Intensive Care Med (2001) 27: 179±186 DOI 10.1007/s001340000747

Michael Imhoff Andrew Webb Andreas Goldschmidt on behalf of the ESICM

Published online: 19 December 2000  Springer-Verlag 2000

)

M. Imhoff ( ) Chirurgische Klinik, Städtische Kliniken, Beurhausstrasse 40, 44137 Dortmund, Germany E-mail: [email protected] Phone: +49-2 31-5 02 00 21 Fax: +49-2 31-5 02 00 81

ESICM STATE MENT

Health informatics

Abstract Health informatics is the development and assessment of methods and systems for the acquisition, processing and interpretation of patient data with the help of knowledge from scientific research. This definition implies that health informatics is not tied to the application of computers but more generally to the entire management of information in healthcare. The focus is the patient and the process of care. The apparent information overload and the imperfection of medical decision making motivate the use of information systems for medical decision support. Health informatics provides tools to control processes in healthcare, acquire medical knowledge and communicate information between all people and organisations involved with

Introduction Health informatics has become one of the new buzzwords in healthcare. In their daily work healthcare professionals are confronted with a growing number of computers and computer applications. Health informatics and information technology provide the tools for the acquisition, preparation and distribution of medical data and knowledge. Unfortunately, medical information technology has not always kept up with current advances although the technological state of current medical information systems is better than it is generally held to be. This review article will try to: · Give useful definitions of health informatics · Describe applications of health informatics in healthcare

healthcare. Although the development of medical information systems may often lag behind the available possibilities, the technological state of the current medical information systems is better than it is generally held to be. Health informatics should help healthcare professionals to provide better and more cost-effective care and enable healthcare systems to be more efficient and to adapt better to our patients' needs. Health informatics may reshape the way we deliver care to meet the demands of the future. Key words Health informatics ´ Evidence-based medicine ´ Clinical decision support ´ Management of information ´ Expert systems ´ Knowledge discovery

· Show how health informatics may affect the process of care · Take a look at some future aspects of health informatics

Definitions Health informatics is the development and assessment of methods and systems for the acquisition, processing and interpretation of patient data with the help of knowledge from scientific research. According to the definition by Haux [1], medical informatics is the discipline concerned with the systematic processing of data, information and knowledge in medicine and healthcare. The domain of medical informatics covers computational and informational aspects of processes and

180

structures in medicine and healthcare. The terminology, especially the distinction between medical informatics and health informatics, is used inconsistently in the literature. For our purposes we will summarise any of the above mentioned processes under health informatics, thereby focusing on the process and the continuum of care. Health informatics is not only the application of computer technology to problems in healthcare but covers all aspects of the generation, handling, communication, storage, retrieval, management, analysis, discovery and synthesis of data, information and knowledge in the entire scope of healthcare. In this respect health informatics is as old as medicine itself. Through the history of medicine four major stages of health informatics can be distinguished: 1. The recording of the first impressions of illness, for communication between physicians and other people involved in the process of healthcare and for teaching medicine. 2. The empirical basis of medicine, where medicine was an art rather than a science. Without some methods to acquire, store, process, analyse and communicate data and information, the present stage in the development of modern medicine could not have been reached. 3. Building the scientific basis of medicine from theoretical conjecture and experimental method. Many of the advances in the understanding of physiology and pathophysiology, the progress of diagnostic and therapeutic methods and devices and the multidisciplinary approach to care could only be realised through the concepts, methods and technologies of health informatics. 4. Today the regulation of medicine on a societal level presents an unrivalled challenge for the handling of medical information, including auditing, quality control, standardisation of care, evidence-based medicine etc. Although health informatics is an ancient discipline, there is currently a much stronger and rapidly growing perception of information technology and concepts in the process of care. Coiera [2] stated ªif physiology is the logic of life then health informatics is the logic of healthcareº. On the basis of this understanding four core elements of health informatics can be identified: · The way healthcare professionals think about patients. · The way diagnoses are made and evaluated and treatments are defined, selected and evolved. · How medical knowledge is created, shaped, shared and applied.

· How healthcare professionals organise themselves to create and run healthcare systems.

The goals of health informatics Two principle goals of medical informatics can be distinguished [1]: · To provide solutions for problems related to data, information and knowledge processing. · To study general principles of processing data, information and knowledge in medicine and healthcare. More specifically health informatics has a role in answering the new challenges for healthcare: · Structures for pooling, communicating and applying clinical evidence. · Organisational processes to minimise resource use while securing maximal benefit. · Development of tools and methods to achieve these aims.

The focus is the patient The patient is the central focus of healthcare. The patient must therefore also be the central focus of health informatics. Without patients there would be no need for health informatics: · The patient generates data and information. · The vast majority of all communication within hospitals is patient-centred. · Administrative and financial data management focuses on billing for healthcare delivered to patients. · Medical data management focuses on patient-centred diagnostic and therapeutic procedures. · Nursing data management focuses on patient-centred nursing activities. · Medical information departments concentrate on the handling of patient records and patient databases. · Evidence-based medicine (EBM) focuses on the analysis of clinical studies involving patients. All healthcare professionals and ancillary departments have to focus on the patient. Communication of data and information takes place between all the parties involved, and between them and the patient (Fig. 1). Patient-related information has to be available at any time, anywhere and in its entirety. Ideally all patient information should be available throughout the lifetime of the patient. In addition to communication within the medical domain, more and more information needs to be provided to institutions such as health maintenance

181

Fig. 1 The patient is at the central focus of health informatics (details in text)

organisations, regulatory bodies, etc. Moreover, these institutions exert a growing influence on medical and non-medical processes within the continuum of care.

Health informatics and the healthcare system The healthcare system spans the entire process of therapy and care for a patient with one or multiple illnesses (Fig. 2). Hospital care encompasses the different treatment and care units, such as emergency room, operating room and general ward. Different modalities of diagnostics (laboratory, imaging, etc.) and treatment are available. Medical documentation and, ultimately, health informatics provide information and communication throughout and are the formal representation of the continuum of care through the hospital [3, 4]. Health informatics may provide a representation of the continuum of care in the healthcare system as a whole.

Applications of health informatics in the process of care The technical applications of health informatics are legion. We encounter them everywhere in the process of care, where they serve, among others, the following purposes: · Hospital administration, billing and accounting. · Resource management.

· · · · · ·

Medical documentation. Diagnostics and therapy. Imaging. Communication. Information management. Clinical decision support.

These applications, or rather the user front end, can be located at different levels of the health organisation: · At the level of instrumentation (patient monitors, ventilators, imaging equipment, etc.). · At the level of the bedside or the point of care (a clinical information system with an electronic patient record). · At the level of the ward or care unit (admission, discharge, transfer and coding systems). · At the level of the hospital (hospital information systems). · At the level of the scientific community (EBM). · At the level of society. This is only a crude classification and it is very difficult to draw a clear line between the different levels. Although health informatics covers all areas of healthcare the technical implementation, in the form of information systems, cannot be seen everywhere. On the instrumentation level we can see broad acceptance of devices. Although more sophisticated devices will replace older ones, for many applications the quantita-

182

Fig. 2 The continuum of care within the hospital (details in text)

tive demands for devices are satisfied in the developed world. Today only few implementations of information systems at the point of care can be seen. Major impediments to the more rapid spread of information technology to support this level are the lack of a sufficient information infrastructure in the hospital and lack of communication interfaces between medical devices and heterogeneous information systems. There is rarely a sufficiently developed conceptual framework at the medical level. Information management at the hospital and/or unit/ward level is essential to run hospitals or other health institutions technically and from the perspective of business administration. Most institutions in the Western world implemented administration systems many years ago. Therefore, it can be expected that at least a rudimentary information infrastructure is present in many places to support this level. As EBM is a methodology and a concept, it is very difficult to estimate the degree of penetration in the medical community. While EBM takes place in the head of the physician, health informatics can provide tools for EBM.

Information overload The amount of information generated in the process of standard hospital care is staggering: Haux [1] estimated that in a major German university hospital treating 50,000 in-patients and 200,000 out-patients annually, about 300,000 new medical records are opened every year containing about 6,000,000 documents. If stored

digitally this information would generate an estimated 2 terabytes of data. In healthcare and especially in intensive care we are facing a data and information overload, as illustrated by numerous examples: · An abundance of information is generated during the process of critical care. Much of this information can now be captured and stored using clinical information systems (CIS) which provide for complete medical documentation at the bedside. The clinical usefulness and efficiency of clinical information systems has been proved repeatedly [5, 6, 7]. While databases with more than 2,000 separate patient-related variables are now available for further analysis [8], the multitude of variables presented at the bedside even without a CIS precludes medical judgement by humans. A physician may be confronted with more than 200 variables in the critically ill during typical morning rounds [9]. However, even an experienced physician is often not able to develop a systematic response to any problem involving more than seven variables [10]. Moreover, humans are limited in their ability to estimate the degree of relatedness between only two variables [11]. This problem is most pronounced in the evaluation of the measurable effect of a therapeutic intervention. Personal bias, experience and a certain expectation toward the respective intervention may distort an objective judgement [12]. · At the level of the hospital, information overload is an issue. Beside the business administration of such a complex enterprise, new and ever more complicat-

183

ed reimbursement and accounting policies lead to an unmatched complexity of data and information [1]. · The volume of scientific literature is growing exponentially. As of 1993 the number of systematic reviews in medicine increased 500-fold over a 10-year period [13]. It is impossible for the individual healthcare professional to keep track of all relevant medical knowledge even in very narrow sub-specialities.

Clinical decision support The apparent information overload and the imperfection of medical decision making motivate the use of decision support systems. Clinical decision support aims at providing healthcare professionals with therapy guidelines directly at the point of care. This should enhance the quality of clinical care, since the guidelines sort out high-value practices from those that have little or no value. The goal of decision support is to supply the best recommendation under all circumstances [14]. This goal may be achieved by the following measures: · Standardisation of care leading to a reduction of intra- and inter-individual variance of care. · Development of standards and guidelines following rational principles. · Development of explicit, standardised treatment protocols. · Continuous control and validation of standards and guidelines against new scientific evidence and against actual patient data. Development of knowledge bases for clinical decision support The foundation for any medical decision support is the medical knowledge base which contains the necessary rules and facts. This knowledge needs to be acquired from information and data in the fields of interest, such as medicine, economics, ethics or sociology. Three general methodologies to acquire this knowledge can be distinguished: · Traditional expert systems. · Evidence-based methods. · Statistical and artificial intelligence methods. Traditional expert systems The traditional approach to the development of expert systems is to gather information from experts in the field of interest. Although this approach has generated some very powerful and successful decision support

rule-bases and clinical pathways [14], it still has serious shortcomings: · Expert knowledge is not necessarily validated against clinical data or evidence-based. · Many underlying pathophysiological mechanisms are not fully understood. · Clinical experiments in intensive care are difficult or not feasible at all. · Prospective trials for validation of the decision support algorithms are costly, time-consuming and may not represent actual clinical practice. Evidence-based medicine Evidence-based medicine (EBM) seeks to make the best possible use of available medical knowledge based on scientific evidence. EBM is not only confined to the results of clinical trials of Phases I-IV. EBM does not offer its own new scientific data, but is able to evaluate available medical knowledge as well as therapeutic procedures, methods and established behaviour with reference to healthcare decisions. It makes use of information technology to provide physicians and other healthcare professionals with information necessary to assist and evaluate critically the benefit of therapeutic interventions. A goal of EBM is to establish databases from which systematic reviews can be accessed. The worldwide network of Cochrane-Centres evaluates medical knowledge and offers it on the Internet. The expansion of health informatics with the facilitation of data transfer and communication, which includes remote data entry via Internet and Intranet, can provide information essential to medical practice and as a basis, as well as support, for future clinical trials. The benefit of efficient data management is evident. (See also: http://www.cochrane.org/ and [15, 16]). However, one must unfortunately realise some shortcomings of EBM in the thorough preparation and interpretation of clinical trials. Relatively speaking, there are only a few controlled clinical trials for the many scientific and relevant questions in intensive care medicine. On the other hand, there are unhelpful controlled clinical studies for irrelevant questions or without useful defined end points. Meta-analysis in most cases does not lead to the essential interpretation, because the conditions, end points, raw data and other facts of studies are not homogeneous and comparable enough. In a lot of cases the desired standardisation for the interesting criteria and parameters is difficult or impossible. But the most relevant problem of EBM in clinical practice may be the gap between the need, actuality and amount of proper information and the actual availability of this. Controlled clinical trials need time for the many reasons mentioned. Thus the consecutive results and clinical im-

184

plementations can often not keep up with the fast innovation processes. Nevertheless EBM is an indispensable tool to support best clinical practice and must be expanded to reduce the information gap. Evidence-based methods help to generate rule-bases for clinical decision support by integrating high-quality clinical research evidence with pathophysiological reasoning, caregiver experience and patient preferences [17]. Evidence-based methods lead to a seven-step process for generating clinical pathways, or rather their underlying rule-bases: 1. Identification and formulation of the clinical problem. 2. Literature research (MEDLINE, EMBASE, Cochrane Library). 3. Evaluation of the literature following strict criteria. 4. Development of the clinical pathway and its rulebase on the basis of the evidence found. 5. Clinical implementation and training of healthcare professionals. 6. Continuous control. 7. Re-evaluation of the pathway after predetermined time intervals. Statistical and artificial intelligence methods While this approach of EBM provides for sound and objective evidence for the rule-base, it cannot use the actual patient data from the clinical data repositories for the discovery and validation of medical knowledge. Moreover, both approaches can be time-consuming and costly. The development and testing of the highly successful computerised ventilation protocol of the LDS Hospital, Salt Lake City, Utah, USA, required more than 25 man-years, which reflects the enormous effort necessary for the development of sound clinical protocols [18]. In a new approach statistical methods for time series analysis were combined with methods for knowledge discovery in large databases (KDD) and applied to a standard clinical information system. This approach, which was recently described in detail [19, 20, 21], allows combining knowledge discovery (i.e., learning explicit rules from actual clinical practice), statistical methods and explicit expert knowledge. At this point the system has been tested in the development of a rule-base for the haemodynamic management of the critically ill. The approach allows the dynamic discovery and validation of rules for decision support systems. Compared to the above-mentioned traditional approaches, the KDD methodology appears to be more efficient by at least one magnitude in terms of time and manpower spent on rule-base development.

It can be expected that modern statistical and knowledge discovery methodologies will have a significant impact on the development of clinical pathways and medical decision support systems.

Management of knowledge as an economic factor In modern industrial society the ªmanagement of knowledgeº is seen as the decisive competitive factor. This requires that hospitals and other healthcare enterprises introduce new management techniques as well as modern information technology with improved applications, where inter-operability can support innovative thinking. In other industries the focus has shifted from centralised to distributed information production and accessibility, from isolated expert knowledge to structured team knowledge, organised learning and workflow management. It can be expected that this development will also take place in healthcare. Moreover, health informatics will also be affected by, and benefit from, the ubiquitous evolution of client-server technology toward browser-based systems, as well as from the Intra- and Internet technology as the predominant information technology paradigm of the future [22, 23]. In hospitals we speak of processes primarily with reference to medical-clinical operating procedures and their implications for management. Optimisation of the governing operating procedures requires increased efficiency, i.e., correct procedures carried out economically. In general, this involves the re-organisation of operational and organisational structures (Table 1). In order to revise and adapt the organisational structures in hospitals, which allow for the assignment of goals and the allocation of incentives, an information and control system must be provided for. The first tier managers of a hospital must be put in a position to receive the important data, in preparation for the decision making process, in time. The diverse data requirements on the different levels (department/unit, area/discipline, profit centre, individual clinics, hospital complex) must be processed, prepared and presented in accordance to the respective need for detail. To quote an information technology representative of the US Military Health Service: ªThe mission is to give the right information to the right people at the right timeº. The requirements relevant to managerial responsibilities are derived from the importance placed on goal-oriented structuring of the enterprise and management. Management information systems (MISs) and tools used in the presentation and modulation of operating procedures support optimisation by means of simulation. While all functionalities of a MIS and of tools for the optimisation of operating procedures are, for the most part, instruments of control, they form the topmost presentation layer of MIS (referred to, at times, as

185

Table 1 Typical criteria and parameters for the optimisation of operating procedures from the perspective of cost-consciousness Recognising and removing bottlenecks involving resources Prevention of disturbances in the area of operations Reduction of peak workloads Covering costs without loss of quality Flexibility and parallel development of operations (both managerial and clinical)

executive information system, EIS) in accordance with the wishes and requirements of hospital leadership down to the responsible senior physician [24]. The complete integration of the existing principal/main and subsystems, even with standardised interfaces, is not an easy task. By collecting the indispensable data, identified by internally conducted analysis, a common format is secured in a separate data base, a data warehouse.

Conclusion Health informatics is the development and assessment of methods and systems for the acquisition, processing and interpretation of patient data with the help of knowledge from scientific research. This definition implies that health informatics is not tied to the application

of computers but, more generally, to the entire management of information in healthcare. The focus is the patient and the process of care. Health informatics provides tools to control processes in healthcare, acquire medical knowledge and communicate information between all people and organisations involved with healthcare. All important tools for acquisition, preparation and distribution of medical data and knowledge are available with the current information technology. However, the development of medical information systems by the manufacturers often does not keep up with the possibilities available. Nevertheless, the technological state of the current medical information systems is better than generally believed. Health informatics is an application-oriented science which must move the technological realisation of methods and concepts of information management in medicine forward. As a result, the gap must be filled between wish and reality. This is a continuous process which always follows the advances in medicine with a certain delay. State-of-the-art here means keeping the gap as small as possible. Health informatics should help healthcare professionals to provide better and more cost-effective care and enable healthcare systems to be more efficient and to adapt better to our patients' needs. Health informatics may reshape the way we deliver care to meet the demands of the future.

References 1. Haux R (1997) Aims and tasks of medical informatics. Int J Med Inf 44: 9±20 2. Coiera E (1997) Guide to medical informatics, the internet and telemedicine. Oxford University Press, Oxford 3. Gardner RM (1986) Computerized management of intensive care patients. MD Comput 3: 36±51 4. McDonald CJ, Tierney WM (1988) Computer-stored medical records: their future role in medical practice. JAMA 259: 3433±3440 5. Imhoff M (1995) A clinical information system on the intensive care unit: dream or nightmare? In: Rubi JAG (ed) Medicina Intensiva 1995, XXX. Congreso SEMIUC. Murcia. Pictographia, Murcia, pp 17±22 6. Imhoff M (1996) 3 years clinical use of the Siemens Emtek System 2000: efforts and benefits. Clinical Intensive Care 7 (Suppl):43±44

7. Imhoff M, Lehner JH, Löhlein D (1994) 2 years clinical experience with a clinical information system on a surgical ICU. In: Mutz NJ, Koller W, Benzer H (eds) 7th European Congress on Intensive Care Medicine. Monduzi Editore, Bologna, pp 163±166 8. Imhoff M (1998) Clinical data acquisition: what and how? Journal für Anästhesie und Intensivmedizin 5: 85±86 9. Morris A, Gardner R (1992) Computer applications. In: Hall J, Schmidt G, Wood L (eds) Principles of Critical Care. McGraw-Hill, New York, pp 500±514 10. Miller G (1956) The magical number seven, plus or minus two: some limits to our capacity for processing information. Psychol Rev 63: 81±97 11. Jennings D, Amabile T, Ross L (1982) Informal covariation assessments: databased versus theory-based judgements. In: Kahnemann D, Slovic P, Tversky A (eds) Judgement under uncertainty: heuristics and biases. Cambridge University Press, Cambridge, pp 211±230

12. Guyatt G, Drummund M, Feeny D, Tugwell P, Stoddart G, Haynes RB, Bennett K, Labelle R (1986) Guidelines for the clinical and economic evaluation of health care technologies. Soc Sci Med 22: 393±408 13. Chalmers TC, Laus J (1993) Meta-analytic stimulus for changes in clinical trials. Stat Methods Med Res 2: 161±172 14. Morris A (1998) Algorithm-based decision-making. In: Tobin JA (ed) Principles and Practice of Intensive Care Monitoring. McGraw-Hill, New York, pp 1355±1381 15. Cooke IE (1996) Finding the evidence. In: Cooke IE, Sackett DL (eds) Clinical Obstetrics and Gynaecology. International Practice and Research, Vol 10, No. 4. Balli›re, London 16. Goldschmidt AJW, Weiûbach L (1999) Spannungsfeld zwischen Signifikanz und Relevanz. Hilfe durch die Methodik evidenz-basierter Medizin? Der Onkologe 5: 1008±1014 17. Cook D (1998) Evidence-based critical care medicine. A potential tool for change. New Horiz 6: 20±25

186

18. Thomsen GE, Pope D, East TD, Morris AH, Kinder AT, Carlson DA, Smith GL, Wallace CJ, Orme JF Jr, Clemmer TP (1993) Clinical performance of a rule-based decision support system for mechanical ventilation of ARDS patients. Proc Annu Symp Comput Appl Med Care 1993: 339±343 19. Imhoff M, Gather U, Morik K (1999) Development of decision support algorithms for intensive care medicine: a new approach combining time series analysis and a knowledge base system with learning and revision capabilities. In: Brugard W, Christaller T, Cremers AB (eds) KI-99: Advances in Artificial Intelligence, 23rd Annual German Conference on Artificial Intelligence, Bonn, Germany, September 1999, Procedings. Springer, Berlin, pp 219±230

20. Morik K, Brockhausen P, Joachims T (1999) Combining statistical learning with a knowledge-based approach ± a case study in intensive care monitoring. In: Bratko I, Dzeroski S (eds) Proceedings of the 16th International Conference on Machine Learning 1999. Morgan Kaufmann, San Francisco, pp 268±277 21. Morik K, Imhoff M, Brockhausen P, Joachims T, Gather U (2000): Knowledge discovery and knowledge validation in intensive care. Artif Intell Med 19: 225±249 22. Goldschmidt AJ, Luz CJ, Giere W, Ludecke R, Jonas D (1995) Multi-dimensional visualisation of laboratory findings and functional test results for analysing the clinical course of disease in medicine. Methods Inf Med 34: 302±308

23. Quade G, Novotny J, Burde B, May F, Beck LE, Goldschmidt A (1999) Worldwide telemedicine services based on distributed multimedia electronic patient records by using the second generation Web server hyperwave. Proc AMIA Symp 1999: 916±920 24. Goldschmidt AJW, Rebholz F (1999) Führungsinformationssysteme für leitende ¾rzte im Krankenhaus. In: Hellmann W (ed) Der Arzt mit Managementkompetenz: Ideenbörse zur Übernahme von Leitungsaufgaben im Krankenhaus. Kohlhammer, Stuttgart, pp 122±132