Building a comprehensive clinical information system from components. The approach at Intermountain Health Care

OBJECTIVES

To discuss the advantages and disadvantages of an interfaced approach to clinical information systems architecture.

METHODS

After many years of internally building almost all components of a hospital clinical information system (HELP) at Intermountain Health Care, we changed our architectural approach as we chose to encompass ambulatory as well as acute care. We now seek to interface applications from a variety of sources (including some that we build ourselves) to a clinical data repository that contains a longitudinal electronic patient record.

RESULTS

We have a total of 820 instances of interfaces to 51 different applications. We process nearly 2 million transactions per day via our interface engine and feel that the reliability of the approach is acceptable. Interface costs constitute about four percent of our total information systems budget. The clinical database currently contains records for 1.45 m patients and the response time for a query is 0.19 sec.

DISCUSSION

Based upon our experience with both integrated (monolithic) and interfaced approaches, we conclude that for those with the expertise and resources to do so, the interfaced approach offers an attractive alternative to systems provided by a single vendor. We expect the advantages of this approach to increase as the costs of interfaces are reduced in the future as standards for vocabulary and messaging become increasingly mature and functional.

The HELP hospital information system: update 1998

The HELP hospital information system has been operational at LDS Hospital since 1967. The system initially supported a heart catheterization laboratory and a post open heart Intensive Care Unit. Since the initial installation the system has been expanded to become an integrated hospital information system providing services with sophisticated clinical decision-support capabilities to a wide variety of clinical areas such as laboratory, nurse charting, radiology, pharmacy, etc. The HELP system is currently operational in multiple hospitals of LDS Hospital's parent health care enterprise--Intermountain Health Care (IHC). The HELP system has also been integrated into the daily operations of several other hospitals in addition to those at IHC. Evaluations of the system have shown: (1) it to be widely accepted by clinical staff; (2) computerized clinical decision-support is feasible; (3) the system provides improvements in patient care; and (4) the system has aided in providing more cost-effective patient care. Plans for making the transition from the 'function rich' HELP system to more modern hardware and software platforms are also discussed.

Inducing practice guidelines from a hospital database

Improving health care quality requires the elimination of unnecessary variation in the care process. Decision support applications already exist that can foster adherence to standards. The challenge resides in developing standards consistent with good medical practice. In this paper we present our efforts in determining where sufficient clinical data are captured electronically to automatically define a care process, and what analyses can be done to identify additional data that would allow a care process to be defined. Data routinely collected by a hospital information system have been examined. The analysis tools utilized include logistic regression, a neural network, a Bayesian network, and a rule induction program.

Development and evaluation of a computerized admission diagnoses encoding system

Hospital information systems designed to support the needs of health care professionals include patient data entered using both freetext and precoded storage schemes. A major disadvantage of freetext storage schemes is that data captured in this format can only be presented as is to the user for review tasks. In the view of many health care scientists, natural language understanding systems capable of identifying, extracting, and encoding information contained in freetext data may provide the necessary tools to overcome this weakness. This paper describes the development and evaluation of a such a system designed to encode freetext admission diagnoses. This system combines both semantic and syntactic linguistic analysis techniques. Evaluation results demonstrate the overall performance of this system to be reasonable, accurately encoding approximately 76% of admission diagnoses. Inefficiencies are primarily due to the inability of this system to generate encodings in roughly 15% of test cases. When encodings are produced, however, accuracy equals that of the current manual coding method. With further modification, this application can partially automate the coding process.

Assessing the behavioral impact of a diagnostic decision support system

This paper describes a prototype for research to evaluate the impact of diagnostic decision support systems on the behavior of physicians. Several indices that can be used to quantify the magnitude of impact are proposed. A large medical diagnostic knowledge base in internal medicine (the Iliad knowledge base) was used in this evaluation. The impact on behavior when different inference models are run against this knowledge base is evaluated for two different case domains and physician's specialties.

Decision support in medicine: examples from the HELP system

Computerized health information systems can contribute to the care received by patients in a number of ways. Not the least of these is through interactions with health care providers to modify diagnostic and therapeutic decisions. Since its beginning, developers have used the HELP hospital information system to explore computerized interventions into the medical decision making process. By their nature these interventions imply a computer-directed interaction with the physicians, nurses, and therapists involved in delivering care. In this paper we describe four different approaches to this intervention. These include: (1) processes that respond to the appearance of certain types of clinical data by issuing an alert informing caregivers of these data's presence and import, (2) programs that critique new orders and propose changes in those orders when appropriate, (3) programs that suggest new orders and procedures in response to patient data suggesting their need, and (4) applications that function by summarizing patient care data and that attempt to retrospectively assess the average or typical quality of medical decisions and therapeutic interventions made by health care providers. These approaches are illustrated with experience from the HELP system.

The use of medical logic modules at LDS hospital

The development of medical knowledge bases for use in a clinical information system (HELP) has been an ongoing goal at LDS Hospital in Salt Lake City, Utah, for the past 25 years. In building our medical knowledge base we felt the need to implement a decision support syntax which could capture the logic of our experts in a way that was not only executable, but also easily read and shared by others. During these 25 years we defined several simple syntaxes to express this medical logic. Our current approach is to cooperate with international standards groups (ASTM) and use the Arden Syntax for medical logic modules. We are working with the 3M Corporation in the joint development of an Arden Compiler for HELP. We plan to use the Arden Syntax initially to support our alert/reminder system and computerized management protocols.

Rationale for the Arden Syntax

The Arden Syntax, a language designed for writing and sharing task-specific knowledge for Medical Logic Modules (MLMs), has been recently accepted as a standard by the ASTM. The syntax is concerned with the critical task of sharing medical knowledge bases across many institutions. Because of the relative lack of agreement on vocabularies and data standards and because of the many other obstacles, the developers of the Arden Syntax took a pragmatic, straightforward approach that has borne fruit in a very short period of time. The syntax provides a vehicle for the health care community to begin sharing, so that we can see what works and what does not work, and we can begin to address the critical obstacles. In designing a language like the Arden Syntax, the authors make many decisions--but the final document gives only the result of these decisions without any explanation. By writing down the rationale behind the design of the syntax, we hope to aid users of the language, implementors of the language, and future designers of new languages.

Interfacing a stand-alone diagnostic expert system with a hospital information system

Few diagnostic decision-support systems are in routine clinical use, mainly because these systems typically require time-consuming manual data entry. This research investigated the feasibility of reducing manual data entry by integrating a stand-alone diagnostic expert system with an existing comprehensive hospital information system (HIS). A knowledge-based intervocabulary mapping technique was developed to map disparate vocabularies. The results of a retrospective study indicate that transferring clinical data from the HIS to the diagnostic expert system at the beginning of workup significantly reduces the manual data entry required for generating the correct diagnoses for patients.

HELP the next generation: a new client-server architecture

A new client-server based system which is centered around a lifetime data repository (LDR) is under construction. The goal of the new system is to maintain the patient centered decision support aspects of the existing HELP* system while providing an open architecture that supports faster application development and allows execution of applications to be distributed across many computers. These goals are achieved by implementing the system with software components that are commercially available or by adhering to national and international standards for software integration. Keys to successful integration include the use of MS-DOS @, OS/2#, and UNIX Section as operating systems, Microsoft OLE 2.0 as a standard interface to the clinical database, the use of TUXEDO as a transaction/communication manager, and the use of ORACLE [symbol: see text] RDBMS as the underlying database management system.

The Arden syntax for medical logic modules

The Arden Syntax for sharing medical knowledge bases is described. Its current focus is on knowledge that is represented as a set of independent modules that can provide therapeutic suggestions, alerts, diagnosis scores, etc. The syntax is based largely upon HELP and the Regenstrief Medical Record System. Each module, called a Medical Logic Module or MLM, is made of slots grouped into maintenance, library, and knowledge categories. The syntax has provisions for querying a clinical database and representing time. Several clinical information systems were analyzed and appear to be compatible with the syntax. The syntax has been tested for syntactic ambiguities using the tools lex and yacc. Seventeen institutions are currently in the process of adopting the Arden Syntax for their decision-support systems. A subcommittee of ASTM has been formed to develop standards for sharing medical knowledge bases. The Arden Syntax has been published by ASTM as a initial standard for sharing medical knowledge.

Sharing MLM's: an experiment between Columbia-Presbyterian and LDS Hospital

The use of Arden syntax for development of shareable medical logic modules (MLM's) has developed as an ASTM standard. To test the feasibility of sharing MLM's between institutions a study was conducted between Columbia-Presbyterian Medical Center and LDS Hospital. In this study seven MLM's clinically executing at Columbia-Presbyterian were used to test the sharing potential of the Arden syntax. The study was limited to measuring the modifications necessary to make executable at LDS Hospital the shared MLM's. Because of the site specific nature of the data variables, multiple modifications were required. Three classes of modifications were necessary. The simplest involved only data variable mappings. The other classes required either minor modifications to the logic or relatively major modifications. Over 50% of the modifications were in the minor or major classes. While the sharing of decision logic was possible and facilitated by the use of the MLM's at the two sites, the absence of standard medical vocabularies limited the utility of the MLM as a mechanism for directly sharing medical knowledge.

ASTM E31.15 on health knowledge representation: the Arden Syntax

ASTM subcommittee E31.15 on Health Knowledge Representation was formed to promote standards for defining and sharing health knowledge bases. Its first standard, the Ardan Syntax, is focused on knowledge bases that can be represented as a set of independent modules called Medical Logic Modules (MLMs). The standard is in clinical use and has generated significant interest in industry and academics. The Extensions task group plans to extend the syntax where appropriate, to expand to other types of knowledge bases. The Validation/Verification task group is approaching the enormous problem of evaluating knowledge bases and the process of sharing them.

Integrating Radiology and Hospital Information Systems: the advantage of shared data

Information management is central to modern patient care. Computerization of information management has resulted in both departmental systems which serve information needs in locations such as the Radiology Department and in hospital-wide information systems which seek to integrate management of clinical data from many departments. For each of these systems to achieve the goal of maximizing both the effectiveness of health care workers and the quality of patient care, they need to share the data that they capture. Below we discuss a variety of applications, both currently available and in the realm of research protocols, that depend on a high level of communication between Radiology Information Systems and Hospital Information Systems. These examples suggest the benefits of integrating the medically relevant data collected by all of the computer-based information systems in the hospital setting.

The potential for medical informatics in the Eastern Mediterranean Region

This article assesses the potential value of an integrated medical/ hospital information system (IMIS) for the members of the Eastern Mediterranean Region of the World Health Organization, and describes the preliminary results of a feasibility study questionnaire done in December 1989 at Salmaniya Medical Center (SMC) in the state of Bahrain.

Pick from thousands: a collaborative processing model for coded data entry

Methods for optimizing coded data entry in clinical systems are a frequent topic of system design. We have developed a new mechanism for this type of data entry that we call "Pick From Thousands" (PFT). It combines several known methods, including menu selection, keyword entry, and initial character matching, but adds a new string matching algorithm. The PFT method is more selective than initial character matching for a given number of keystrokes if entries in the coded list have more than one word. Collaborative processing between a PC workstation and the central HELP system computer is used to optimize ease of maintenance and increase the flexibility and performance of the system.

A computer method for visual presentation and programmed evaluation of labor

Manual graphing of the progress of labor is considered useful but is not often done. The early detection of some deviations requires special graphics aids. Our objective was to develop an easy-to-use computer program for the integrated visual presentation of information characterizing the progress of labor. Through the use of inexpensive personal computers equipped with graphics monitors, the program provides a combined graphics display of timed progressive cervical dilatation, fetal station, and stimulation of uterine activity (oxytocin infusion). For the early detection of abnormalities, phase-specific normal ranges (reference areas) are displayed. In addition, protraction/arrest as well as precipitate labor disorders are highlighted and computer messages are displayed. The program was evaluated through the assessment of 405 labors entered into a local area network of computers. On average, the program identified 1.5 abnormalities per recorded labor (2.0 for labors resulting in vaginal delivery). The graphic presentation of the labor curve, produced within 3 seconds, displayed 27% more information than the tabular format on the same screen area and provided a single-screen display of the labor curve even for patients with excessive data. The computer-generated display of labor curves facilitates visual presentation and interpretation of labor progress and can also help to translate quality assurance criteria into clinical practice.

Older age and elevated blood pressure are risk factors for intracerebral hemorrhage after thrombolysis

Intracerebral hemorrhage is an important concern after thrombolytic therapy for acute myocardial infarction, but risk factors are controversial. Accordingly, we assessed risk factors in 107 treated patients of whom 4 had intracerebral hemorrhage. Intracerebral hemorrhage occurred at a mean of 25 hours (range 3.5 to 48) after therapy and was fatal in 2 patients. Significant differences were found between patients with and without intracerebral hemorrhage for age (77 +/- 7 vs 62 +/- 11 years, p less than or equal to 0.01), and initial (161 +/- 23 vs 135 +/- 23 mm Hg, p less than or equal to 0.03) and maximal (171 +/- 30 vs 146 +/- 20, p less than or equal to 0.02) systolic blood pressures. Initial and maximal diastolic blood pressures also tended to be higher (101 +/- 25 vs 86 +/- 16, p less than or equal to 0.07; 104 +/- 24 vs 90 +/- 13, p less than or equal to 0.06). Differences did not achieve significance for comparisons of gender, height, weight, site of infarction, time to therapy, specific thrombolytic agent used, concomitant therapy, interventions and partial thromboplastin time. It is concluded that age (greater than or equal to 70 years) and elevated blood pressure (greater than or equal to 150/95 mm Hg) are important risk factors for intracerebral hemorrhage. The overall balance of benefit and risk of thrombolysis should continue to be assessed by large mortality trials.

Evaluation of an SQL model of the HELP patient database

We tested a new model of the HELP patient database that makes use of relational tables to store patient data and provides access to data using SQL (Structured Query Language). The SQL database required more storage space and had many more physical records than the HELP database, but it was faster and more efficient in storing data than the standard HELP utilities. The HELP utilities used disk space more efficiently and were faster than the SQL tools when retrieving data for typical clinical reports. However, the SQL model provides networking capabilities, general report writing tools, detailed user documentation, and an ability for creating secondary indexes that offset its poorer performance.

Development of decision support systems

Use of hospital information systems (HIS) are no longer limited to administrative functions. The addition to these systems of decision support capability is now a necessity. Development of the decision support modules requires a different software architecture than that employed by most HIS systems today. This paper describes the generic uses of decision support throughout the many hospital applications. Several levels of decision support are outlined with examples to illustrate the many areas where decision support is useful. At LDS Hospital in Salt Lake City, Utah we have developed an HIS using a new software architecture which supports the creation of decision support applications. This system uses a frame structure to represent knowledge. Examples of the frames and their syntax is presented. Using the frame tools which are provided, an application developer can easily develop and test decision support modules which interact directly with the clinical user and the patient database.

Development of a computerized laboratory alerting system

Using the capabilities of the HELP medical information system at LDS Hospital, a Computerized Laboratory Alerting System (CLAS) was developed. CLAS monitors and alerts for the presence of life-threatening conditions in hospitalized patients which are indicated by laboratory test results. Alerts are posted on computer terminals on the hospital's nursing divisions, where they are reviewed and acknowledged by hospital staff so that appropriate treatment can be rapidly instituted. CLAS was evaluated to determine its effectiveness in relaying alerts to the clinical staff, and improvements were made to develop an effective user interface. Initial average alert response times on nursing divisions ranged from 5.1 to 58.2 hr. The average alert response time dropped to 3.6 hr when alert review was integrated with laboratory result review, and to 0.1 hr after installation of a flashing light to notify hospital staff of the presence of new alerts.

Computer-based data entry for nurses in the ICU

Efforts are being made to improve the efficiency and quality of care in intensive care units (ICUs) at Latter Day Saints Hospital. The ICUs have been computerized, and the collection, storage, and presentation of patients' data have been improved. Nurses use computers for entering clinical data and plans for nursing care, and the effects of these changes on the work patterns of nurses in the ICU have been evaluated. Contrary to our expectation, our studies showed a decrease in the proportion of time that nurses spent in direct patient care (from 49.1% to 43.2%) and an increase in the proportion they spent entering clinical data (from 18.2% to 24.2%) after computerization. These changes, however, were attributed to a decrease in the severity of patients' illnesses, rather than the availability of the computer. There was no measurable difference in the proportion of time spent at other nursing activities.

Computerized nurse charting

Computerized nurse charting programs have been used at LDS Hospital for over two years. These programs allow the nurse to create nurse care plans for the management of the patient, and chart on the computer actions and information which support the documentation of the management of the patient according to the care plan created for the patient. Computer terminals have been placed at the patient's bedside to facilitate the use of these programs. This paper describes the programs available at LDS Hospital and several evaluation studies which have been performed to measure the efficacy of the programs. The evaluation studies indicated an increase in the level of documentation completeness and accuracy by the nurse but at some minor expense to time available to the nurse for patient care. Evaluation of the need for bedside terminals versus centrally located terminals showed an overwhelming desire by the nurse in favor of the bedside terminal. It was also found that data was entered more timely with less waiting when bedside terminals were available. Physician acceptance of the nurse charting system was found to be favorable.

Application of a computerized medical decision-making process to the problem of digoxin intoxication

A computerized medical decision-making system was used to monitor signs and predisposing factors of digoxin intoxication in patients receiving digoxin. This process automatically reviewed the patient's data base nightly for drug interactions, laboratory data and electrocardiographic findings with known association with digoxin intoxication. These decisions were formated into a "digoxin alert report" and sent to line printers in the nursing division to be placed on the individual patients' charts. To assess the effect of these reports on patient management, a randomized double-blind study was undertaken. Patients were assigned to an alert or nonalert group. Alert reports were withheld from charts of patients in the nonalert group. A medical record review was subsequently carried out, wherein the physician's orders were searched to identify actions taken with possible relation to the digoxin alerts. The computer monitored 396 patients over a 3 month period. Of these, 211 (53%) were randomized to the alert group and 185 (47%) to the nonalert group. Seventy-two percent of patients received at least one alert. The most frequently occurring alerts included: hypoxemia, hypokalemia, concurrent use of a beta-adrenergic blocking agent, renal insufficiency and ventricular arrhythmia. Results from the record review demonstrated a 22% increase in physician actions for the alert group. Specifically, patients in the alert group were 2.7 times more likely to have a serum digoxin determination ordered and 2.8 times more likely to have digoxin withheld on the day of a digoxin alert than were patients in the nonalert group.

Evaluation of size and dynamics of the inferior vena cava as an index of right-sided cardiac function

To define normal criteria of size and dynamics of the inferior vena cava (IVC) and its clinical value in assessing right-sided cardiac function, 2-dimensional (2-D) and M-mode echocardiography (echo) were performed in 175 subjects, who were classified into 3 groups: group 1-80 normal subjects; group IIA--65 patients with documented right-sided cardiac disease, and group IIB--30 patients with cardiac disease but no right-sided abnormality. The IVC was adequately imaged in 175 of 185 subjects (95%). There was good correlation between M-mode and 2-D echo (r = 0.84) and long- and short-axis (r = 0.88) measurements. The IVC diameter during expiration was: group 1-9 to 28 mm (mean 18.2 +/- 4.6); group IIA--15 to 40 mm (mean 23.1 +/- 4.8) and group IIB-8-24 mm (mean 15.6 +/- 3.7). Collapsibility index (inspiratory decrease in diameter) was: group I-37 to 100% (mean 55.8 +/- 15.9); group IIA--0 to 39% (mean 13.5 +/- 10.5); and group IIB--44 to 100% (mean 60.4 +/- 13.1). A and V waves could be measured in 120 of 151 cases (79%). Both A and V waves were less than 125% of its diameter in group I. The A wave was absent in 34 patients; 30 (88%) were in atrial fibrillation. Among 8 patients with tricuspid regurgitation, 5 (63%) had V waves greater than 125%. There was no correlation between diameter or collapsibility index and age, sex, rhythm or body surface area.(ABSTRACT TRUNCATED AT 250 WORDS)

Multifrequency and nonuniformly-spaced arrays: effects on grating lobe amplitude

A computer simulation study on the effects of multifrequency and aperiodic ultrasonic transducer arrays has shown that both methods improve grating lobe response when compared to a periodic array of the same aperture and element spacing. The multifrequency array produces a better beam pattern than either the aperiodic or combination aperiodic/multifrequency array. Both the aperiodic and multifrequency arrays have a narrower main lobe and higher side lobe response than a periodic array with reduced element spacing and an equal number of elements.

Analysis of B-scan speckle reduction by resolution limited filtering

Much of the signal energy due to speckle in ultrasound images is shown to be of higher spatial frequency than the intrinsic pulse shape limited resolution in B-scan ultrasound images. A significant increase in signal to noise ratio can therefore be obtained by resolution limited spatial filtering which selectively removes energy of higher spatial frequency than the pulse envelope resolution limit. The concept is illustrated by resolution limited filtering ECG gated B-scan echocardiographic images. Signal to noise improvement is illustrated by comparing time-motion displays generated from both processed and unprocessed images.

Noninvasive evaluation of exercise training in college-age men

The purpose of this study was to assess noninvasively the effects of intense aerobic training on cardiac structure and function in a group of healthy, college-age men (25 experimental and 11 control, mean age 22 years). Echocardiographic, electrocardiographic (ECG), and fitness measurements were obtained before and after a 3-month endurance training program and compared with similar measurements obtained in nonexercising subjects. The supervised training program consisted of 50-minute jogging sessions 5 days a week at 85% of maximal heart rate. Compared with the control group, echocardiography after training showed an increase in left ventricular (LV) end-diastolic dimension (p less than 0.05). LV posterobasal wall thickness, septal wall thickness and ejection fraction did not change significantly. ECG measurements revealed a decrease in resting heart rate (p less than 0.05) and an increase in R-wave voltage in leads V5 and V6 (p less than 0.01). The measured maximal oxygen consumption increased by 16% (p less than 0.001). These data indicate that intense aerobic training in college-age men results in a significant increase in resting LV end-diastolic dimension and volume. The increase in maximal stroke volume associated with exercise training may be partially explained by these changes in cardiac dimensions.

Evaluation of a computerized arrhythmia alarm system

A prospective study of a computerized arrhythmia alarm system was carried out in the coronary care unit during 200 patient hours of monitoring. The computer system was designed to activate an alarm on the development of rhythm and conduction disorders including asystole, ventricular tachycardia, atrial tachycardia, sinus tachtcardia, bradycardia, frequent premature ventricular beats, atrial fibrillation and bundle branch block. Study patients were simultaneously monitored by the computer system and a conventional analog heart rate alarm system. All alarms generated by the two systems were evaluated. Of 79 computer alarms, 42 (53 percent) were true positive alarms; during the same period there were 167 analog alarms of which only 13 (8 percent) were true positive alarms. In both systems, false positive alarms were primarily due to patient movement, but they occurred only 25 percent as often with the computer system as with the analog system. These results indicate that computerized arrhythmia monitoring systems offer significant advantages over conventional monitoring techniques.