CLINICAL DECISION SUPPORT SYSTEMS

Fahhad Farukhi

 

 

INTRODUCTION

 

The health care market is the largest industry in the United States, with expenditures expected to reach $2.6 trillion, or 15.9% of the national GDP by 2010 [1]. In addition to the rapid increase in health care expenditures, technology has developed to provide support to health care delivery staff.  The primary care delivery entity is the physician or the specialist.  Throughout the evolution of medical technology, the development of an efficient, useful, and practical clinical information system has become a significant focus in the vendor market.  However, the resistance to the implementation of helpful technology that is common within the health care market has limited the maximization of the potential of clinical information systems.  Through the satisfaction of aggregate physician needs in conjunction with the needs of health care managers, the implementation of a clinical information system can be advantageous to the health care delivery process.

 

Why create a CIS?

The creation of a clinical information system requires immense financial and intellectual investment.  Therefore, the justifications for such an investment must be quantitatively and qualitatively measurable upon implementation.  The development of the clinical information system arose from needs within the health care field, from health care managers, quality control, and care providers, such as physicians.  The needs that motivated the development of clinical information systems include; clinical task support, clinical management control, competition support, and clinical decision support [2].

            The primary source of support needed is in the clinical task support domain. Through the development of the electronic medical record, the maintenance of the patient record will be simplified, providing information support to the clinician.  The electronic medical record will allow for a longitudinal source of information regarding patient history, previous encounter history, drug allergies, and other relevant information.  The development and use of such a system will allow for a significant decrease in medical errors.  The National Academy of Sciences’ Institute of Medicine estimates that the deaths caused by medical mistakes are greater than deaths caused by AIDS, breast cancer or car accidents.  The number of deaths caused by medical error has been estimated to be 98,000 individuals per year [3].  Through effective data entry and use of the electronic medical record, the amount of deaths caused by lack of data or erroneous data entry will be significantly reduced.

            Furthermore, the development of an effective clinical information system should provide the clinician process support, or the ability to share information.  Through the creation of a standard electronic medical record, numerous users involved in the care delivery process can utilize the information stored on the record[1].  This accessibility will allow for all users, nurses, administrative staff, laboratory staff, pharmacists, and physicians to update and extract information for their specific usage needs.

            A significant contributor to the creation of a universal electronic medical record is the development of SNOMED, or the Systemized Nomenclature of Medicine, and Arden Syntax[2]. SNOMED allows for a consistent compilation of clinical information that allows various specialists, researchers, and even patients to share their knowledge based on a common linguistic description of health.  This knowledge sharing can be conducted across care sites and differing computer systems.  Furthermore, privacy is maintained through the use of authorization privileges.  This expandable nomenclature system can be customized for individual clinicians or nation-wide health care delivery organizations [4]. Arden Syntax is a medical language system that is specifically designed for medical logic systems, or medical logic modules.  Each medical logic module possesses ample information to create a single medical decision based on information entered.  The use of medical logic modules allows for the constant maintenance of patient status and clinician alert of any changes in the status of the patient [5].

            In addition, the development of the clinical information system arose due to a need to maintain clinical management control.  This management of clinical practice is applicable to individual patient management, intra-practice management, and inter-practice management.  On an individual patient basis, the measurement of quality of care, the monitoring of care provided, and feedback regarding the care that has been provided allows for improved quality care provision.  The clinical information system would allow for the maintenance of a standard practice pattern, which includes the provision of care based on specified care paths and/or flow sheets.  Furthermore, the treatments and protocols chosen for care provision can be compared to an industry standard, or a “best practice” methodology.

            This clinical management support also applies to inter-practice management situations.  By standardizing “best practice” care methodologies; the clinician has a database against which their care path decision may be measured.  Also, the integration of the numerous care contributors into a single information system will allow for cost management and quality control across care sites.  Such inter-site care monitoring is especially significant for large care delivery organizations such as Kasier-Permanante, where cost control, patient health, practice pattern variation, and quality control is integral to the business operations.

            In terms of competition support, the development of an effective clinical information system will place health care organizations in an advantageous position relative to other competitors.  The benefits of a clinical information system are evident in terms of the reduction of medical error, the standardization of medical protocol, knowledge sharing, cost control, quality control, and decision support.  Therefore, firms that are able to develop facilities in these areas will have a competitive edge over other care delivery organizations.  Furthermore, a clinical information system will allow for the comparison of numerical data from numerous fields with the data obtained from other firms in the industry.  Such comparisons will allow delivery organizations to increase the quality of care accordingly, learn from other firms, adjust price according to competitor levels, and adjust to community demand levels.

            However, the domain within which the clinical information system may provide the greatest support is within the clinical decision support domain.  Through the development of the electronic medical record and the increased SNOMED usage, clinicians will be able to share knowledge more rapidly on an interactive level.  Furthermore, the clinical environment requires an extensive knowledge of rare complications and revolutionary research that could dramatically alter the diagnosis of diseases in patients.  However, the breadth of information and the specialization of practice prevent physicians from understanding or even learning that such findings may exist.  Through the collaboration technologies that would be available within a clinical information system, clinicians may be able to rapidly contact “experts” within a specific field who would be better able to assess the patient situation through a brief analysis of the patient’s electronic medical record and other information as provided by the physician.

            Currently, the adaptation of new technology within the health care sector is fairly slow, with change hinging more on cultural features of the organization or entity than on the technological benefits that are produced by the implemented system.  Therefore, the clinical information system technology has become increasingly demand driven, in that the usage of the systems will only increase as long as there is an acceptance of the new technology within the environment by the physicians.  In order to ensure successful integration, the organization must ensure that the key players, mainly the physicians, are informed of the benefits to the individual physician, the health care organization, and health care as a whole [6].

 

MD benefits of the Clinical Information System

            The primary benefit of the clinical information system will be within the domains of task support and decision support[3].  The task support field has been revolutionalized through the implementation of the electronic medical record in conjunction with SNOMED.  The universalization of the electronic medical record will increase the accessibility of patient information to clinicians as well as increase the amount of data available for clinical use, reducing medical error significantly.

            However, the greatest tool to increase the standardization of care, reduction of practice pattern variation, successful and effective diagnosis, and correct care path choice will result from the development of the clinical decision support domain of the clinical information system.  Clinical decision support software offers the possibility to improve the quality and reduce the cost of care by influencing medical decisions at the time and place that these decisions are made.

            An ideal clinical information system would alert physicians when outlier results are returned from data entry of laboratory testing.  The data attained for a specific patient can then be compared to the general population to indicate whether the data is within the normal fit or is an outlier that may require further analysis.  Such a practice would induce the physician to notice certain data that may otherwise go unnoticed, and therefore, alter the diagnosis of the patient.  Also, the physician may interact with the system on a hypothesis-testing basis.  A physician may enter a possible diagnosis into the system and then receive feedback from the system regarding the plausibility of such a diagnosis being true.  This allows physicians to receive guided feedback during their consideration of similar diagnoses, which may be significantly different based on their appropriate care path.

            The development of an effective clinical decision support system will have a significant impact on practice methodology.  Clinical decision support systems are intended to receive patient data and utilize that data to propose a series of possible diagnoses and a course of action.  The advent of such a system will provide physicians with a guideline through which the physician can model their decision.  Furthermore, the clinical decision support system can lead to a reduction of the practice pattern variation that plagues the health care delivery process.  By reducing practice pattern variation, the overall cost of healthcare may be reduced.  Also, an effective clinical decision support system can recognize drug-drug interaction and patient complications that would otherwise be unrecognized by the physician to provide a valid, efficient, and “best practice” solution to the patient diagnosis process.

            The dynamic environment surrounding patient diagnosing complicates the process of diagnosis.  Numerous significant input variables, special patient circumstances, and the basic complexity involved in the diagnosis process limits the accuracy of a given clinical decision support system.  However, the potential advantages that are introduced through a successful system are significant.  Primarily, possible diagnoses could save numerous patient lives, since an information system may be better able to synthesize vast amounts of information.  Furthermore, the diagnosis suggestion may allow for better standardization of care delivery and the care path followed.

Ever since computers were first coming into use, it was believed that computers could model the clinical problem solving techniques used by physicians.  As early as 1970, William Schwartz of Tufts University School of Medicine wrote: “Computing science will probably exert its major effects by augmenting and, in some cases, largely replacing the intellectual functions of the physician.” [7]

However, the dynamic, complex, and unique health care environment has hindered the development of a standard clinical decision support system that would enable the universalization of any clinical decision support system.  Several factors including lack of investment, lack of leadership from practicing physicians, medical schools are responsible for the dreams of an ideal clinical decision support from being realized.

 

DEVELOPMENT OF DECISION SUPPORT SYSTEM DOMAIN

            Medical diagnosis is a complex human process that is difficult to represent in an algorithmic model.  Not only does medical diagnosing require the understanding of symptoms, drug-drug interactions, and patient history, the diagnosing process requires knowledge of diseases in general as well as the general population.  Furthermore, the system would have to be updateable to constant changes that accompany the scientific development that is a result of the extensive research within the medical field.  Also, the system would have to be able to utilize varying levels of data in order to diagnose an individual.  While one patient may have data showing high cholesterol, chest pain, higher blood pressure within an arterial section, and previous heart attack history within the family, another patient may only show high cholesterol and chest pain.  While both patients may require a catheterization, the limited data of the second patient may limit the ability of the diagnosis, and therefore, could lead to the misdiagnosis of the patient.

            Furthermore, it is imperative that the diagnosing systems provide reasoning for the medical diagnosis provided.  Such a process would allow the physician to understand the reasons the system may have had for a specific decision that may have been made.

            Clinical information systems have developed from rudimentary data entry and retrieval on an intra-hospital basis, to real-time data retrieval, multi-user data entry, multi-access data retrieval, knowledge sharing, sophisticated consultation, patient and inter-practice management, competition support, and enhanced decision support functionality.  With the advent of the physician workstation, hand-held data entry systems[4], voice recognition systems, and real-time clinical data retrieval and electronic medical record update, the clinical information system is becoming a comprehensive system integrating many aspects of the care delivery process.  Innovative point-of-care support, such as vital sign monitoring, medication administration monitoring, basic chart maintenance, lab and drug orders administration, and alerting, are reducing labor needs while increasing accuracy and quality through the continuous update of the electron medical record.  The development of technology that has led to greater “alerting and protocol support, utilization control, case management, outcome management, and executive decision support” [8], have enhanced the care delivery process, particularly the decision support aspect of the clinical information system.

            Clinical decision support systems have evolved from a foundation based upon statistical algorithms to complex artificial neural networks.  The early decision support systems, also termed medical diagnostic decision systems, were based on Bayesian statistical theory[5], providing crude probability diagnoses based on certain critical variables [9].  In 1994 Berner et al published the results of a study in which four commercially available medical diagnostic systems were challenged to diagnose a series of 105 patients each of whom had been referred to a consultant and in which of whom a diagnosis had been established.  The programs studied included Dxplain, Iliad, Meditel and QMR.  The proportion of correct diagnosis ranged from 52% to 71% and the relevant diagnoses ranged from 19% to 37%.  Thus it appeared that computer aided diagnostic systems had failed to deliver on their promise [10].

The complexity of the health care environment requires significant adaptations in order to maintain legitimate diagnoses.  Clinicians do not combine clinical data using probability used by the computer but use case specific knowledge and heuristics based on their experience.  Hence earlier systems that attempted to replace the clinician were largely unsuccessful.  Ralph Engle, one of the pioneers of computer assisted diagnosis wrote: “ Our experience confirms the great difficulty and even impossibility, of incorporating the complexity of the human thought into a system that can be handled by a computer.  We concluded that we should stop trying to make a computer act like a diagnostician and concentrate instead on ways of making computer-generated relevant information available to physicians as they make decisions.” [11]

 

Current Practices in Decision Support Systems

Previous decision support systems have utilized the Arden syntax in conjunction with HL7 standards and medical logic modules to create decision systems ranging from single decision models to complex, sequenced decision models.  Each medical logic module contains, within itself, the ability to make one single medical decision based on data entry.  However, through the sequencing of various medical logic modules, fairly complex models may be created.  Numerous clinical information system vendors have developed such decision support systems, including; HBOC, IBM, Siemens Medical Systems, and Health VISION.  These systems have been implemented within numerous care delivery sites, including; Columbia-Presbyterian Medical Center, JFK Medical Center, Ohio State University, and Meridian Health Systems.

 

Recent Shifts in CDSS:

In recent years, clinical decision support systems have begun utilizing artificial neural networks.  However, the shift to the artificial neural network continues to utilize the Bayesian statistical theory, but is able to integrate greater decision variable than previously possible.  Previously, medical diagnosis systems were effective in evaluating outcomes for a group of patients, yet failed to perform as successfully on a patient-by-patient basis.  Through the application of the artificial neural network system, this success rate on an individual patient basis may be increased significantly.  The neural network is able to accept numerous input variables, including demographic information, admission information, previous diagnosis information, and patient history to rapidly create possible diagnoses [12].  Rather than provide a single diagnosis for a specific patient, the system returns a set of possible diagnoses from which the clinician may choose based on their own discretion.  Advanced systems are able to return probability estimates of the likeliness of a particular diagnosis.  Furthermore, these advanced systems are able to propose “best practice” methodologies for care delivery or flow sheets that dictate action steps in the care delivery process.

However, in order to effectively determine “best practice” methodologies, clinical practice guidelines must be set.  Clinical practice guidelines that define what steps are necessary in order to ensure quality care provision can be separated into decisions, actions, and processes.  The decision model includes selection of which variables to consider and at what weights, selection of diagnosis, and consideration of alternative diagnoses.  By utilizing such a flexible system, the patient and the physician become the “chooser” in the environment, being able to control what information is relevant and which result to act upon.  Furthermore, the action model specifies which actions need to be performed.  These actions include the specification of type of action and temporal limitation (i.e. take dose for 3 months) through the standardized medical terminology available.  Finally, the process model organizes actions sequentially and hierarchically in order to determine which actions are crucial to the care process and in what order the care should be delivered [13].  The creation of clinical practice guidelines is necessary in order to have a template with which the system may prescribe diagnoses, actions, and processes.

 

Regulation

With the increasing number of vendors producing such systems there is increasing variability in their quality.  This is a cause for concern as a simple mistake in a clinical decision support can lead to the loss of a life. Hence the need for enhanced oversight and regulation is needed.  The FDA has regulated that CDSS’s are similar to medical devices. However the legal responsibility for the treatment and advice given to the patient will rest with the clinician regardless of whether he was assisted by the CDSS [14].

 

XML

XML is a type of computer language that stands for eXtensible Markup Language.  It is considered to be the future of computer applications in health care.  Its acceptance and widespread use as a method for defining clinical content in text documents depends on the establishment of standard vocabularies so that healthcare organizations can exchange electronic information with one another.  The design of XML is to highlight the content of information rather than its appearance, which is the case with HTML (Hyper Text Markup Language).  This technology allows uses to create queries and defined structures of information based on relevant content.  This means that health care providers and look at a chart, discharge summary, or other hospital records and isolate only the information they need.  Users can search for information from a number of sources and bring it together in one place.  They can sort the relevant information and group it by a wide array of classification systems as befits their needs.  This data can then be extracted to fit any type of order, form, or document that the user requires.

XML can also be used to create documents that pursue set pathways according to criteria such as content, intent, and origin.  One example of this potential is having a document system that requires an attending physician’s signature.  Any chart signed by a resident without the attending’s signature would be routed back to the attending physician for review.   Thus, this technology creates efficiency and speed by creating a system that minimizes errors and checks for complete and full workup in clinical documentation. 

On the technical side, XML files can typically be used in conjunction with HTML and with other computer systems.  This flexibility will allow users to uniformly obtain all the information they need from a wide range of sources in order to filter out what information they precisely need.  Hence, XML appears to be the binding force that culminates clinical data from multiple systems and presents it in a form that is easy to filter and utilize [15].

 

Increasing Acceptance

The trend for the future shows increased dependence on clinical decision support systems by physicians.  Constant contact with such systems will ensure that the most optimal level of care is provided utilizing both physician judgment and technological innovativeness.  Such a future will mean that physicians and other health care providers will have to change the way they collect, sort, and use health care data.  There are many benefits to such systems as well as barriers to use.  It is important to look at such barriers in order to understand how to increase support and acceptance to ensure successful implementation.  It is only a matter of time before these systems are common place in all health settings.

 

The Role of Physicians and Barriers to Use

It is extremely important for physicians, caregivers, staff, administrators, and technical experts to work in collaboration in the design, implementation, and improvement of decision support systems.  The physician role has expanded significantly in today’s managed care environment.  Physicians and other clinicians need to be involved in throughout the whole process of finding, designing, implementing, and improving such a system.  Their input and support is critical because of the significant role that physicians play in the health care delivery process [16].

The acceptance of these systems has not been as easy as most other healthcare technologies.  One reason for this is the fact that these systems including decision support systems affect the long history of traditional medical practices.  New systems are changing the ways physicians think and behave [17].  Failure to accept these systems among physicians occurs when implementation does not provide direct benefits to their users and the process of implementation itself changes the traditional practices of the clinical environment [18].

 

Organizational Environment and Barriers to Use

The culture in a health care setting can be very complex.  This is because there are a many types of people involved in the system.  Not only are there patients and physicians, but there is also the administration, technical experts, and other staff that add to the web of issues, tensions, and interests that exist.  There are many values and mode of practice that exist in such a web that often complicate the culture of the health care environment as a whole.  It is natural to conclude that with such a large group of diverse people, there are often a difference of values and interests.   Therefore, adopting a major acquisition such as a clinical decision support system requires that there is overall acceptance and support from all relevant parties to ensure successful implementation.  The benefits and how people evaluate them with regards to how they fit within their practices are the key factors indicative of success [19]. 

Despite evidence that clinical information systems can improve patient care, there have been a number of instances where implementation has not been successful [20].  Barriers here are due to organizational stigmas that the systems are not noticeable benefits to the practice of care.  Others argue the cost benefits of certain systems within their operating margins.  Like physicians, those medical staff and administrators who are not properly acquainted with new systems argue that they negatively affect care in areas of patient waiting times and work load.  However, negative results in these areas occur when those using the system do not have adequate training and support for such a system whereby the use of it turns out to be negative rather than positive. [21, 22, 23] 

Since the acceptance of decision support systems depends heavily on physicians, it’s the values that they hold which determine if they view new systems as beneficial. These values concern the patient-physician relationship, the quality of patient care, the balance between clinical guidelines and decision support technology, and physician autonomy. 

 

Quality of Treatment

Physicians’ attitudes about what a piece of technology can do has more of an effect on acceptance of that technology than what it can do in a realistic setting.  This belief is confined towards the physician’s own practices and how they operate.  If there is a direct benefit here, then a system has improved the quality of care.

Clinical Procedures and Decision Support Technology

The balance between the science of medicine and the art of it is a pressing issue that will influence acceptance.  There is natural tension between the capabilities of technology and the method of medical practices that have existed for years.  Physicians struggle over whether to let technology based on population characteristics guide them in their decision making as opposed to making decisions based on their own experience and the novelty of each patient’s case.  The advent of providing accurate information through clinical information has been essential in the improvement of patient care.  However, just because information is provided does not mean that it is used.  Physicians may ignore information offered up by some clinical decision support systems.  Those who do use it may not use it to its fully capability or may only use it for a short time before discontinuing its use.  Physicians are worried about the legal ramifications of not following the advice of decision support systems.  However, the issue that decision support systems are only a tool for advice and not a computer making the final decision has increased physicians’ comfort level with decision support systems [21].

 

Relationship between Patient and Physician

The relationship between a physician and his/her patient is crucial to quality care.  The acceptance of decision support systems depends on their ability to cater to this need.  Systems have to make the physician feel more equipped to provide better care while instilling trust and confidence in the patient that the technology is not a replacement of the physician, but a tool that enhances the treatment process. 

 

Physician Autonomy

Medicine is characterized by a simple, but long lasting philosophy.  The ability to treat patients requires not only individual expertise, but experience in being able to effectively address the needs of patients.  Therefore, physicians have favored their own clinical judgment in determining the needs of particular patients over broad policies, administrative guidelines, and the reliance on technology.  Therefore, promoting awareness about decision support systems capabilities has to be done in a way to break down social barriers and stigma that are associated with physicians’ egos [24].

 

CONCLUSIONS

Current trends in the United States will not only lead to increased spending in the health care arena but also an accelerated growth of the acceptance and spending of health care information technology.  The necessity for innovative and dependable clinical information systems with decision support capabilities is crucial.  Increasing systems’ acceptance depends on the culture of the hospital as well as the involvement of physicians.  The involvement of all health care professionals especially physicians in the selection and implementation of the system from the outset is essential.  Not only will this ensure support, but as a result the frequency of communication is likely to increase amongst one another.  In turn, this frequent consultation and communication is likely to better the quality of patient care as well as the patient-physician relationship [25].

Secondly, it is essential to consider in advance the new system’s effects on the culture, practices, and attitudes of the people in the organization.  Explicitly identifying how people and groups in the organization will benefit specifically will lend support for its implementation.  The use of information systems by physicians will occur if that system will allow them to provide better care for their patients.  Benefits to the organization, in general, will not be as successful to motivate or even inspire physicians to alter the way they have practiced medicine.

Finally, healthcare organizations must anticipate and be prepared to handle a diverse array of changes that will occur during the implementation process itself. Therefore the organization needs to be able to operate normally on schedule while operating the implementation process as well.  By phasing in the system's implementation and anticipating problems proactively, the hospital should be able to reduce the number of negative experiences associated with the introduction of a new system [26].

 

References:

 

1.      http://www.hcfa.gov/stats/NHE-Proj/proj2000/default.htm. Financing forecasts from the Health Care Financing Administration.

2.      Dowling, Alan. Class Notes MIDS/HSMC 432. Notes 5-6.

3.      Woody, Todd. “Getting the Record Straight”. The Industry Standard. 10 April 2000. http://www.thestandard.com/article/0,1902,13405,00.html

4.      http://www.snomed.org

5.      http://www.cpmc.columbia.edu/arden/

6.      Burge, Lucy; Creps, Linda; Wright, Brenda.  If you build it, will they come? Health Management Technology. Volume 22. Issue 9. Sep 2001. 14-20.

7.      Schwartz, WB. Medicine and the computer: the promise and problems of change. NEJM1970;283:1257-1264.

8.      Dowling, Alan. Class Notes MIDS/HSMC 432. Notes 5-6.

9.      Miller, RA. Medical Diagnosis Decision Support Systems-Past, Present, and Future. JAMIA. 1994:1;8-27.

10.  Berner ES, Webster GD, shugerman AA, Jackson JR, Algina J, Baker AL, Ball EV, Cobbs CG, Dennis VW, Frenkel EP, Hudson LD, Mancall EL, Rackley CE, Taunton OD. Performance of four computer based diagnostic systems. NEJM1994;330:1792-6.

11.  Engle RL. Attempts to use computers as diagnostic aids in medical decision making.

12.  Frize, M; Ennet, CM; Stevenson, M; Trigg, HCE. Clinical decision support systems for intensive care units: using artificial neural networks. Medical Engineering & Physics. 23;2001: 217-225.

13.  http://smi-web.stanford.edu/people/tu/Talks/2001HL7International.pdf.

14.  Hunt, Dereck; Haynes, R. Brian; Hanna Steven E.; Smith Kristina. Effects of computer based Clinical Decision Support Systems on physician performance and patient outcomes. JAMA 1998;280:1339-1346.

15.  http://www.techapps.co.uk/pdfs/xmlseminar/xs-h2-applicationsxmlhealthcare.pdf

16.  Coddington, Dean, et al., Integrated Health Care: Reorganizing the Physician, Hospital and Health Plan Relationship, Center for Research in Ambulatory Health Care Administration, Englewood, Colorado, 1994.

17.  Anderson JG. Computer-based patient records and changing physician practice patterns. Top in Health Information Manage 1994; 15: 10-23.

18.  Schoenbaum SC, Barnett GO. Automated ambulatory medical records systems: an orphan technology. J Technol Assess Health Care 1992; 8 (4): 598-609.

19.  Joel D. Howell. Technology in the Hospital: Transforming Patient Care in the Early Twentieth Century (Baltimore and London: The Johns Hopkins University Press, 1995).

20.  Kleinke JD. Release 0.0: clinical information technology in the real world. Health Affairs 1998 17 (6): 23-38.

21.  Lorenzi NM, Gardner RM, Pryor TA, Stead WW. Medical informatics: the key to an organization's place in the new health care environment. J Am Med Informatics Assoc 1995; 2 (6): 391-392.

22.  Barnett GO, et al. A computer-based monitoring system for flow-up of elevated blood pressure. Med Care 1983; 21: 400-409.

23.  Campbell JR, Givner N, Sellig CB, et al. Computerized medical records and clinic function. MD Comput 1989; 6 (5): 282-287.

24.  Charles Bosk. "The Impact of the Place of Decision-Making on Medical Decisions." In Proceedings of the Fourth Annual Symposium on Computer Applications in Medical Care, ed. Joseph T. O'Neill (Silver Spring, MD: IEEE Computer Society, 1980) pp. 1326-1329.

25.  Massaro TA. Introducing physician order entry at a major medical center: I. Impact of organizational culture and behavior. Acad Med 1993; 68: 20-25.

26.  Massaro TA. Introducing physician order entry at a major medical center: II. Impact on medical education. Acad Med 1993; 68: 25-30.

27.  http://www.cerner.com/swf/

28.  http://www.per-se.com/home.htm

29.  http://dxplain.mgh.harvard.edu/dxp/dxp.sdemo.pl?login=demslcshome

30.  http://www.firstdatabank.com/medical/index.html

31.  http://www.clineanswers.com/home/index.html

 

 


 

[1] The EMR may be standardized through the use of XML.

[2] Used in conjunction with HL7.

[3] Both management and competition support apply to higher-level management support.

[4] Such as a PDA.

[5] For Bayesian statistical info, see http://astrosun.tn.cornell.edu/staff/loredo/bayes/