"BEST OF BREED" VS SINGLE VENDOR SYSTEMS -OPTIMIZING CLINICAL INFORMATION SYSTEMSIN COMPLEX COMPUTING ENVIRONMENTS
Information Technology cannot usually be justified in its entirety purely on the basis of productivity gains or even quality improvements, even though significant gains in productivity and quality can be achieved. More importantly, I.T. is an enabling vehicle to support the strategic and business objectives of your institution. It is important, therefore, to consider the institutional information system as a whole and discrete entity, even if it is made up of a number of separate, dissimilar, and discrete subsystems interoperating (or not) among themselves to support the various user constituencies that exist within the institution. Effective integration in some form is a pre-requisite for long-term success.
Results of the February 1994 HIMSS/Hewlett-Packard Leadership Survey (1,033 respondents) indicated that integration is the most important I.S. priority in the next two years. Survey results indicated that for "first most important", 31% of the respondents listed "integrating systems across separate facilities", 13% "integrating disparate departmental systems", for a total of 44% of all respondents. In addition, implementing a computer-based patient record, clearly an integration issue, was cited by 19% as their most important priority. Therefore, nearly two-thirds (63%) of those surveyed felt integration to be their most important initiative.
Installing and effectively operating and managing information systems in today's complex healthcare environment is not a trivial undertaking. Many healthcare CIOs find themselves in a quandary with various clinical and operational departments seeking "best-of-breed" systems and the increasing, more global need for higher levels of integration of information from such systems in order to provide managers/administrators with correct and comprehensive data from which to make appropriate financial, operational, and patient care decisions.
To some of us the term "best-of-breed" is more appropriate for dog shows than information systems. With the advent of "open" systems, it is generally perceived that various specialized departmental systems (actually sub-systems) can be acquired independently from a variety of "boutique" vendors. It is further presumed that such an assemblage of independently developed systems can, through the miracle of "open" systems and data exchange standards (e.g., HL7, ASTM), function harmoniously and almost "seamlessly". This has rarely been the case.
Granted that the evolving standards and technologies are making it easier than ever before for different systems to communicate with one another. But in the health care setting, is this sufficient? For example, in a multi-vendor, open architecture with multiple "best-of-breed" systems, is the appropriate clinical, financial and/or demographic information available with the right content in the right format, at the right time, in the right place for a particular type of user; e.g., physician, nurse, hospital administration, business office, medical records, etc.? Can advanced "expert" systems apply "decision support" to data which exists in the data bases of the various systems? In passing data from system to system, how can we assure that all systems are consistent? What happens if one or more systems go down? Sound complicated? It is.
Too often, in our experience, clinical department management, in their consideration of a specific departmental information system; e.g., laboratory LIS, radiology RIS, etc., often underestimate the importance of system integration. Their goal, appropriately enough, is to optimize their system (more appropriately their sub-system). What results is usually sub-optimal for the institution as a whole.
Unfortunately, without integration with other hospital systems many benefits can be lost and in some cases additional, unanticipated manpower needs arise; e.g., to manually enter data into the departmental system which could, and should, be automatically supplied by other complementary hospital information systems. Further, lack of effective integration for the purpose of access to and distribution of clinical departmental information can adversely affect user constituency's perceptions of the department.
The consequences of inadequate integration include:
- Physicians' view of departmental data is limited and often cumbersome to access and manipulate
- Data interchange among systems is limited and/or costly
- Expert systems ("rules-based", "artificial intelligence") are limited in scope and accessibility of data
- Concurrent and dynamic correlation of related clinical data is not possible
- Administration, operations, and patient care managers lack a complete picture
- The system architectural strategies employed to provide higher levels of integration can be considered in the context of two polar opposite approaches:
- Interconnection of "best-of-breed" systems in some network topology
- Use of a single vendor, "highly integrated" common software system.
These two extremes are graphically depicted in Figure 2. Admittedly, the illustration is intended to be somewhat "tongue-in-cheek" but is a good depiction of our dilemma. Do we:
- Pick an assemblage of "best-of-breed" systems and have the complexity of managing a variety of hardware platforms, operating systems, programming languages, and databases - not to mention dealing with a multiplicity of vendor sales and support representatives. In the best-of-breed" dogsled analogy, the great challenge is to make all the different dogs, each with their unique strengths and weakness, pull together effectively. Even the Chihuahua, whose feet cannot reach the ground, has a contribution to make - but how do we do it?
- Acquire a "homogeneous, integrated" system from a single vendor and attempt to accommodate, work-around and/or try to ignore those deficiencies perceived as shortfalls in functionality (rightly or wrongly) by various departments. For the "highly integrated" dogsled, the trick is to provide sufficient capabilities to meet the needs of various unique hospital departments and services while retaining common software, database and "look and feel".
Table 1 indicates some of the positive and negative characteristics intrinsic to the "best-of-breed" open systems approach in contrast with the "Homogeneous" single vendor strategy. There are, obviously, pros and cons for both "open, best-of-breed" systems and more homogeneous, single vendor integrated systems. Open systems provide for departmentally optimized systems, flexibility, and control at a departmental level and the ability for an institution to replace any departmental system (or subsystem) without replacing the entire hospital/financial system structure. In doing so, we incur the burden of dealing with multiple different vendors (and the potential for "finger-pointing" in the case of problems of undetermined origin), varying user view and access methodologies - which can compound already significant training needs and the potential for database inconsistencies among the various unique system components. Integrated systems, as indicated in the table, are the "mirror-image" with regard to pros and cons.
Fortunately, modern, evolving technologies are providing mechanisms to assist the integration process. More and more healthcare professionals, facing the challenges of health information networks, managed care, increasing regulations, reduced budgets, and more, are accepting the old adage that "the whole is more than the sum of its parts" and are taking steps to increase levels of integration within their institutions and in community and regional networks.
The following represent some (but by no means all) of the things that could be done to create a more integrated systems environment within your hospital.
Establish multi-departmental teams, including I.S.
PRO Creates framework for integration. Improves understanding of inter-departmental informational needs.
CON May be difficult to find active participants. Takes time away from other duties. Requires substantial commitment.
Select vendors who are "Open" to integration.
PRO Permits cooperative networking and a "team" approach.
CON "Openness" is more than adherence to standards - it is also a frame of mind and a willingness to cooperate.
Adhere to common standards (ASTM, HL7, etc.)
PRO Provides at least a basis for commonality.
CON Standards not complete or universally accepted and/or implemented.
Utilize an Interface Engine.
PRO Reduced costs to exchange data among multiple systems. Easier response to changes. Easier to maintain than point-to-point interfaces
CON May be a single point of failure. Throughput and performance need to be assured. Care should be taken to assure synchronization among participating sub-systems.
Develop a clinical data repository.
PRO Global data base. Sub-systems as "feeders". Expert systems can apply. Fits client/server model.
CON May be a single point of failure. Approach still largely unproven.
Integration is important!
Integration doesn't "just happen"! Just because something is called an "open" system doesn't mean that it is - or that it is a panacea. Information Systems professionals working in concert with teams of other departmental professionals can, collectively, optimize information systems within their institution and achieve levels of information integration, access, distribution, timeliness, and usability never before possible. The Hospital, its administration, physicians, staff, and its patients will all benefit.