Making public health data available to community-level decision makers--goals, issues, and a case report

Effective community health assessment and planning depends on the availability of appropriate public health data. Web-based technologies have created an unprecedented opportunity for making data available to community-level public health decision makers. An interactive data retrieval system targeted to the community-level user must provide an intuitive and easy-to-learn user interface with functionality and statistical complexity appropriate to the expected users while maintaining confidentiality of personal health information. The authors use the Missouri Information for Community Assessment Web site as an example to discuss goals and issues involved in the development of such systems.

Prediction accuracy and financial savings of four screening tests for a sequential test of clinical performance

BACKGROUND

Sequential testing of clinical performance is an effective strategy to reduce cost of testing.

PURPOSE

To evaluate prediction accuracy and financial savings of 4 screening tests of clinical performance.

METHODS

Screening tests were created from a 13-case examination taken by 434 medical students at 4 schools. Regression analysis determined prediction accuracy for 2 test outcomes. Financial savings were computed from published estimates.

RESULTS

Zero false passes were obtained with the "Total Number of Cases Passed" screening test, but it saved only 27%. Sixty-two percent savings with 5% false passes occurred with the "Classification" screening test. The "Scale" and "Mini Test" screening tests would have excused 79% and 67% examinees with 5% and 1% false passes, respectively.

CONCLUSIONS

Prediction accuracy varies with screening test and outcome measure. Sequential testing of clinical performance can save 40% to 60% with low false pass rates. However, programs need to consider loss of information for curriculum and individual feedback relative to financial savings.

Medical informatics in healthcare organizations: a survey of healthcare information managers

OBJECTIVE

To assess the medical informatics needs of healthcare organizations and the work roles for informaticists in those organizations.

METHODS

A 128-item survey was developed and administered as a structured interview to thirty-two information managers in eighteen organizations. The survey included items about medical informatics training, prior work experience, skills for informaticists, and programming proficiency.

RESULTS

There was a strong preference for informaticists with prior clinical work experience and an understanding of healthcare. Project management and data warehousing were highly rated skills. Informaticists were expected to know about healthcare processes, clinical guidelines, and outcome management. They were not expected to be expert programmers.

CONCLUSION

There is a role in healthcare organizations for interdisciplinary workers who understand clinical medicine, healthcare management, information technology, and who can communicate and work effectively across these organizational boundaries.

Barriers to evaluation of clinical vocabularies

BACKGROUND

The importance of standard vocabularies for representing clinical data is widely accepted. The selection of suitable vocabularies for a given task relies upon criteria for systematically differentiating vocabularies. However, methods and measures for comparing vocabularies are only now emerging. Substantial barriers inhibit progress in the development of these measures.

METHOD

As part of a larger project, the authors are developing quantitative measures for characteristics of controlled vocabulary as identified in the literature.

RESULTS

Seven barriers have been identified that inhibit the development of quantitative measures of the characteristics of controlled vocabularies: 1) application dependence; 2) empirical v. independent assessment; 3) dichotomous v. continuous measures of characteristics; 4) poor definition of characteristics; 5) number of characteristics; 6) multiple levels of significance of characteristics; and 7) interdependence of characteristics.

CONCLUSION

Progress toward quantitative assessment of clinical vocabularies is dependent upon overcoming barriers to the development of appropriate measures. Such progress requires innovative solutions and collaboration among investigators.

A Web link management tool for optimizing utilization of distributed knowledge in health care applications

The number of health-related Web sites on the Internet is increasing. Incorporating these sites into clinical decision support systems and other health care applications can significantly enhance the educational and instructional value of such systems. While search engines exist for finding sites and criteria are available for assessing site quality, few tools are available for managing Web-based health care information. Management of Web-based information is particularly challenging because the information is continually changing and new resources are continually being added. In this paper, we describe the development and use of a Web-link manager for health care applications. This system retains search strategies for repeated use, catalogues search results in a search results database, accommodates tracking of site review and use status, and provides periodic checking of link integrity for sites that are used in local applications. The Web-link manager is currently in use to manage the links used in a clinical decision support system that presents clinical practice guidelines interactively to clinicians at the point of care.

Siegfried: System for Interactive Electronic Guidelines with Feedback and Resources for Instructional and Educational Development

The proliferation of clinical practice guidelines (CPGs) has necessitated computerized solutions for guideline distribution and implementation. In this paper we describe a Web-based system that interactively presents CPGs at the point of care. Our system, known as Siegfried, provides a generalized solution for implementing CPGs by maintaining the guideline knowledge base separate from the application that presents the guidelines. As a result of this design, new CPGs can be easily added and existing CPGs can be expeditiously modified without additional programming. This system also solicits feedback from users regarding guideline recommendations and provides hypertext links to relevant Web-based instructional and educational resources.

Extracting medical knowledge for a coded problem list vocabulary from the UMLS Knowledge Sources

INTRODUCTION

The Unified Medical Language System (UMLS) Knowledge Sources embody a rich source of medical knowledge. We sought to extract a portion of this knowledge by incorporating information about relationships between UMLS concepts into an existing problem list vocabulary.

METHODS

We matched terms from the coded problem list of The Medical Record (TMR), a computer-based patient record system, with those found in the UMLS Metathesaurus. Those UMLS concepts that participate in 'parent' relationships with the matched TMR concepts were translated back into TMR codes and the relationship information was retained for integration into the coded problem list of TMR.

RESULTS

Of the coded problems currently in use in TMR, 67% (1627/2436) could be matched by normalized string matches to the UMLS Knowledge Sources. Of these matched TMR concepts, 91% (1488/1627) participated in at least one UMLS-identified parent relationship but only 28% of the matched concepts (454/1627) participated in parent relationships that already matched to a TMR code. As a result, although 67% of TMR codes were matched to UMLS concepts, only 19% of our original problem list (454/2436) could be augmented by relationship information contained in UMLS without improving the rate of matches or adding additional UMLS concepts as coded problems in TMR.

CONCLUSION

This study illustrates the rapid decline in overall rates of matching that result from a multiplicative effect of successive matches of terms to concepts, concepts to relationships and concepts back to entry terms. This effect will hamper any effort to extract relationship knowledge from the UMLS for incorporation into an entry vocabulary that is not already one of the source vocabularies of the UMLS Metathesaurus.

Integration of a computer-based patient record system into the primary care setting

The use of computer-based patient record systems (CPRS) in the primary care setting will increase significantly over the next few years. Real-time, point-of-care use of such systems must provide adequate payback to justify the intrusion into the provider/patient relationship. The authors describe, through the transition of a legacy system into a state-of-the art system, how such a system might be integrated into the primary care setting. Information flow is organized around an event, such as a patient encounter, or around the patient. The Medical Record (TMR) optimizes the provider/computer interaction through the use of protocols and clinical guidelines. Documentation is enhanced through the use of computer-generated progress notes.

Development and use of a Guideline Entry Wizard to convert text clinical practice guidelines to a relational format

Computerization of clinical practice guidelines (CPGs) has been proposed as one solution to enhance the use of guidelines in influencing standard clinical care. However, the conversion of text guidelines to the format required by a computer program is a major barrier. Clinicians who best understand the content of CPGs are typically ill equipped to convert textual guidelines into a computer accessible format. The potential of knowledge acquisition tools to assist in this process has been documented in the literature. In this paper we describe an application prototype, the Guideline Entry Wizard, created to assist in the conversion of text CPGs to a structured format within a relational database. We have tested this application through the input of information from several CPG. The application is a prototype for a more advanced tool. We have used this prototype to enter several CPGs and have demonstrated its effectiveness in inputting guideline content into a knowledge base.

Medical data mining: knowledge discovery in a clinical data warehouse

Clinical databases have accumulated large quantities of information about patients and their medical conditions. Relationships and patterns within this data could provide new medical knowledge. Unfortunately, few methodologies have been developed and applied to discover this hidden knowledge. In this study, the techniques of data mining (also known as Knowledge Discovery in Databases) were used to search for relationships in a large clinical database. Specifically, data accumulated on 3,902 obstetrical patients were evaluated for factors potentially contributing to preterm birth using exploratory factor analysis. Three factors were identified by the investigators for further exploration. This paper describes the processes involved in mining a clinical database including data warehousing, data query and cleaning, and data analysis.

Structuring clinical practice guidelines in a relational database model for decision support on the Internet

The rapid proliferation of clinical practice guidelines (CPGs) has made computerization increasingly useful to clinicians. Computerization, however, requires transformation of the content and logic of each guideline into a computer-accessible form. In this project, we sought to use a relational database to construct a generalized guideline knowledge base for use with Internet-based decision support applications. We hypothesized that knowledge representation schemes could be developed to capture guideline content and logic within the constraints of a relational database model. In this paper we describe a database schema based on a relational model for computerizing CPGs using a hybrid of structured and procedural knowledge representation schemes. We developed and refined this model in the context of five diverse CPGs and found it accommodated all necessary representational requirements.

A clinical economics workstation for risk-adjusted health care cost management

This paper describes a healthcare cost accounting system which is under development at Duke University Medical Center. Our approach differs from current practice in that this system will dynamically adjust its resource usage estimates to compensate for variations in patient risk levels. This adjustment is made possible by introducing a new cost accounting concept, Risk-Adjusted Quantity (RQ). RQ divides case-level resource usage variances into their risk-based component (resource consumption differences attributable to differences in patient risk levels) and their non-risk-based component (resource consumption differences which cannot be attributed to differences in patient risk levels). Because patient risk level is a factor in estimating resource usage, this system is able to simultaneously address the financial and quality dimensions of case cost management. In effect, cost-effectiveness analysis is incorporated into health care cost management.

Converting a legacy system database into relational format to enhance query efficiency

The analysis of clinical data collected over time can provide important insight into the health care process. Unfortunately, much of the electronic clinical data that exists today is stored in legacy systems, making it difficult to access and share the information. An approach is needed to improve the accessibility of electronic data stored in legacy system databases. In this study, a legacy database is converted into a relational format in the personal computer environment. The impact of such a conversion on query performance is evaluated, and issues that need to be considered when converting a legacy system database are identified.

Evaluation of an object-based data model implemented over a proprietary, legacy data model

Most computerized medical information today is contained in legacy systems. As vendors slowly move to open systems, legacy systems remain in use and contain valuable information. This paper evaluates the use of an object model imposed on an existing database to improve the ease with which data can be accessed. This study demonstrates that data elements can be retrieved without specific programming knowledge of the underlying data structure. It also suggests that underlying data structures can be changed without updating application code. Programs written using the object model were easier to program but ran greater than one order of magnitude slower than traditionally coded programs. In this paper, the legacy information system is introduced, the methods used to implement and evaluate the object-based data model are explained, and the results and conclusions are presented.

Alternative management strategies for maximising productivity in beef cattle in the subtropics

Two concurrent factorial experiments investigated the relative merits of the following management options to maximise liveweight gains in finishing cattle: breed [Hereford (Bos taurus) v. Belmont Red (B. taurus x B. indicus)]; implantation with zeranol; feed energy source (sorghum grain, molasses + urea, tropical grass pastures); feed protein source (sunflower meal, Leucaena leucocephala, tropical pasture legumes). These combinations of treatments were delivered by using nil, intermittent, or continuous grazing systems. Sixty steers of initial liveweight (� s.e) 321 � 3.6 kg were held on their respective treatments for 77 or 98 days, by which time all except those on pasture without supplement had attained at least 400 kg liveweight. Steers were slaughtered in a commercial abattoir and carcass characteristics recorded. Zeranol-implanted cattle grew faster than untreated (0.84 v. 0.66 kg/day), and there was a trend (P = 0.14) for Belmont Red cattle to grow faster than Herefords (0.83 v. 0.67 kg/day). In all systems, cattle given ad libitum sorghum grain grew significantly (P<0.05) better than those given molasses + urea (1.54-0.65 v. 0.72-0.45 kg/day). The poor performance of molasses-fed cattle was associated with low voluntary feed consumptions. The highest rates of gain (1.54 kg/day) were for penned cattle on a sorghum grain + sunflower ration. The provision of either supplement promoted better gains than that found for unsupplemented steers grazing tropical grass-legume pastures (-0.29 to +0.26 kg/day). However, cattle continuously grazing (10 steers/ha) leucaena-grass pastures and those given sunflower meal had significantly (P<0.05) higher rates of gain (0.82 and 0.91 kg/day, respectively) than cattle given molasses supplements or grazing alone. Cattle given sorghum had higher carcass weights and fat contents than those fed molasses or with grazing alone, although only those in the feedlot had acceptable fat cover at slaughter. There was no significant effect of treatments on the fatty acid composition of subcutaneous fat.

Factors impacting the success of computerized preadmission screening

Many approaches to controlling costs under managed care rely on the ability to prospectively identify the type or level of service a patient requires at the time of presentation. Although computers may effectively predict these factors, the impact of such a computer system is greatly dependent on its integration into the admission process. Three factors that influence the effectiveness of predictive screening using a computer were identified. They are detection, intervention and compliance. The effect of these factors was then measured in a prospective randomized trial evaluating the effectiveness of computerized preadmission screening for predicting the appropriateness of inpatient care. This paper examines the three factors and their impact on the effectiveness of the system. A mathematical model that relates the factors to the overall effectiveness of computerized preadmission screening is proposed and considered in a more general context.

Using the Internet Gopher Protocol to link a computerized patient record and distributed electronic resources

At Duke University Medical Center, we are developing a prototype clinical application for automated patient care plans with integrated links to electronic documents and other electronic resources. These links are implemented using the Internet Gopher Protocol, an emerging standard for distributed document search and retrieval. Use of this protocol permits storage of electronic documents in an open, nonproprietary manner. This paper discusses the architecture of the link mechanism and presents some of the advantages and disadvantages of the proposed method.

Integration of a stand-alone expert system with a hospital information system

A stand-alone PC expert system for evaluating the appropriateness of inpatient admissions has been integrated with an existing hospital information system. The expert system supports preadmission screening for appropriateness of inpatient admissions. The HIS provides extensive clinical data in a coded electronic form, permitting high-level decision support. The integrated system was developed for a 20 week randomized clinical trial to evaluate the effects of preadmission screening on inappropriate inpatient admissions. Three factors of the integration are considered: programmatic integration of the expert system, seamless presentation of mixed platform applications, and integration of coded data from the stand-alone application into the HIS data structure.

Predicting discharge diagnoses using a computerized preadmission screening tool

The authors integrated into an existing hospital information system a preadmission screening system, intended to reduce payment denials and reimbursement shortfalls that result from improperly classifying the billing status of patients. They report the results of efforts to validate the expert system upon which the preadmission screening system is based.

Grazing management studies with Australian cashmere goats. 1. Effect of stocking rate on the growth and fleece production of weaner goats grazing tropical pastures

Australian cashmere weaner goats were continuously grazed on either nitrogen (N) fertilised Pangola grass or mixed tropical legume-pangola grass pastures at 3 stocking rates (15, 30, 60 goats/ha). Grazing commenced in January and was terminated in May (18 weeks). Three parasite control treatments (Seponver, Avomec, Systemex) were superimposed on each treatment. Parasite egg counts, liveweight change, cashmere growth and changes in pasture components were measured. Intestinal parasite burdens were low (<400 eggs/g faeces) in all animals throughout the trial, and there were no significant effects of parasite control treatment on goat productivity. The liveweight gain of goats decreased with increasing stocking rate and time on pasture. Mean values for the liveweight change of goats on the legume-grass pasture were 66, 37 and 14 g/day at stocking rates of 15, 30 and 60 goats/ha, respectively. Comparable values for goats grazing N-fertilised pangola grass were 40, 39 and 23 g/day for stocking rates of 15, 30 and 60 goats/ha. Goats grazing the legume-grass pastures at the lowest stocking rate (15 goats/ha) had significantly higher liveweight gains than did goats on all other treatments. There were no significant effects of stocking rate, pasture type or worm control treatment on fleece growth in these goats. Mean values were 191 g, 39 g and 15.0 �m for fleece weight, cashmere weight and mean cashmere fibre diameter respectively. The major legume in the legume-grass pasture was siratro, with axillaris, greenleaf desmodium and cassia forming minor components (10-300 kg/dry matter (DM)/ha). During the first 6 weeks of grazing, all legume components increased under stocking rates of 15 and 30 goats/ha. Thereafter, there was a significant decline in siratro yield to less than 200 kg DMha at all stocking rates. During this same period, the yields of the other legumes generally increased, suggesting that at low stocking rates the goats preferred grass to legume, and when grass was limiting, siratro was the legume preferred over all others. Weed yields increased with time in all except the high stocking rate treatments. It was concluded that weaner goat growth can be maximised on tropical legume-pangola grass pastures when stocking rates and intestinal parasite burdens are low. Protein intake may be limiting growth on N-fertilised grass pastures.

Grazing management studies with Australian cashmere goats. 3. Effect of season on the selection of diet by cattle, sheep and goats from two tropical grass-legume pastures

Oesophageally fistulated cattle, sheep and I goats were grazed together to study species diet preferences from 2 different tropical grass-legume pastures during 3 seasons (summer, autumn, winter). In one pasture, Brachiaria decumbens (signal grass) was the dominant grass and in the other pasture, Paspalum plicatulum. Each pasture had a common suite of tropical legumes, mainly Desmodium intortum (greenleaf desmodium), Macroptilium atropurpureum (siratro) and Macrotyloma axillaris (axillaris). Axillaris was the major legume available in summer and autumn and greenleaf desmodium was the major legume in winter. There was no significant effect in any season of the basal grass available on the composition of the diet selected by the different species. Both goats and cattle had a high preference for legume leaf during summer and autumn, with goats having a higher preference (34-75% of diet) than cattle (1245%). Sheep selected against legume (5-19% of diet) and had a high preference for grass leaf (58-74% of diet) in all seasons. Cattle selected grass in preference to legume (greenleaf desmodium) in winter. Goats showed the greatest discrimination against both grass and legume stem. In summer sheep selected a higher quality (in vitro digestibility) diet (62.7%) than did cattle (53.4%) and goats (51.2%). In both autumn and winter cattle and sheep selected diets of similar digestibility (range 58.6-63.6%). In vitro digestibility of the extrusa samples was significantly (P<0.05) correlated (r = 0.74) with the proportion of grass selected. It was concluded that in their grazing preferences, goats were more like cattle than sheep. The different preferences of sheep and goats for tropical legume leaf was associated with location of these legumes in the sward, goats browsing from the top down and sheep grazing from the bottom of the sward.