magnitudes greater than that covered by non-medical standards. We do not yet have the tools that would be needed to manage such a large base of concepts.
Second, each record requires significant data entry. Data must be captured at the source using a non-ambiguous data standard if it is to be mapped to other standards or exchanged between systems. Our existing data capture techniques are not rapid or accurate enough to permit wide spread deployment into practice environments that are already overburdened by demands for increased patient volume and response to administrative requirements.
Given these realities, I think that there are several areas where government action is essential to moving the country toward the long term vision for the electronic patient record:
Support for Medical Informatics Research – Key areas include: structures for representing data and information that make relationships between terms and concepts explicit; methods for data mining and limiting retrieval to context; workflow; change management and human-computer interface issues.
Provision of incentives for use of electronic records in practice, such as requiring that the clinical reports for billable services be accessible electronically in an open format; and requiring practices to show how they incorporate best practice into their work.
Support for demonstration projects, such as case studies to demonstrate effective implementation strategies, or to develop metrics that can be used as bench marks to compare strategies.
Certification of a product’s functionality regarding: use of data exchange standards; export of patient data into non-proprietary formats without loss of information; provision of full function through application programming interfaces; and linkage of external information resources for operation and decision support.
Provision of a standard way of identifying people and facilities that give or receive care; products used in clinical evaluation or care.
Requiring government or commercial payers to make contract terms and authorization requirements available in a standard way as profiles that can be easily referenced by practice management and billing systems.
In the near term, I would focus upon standards that allow any report or block of clinical content to be treated as a package. The key is a header that identifies the patient, provider, facility, type of report, etc. The content behind the header should be in an open format, so that a human can determine its content with a viewer, or so that it can be processed through data mining. An example of this type of standard is provided in a paper by Dean Bidgood et al, that will be appearing in the January 1999 issue of JAMIA, outlining a framework for representing image acquisition context for images that are themselves represented with the DICOM.
In parallel, I would promote development in three areas that are essential to the long term vision for the electronic patient record. First, I would promote development of tools such as interlinguas and other data modeling techniques following the lead of the National Library of Medicine’s Unified Medical Language System project. These tools will be needed to manage content rich standards. Second, I would promote development of standard software architectures, such as CORBA, to support a “services” (as contrasted to “data”) view of the problem. Third, I would promote use of emerging E-commerce technologies to support patient (as contrasted to facility) ownership of the record.
For the short haul, I question whether we should place the development of an electronic patient record that is coded according to a uniform data standard in the critical path to evidence-based medicine and disease management. At Vanderbilt, we have been able to obtain a 40-50% reduction in cost of procedures, while documenting improvements in patient satisfaction, by helping care teams work with data to identify where there is variance in practice, and to then look at specific cases and the literature to agree on best practice. Our severity adjusted charge per case for Medicare patients based upon a statistical sampling of DRGs is now below regional, non-academic market, despite inclusion in our charge structure of the costs of providing over 40 services that are not provided anywhere else in the region. This change in practice was accomplished with data from the billing system.
In a similar vein, we have used a brute force approach to providing access to patient information for direct care. Any computer readable text, whether generated by a system such as the LIS, or by a specialty clinical application such as an automated pre-admission work-up, or by transcription, is imported into an indexed text repository that uses massively parallel processing to construct a patient record on the fly from a retrieval mask. The information can then be accessed via a browser over our secure intranet without querying proprietary systems. With this technique we were able to implement a computer-based record for our inpatient and outpatient complex in less than a year for less than $500,000.
I suggest that the working group direct recommendations to areas that will let the country reap the benefits of this type of low hanging fruit in the near term, in addition to making recommendations for efforts that should be carried out in parallel to lead toward our long term vision for electronic patient records.