S Teng Liaw
In the September 1992 issue of Informatics in Healthcare Australia, Ball & Douglas (1992) outlined the concept of the computer-based patient record (CPR) as proposed by the USA Institute of Medicine Committee on Improving the Patient Record and the CPR Institute. This paper will expand on this CPR concept (Weed 1969, RCP 1991, Dick & Steen 1990) by broadly outlining some of the historical and international developments that have led to the current recognition and consensus that the patient record serve to support patient care by facilitating:
Evolution of the paper-based patient record
Clinicians[1] have always collected and stored information on their patients as an aide memoire for ongoing diagnostic formulation and management in the provision of continuing care. The traditional medical record merely stored information with an emphasis on privacy and ethical/legal issues: accessibility of the information to others was neither desired nor desirable. This lack of emphasis on the communication of health information in the provision and evaluation of health care (Nightingale 1859) and medical and surgical interventions (Codman 1914, Hey Groves 1908) has been recognized for a long time.
The organization and presentation of clinical data has traditionally been based on the task, the person performing the task and/or the encounter. Problem-oriented recording, which links the patient's clinical data with his or her problems (Weed 1968,1969), and time-oriented recording, which captures data in a chronological sequence (Whiting-O'Keefe et al 1985,1988) were relatively recent develop-ments. Records (and data entry) may be structured (Duggan et al 1990) and/or include protocols for management (Cook & Heidt 1988). The A4 folder problem-oriented medical record (SOAP[2]) system, with structured patient summary and data entry sheets for various aspects of health care, has been adopted by most clinicians in Australia, UK/Europe and North America as the standard for record keeping on paper (Simbourg et al 1977) and on computer (Johns et al 1977).
With increasing multidisciplinary care and group practice, the patient record must integrate lengthy notes from medical, nursing, and allied health workers as well as the results of investigations and reports into one source that may be shared by the health care team. The patient record has been described as the principal instrument for ensuring coordinated, continuous and comprehensive care (Starfield 1992, Barnett 1984). The trade-off is the increased time, personnel and financial resources needed to collect accurate and reliable information and maintain its accuracy, reliability, confidentiality and availability to clinicians, non-clinicians, administrators, consumers and the community.
Many studies have criticized the variable quality of existing patient records (Mansfield et al 1991, Barnett 1984, Romm & Putman 1981, Thompson & Osborne 1976, Chamberlain 1971) which may be a reason why an estimated 70% of physician information needs are unmet during a patient visit (Covell et al 1985). Patient record data are often missing, illegible or inaccurate (Tufo & Speidel 1971, Bentsen 1976, Romm & Putnam 1981, Pories 1990). Errors can be introduced by clinicians, patients or equipment (Burnum 1989). Excessive redundant information is a problem (Korpman & Lincoln 1988). Ambulatory care records in the USA frequently contain poorly organized data, lack of documentation of key aspects of care, and exhibit inaccurate diagnostic coding (IOM 1990); they also lack standardization in content and format and are inaccessible, incomplete, and inaccurate (Davies, 1990).
Despite its familiarity, portability, flexibility for recording data and easy browsability in the one-to-one clinical situation as well as independence from technology requirements and computer 'downtime', paper-based records have significant limitations like illegibility, unavailability, sheer physical volume over time, difficult transferability and integration between providers and institutions, and the necessity to record the same data many times (double-entry) on different documents (Fries 1974, Pories 1990).
The increasing complexity of health care, information requirements and legitimate demands for patient-based data for clinical and epidemiological purposes has exposed the inadequacy of paper-based records to carry out a systematic programme of health care. The proposed answer to this information management inadequacy is the computer-based patient record (Barnett 1984, Dick & Steen 1990).
Evolution of computer-based health information management
It has long been recognized that 'computers could and should be used in medical record keeping, mass screening, description of the natural history of disease, and national data-banks' (Valbona 1968, Weed 1969). The ability of a computer to use a single datum in multiple ways - for patient care, decision support in management and diagnosis, communications, laboratory requests and reports, practice management, accounting, billing, and planning, research, and understanding the natural history or clinical course of patient problems - eliminates double entry, saves time, eliminates the tedium of writing referral letters or laboratory requests and reports, meets the increasing demands for data by the 'health care team' and requests by patients for access to their records, and enables quality control efforts to be focused at the single point of data entry.
Pilot projects using mainframes in academic and hospital-based centres in North America, UK and Europe began in the early 1960s. The fragmented user-based development of public health, hospital, departmental (laboratory and radiology) and clinical information systems in the USA (Weiderhold 1988, Barnett 1984) and the UK (Knox 1991) had resulted in a multiplicity of applications for numerous data sources as outlined below (only one example per application is quoted):
Day-to-day clinical management of individual patients required more real-time online support for professional and scientific judgements rather than administrative, legal and management interventions. Early obstacles were high cost, limited capacity of hardware, lack of flexible and compatible software, the embryonic state of computer science, and an inadequate and unfriendly computing-medical interface.
The fundamental concepts of automated information systems emphasised the individual health profile, lifelong medical record, general accessibility of medical knowledge, and expert systems (Lindberg 1985). Automatic Multiphasic Health Testing (periodic measurements monitored over an individual's life time) is controversial. However, the readily available, adequately structured lifelong medical record, either distributed on a 'soft network' linked by an electronic smart-card (Hopkins 1990) or a 'hard-network' permanently linked by telecommunication lines, is believed to improve individual patient care as well as the health care system (Branger & Dulsterhout 1990). General accessibility of medical knowledge has been the subject of the Integrated Academic Information Management Systems (IAIMS) which link communications, knowledge bases, patient record keeping and decision support. HEALTHSAT transmits medical journals electronically to medical schools in the poorer Third World countries (International Physicians for the Prevention of Nuclear War 1992). Expert and decision support systems are considered desirable in and integral to any health information management system (Shortliffe et al 1990)[3]. Prototype decision support systems are now available for nearly every aspect of clinical practice (van Bemmel & McCray 1992).
Evolution of automated primary care patient information systems
It soon became obvious that the only difference between the fragmented user populations lay in the frequency with which a function was used and therefore considered important. The primary health care practice setting, a microcosm of these administrative, departmental (investigations, prescribing), epidemiological and clinical settings, was a logical place to attempt to combine and share information among these functions. The computer applications desired by GPs are fairly uniform in most countries as demonstrated by surveys in the UK (DoH 1986), Netherlands (Hasman & Westerman 1985) and Germany (Szecsenyl 1993).
In the 1960s individual GPs in the UK developed and used computer-based systems for immunization and screening (Hodes 1968) and morbidity recording (Dinwoodie 1969). By the mid 1970s, various sites in the USA (Henley & Wiederhold 1975, Froom 1974, 1975), Canada (Newell et al 1976, Farley et al 1974), UK (Preece et al 1970, Abrams 1972, Grummitt 1977, Bain 1983), and Australia (Day & Larsen 1978) had some form of automated primary care (ambulatory) patient record system with varying levels of comprehensiveness. Problematic issues like cost, data security, undisciplined record keeping and data collection, and a lack of direction or focus led to a stagnation and low point in GP computing until the microcomputer led the rejuvenation in the 1980s.
The initiatives of the RACGP into general practice computing in the early 1980s, particularly the Computer Fellowship, Computer Assisted Practices Project (CAPP) in 1985 (MacIsaac et al 1990) and Standards Project (Crampton & Lord 1988) had been a significant step forward. It was also instrumental in the formation of the Australian Health Informatics Association for the 1990s. The activities of the RACGP Information Management Committee include establishment of accreditation criteria for software com-panies, organising the RACGP Computer Conference, coordinating various informatics projects and liaising with relevant government agencies and projects. The Practice Manage-ment Committee of the RACGP is particularly involved in computer based office manage-ment.
The Canadian Ontario Data Standards in Family/General Practice Project reviewed its medical informatics activities since the 1970s and emphasised a planned and methodical approach to the collection, processing, analysis and reporting of accurate and reliable data at both micro and macro levels within the framework of a Care Management Process Model and a Clinical Data Model (Ontario Ministry of Health 1992). A Primary Care Information Institute was proposed to conduct the implementation and ongoing evaluation and revision of the models.
The British Medical Association (BMA) and the Royal College of General Practitioners (RCGP) published 'Computing in General Practice' (Palmer & Rees 1980) and established the Joint Computer Policy Group in 1982 (Lucas & Metcalfe 1980). The 1982 joint Department of Industry/DHSS #2.5 million 'Micros for GPs' scheme (Jones 1987, DHSS 1985) and G-PASS in Scotland (Ryan 1987) helped GPs to further identify computing needs and issues like data loss and corruption as a result of 'ageing' of computer systems (McWilliams 1986). The 1990 UK National Health Service GP 'performance-related contract' (DOH 1989) placed primary health care (PHC) at the centre of the development of health care services as a whole and demonstrated that IT can support primary care clinical, administrative and financial activities (Bradley 1992). A strategic alliance between the Family Health Services Authorities (FHSA), District Health Authorities (DHA) and general practitioners was proposed to release the population and demographic information held within the general practice database to the NHS as a whole. While its reliability may be suspect (Johnson et al 1991), this information will be a better guide to real population needs than the existing incomplete and fragmented statistics.
The European Community Advanced Informatics in Medicine (AIM) project was established in 1989 to promote collaboration and a common conceptual framework for cooperation across systems and language barriers as well as to address the future requirements and options in health care information infrastructure and services (Roger France 1990). The range of projects cover most aspects of primary care clinical practice. The Systems Accreditation Project for Primary Healthcare Informatics Requirements and Evaluations (SAPPHIRE, Kay 1992) and the GP Minimum System Specification (GPMSS) project (Gardner & Smith 1992) aim to normalise the chaotic market place by accreditation and facilitate the development of inter-operable systems to allow electronic communication links among GPs, laboratories, pharmacies, Family Health Services Authority (FHSA) and hospitals. The trend is to increasing use of EDI and decision support systems (DSS) for diagnosis and therapy (Hasman & Westerman 1985, Westerhof & Dupuits 1992). Projects involving laboratory information in Oxfordshire (Daniels et al 1992) and Gloucester (Pill 1992) as well as electronic links between FHSAs and GPs in England and Wales (Treleaven 1992) are underway using the FHSA Racal Healthlink network. There are large scale trials on electronic communication between different health services and providers (Hasman et al 1990, Ament & L'Ortye 1990) as well as the use of 'smart-cards' and 'laser-cards' (Hall & Hopkins 1990).
Scepticism remains as to whether the CPR will ever replace paper-based records. The pioneering systems of the 1960's and 1970's have either disappeared or are used only by small core user-groups based in or associated with an academic institution (Kuhn & Wiederhold 1981). The current state of the art and level of penetration of computer-based medical records through the USA health-care system were much less than indicated by the advances in the 1960s (Stead 1989). Only about 17% of general practices in New Zealand (Walls 1990) and 41% of general practices in Australia (Crampton 1990) used computers to any extent. No more than 5% of GPs in Australia (Crampton 1989, Liaw 1992) and North America (Froom 1987) use computer-based clinical applications. The more socialised European Community countries, especially the UK and Netherlands, have had more success with about 65% utilization of computers. Most information technologists are agreed that the main barriers to the CPR are political, organisational and educational rather than technological (Walker et al 1991, Dick & Steen 1990, Difford 1987).
The successful AHA Meditel and Vamp Health data collection scheme in the UK, an attempt to trade in health information, highlighted the lack of incentives for GPs to invest in computer-based information management systems[4]. The multiplicity of computer systems in most countries, about 60 computer system suppliers in the UK (McWilliams et al 1992) and 50 computer system suppliers in Australia (RACGP 1992), raises problems of compatibility and interoperability. The trend for the market place to become dominated by a few large companies is cautiously welcomed as the reduction in the number of usually incom-patible systems in practice may minimise costs and maintenance problems like the 'sixth installation syndrome'[5].
A review of the proceedings of the major health informatics conferences since 1989 (Salamon et al 1989, SCAMC Proceedings 1989-1992, IMIA MEDINFO 1989-1992, O'Moore et al 1990) highlighted the depth and diversity of research and development in Europe and North America. The European Community AIM projects emphasised more practical clinical applications in primary health care while the USA projects emphasised more theoretical hospital-based projects. There is growing recognition that standards in clinical information systems, nomenclature and classification frameworks that will support local information processing and link to other levels of the health care system must be defined at the primary care level where the patient enters the health care system. There is also a growing trend to generic systems to avoid the problems associated with proprietary commercial systems. The South Oxford Health Centre Patient Record (Randall 1993) and MUFFIN (Liaw & Chan 1993) are examples.
The approach to the design of a CPR may be top-down from a molecular (divisible) concept or bottom-up from an atomic (indivisible) attribute. Among the AIM projects, ADAM (Advanced Architecture in Medicine) reflected the top down approach (Frassine et al 1992) and has evolved to ISAAC (Integrated System Architecture for Advanced primary Care) which is focused on information technology and telematics to support the framework of general practice (Fisher & Marshall 1992). The Good European Health Record (AIM Project A2014) is another. PEN and PAD (Nowlan et al 1990), which has evolved to the GALEN (Rector et al 1993) project, represents the atomic approach to patient record design. The North Americans are basically doing the same things in a more commercial and laisse faire fashion. The ASTM Standard 1384 Standard Guide for the Content and Structure of an Automated Primary Care Record of Care (1992) is an example. The problem with the voluntary nature of the process of standards development (ASTM 1990, CEN 1992) is the potential for non-compatibility. For instance, EDIFACT (ISO 1988) is being proposed for Electronic Data Interchange (EDI) in Europe, while the North Americans plan to adopt the HL/7 and ASTM 25 standards (McDonald 1992, Ontario Ministry of Health 1992).
In summary, it would appear that we have not only been keeping records for the wrong reasons, but were also doing it poorly under difficult circumstances. A paradigm shift to emphasise communication in patient record keeping, driven by incentives and structured entry of a minimum and optimum data set as part of routine clinical practice, is needed. This broad historical review suggests that we are moving steadily in the right direction.
Available on request (four pages). Contact S. T. Liaw on e-mail: pcstl@gamgee.cc.flinders.edu.au.
S Teng Liaw, FRACGP is a NH&MRC Research Fellow, School of Medicine, Flinders University of South Australia.
