Information systems for health services

PUBLIC HEALTH IN ..................... -.3-. __ , REr- -, 17 JUL 1980 IPPt- tu,,vPE REGIONAL OFFICE leformation systems for health services Edited and introduced by G. McLACHLAN Secretary Nuffield Provincial Hospitals Trust London, England Foreword by D.D. VENEDIKTOV Deputy Minister of Health Ministry of Health of the USSR Moscow, USSR REGIONAL OFFICE FOR EUROPE World Health Organization Copenhagen 1980 ISBN 92 9020 132 0 © World Health Organization 1980 Publications of the WHO Regional Office for Europe enjoy copyright pro- tection in accordance with the provisions of Protocol 2 of the Universal Copy- right Convention. For rights of reproduction or translation, in part or in toto, of this publication application should be made to the WHO Regional Office for Europe, Scherfigsvej 8, DK-2100 Copenhagen</), Denmark. The Regional Office welcomes such applications. The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of the Secretariat of the World Health Organization concerning the legal sta- tus of any country, territory, city, or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. The mention of specific companies or of certain manufacturers' products does not imply that they are endorsed or recommended by the World Health Organization in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of proprietary products are dis- tinguished by initial capital letters. The views expressed in this publication are those of the authors and do not necessarily represent the decisions or the stated policy of the World Health Organization. PRINTED IN DENMARK Reissued under ISBN: 9789289024068 (print) in 2025. Originally published under ISBN-10: 9290201320. ISSN 0300-4880 CONTENTS Page Foreword. . . . . . .. . . . . . . . . .. ... . . .... . . . .. . . . . . . . . . . . v Introduction - G. Mclachlan . . ... ........ .. .... .. . . . .. .. . PART I: BACKGROUND AND REPORT Chapter 1. Information systems for health services at the national health level - A.S. Hiirii. . . . . . . . . . . . . . . . . . . . . . . . 7 Chapter 2 . The WHO information system and the interaction between national health information systems - A.S. Hiirii. . . . . . . . 17 Chapter 3. Information systems in the health service of the Federal Republic of Germany - G. Griesser . . . . . . . . . . . . . . . . 27 Chapter 4 . The experience of the USSR - A.S. Kiselev . . . . . . . . . . . 37 Chapter 5. The strategy in Scotland - M.A . Heasman . . . . . . . . . . . . 41 Chapter 6. The general situation in Europe in 1977 - A.A. Weber. . . . 45 Chapter 7. Report on the technical discussions , Munich, 6 - IO September I 977 . . . . . . . . . . . . . . . . . . . . . . . . . 51 PART II: SOME TECHNICAL AND OTHER PROBLEMS Chapter 8. Data security and data protection in health information systems - G. Griesser . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Chapter 9 . Advantages and disadvantages of centralized and decentralized data storage and processing - A.S. Kiselev . . . 69 Chapter I 0. Interview surveys for decision-makers - F. Sawicki . . . . . . 73 Chapter 11. Models of the health system as a basis for data collection - E.N. Shigan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Chapter 12. Criteria fo r the effective planning and establishment of computerized health information systems - P. l e Beux & F. Gremy. .... . . .. . .. .. .... . . .. . . . . . . . . . . . . 95 Chapter 13. True morb idit y and the use of health care services - A. Fenton l ewis . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Chapter 14. Liaison of environmental exposure data with morbidity data - A. Sjostriim & P. Westerholm . . . . . . . . . . . . . . . . 115 Chapt er 15. A method of measuring hospital activit y in relation to the patient - B. Marielle . . . . . . . . . . . . . . . . . . . . . . . . . . 123 iii FOREWORD In September 1977 the twenty-seventh session of the WHO Regional Com- mittee for Europe devoted its traditional technical discussions to "Information sy stems for health services ''. The selection of this topic reflected a growing need, both in the European Region and in all other WHO regions, to under- stand and give new meaning to a number of important and controversial prob- lems relating to the analy sis and control of services and sy stems for protecting and strengthening the health of populations. Without doubt, Europe has seen a more active and pro longed development than o ther WHO regions of health institutions and services of various kinds which, while reflecting the socioeconomic and other characteristics and condi- tions of each country, nevertheless y ield satisfactory results on the whole. Th e populations of most European countries enjoy a relatively high standard of health and average life expectancy, and the various medical and health institu- tions are justly proud of their work, both past and present. At the same time, there also exist in Europe a sizeable number of hither- to unsolved problems and difficulties, both "old" and new, in the develop- ment of health services, and on which complete and accurate information is oft en needed. One need only cite the consequences of demographic change, the increase in chronic and degenerative diseases, accidents and drug de- pendence, and the uneven coverage of certain population groups (particu- larly the elderly , the disabled and the migrant worker) by health care ser- vices in a number of countries. Th ese are undoubtedly connnected first and foremost with the radical socioeconomic and political changes that have occurred among the nations of Europe, which have experienced two world wars in the twentieth century, and with the existence on the European sub- continent of two different socioeconomic systems - th e socialist and the capitalist - with , as a result, different structural principles and trends in health service development. Th ey are also linked with the general develop- ment of science and technology, the growing complexity of economic, politi- cal, cultural and other relations between countries, the rise in the standard of living and educational level of the population, the increasingly numerous attempts to determine the prospects for mankind 's development up to the y ear 2000 and beyond, and many other factors. In Europe there has also been a substantial deepening and growth in the understanding of health as a biological and social category , of the right of everyone to health protection and promotion, and of the major responsibility V of governments and States for the health of their peoples. An important stimulus in this respect was provided by World Health Assembly resolution WHA28.88 on the development of primary health care as "the point of entry for the in- dividual to the national health system•; and by resolution WHA30.43, which noted that "the main social target of governments and WHO in the coming decades should be the attainment by all the citizens of the world by the year 2000 of a level of health that will permit them to lead a socially and economically productive life''. These resolutions, together with the decisions and results of the International Conference on Primary Health Care, held in Alma-Ata in 1978, constitute an important milestone in the work of the World Health Organization. The technical discussions had as their basis a report on information systems in health services. This report, along with the answers submitted by European countries to a questionnaire compiled by the Regional Office, formed a fruit- ful basis for discussion. What, then, were the results of the technical discussions? First of all, the debate revealed an increasingly profound understanding in European countries of the systemic nature of health services as an expression of a whole combina- tion of measures aimru. at maintaining and strengthening health and prolonging the active, creative life of every individual and of the population as a whole. The discussions also illustrated the growing need for increasing purposeful and rational management of health services and systems, the effective coordina- tion of all links in these systems, the correct use in health practices of the entire body of scientific and technological expertise available, the assessment and cor- rect use of financial, personnnel, material-technical and other resources, and the combining of the interests of both the general population and workers in the medical professions. The present-day management of health services calls for consideration to be taken of the extensive medicoscientific links and collab- oration between many States under bilateral, multilateral, pan-European, inter- regional and global agreements, and especially within the framework of the World Health Organization. If this is not done, no country can effectively ensure that the health of its population will be protected and strengthened. Under current conditions health service management cannot take place without the co llection, processing and analysis of extremely voluminous and varied masses of information. The methods usru. for processing this informa- tion differ substantially and are employed professionally by different spe- cialists in the various countries. Some of these methods have gradually gained general recognition, while others are used only in individual countries. A de- cisive factor now is the construction of health information systems or, more precisely, the rational supply of information to health services on the basis of the widespread use of up-to-date computer technology. The discussion made clear the chief lines already taken in the construction of such information sys- tems in health services. First of all, it involves the same comprehensive systems approach to the provision of information as that adopted in respect of health services as a whole, namely, the consideration in their entirety of all previously fragmented sources of information, their content, forms and methods of processing them. Under this heading are included data on the health status of individuals, their vi case histories, and data on the work of individual medical specialists and health in- stitutions. Also included here are data on the status and development prospects of medical science as a systematized body of objective knowledge in the field of biol- ogy and medicine and as a process of accumulating new knowledge. We are also considering here general statistical information on the health status of the popula- tion, as well as the work of health organizations, institutions, services and systems in various areas and at different levels in various countries and regions. In addi- tion, we are dealing with various registers, regulations and items of legislation, as well as different kinds of plans, forecasts and proposals for the development of the health system as a whole and of its individual services in particular. The volume of all types of information is rapidly growing; however, it is necessary to distinguish co"ectly the nature of these data, their interconnexions and interdependence, and the qualitative differences between them In essence, information circulating within the health system originates from isolated, individ- ual clinical or experimental observations. However, when these observations are aggregated, a qualitative transition is made to a deeper and more widely based understanding of the most important biomedical and social principles. All this, taken together, should and does form a single body of informa- tion, reflecting everything that needs to be known about the health status and trends towards change in individuals and the population as a whole, about the factors that present or may present a hazard to health, and about methods of disease prevention and treatment. In this regard, the health of an individual cannot be regarded as a static "condition•: but rather as a continuous dynamic process whereby the optimum balance is maintained between the organism and its natural and social environment. In the same way, "public health" is not some kind of arithmetical sum of individual "healths':· it is a qualitatively more general, integrated reflection of the sum total of the individual levels of health of all members of society. Unfortunately, all the existing methods for assessing public health are rela- tive and incomplete. However, there are favourable prospects in applying methods for the dynamic modelling of public health, taking into account both the probable fluctuations in the health of individual members of society during their lives, and sociodemographic and o ther social processes. A second condition for an effective health information system is the availa- bility to and rational use by managers of the basic forms of information that are of great value in directing a health system. In other words, it is essential for infor- mation systems to be genuine storehouses or depositories of information and not "information graveyards''. In order to achieve this, it is necessary to reconstruct and reorganize the separate bodies of information in the health system in such a way that they can cope with the information needs of all kindsofhealth worker, including any new requirements that may be dictated by circumstances. The reason for this is that the most acute problems in health systems are con- stantly arising at al/ levels, ranging from individual doctor-patient relationships and the solution of questions concerning operative and other medical inter- ventions in critical conditions, to medicosocial and organizational problems at the local, regional, national and even the international level. At all these levels appropriate decisions are being made. Th e decision-makers may be workers in the primary health sector, medical practitioners, scientists, heads of health vii - l I departments, ministers of health or finance, directors of scientific and interna- tional organizations, etc. It is precisely this factor that makes it impossible to foresee what questions may prove to be important in various circumstances, and furthermore it is impossible to envisage all the answers to all new questions. It would appear that the most suitable ty pe of organization for an in - formation system is the "pyramidal" form, in which the greatest and most serious attention is given to primary medical data ( at the level of the individual patient, speci11list or institution), while more general data and indices are supplied to the higher levels of health service administration. At the same time, it should always remain possible to have direct recourse to the primary data and to obtain them in a very short time for additional analysis. It should be the task of the information system to give a well-defined idea of what information actually exists regarding the health or diseases of individuals, where that information is located, and how it can most speedily be obtained when neces- sary so that an adequate decision can be taken. A matter of considerable importance here is that of being able rapidly to use basic medical data, while at the same time maintaining confidentiality regarding the health status of individual patients and even of certain population groups. Data secun·ty is perfectly logical and understandable for a number of reasons, ranging from the purely psychological to the organizational and socioeconomic. This problem can be solved only through a realization that th e interests of the individual and of society with regard to health and health protection coincide and are in no way contradictory. Both the individual and society have an in- terest in each member of that society possessing the maximum level of health; from this point of view, subject to the guarantee of confidenti11lity of data concerning the health status of an individual, any information regarding this or that hazard to public health must immediately be brought to the attention of all the persons and organizations concerned. A lthough in contemporary in- fomwtion systems a technical solution to this problem presents certain diffi- culties, it is nevertheless quite capable of achievement. Finally, it is fully apparent that, as health systems develop and informa- tion improves, these systems will make new demands on existing health care services and methods for the co llection and analysis of completely new types of information that do not exist at present. For this reason, speakers in the technical discussions frequently emphasized th e importance of further collab- orative work by speci11lists in such fields as health management and administra- tion, medical statistics, systems analysis and the promotion of biomedical research. Only through the co llective efforts of all these specialists can we hope to achieve the successful construction of fundamentally new types of health information system. D.D. Venediktov Chairman of the technical discussions viii INTRODUCTION G. McLachlan° In today's world, even in those national economies in which the free market is dominant, health services operate within a system involving some form of bureaucratic authority. This is primarily for reasons of efficiency and accountability, which pose questions of measurement. Thus, added to the re- quirements for the basic clinical information fundamental to medical science are those of information for management, the function of which is to ensure the optimum deployment of resources with the general aim of providing ef- fective services for prevention, cure, and care. Information is valueless without the ability to use it purposefully, which presupposes clarity of role and objec- tive on the part of the user, whether it be for diagnosis, for treatment, or for better management at the various levels. A modern health information system itself constitutes a subsystem within the health care system. It also involves the employment of staff with high levels of ability and qualification. It thus raises the questions of education and train- ing, both of which require their own forms of subsystem. International com- parisons are difficult to make under any circumstances, but in the case of health the picture is even more confused because of the different management and educational structures that have been evolved in different countries for the delivery of medical care services. These have stemmed from the operating poli- tical, social , economic, and cultural conditions, the relative weights of which with regard to policy making differ from country to country. Universally, however, the major management problem is one of devel- oping suitable information and the capabilities for using it (a) at the central level to create strategies for improving the health of the population at large, and (b) at the operational level for applying the overall plan to local situations in order to produce effective services for the population, without surrender of the individual's right to privacy and at a reasonable cost. This last point is currently of immense importance, not just because health care is now one of the most expensive forms of service to individuals but also because the govern- ment of practically every country in the world is deeply involved in its health system and the way it is financed. The requirements of public accountability and control are becoming ex- tremely sophisticated and therefore complex. The forging in the last few years 0 Secre tary , The Nuffied Provincial llospitals Trust, London, England. of policies designed to control aspects of service, either for quality and effec- tiveness to ensure the best use of resources or simply as part of the drive for cost containment , has created large requirements for information as well as for mechanisms to use it effectively at all levels. Indeed , it is the objectives and limitations of the various levels of manage- ment (including that of patients) that largely determine the shape and content of the information system. Information for clinical purposes is a prime require- ment, but the extent to which data of all kinds are generated and collected at the operational level and how much can usefully be aggregated for passing to the centre is, in most countries , an unresolved problem. Although the health systems of European countries differ, they have in common the fact that basic data requirements generated in the patient/doctor encounter are private, being concerned directly with the therapy and care of individual patients. With the present generation of processing equipment, the principles for the col- lection and processing of such infonnation do not differ greatly from those that apply to management information, but not all information collected spe- cifically for each purpose can be commonly used. The development of com- puters has opened up a new world for data processing but the types of system currently in use make the cost of "total" collection prohibitive and the selective use of a restricted data base has tended to be favoured. Much of the development in management information in the last few years has been towards the presentation of data to compare the efficiency of functional services or local systems within a single economy. More recently, the development of the idea that there should be better communication of information between countries has provoked interest in international comparisons, and consequently in the setting up of certain basic principles to enable such comparisons to be useful to the countries concerned. The present publication reflects the policy of WHO in establishing common prin- ciples and providing basic information to help Member States towards the settlement and implementation of their own national policies. The intention is to present an overall picture of the present position, together with a selection of technical problems. * * * The rationale for this publication was the decision by the WHO Regional Committee for Europe to take up the subject of information systems in health services in its technical discussions in 1977 . Part I is largely based on informa- tion presented at the Regional Committee and consists of essays from some of the technical experts with practical experience of health information systems who were assembled to consider various aspects of information . The dis- cussion documents presented were such that, with their concentration on the general principles ruling at both national and international levels, filled out by specific examples, they provided criteria for national systems against which shortcomings in policy and performance can be measured. This is the reason 2 for focusing special attention on the role of WHO as well as the experience of a number of countries, particularly the Federal Republic of Germany, Scotland, and the USSR, where special efforts have been made in the last few years to harness available techniques to the production and use of information. The report of the technical discussions indicates some of the many weak- nesses in the major aspects of policy set out in the original documents, notably failures in internal communication involving epidemiologists, administrators, statisticians, and physicians. It is not difficult to deduce that this may, how- ever, arise because the statisticians and epidemiologists are very much aware of the shortcomings of the data with which they are called upon to work. In addition, the expectations raised by the development of computers and related technology, which has been a major feature of the last 20 years and which initially gave rise to many high hopes for the use of information in manage- ment, including clinical management, have not been realized. It is not claimed that any solutions have been found to the daunting problems that emerged as a result of the discussions, except in a sharpened awareness of many of the problems identified and in the general perspective. Perhaps one of the reasons is that some of the more sophisticated aims, such as outcome measurement and health indicators, may be well beyond present competence to gather the elements and to collate, assess, and relate them in a quantifiable way. Part II, which was planned after the technical discussions, consists of a selection of the more complex general problems, some of which of course overlap. Professor Griesser writes with great understanding about one of the pressing problems of civilized societies, that of security and protection of data to ensure that the most personal information gathered for clinical purposes is available only to those authorized to make use of it. Professor Kiselev discusses the universal dilemma of centralized and decentralized data storage and pro- cessing. At the same time he points to the different aims of each level and the fundamental restraints. The methodology is particularly important and Pro- fessor Sawicki gives a revealing account of the techniques involved in the use of interview surveys in the promotion of basic material for administration of health services, for planning and evaluation purposes. The chapter by Professor Shigan illustrates the capacity of operational research techniques in the modelling of health system(s) as the basis for data selection and collection and of the feedback requirements of any data system necessary for the con- tinual improvement and updating of information. The development of criteria for the production of specific information for effective planning, in a form capable of being processed by computers, is dealt with by Mr Le Beux and Pro- fessor Gremy in a way that shows the absolute necessity for defining the prob- lem initially and setting out the issue involved with clarity. It also gives an indica- tion of the vastness of the effort needed to understand health issues in the modern state as a prerequisite of planning. More specifically, Dr Fenton Lewis highlights the data requirements for an important aspect of community care and the great problem of accumulating such data. The fact that we are now in a period in which it is recognized that the environment has a great influence on health is the background of the account by Dr Sjostrom and Dr Westerholm. Their observation that an improved understanding of disease in populations 3 constitutes a basis for a better understanding of the underl y ing mechanisms, and constitutes a technological base for various types of policy options in planning for medical services, is in the true epidemiological tradition. Finally, Mr Marielle discusses the measurement of hospital activity in relation to the patient, mainly in the context of the French hospital system. * * * Given the technical success of the age, if one of the main impressions is sadly of a vast shortfall in achievement from what nlight be expected at this particular time, it needs to be remembered that the production of information is itself a young activity , concerned with issues and techniques of great com- plexity and involving many disciplines, some old , some still developing. It re- quires a comnlitment by many to an understanding, not just of up-to-date methods of collection and use of basic infom1ation for the benefit of health , but also of how people can be educated to its use. This raises the need for policies, concerning not on ly those dealing directly with patients but also those on the middle and higher eche lons of management as well as the politicians. In management a fundamental requirement fo r planning, as well as service, is an appropriate level of numeracy to understand relative measurements. Yet, equally, the ability to distinguish effect from cause is more important than a knowledge of, for example , calcu lus or the binary principle. Data are not in- fonnat ion. The main need is for the translation of data into in fo rmation, which ultimately should be presented to managers in as simple a form as possible within the linlitations of the presentation. Even if much effort and sophistica- tion are required to transform the raw data into the fmi.sh ed product of infor- mation, the manager should not be "blinded with science" so that he misses the essential points. Managers are likely to benefit greatly from what is de- livered to them only when they have understood the real implications of the information, without necessarily working through its detailed analysis . Central to confidence in the collection, analysis, and use o f information, which applies both to clinical and to management purposes in general, is the be- lief that technological progress has been a major factor in revealing the oppor- tunities of medical intervention , both to cure certain diseases and to alleviate others. The modern health sys tem is gea red to prevention, cure, and care, and as more information has come to hand about the causes of illn ess, the more has human ingenuity been stimulated to develop mean s to help the sick. * * * The two main streams of information fl ow have been for clinical pur- poses and for managerial effectiveness. It has become evident that both these 4 streams mingle at certain points, and in an ideal world a total information sys- tem based on data relating to the health and diseases of the population would seem to be a prime objective in order to plan health care programmes. How- ever, the mechanisms needed at present to make use of all health information would be colossal and the task seems almost beyond financial or human pos- sibilities. This has posed an even greater need for the development of sensible strategies in each system to select the kind of information that will be of maximum operational use in both a clinical and a managerial sense. Indeed, a major problem in industrialized societies today is the need for reconciliation between the clinician's requirements for information and those of the manager, and how to achieve this without surrendering rights to privacy. This is by no means a theoretical consideration ; for example , the epidemiologist, depending on how his role is conceived, has often the dual function of serving the clinician and the manager, as well as being a specialist in his own right. It has become evident in recent years that epidemiologists are aware of the inconsistencies of their roles in all these functions, but unfortunately epidemiologists are in very short supply. Part of the health manpower strategy for the future must be the education and training of a considerably larger number of epidemiologists than exists at present. There are, however, questions that go beyond the narrow professional one. The epidemiologist is a key figure in the development of comprehensive policies to take account of the funda- mental problem of selection, extraction, and collection of information of various kinds relating to health, and their effective integration into the manage- ment process. It cannot be denied that this is a complex operation, since all the main-line professions, including those directly concerned with the patient as well as those engaged in management, regard their own particular require- ments in information as essential tools for playing their several parts in the total system. If it means that total information is an impossible goal, selec- tivity is necessary with all its attendant limitations. To understand what is re- quired is a major objective of all concerned in the decision-making processes - clinicians, administrators, and politicians - who are now actively wrestling with the problem of priorities in health services. * * * The variety of the experiences revealed in this publication by individuals in but a small number of countries indicates the richness of approach and the human skills and resources that have to be involved . These contributions taken together as a whole accentuate the need for a general strategy embracing re- cruitment and education of those who are necessary not only for the produc- tion and processing of information for a variety of immediate management purposes, but also for general planning, whether it be for new regimes of therapy or for better organization to make the best use of the manifold re- sources used in a modern health care system. We cannot hope to develop health 5 services at an appropriate level without providing for the accumulation and deployment of such skills, either for management in the planning process or the monitoring operation or as a special support to clinical work and research. Thus, part of the immense problem involved in sorting out and selecting the information necessary is a need for a policy for the education, training, and deployment of suitable manpower. Although the problems sometimes seem not only formidable but almost insurmountable , in the long run the development of a suitable comprehensive strategy and a series of tactical steps are essential requirements of public policy on health in any society. 6 PART I BACKGROUND AND REPORT INFORMATION SYSTEMS FOR HEALTH SERVICES AT THE NATIONAL LEVEL A.S. Hiiro 0 For the purposes of this publication, a national health information service is a broad concept aiming at the provision of re le van t and timely information for health managers at all levels , and at the sharing of technical and scientific information by all health personnel participating in the health services of a country. Possibly, no country at the present time can claim to have such a sys- tem in operation, but the concept is receiving much attention and some pre- liminary trials have been initiated . There is no doubt that technological progress has markedly increased the possibility of intervention in the natural dynamics of diseases and in all aspects related to health, but it is also obvious that over the years the management of health services has fallen far behind these technological advances. The pre- dominant need is for the development of new strength in management, based on strategies, goals, and priorities that will satisfy the human aspiration to health as a basic right. Managerial goals should reflect all available knowledge of the needs of the population as a whole and a general belief in a responsi- bility to meet those needs. The two key words are "responsibility" and "knowledge". Great gaps in knowledge can be compensated for by detemlined action and political will, but the decisions that guide those actions inevitably reflect the infor- mation on which they are based. A prerequisite for efficient management is a reasonable amount of information concerning problems, resources , and the possibility of intervention in an "environment" in which specific actions are encouraged. TH E NEED FOR INFORMATION Information available to guide policy formulation and the planning of health care is always incomplete and it is exceptional to find that information is both completely relevant and covers all questions that arise, even for the 0 Director , Department of Planrung and Evaluatio n , Natio nal Boa rd o f Health , Hel- sinki , 1:inland . 7 solution of relatively well-defined problems. Many critical comments and recom- mendations concerning statistics and related sources of information have been made. The First World Health Assembly in 1948 endorsed a recommendation that Member States should establish special committees to plan and coordinate vital and health statistics. ln 1973 the Twenty-sixth World Health Assembly stressed a need for the complete reconstruction of health information sys- tems(/). The reason for this is clear. The problem of inadequate information is well recognized by both users and producers of information. Expressions like ''un- reliable", "irrelevant", "difficult to read", and "available too late" belong to the standard vocabulary of critical comments. Administrators are dissatisfied with the in formation they obtain and statisticians with what they are asked to pro- duce, while neither is entirely aware of the other's dissatisfaction. As a result, decision-makers tend to do without information, while producers feel that their skills are not understood and that information is not used properly. This means that there is considerable room for improvement even though there may be no disagreement between the users and producers about the value of health informa- tion. The user- producer dialogue, which is a prerequisite for progress, is either not initiated at all or is carried on without mutually shared interests. Part of the difficulty is that users and producers of information often do not fully under- stand one another. The technical expressions of statisticians and information scientists are not always comprehended by problem-oriented decision-makers and managers, while the complexity of health services and their terminologies are often only partly understood by those concerned with producing information. Competent managers often have to deal with problems without the benefit of systematic quantitative measurements. Taking risks is one of the elements of management; on the other hand, it is generally accepted that erroneous decisions are not acceptable if those making them have not done their best to diminish un- certain ties by seeking the help of information services. While it is primarily the duty of decision-makers to indicate the kind of information needed, technical experts should participate in discussions about the collection, analysis, and pres- entation of the information. The constraints of time on decision-makers often re- quire compromises that cannot be achieved without mutual understanding. There are exceptions, but many health service systems have grown step by step without systematic planning and without due attention to rational management. Modern management requires skills that are not systematically included in the education of physicians, statisticians, economists, social policy experts, etc. Orientation to biological systems in which only a few factors can be controlled often limits the ability of management to notice that in social systems there are numerous different courses of action. SOME COMMON MISCONCEPTIONS ABOUT INFORMATION There are some widespread misconceptions concerning information and its production and presentation. One of these is the belief that there can be a 8 minimum set of data to serve all purposes at all times. However, this would not permit a dynamic and flexible approach towards the planning ofhealth services. The more active and innovative the health policy, the less predictable the in- formation needs. A health information service should be able to translate broadly formulated requests for information into definable tasks, clearly presented within a relevant time scale. On the other hand there is value, at both the national and international levels, in the concept of a common minimum set of basic data for providing comparative information. Another misconception results from a belief that it is relatively simple to know what information is needed and how it should be used . In most cases decision-makers are not aware of their information needs and much effort is needed to discover what should be known. Sometimes, data are seen as unim- portant because they have not been used. Undeniably, there exist useless data but, on the other hand, the activities of organizations fluctuate. During quiet periods, when there are no major managerial problems, information needs are minimal; in crisis situations, facts that are hardly relevant during quiet periods are urgently needed. Changes in strategy, acute manpower problems, financial crises, and epidemics are examples of such situations. Reliability of information is a natural requirement, but even in this mis- conceptions are common. Formal exactness in the mathematical sense is not always the most important quality from the point of view of the decision process. Much more dangerous are weaknesses in validity, for example, when basic definitions are not fully applicable or have never been documented. Dif- ferences in coverage and lack of reliable population data are examples of com- mon limitations. Errors and weaknesses are not to be defended but, on the other hand, rather unreliable but relevant information can be most helpful. The main requirement is that the direction and magnitude of errors are known or can be estimated. SCOPE AND CONTENT OF HEALTH INFORMATION SYSTEMS Although statistics form an important part of any health information sys- tem , the latter concept is much wider and comprises other types of health and biomedical information and also includes centres for information services, mechanisms for obtaining information, and information flow and information processing, together with appropriate technologies for these activities. At present, a health in formation system may not come entirely under one ad- ministrative authority; indeed, it may never be possible or desirable to achieve a unified administration, but there should always be close links between the various parts of a system. It is this wider meaning of the term "information system" that is used here , although it is recognized that in some countries the use of the term has been confined to statistically oriented services and in others to technical data processing matters. A health service can be understood as a system. Its components include concepts (e.g., health and disease), ideas (e.g., equity), objects (e.g., hospitals), 9 and persons (e.g., physicians). Together these form a whole in which all the components interact to support or control one another. The elements that make interaction possible are information and communication, using informa- tion in its widest sense. The components of a system can also be thought of as subsystems where any system is a part of a grea ter one, or of several. A total management system can be divided into subsystems, one or more of which specialize in the collection, processing, analysis, and presentation of data and information. All such subsystems together form an information system, which is the generic term for all activities that ensure that management has relevant information when making decisions. It is natural to speak about information services because the purpose of an information system is primarily to serve management. The main task of an information service should be to minimize the amount of uncertainty in relation to decisions and the ultimate objective is to help a health-related organization to achieve its goals. Accordingly, the information produced should be relevant to this objective. In addition, there are numerous important information needs, such as research, in which the direct managerial interests are much smaller. In this connexion there is no need to itemize that information which can be useful or that which must be avail- able. Generally speaking it falls in to three main categories - all equally im- portant - namely information on: health related risks; health problems; and use of resources. ELEMENTS OF A HEALTH INFORMATION SYSTEM A health information system can be divided into parts or subsystems in different ways, and in this connexion it is practical to think about the different methods of collecting and processing data as basic elements. Traditional statistical routines are naturally a typical element in any national health in- formation system and form the main part of systems operating in many coun- tries at present. Aggregation of data into summarized reports or forms for dif- ferent persons, events, etc. during a given period is the basic principle. These are again summarized and national summaries are prepared from regional data. This method closely follows hierarchical lines but its main weakness is its rigidity. Answers to questions are, to a large extent, limited to those that were required when the basic form was introduced. Individual data systems apply a different principle: each person, event, etc. is handled separately and summaries are made on an ad hoc basis. The advantage of such systems is flexibility ; their limitations are the amount of data collection involved and the correspondingly greater demands on resources. The routine statistics, and to a great extent also the registers, are oriented to provide continuous control, but there are situations in which the focus of interest is a critical situation or problem that cannot be solved without a specially planned survey or study. Such an ad hoc approach can give a very reliable response. It is occasionally the most economical way of obtaining information but there are many exceptions. A serious difficulty is that ad IO hue studies require time for planning and execution, and results may not be ready with.in the time available ; in addition, they usually require competent personnel. lnfomrntion from control activities, such as those concerning the quality of drugs, the efficacy of applied methods, and the safety of technical pro- cedures belongs in principle to the health information system. In practice, how- ever, this type of information is mainly safeguarding the environment in which managerial decisions are made. Another relevant element of a health information system is that provided by experts or informants, which is the traditional administrative method of accumulating knowledge and opinions. The decision process often needs infor- mation about nontangible or at least nonquantifiable aspects. ActuaUy it seems to be true that in the most complicated problems the nonquantifiable, "soft" information has a decisive role. Personal contacts, reports , and memoranda are important in health manage- ment and essential for organization. Published information such as that provided by a library service is also very useful. These elements are defined on the basis of the general principles applied in the coUection and processing of information. A health information system also has components with quite different functions. One of these is the technical processing of data in which electronic devices form the vital element, and these are so in1portant that very often the term "information system'' is used more or less as a synonym for a computerized information se rvice. Information services cannot fulfil their purpose without communication. Organized communication in the form of statistical reports is a natural element of any information system. The material reported and the form and frequency of reporting vary according to the situation, and the intended use dictates the form of publication. However, if reports are not readable, owing to excessive use of mathematical jargon or inclusion of too detailed tabulations or opera- tionaUy meaningless classifications, or if they are not timely, real communi- cation is not achieved. Such experiences are not infrequent and one should not expect management to value information systems or their services highly in such circumstances. Indeed, such a situation is the cause of some of the dif- ficulties that make the producer-user dialogue so difficult. GENERAL PRINCIPLES or PLANNING A HEALTH INFORMATION SYSTEM A health service is a complex dynamic system influenced by different historical, economic, political, and other factors. The exact nature of the in- formation system required and its links and relationships with social and eco- nomic information systems vary so much in different countries that there can never be any generally valid model. The best system for any country is one that "takes account of these factors in appropriate measure, giving consideration to aU and presenting the proper admixture of co-ordinating authority and delega- tion of responsibility" (2). 11 It is not at present commonly assumed that the management, and especially the top management , of health services is directly responsible for the planning and functioning of the health information system. The elementary function of management is sometimes described as being the translation of information to instructions. It is impossible to do this efficiently without the help of a well organized information service. One common misconception, reflecting past times when development was very slow, is that information services are for the specialist in statistical methods. Such an expert cannot advise on what should be known, but the role of the technical expert is fundamental when decisions are made about how data should be collected, processed, and presented. It is primarily the task of decision-makers to ensure that the following three important aspects of health information systems planning are properly balanced: the planning of the in formation content; the planning of technical and methodological aspects ; and the planning of the organizational structure, location, etc. of the health information system. PLANNING THE INFORMATION CONTENT Planning the content of a health information system is a difficult task and involves both users of information and technical experts. To analyse and des- cribe in detail decisions as they are actually made is obviously impossible, but much can be achieved if the most important items are selected and presented. This aim can be achieved in different ways but one recommended procedure is to analyse systematically the programme, activity, or administrative sector and to describe these in the form of a model. The word "model" has many meanings but in this connexion it simply means a systematic way of indicating which elements belong to the system, which can be controlled, and which indicate success or failure. For this limited purpose, relatively crude models are often adequate. A simple model for vaccination is shown in Table 1. Development of models requires teamwork by experts and indicates what information is relevant, what degree of accuracy is required, and the risks in- herent in using estimates instead of measurements. In tum, the model should assist in the design of the information system itself. It must be stressed that decision-makers, being users, are the leading group in model building; with- out their participation this activity is meaningless and without their help the other participants do not know what information is essential and what is not needed. External experts can be useful but detailed planning of the health informa- tion system should be carried out inside the organization and not by visiting consultants. This kind of modelling is applicable at any level of government - local, regional, or central - but it is necessary to plan for compatibility, com- parability, and capacity to aggregate at each level. The recent report of a WHO Expert Committee on the application of systems analysis to health manage- ment should be of value in this connexion (3). 12 Table 1. Information required in a vaccination modela Item, problem Theoretical efficacy of the method Applicability of the method in practice Control of the vaccine, etc . Risk of the disease in the society in question Population at risk Req uirements of the procedure in terms of: (a) training in the procedures (b) vaccinations performed per unit of time (c) other technical requirements Effect of the activity on the services Attitude of the personnel and population Side-effects and accidents related to pro- cedure Participation Refusals Cases of disease Other relevant items such as resources re- quested, cost, etc. l Information controlled trial operational study special studil!s epidemiological study census data from demonstration areas and expert opinion expert opinion (mainly from other infor- mation subsystems) from expert opinion;can also be estimated from opinion surveys and participa- tion rates individual reports to be stored and studied data bank (as ad hoc study) that provides answers to questions such as which persons should be specially invited sample study of characteristics of refusals, rate of vaccinated persons by locality, cohort, etc . from notification and follow-up reports stored in a data bank a For simplicity, only that part of a model showing the type of information required for the planning or monitoring of a vaccination programme is used in this example ; re- lationships between the various elements (items) are omitted . 13 PLANNING THE TECHNICAL ASPECTS Before they can use their skills to assist in the decision-making process the experts (statisticians, data processors, etc.) have to know and understand the objectives and strategic problems of management, and these in turn must be put into terms that the expert can convert into information requirements. At present it is hardly possible to think of a health in formation system with- out the help of computers. It is true that many statistical routines are nearly im- possible without computers, but there is reason to stress that they are not the sol- ution to all information problems . A mass of "interesting" data which, it is hoped, the computer will process into information helpful for managerial pur- poses in many cases proves useless. The fault lies in a lack of dialogue and mutual understanding, and not in the data processing experts or the computers. On the other hand, there exist in most countries computerized sources of information that are primarily serving other purposes but could be made useful for health management. This is especially true of health insurance files, which are usually kept with the help of modern data processing facilities. The fuU capacity of these systems is not always explored by experts in health management. One of the problems that the statistician must solve is the presentation of results in such a form that the user can make useful conclusions. Too much scientific jargon and listing of all possible theoretical sources of error may prevent the reader from using the information at aJI. Good presentation means much more than statistical tables and elaborate mathematical treatment of findings. Most important are analyses and interpretations that give logical explanations of managerial problems and indicate to what extent the goals of health services have been achieved. ORGANIZATIONAL ASPECTS One of the natural requirements of a health information system is that it should serve all levels -· local, regional, and national - without undue duplica- tion. In addition, different institutional and sectoral interests should be satis- fied . It is more the rule than the exception that the same or rather similar in- formation is collected separately for different users. This is unavoidable to some extent but if proper attention is devoted to organizational structuring and rationalization much can be saved and improved. From this point of view it is therefore advisable to locate the system close to the decision-making function . This also makes it easier to integrate information generation with legal, financial, and other managerial activities in such a way as to avoid the need for special forms for "statistical" purposes. Integration with the general information system has technical advantages and , at least in theory, large national policy problems can be analysed in a more comprehensive way. Health in the well-known WHO definition has physi- cal, mental, and social aspects and the majority of decisions concerning health 14 in its broader sense are made outside health management. There are examples of both types of organizational solution in Europe. In most cases a part of the health information system is located inside the service organization but there are also integrated elements. No detailed model can be recommended for universal use since this depends on the culture, economic system, and popula- tion structure of the country . SUMMARY AND DISCUSSION The scope of health infonnation systems The concept of a health information system presented here is wide in scope . It includes not only statistical information but also technical/biomedical information and bibliographical services, together with information on the availability of expertise, statements on policy, etc. An important element is the information service that assists the conversion of data into relevant and usable information . Also included are the mechanisms for information flow, informa- tion processing, and the appropriate technologies for these. A properly functioning health information system should be flexible , modifying its activities according to information needs . Modelling, simu- lation, and related activities should be conducted continuously. Routinely collected series can be , and usually are, basic to an efficient system, but future needs cannot be met without changes that will allow limited resources to be devoted to more relevant purposes. Large investment and a great num- ber of personnel are not needed, though without some resources very little can be done. In few other areas in the health field are relatively small invest- ments so rewarding. This chapter discusses a "functional" system rather than an all-embracing administrative organization for the production of information . Nevertheless, such a concept will be unfamiliar to many. The participants in the technical discussions were invited to consider this concept as well as the desirability and feasibility of achieving close links between the various existing and planned information systems in the health services and how this might best be done at the national level. The need for a user-producer dialogue Users of information often find it difficult to obtain infomiation in a rel- evant form and within specified time limits. Why is this? Both users and pro- ducers undoubtedly encounter problems. It is suggested that much of this results from a failure of communication between the two groups. How best can such communication be established and maintained? Is it preferable to arrange top- level discussions between users and producers or is there more hope of progress by discussion at a lower level dealing either with specific subsystems or even with specific topics on which information is currently required? Obviously all l 5 approaches have merit but it would be valuable to have a means of obtaining experience of the best ways of fostering a beneficial user-producer dialogue as a basic requirement for a systematic approach. Planning the development of a health information system Most health information systems have evolved over many years, often as a set of discrete isolated systems designed for different purposes at different levels and often without either formal or informal links between these sys- tems. Is development facilitated by the production, for example, at the na- tional level, of an approved master plan? If so, what should be the basic el- ements of this plan? Evaluation of health information systems This is not an area in which one can calculate cost-benefit ratios or other exact measurements of benefit, since both costs and benefits are difficult if not impossible to calculate, even approximately. In most cases the satisfaction of the greatest number of users is the best indication that can be obtained . REFERENCES 1. Report of the Technical Discussions at the Twenty-sixth World Health As- sembly on organization, strncture and functioning of health services and modem methods of administrative management. Geneva, 16 May 1973 ( document A26/Technical Discussions/5). 2. Handbook of statistical organization. New York, United Nations, 1954 (Studies in Methods, Ser. F, No. 6). 3. WHO Technical Report Series, No . 596, 1976 (Application of systems analysis to health management: report of a WHO Expert Committee). 16 2 THE WHO INFORMATION SYSTEM AND THE INTERACTION BETWEEN NATIONAL HEALTH INFORMATION SYSTEMS A.S. Hiiro 0 The World Health Organization's priorities cannot differ from those of its Member States expressed collectively. The Organization's role is that of a co- operative partner in the national health programmes of Member States, seeking the most effective use of health resources on intercountry , regional, or global levels. It is in this context that the interaction and interface among national health information systems, and between national systems and the WHO infor- mation system (WHO/IS) should be understood. "Interaction" between sys- tems means the exchange of concepts, ideas, and experiences for developing and administering the systems ; "interface" between systems means the transfer of data and/or information in a form that can be processed by both systems. In recent years it has become clear that the development of a health information system faces problems of (a) integrating or coordinating separate health information systems and services at the national level and (b) contri- buting to, and using data and/or experience of, other national, regional, or global information systems. This chapter deals primarily with the international aspects, i.e., the relationships between WHO and Member States vis-a-vis the requirements of intercountry, regional, and global coordination of information exchange, and the direct service by WHO to individual countries. COUNTRY EXPECTATIONS AND RESPONSIBILITIES From the point of view of a national health information system, it is most important that the WHO/IS makes up-to-date information readily available, thus enabling a Member State to study its own position in relation to that of other co untries now and in the recent past, and to determine how it might be ab le to improve that position . Statistical in formation such as crude, standardized, or age-specific mortality tables, numbers of physicians or other health per- sonnel, and technical and other health-related information could be included, together with notes on qualifications and limitations as to its use . It is extremely 0 Director, Department of Planninl! and Eva luation, National Board of Health , Hel- sinki, Finland. 17 important, therefore, that the terms used should be clearly defined for users. This is all the more important where achievement of international standard- ization may not be possible in the present circumstances. Standardization of concepts and terminology is a desirable objective, but awaiting its achievement may prevent much useful information at the international level being produced and communicated. The activities of WHO are based in principle on the decisions of the World Health Assembly, which comprises delegations from all the Member States. It is clear that if the Organization is to implement those decisions efficiently the Member States must meet their responsibilities in providing information. TECHNICAL INFORMATION The technical programmes of WHO have numerous self.evident information needs. Some examples of the types of technical information are as follows. Information on health strategies. For the development of comprehensive health services, information is needed on how different strategies have been tried, and with what resources, to solve specific health service problems in dif- ferent stages of national development. Information on control of communicable diseases. A planned network for epidemiological surveillance on a regional or global scale is necessary to determine the status of specific diseases. Information on efficient prevention and control measures for noncom- municable diseases. These are a major health problem in the developed coun- tries and a growing problem in the developing countries. Information on health manpower. Experience of health manpower planning is important as an integral part of country health planning. An- other area is information on educational and training practices, methods, and materials. Information on adverse effects of drugs. Information on acceptable standards for water and air quality and other environmental conditions. These types of information cannot be collected without effective collab- oration between the information source and the processor (i.e., between national bodies and WHO technical programmes), and those responsible for the latter should ensure that the produced information is properly disseminated . The whole process should be supported by adequate information technol- ogy. The problem with this kind of information system is the handling of input 18 information, wltich may vary within a programme. Tltis is especially true con- cerning the processing, storage, and retrieval of such information. Another problem is the dissemination of relevant information and evaluation of its use. In formation must be available to those needing it in the right form and at the right time. ROUTINE ST A TISTICS WHO collects, processes, and disseminates a great amount of traditional statistical material about health and other matters related to health . The most important items are: (a) population and vi ta! statistics; (b) mortality, including late fetal mortality, by cause, age , and sex ; (c) morbidity and mortality from notifiable infectious diseases; (d) cancer registrations by site, age, and sex; (e) hospital establishment statistics, including catego ries of establishment and occupancy; and (J) health manpower statistics by categories. In addition, survey data on specific topics, such as information on the incidence and prevalence of cardiovascular diseases and congenital malformations, are obtained on an ad hoc basis. The WHO/IS shares numerous common interests with national health infor- mation systems. One very natural expectation is the availability of up-to-date data. The system should enable the Member States as well as WHO to compare the position of different countries in a reliable way. WHO also expects that Member States will develop methodology in order to link health statistics with socio- economic statistics for their long- and medium-term planning. One common problem is the definition of indicators against which trends can be measured and presented. It should also be beneficial for all parties concerned to agree on which methodology and statistical tools should be used to analyse specific prob- lems. Inessential variations often make useful comparisons impossible. In addition to the items of information listed above, Member States have shown marked interest in many others. The following expectations should be mentioned in this connexion : (a) up-to-date information on objectives and targets of current or planned health programmes and projects in other countries, includin g those in which WHO takes an active part, as well as information on their organization, operating methods, and financing ; (b) information on health and health-related institutions undertaking speci- fic research and educational activities related to the Organization's programme ; 19 (c) information on, or a referral service to, health and biomedical litera- ture (e .g., a bibliographical retrieval service); and (d) information on health legislation in other countries, and comparative studies on specific legislative subjects. From the point of view of an information system, these expectations call for : comparability in information storage and efficient retrieval capabil- ities; close coordination between those responsible for technical programmes, public information services and publications, and health and biomedical literature services; and a system of feedback to contributors of information. The most important requirement is the consolidation of a wide range of national health information into a synopsis of the world health situation. This process is the key to interface between national health information systems and the WHO/IS and must resolve problems of relevance, timeliness, com- parability, and accuracy of data. THE ROLE OF WHO Broadly speaking, the rmss10n of WHO is technical cooperation and information transfer. These aims influence, and can be influenced by, health in- formation systems. The main points to be considered are coordination,informa- tion dissemination, idea production, and direct services. WHO has responsibility for coordinating international health matters, including coordination of country efforts in developing linkages among national health information systems. Such coordination should go much further than sharing of methodological procedures and should involve, where possible, the basic concepts and strategies of system development. Strengthened links among national health information systems should avoid costly development of single- purpose information systems by many individual countries; a good example is the setting up of an intelligence/surveillance network in the form of WHO col- laborating centres for specific diseases. WHO serves as "neutral ground for absorbing, distilling, synthesizing and disseminating information that has practical value for countries in solving their health problems".0 Information on health may take the form of: (a) production of publications and documentation (monographs, period- icals, reports, etc.); (b) other forms of presentation (films, radio programmes, slides, etc.); (c) answering of enquiries and requests for specific information; and (d) holding of meetings and other forms of interactive communication. The WHO/IS will have to give technological support to these activities. a WHO Official Records, No . 236 , 1976 , p. XIII. 20 Observation of a wide spectrum of international health issues and national programmes places WHO in a unique position to advance new approaches to health problems ; the WHO/IS supports this with information technology. A good example is the new Special Programme for Research and Training in Tropi- cal Diseases, where multinational experience and research resources are being brought together. Another example is monitoring of side-effects of drugs. Even in routine data collection/dissemination activities there are abundant opportunities for WHO to rationalize its own information requirements, to the consequent advantage of national health information systems. Steps to be taken might include: (a) reassessment of the value of information fed back to countries ; (b) determination of what modification or rationalization, if any, is required in the content of input information, and methodological aspects of collection and transmission of information; duplication of requests for similar information from different countries is a problem that still requires consider- able attention; (c) the drawing up of careful plans for requesting information so as not to create unnecessary financial and administrative burdens for Member States; and (d) increasing collaboration and cooperation with other international agencies in respect of health-related information. EST AB LIS HM ENT OF A ST ANDING COMMITTEE ON IN FORMATION Within WHO, information systems are the responsibility of the informa- tion systems programme. However, almost all other programmes, and in particular those concerned with health statistics and health and biomedical information, which includes WHO publications and health literature services, deal with information in one way or another. ln order further to develop and strengthen coordination of WHO's activities in these fields, both in Member States and in the Secretariat, the Director-General has set up a Standing Com- mittee on Information to : (a) advise on and develop principles for the further development and rationalization of information systems and services, both in Member States and in the WHO Secretariat at all organizational levels; (b) coordinate all WHO information activities ; (c) review the necessity, relevance, and usefulness of information being collected and disseminated , or proposed for collection and dissemination, by WHO; (d) coordinate the Organization's technical cooperation activities with respect to the development of national health information systems and their relationships with the WHO information system ; and 21 (e) solve urgent and day-to-day problems ansmg from all information questions in which WHO is involved with a view to promoting concerted action and preventing conflicts and overlapping. In carrying out the above responsibilities the Standing Committee collaborates fully with the regional offices, which have similar mechanisms. RATIONALIZATION OF SPECIAL -PURPOSE INfORMATION SUBSYSTEMS In support of some aspects of the role ofWHO , anumberofspecial-purpose information (SPI) subsystems have been developed over the years, without close coordination existing between them. An SPI subsystem is technically oriented to support one or more aspects of the Organization's programmes, in- cluding international exchange of health information and information transfer. Strategy for the implementation of the new WHO/IS includes rationaliza- tion of existing, and the development of needed, SPI subsystems. A recent survey at WHO headquarters of all SPI subsystems identified a number of these subsystems and quantified related problems. Based on an analysis of the fmdin gs of this survey , a number of significant recommendations were made , and their speedy implementation was approved. They include: (a) development of a central service for the acquisition , storage , and retrieval of country health information ; (b) study of the characteristics and requirements of scientific surveys carried out in response to World Health Assembly or Regional Committee resolutions with a view to developing standardized procedures for the collec- tion , validation , storage , and analysis of country and other data ; and (c) rationalization of the rosters , directories, and nonbibliographical reference subsystems maintained in WHO into fewer centrally accessible ones through development and implementation of a common technological, or other , support for these. TH E WHO HE ALTH STATISTICS PROGRAMM E The main objective of the WHO health statistics programme is cooperation with Member States in the development of adequate statistical information support required for the management of their health services. National health statistical services have often tended to operate as independent information services relying heavily on routine statistics collected through established ad- ministrative channels. In recent years it has been increasingly emphasized that statistical services should work in a more dynamic and flexible manner to 22 assist directly in health service management. In order to fill the serious gap between the information requirements and the statistics currently available , countries have been reorienting their statistical services . In line with these changes, a WHO programme for the promotion of better national health information systems is being formulated, and health statistical services should be integrated into these systems . WHO's programme of technical cooperation in tJtis field, now operating in about 50 countries, has been reoriented in this direction. Part of the WHO health statistics programme is devoted to a methodological support service for various health programmes, covering statistical and math- ematical methodology and systems research. WHO also assumes responsibility for developing, establishing, and improving international standards in health statistics, among wltich revision of the International Classification of Diseases is most significant. Another statutory function of WHO in statistics concerns collection and dissemination of international health statistics. Two major publication series are issued regularly, namely World health statistics annual and World health statistics report (as from 1977 published quarterly instead of monthly). The former contains basic statistics while the latter puts emphasis on analysed and interpreted information. In addition, unpublished statistical information is dissentinated to Member States on request. NONSTATISTICAL PUBLICATIONS The WHO technical publications programme has two parts: periodicals (WHO chronicle, Bulletin of the World Health Organization, and International digest of health legislation) and other technical publications (Technical report series, Monograph series, Public health papers, Offset publications, etc.). These are all placed on sale and issued in both English and French, either in separate editions or bilingually, and there are Russian and Spanish editions of a substantial proportion of them. In addition , WHO also produces a large volume of technical information in the form of documents for meetings, study groups, consultations, and other purposes. These are for internal use, although they may also have a lintited distribution outside WHO ; they are not for sale to the public. This output from headquarters is augmented by both publications and documents from the regional offices and from the International Agency for Research on Cancer (!ARC) in Lyons. Taken together, this body of periodicals, publications, and documents constitutes a vast storehouse of technical information ac- cumulated and dissentinated by the Organization. One problem has, however , grown to considerable proportions and a solution to it is urgently awaited: an appropriate mechanism for the systematic indexing and retrieval of all information produced in written form by the Organization. At present all WHO publications can easily be identified by means of the cata- logue of publications, which is issued periodically together with supplements, or still better through the bibliography of WHO publications that appears at 5-year 23 intervals. There is, however, no cumulative list for all documents produced at headquarters or in the regional offices, and as a consequence a large proportion of the information issued by WHO is difficult to identify and retrieve. BASIS FOR INTERACTION AND INTERFACE Interaction and interface among national health information systems and between national systems and the WHO/IS can be considered on two levels: 1. Information systems technology (means, devices, methodologies). 2. Information content, i.e., the scientific, technical, or managerial information processed by these systems. Interaction presupposes communication and the capability to react; it means influencing and being influenced . On the information systems technology level , interaction first means learning from the failures and successes of others in building information systems. It further means building the different systems in such a way that they can communicate with one another. On the information content level, interaction means that information provided by one system influences not only the decisions of the immediate users of that system (inside a country) but, in conjunction with information from a second system (another country), it may also influence users of the second system. It would be realistic to recognize, however, that at present much potential interaction is compromised by generally "uncoordinated" developments of national health information systems both within and between countries. Interfaces between information systems provide the basis for, or facilitate, interaction. On the information systems technology level interaction may take the form of common questionnaires or compatible computer-readable storage media such as magnetic tapes. On the information content level there may be agreed nomenclature, standard units of measurement, or agreed rules. SUMMARY AND DISCUSSION Communication from and to health information systems Generally speaking, the expectations of Member States as well as WHO have been met. Improvements in these activities are continuously being sought and are therefore part of the ongoing responsibilities of WHO. An indication of current and planned developments in WHO towards closer interaction between national health information systems and the WHO/IS is presented . 24 The importance of communication has been stressed throughout this chapter, since without communication no effective health information system can exist. The user is concerned to obtain the information he wants in a suitable form, but it is also important that data should be provided in an appropriate manner to the system. In seeking data, the user must not place an excessive burden on those providing it. This implies that data should wherever possible be collected from , and used for , as many purposes as possible, and that there exists an efficient information rationale within country health services and also between national administrations and the WHO/IS. Relationships between national health information systems and the WHO/IS The WHO/IS can be divided into two major parts : I. Information subsystems about the Organization 's programmes (i.e., management-oriented subsystems) referred to as the programme information system. 2. Information subsystems for the Organization's programmes (mainly technical and scientific information subsystems). Interface and interaction between WHO and Member States are particularly emphasized in the latter category of subsystems, because only when there is good communication can the WHO/IS fulfil its main function , and at the present time such communication is not entirely satisfactory. Interaction on - the collection and use of information, and interface of technological support and exchange of experience in overall systems development , also deserve atten- tion from policy-makers and information system planners. 25 3 INFORMATION SYSTEMS IN THE HEALTH SERVICE OF TH E FEDERAL REPUBLIC OF GERMANY C. Griesser0 Because of its decentralized political structure and three governmental levels (federal, state, and county or large city), the Federal Republic of Ger- many has divided responsibilities for health affairs (Table 1). Responsibility for the public hospitals rests with the authorities of the counties and cities that administer them, whereas the federal states are responsible for the uni- versity and mental hospitals. In addition, there are numerous hospitals run by religious and other charitable organizations, as well as other private hos- pitals . Outpatient care is provided by panel doctors in free practice, organized in regional panel doctors' associations as autonomous public-law corporations. These doctors are partners of the various institutions that run the statutory health insurance schemes, such as local and other health insurance funds, which are also organized into regional and federal associations . The wide distribution of responsibilities within the health care system is exemplified by the structure and function of the health information system, which is organized hierarchically and run by the official statistical services of the counties/cities, the states, and the federal government. These statistical services collect health care data from various sources and produce health statistics based on federal or state law. Although electronic data processing is widely used, there is a considerable delay between the generation of the data and the publication of the aggregated health information. Because of this, its value to health care planners is rather limited and consequently the necessary feedback is reduced (1). Within the field of health information, two different groups of statistical reports can be distinguished : those based on federal health law and those based on the statutory requirements in the Sozialgesetzbuch-Reichsversiche- rungsordnung (RYO) (2). a Director , Department of Medical Statistics and Documentation, Hospital of the University of Kiel , federal Republic of Germany. 27 Table 1. Distribution of responsibilities in the health care system of the Federal Republic of Germany Authority Federal level Ministry for Youth, Family Affairs and Health Ministry of Labour and Social Affairs State level Ministry of Social Affairs Ministry of Education Community (county /city) level 28 Responsible for: Preparation of federal laws concerning hospital affairs Legislation for: 1. medical education 2. education of nurses and paramedical personnel 3 . scales of medical fees Supervision of public health services (con- trol of epidemics) Federal statistics of health Preparation of laws for and supervision of : 1. statutory health insurance including ambulatory health care and preven- tive measures by panel doctors 2. statutory pension insurance 3 . statutory occupational accident in- surance 4. rehabilitation 5 . assistance to war victims Preparation of state laws or other legal regulations of hospital affairs Supervision of education of nurses and paramedical personnel Supervision and performance of the pub- 1 ic health service Supervision of : 1. State Board of Physicians 2. Regional panel doctors association 3. Health services of counties/cities 4 . State statistics of health Medical education by universities and supervision of the university hospitals Performance and control of : 1. hospital care 2. public health service HEALTH STATISTICS The following health information is provided by the state and federal offices of statistics and published annually in their statistical yearbooks (3, 4). Information on births, deaths, and morbidity 1. Statistics on causes of death (5) using the 3-digit code of the Inter- national Classification of Diseases. These are based on confidential death certifi- cates and indicate the absolute nwnbers of deaths and the rate per 1000 per- sons by age and sex, including (a) tabulations of maternal mortality by causes of death and age; and (b) tabulations of infant mortality by age, sex, and selected causes of death. 2. Birth statistics according to the age, family status, and nationality of the mother. 3. Statistics on communicable diseases, including tuberculosis (incidence of acute forms). 4. Statistics on venereal diseases, based on anonymous notification by affected persons. 5. Statistics on morbidity. Some insight into the morbidity of the popula- tion of the Federal Republic of Germany is achieved by a periodic census on a 0.1 % random sample, during which additional questions are asked on health conditions (previous and present diseases and accidental injuries, and accidents and intoxications in children up to the age of 15 years). In a few states, such as Hamburg and Saarland, there are cancer registries. Few other statistics of mor- bidity are produced, except in Schleswig-Holstein. In that state, data on inpa- tients are collected anonymously from the participating hospitals by the state of- fice of statistics, using a discharge summary form as a minimum basic data set (6). These statistics of morbidity, collected with the agreement of the hospitals taking part, contain data on sex, age, area of residence and paying institution, and the discharge diagnoses of patients treated in community, denominational, some private and state hospitals (mental hospitals and one university hospital). About 70% of the beds for acute hospital care and about 85% of the inpatients treated in Schleswig-Holstein are covered. These statistics of morbidity, when related to the general population statistics, give a very good indication of the demands of the population for hospital care and of the efficiency of the hospitals (7; Bridgeman, R.F., unpublished data, 1975). 6. Statistics on disabled persons are a good example of a useful com- bination of health information and its practical application to health care. 7. Statistics on compulsory examinations of young people starting work for the first time give some insight into the health of adolescents. 29 Information on health facilities and personnel 1. Statistics on health personnel, giving numbers of: - physicians by age, sex, specialty, and place of work; - dental surgeons, classified in the same way; - pharmacies; - pharmacists; - veterinarians ; - hospital nurses (female and male); - paediatric nurses; - midwives; - medical technicians ; - others. 2. Statistics on hospitals, reporting: - number of hospitals according to function, size and administra- tion (government, communal, denominational, private); - number of hospital beds for acute and chronic care; - nursing homes; - number of discharges and bed days ; - average length of stay; - number of deaths and autopsies in hospital ; - number of births in hospital ; - numbers and composition of the hospital staff. 3. Statistics on the schools of the health care system, containing: - number of approbations of physicians, dental surgeons, veterinarians, and pharmacists ; - number of nursing schools ; - number of hospital nurses (female and male) in education; - number of midwifery schools; - number of midwives in education. The activities of the public health care system are reported by the public health officers of the counties/cities and states to the Federal Ministry for Youth, Family Affairs and Health in the form of an annual report. 30 Medical literature information system A medical literature information system has been made available by the Deutsche Institut fiir medizinische Dokumentation und Information (DIMDI) in Cologne, which is supported financially by the Federal Ministry for Youth, Family Affairs and Health. The DIMDI computer-supported information sys- tem is based on the MEDLARS system and offers either off-line services for answering medical literature requests from research workers or an on-line serv- ice by direct connexion to the libraries of some universities. THE HEALTH INfORMATION SYSTEM WITHIN THE STATUTORY SOCIAL SECURITY SYSTEM "The extent and the complexity of the tasks to be carried out in the interest of increased social benefits" (8) led to the construction of an informa- tion system which is extensively computer aided. The insured person is identi- fied by a unique number applicable to all branches of the social security system. The information system is regulated by law and its structure reflects that of the social security system. Statutory health insurance The statutory health insurance institutions are organized in the form of a well-distributed system following various principles of classification. The sys- tem serves salaried employees, workers, and their dependants (next of kin). Owing to the statutory nature of health insurance up to an annually defined premium assessment limit, and the possibility of voluntary insurance, about 55 million inhabitants of the Federal Republic of Germany participate in the health insurance system. This is offered by 1450 health insurance funds organized in the form of self-reliant public-law corporations. They are classified as follows: Local health insurance funds, organized according to area (county/city). As the affiliation of the members depends on the employer's address, the politically defined population does not necessarily correspond with the mem- bership of the sick funds. Guild health insurance funds for craftsmen, organized regionally. Company health insurance funds, mostly arranged by large, often supra- regional private or public enterprises. Farming health insurance funds, organized regionally. Health insurance societies, which are organized corporately and cover the whole country. The members of certain professions, such as commercial 31 clerks, employees of banks , civil servants, technicians, civil engineers, and gardeners, are entitled to membership of these societies instead of the ap- propriate local health insurance funds . The system of statutory health insurance is complemented by numerous private health insurance companies working on a commerical basis. Statutory pension insurance The regionally organized pension insurance funds for workers (Landes- versicherungsanstalt, LVA) and the centralized federal pension insurance fund for salaried employees (Bundesversicherungsanstalt fiir Angestellte, BfA) also play a role in the health care system, especially in the treatment of tuberculosis. They are also involved in certain rehabilitation measures designed to avoid an impending, or to overcome an existing, permanent incapacity in pursuing certain (learned) professions. Employers' liability associations These associations (Berufsgenossenschaft, BG) are organized by regions and industrial branches and are financed by the employers. They ensure indus- trial safety for employees and workers and provide outpatient and hospital care for members suffering from occupational disability or disease through specialist physicians, surgeons, and hospitals. They also provide rehabilitation facilities. Miners' provident fund This fund provides insurance against occupational injury and disease, as well as being the miners' health insurance fund. THE HEALTH INFORMATION SYSTEM OF THE STATUTORY SOCIAL SECURITY INSTITUTIONS Social security statistics From acquired data, statistics are produced periodically containing certain health information. Health insurance statistics are derived from medical certifi- cates of temporary disablement or from hospital discharge forms. Since the certificates of disablement only apply to those employees covered by health insurance, the statistics so derived are incomplete and biased as indicators of morbidity. Bias also arises from the fact that the population insured by the various sick funds does not correspond with that of the political areas. Any statistical inference about the health status of a given population is therefore impossible and , if applied, misleading. Another factor that may limit or distort the information obtained is the diagnosis supplied by the treating physician on the medical certificate used by 32 the health insurance institutions. First, the physician may be moved to observe confidentiality under certain circumstances, especially in cases with so-called "sensitive" diagnoses ; there is an understandable conflict between an individual's right to privacy and the need to know the characteristics of the insured popula- tion. Second, the ability to make consistent and adequate diagnoses undoubtedly differs among physicians. Statistics on health insurance funds are collected and compiled by the re- gional and/or federal associations and reported to the Federal Ministry of Labour and Social Affairs (9). Statistics on pension insurance funds (10) as related to health care concern the number and results of rehabilitation measures, including costs, and the number of cases of premature permanent disablement. Whereas the amount of medical data used in the health insurance institutions is rather small (diagnoses and rendered health care services) , the pension insurance funds and the em- ployers' liability associations need and use more detailed medical information, such as data from the medical history and medical findings. These data enable those institutions to rehabilitate the insured person as efficiently as possible. However, the medical information used is more case-related and does not necessarily enter the health information system. Statistics on employers' liability associations (J J) contain the numbers and types of occupational injuries and deaths, including causes of injuries ; the numbers and types of occupational diseases ; and the results and expense of rehabilitation measures. Statistics on the LV A, Bf A, and BG cover a more selected portion of the total population than those on the health insurance institutions and therefore permit only very limited conclusions with regard to morbidity. However, the statistics on occupational accidents and diseases are complete and give very valuable information on the causes of occupational accidents and diseases as well as on medical (diagnoses) and personal (sex, age , occupation, etc.) facts. TH E INFORMATION SYSTEM FOR OUTPATIENT CARE Each patient insured by a health insurance fund uses a medical card for treatment by a panel doctor which is valid for three months in each year. These cards are used by the physicians as a kind of short documentation of medical services rendered and for accounting purposes. The medical cards contain, be- sides the identification of the patient (name, sex, social security number), the diagnosis and the nature of the medical services rendered to the patient. At the end of each quarter the completed medical cards are sent to the responsible health insurance institutions via the panel doctors associations. Statistics on the panel doctors associations Statistics are made up from the derived patient data , which has been made anonymous , and used in surveys of the services rendered both by individual 33 panel doctors and by certain categories, such as general practitioners ( es- pecially in the countryside), surgeons, and gynaecologists. These surveys cover the number of cases treated every quarter, the services rendered, the individual panel doctor's fee and a calculation of the average values and standard devia- tions of these variables. These results, related to the individual panel doctor and not to the patient, are computed and a projection is made for the sub- sequent quarters of the year. Furthermore, the aggregated statistical health information is collected periodically by the Federal Panel Doctors Association and used, inter alia, to prepare reports to the Federal Ministry of Labour and Social Affairs and for the yearly negotiations with the different health in- surance associations. Statistics on preventive measures According to the RVO, insured persons are entitled to preventive examina- tions (infants for malformations, etc., adults for cancer). The examining panel doctor reports the results to his association, which derives statistics from the medical data and sends them to the corresponding associations of the health insurance funds according to contractual regulations (12). Combination of health care and social security information The health insurance funds carry out ongoing trials to acquire person- related health care information to construct data bases. These are intended to provide a better insight into the medical and financial demands of the insured community and on the morbidity of the population. However, when it is not the patient himself but his medical card that is counted, there may be multiple countings since, for instance, a person suffering from a chronic disease may need more than one medical card per year or a patient may be transferred by his family doctor to one or more specialists during a period of treatment. CONCLUS IONS Because of the organizational structure of the Federal Republic of Ger- many and the resulting distribution of responsibilities in the health care system and the system of social security, there is no single comprehensive health in- formation system but a variety of single-purpose health information (sub-) systems. The type of data collected differs according to the purpose for which it is to be used . In addition, the reference population may change; for example, the number of inhabitants may be used in statistics of mortality whereas the number of insured persons (with or without their dependants) is used by the health insurance institutions. These facts make it impossible for the various health statistics produced to be compared with one another and lead to different interpretations of the 34 statistical results. Although all the information necessary for a reasonably com- prehensive multipurpose health information system is available, there are serious political objections to such a system, one of which is the possible violation of privacy and confidentiality. REFERENCES I. Griesser, G. General statistics and hospital information. In: Kool, G.A., ed. Hospital statistics and a minimum basic data set. Leiden, Netherlands Institute for Preventive Medicine, 1976, pp. 63- 79. 2. Aichberger, F. Sozialgesetzbuch-Reichsversicherungsordnung. Textsamm- lung. 35.-37. Auflage, Munich, Beck, Ergiingzungslieferung, Septem- ber 1977. 3. Statistisches Jahrbuch fur die Bundesrepublik Deutsch/and. Stuttgart, Kohlhammer (published annually). 4. For example: Statistisches Jahrbuch Schleswig-Holstein. Kiel, Statistisches Landesan1t Schleswig-Holstein (published annually). 5. Handbuch der internationalen Klassification der Krankheiten (/CD) 1968. Band/: systematisches Veneichnis. Stuttgart, Kohlhammer, 1968. Band/I: alphabetisches Verzeichnis. Stuttgart, Kohlhammer, 1971. 6. Die Krankheiten der Krankenhauspatienten in Schleswig-Holstein. Kiel, Statistisches Landesamt Schleswig-Holstein (published annually). 7. Griesser, G. & Hedderich, J. Some considerations of multivariate hospital statistics. In: Proceedings of the Symposium on Medical Informatics, Toulouse, March 14-17, 19 78. 8. Schmidt, H. Social information system in the Federal Republic of Ger- many (summary). In: Proceedings of the Second International Conference on Dynamic Modelling and Control of National Economics, Vienna, January 24-27, 1977. Oxford, Pergamon Press, pp. 114-119. 9. For example: Statistik der Ortskrankenkassen in der Bundesrepublik Deutsch/and. Bonn, Bundesverband der Ortskrankenkassen (published annually). 10. Statistik der deutschen gesetzlichen Rentenversicherung. Frankfurt/Main, Verband der Rentenversicherungstri.iger (published annually). 11. For example: Ubersicht uber die Geschiifts- und Rechnungsergebnisse der gewerblichen Berufsgenossenschaften mit Erliiuterungen. Bonn, Haupt- verband der gewerblichen Berufsgenossenschaften (published annually). 12. Vertriige der kasseniirztlichen Bundesvereinigung mit Sozialversicherungs- und anderen Kostentriigem. Koln-Lovenich , Kasseniirtzliche Bundes- vereinigung, Deutscher Arzteverlag, 1976. 35 4 THE EXPERIENCE OF THE USSR A.S. Kiselev 0 In the USSR the Government is responsible for protecting the health of the population, a fact that has made it possible to provide the entire popula- tion of the country with accessible, skilled care, to create an extensive network of public health establishments, and to conduct scientifically based preventive and therapeutic measures on a nationwide scale. The stability of the system, its ability to adapt to internal reorganization, and its dynamic development are founded on the basic priniciples of public health in the USSR. Information systems for the management of public health are constructed in accordance with government plans for developing the national economy and are promoted by the system of health statistics within the general system of statistics developed in the USSR. This makes it possible to provide manage- ment at all levels with information on the state of health of the population and the development of the public health services. Standard registration documents and report forms have been designed for use within the statistical service . In recent years there has been a growing need for a substantial increase in the volume of information used in making management decisions, and the use of computers in public health has considerably influenced this trend. DEVELOPMENT STRATEGY Over the last IO years modernization of the health information system has been proceeding under the aegis of the Ministry of Public Health, in which specialized subdivisions have been set up to coordinate and finance these activities . There are three main lines of development : I. Establishment of a network of health computing centres at national , republic, regional (within republic), and establishment levels. Personnel are trained in analytical computer processing of health information, and standard designs and norms are prepared. In addition, standard procedures and methods of collecting and analysing data and programme packages are developed . a Director, Statistical Centre for Psychiatry, Moscow, USSR. 37 2. Several subsystems of the information system at al] levels of manage- ment (national, republic, regional, establishment) have been built or are in the process of being built. 3. Research is in progress to develop a simulation model of public health. THE PRESENT SITUATION National and republic levels Nonspecific (auxiliary) subsystems were built first. These include the following: 1. A subsystem for recording information about movements of personnel working in the health services. According to a broader programme, information is collected about management personnel. On request, information is issued concerning individuals and groups. 2. A planning and financial information subsystem that stores different information on the provision of finance. On request, different planning tasks are resolved algorithmicalJy. 3. A subsystem for planning drug provision and management of phar- macies that wovides for the processing of requests by consumers for drugs, stocktaking of reserves, control of consumption, drawing up of long-term plans, and so on. Specific health information subsystems have been built or are under development. The health statistics subsystem. This subsystem has been planned in several stages. In the first stage, on the basis of the existing system of registration- report documents, automatic calculations of rates and indices are carried out and an analysis made of the state of health of the population and the state of the public health services. On request, a variety of reference material is issued . The sanitary- epidemiological subsystem. This keeps account of mor- bidity from infectious diseases and analyses the epidemiological situation. On the basis of these data, preventive and antiepidemic measures can be planned. The psychiatric information subsystem. This subsystem is used to collect primary registration documents for each patient in the country. These are used to draw up statistical reports , and various analytical calculations are made in order to assist in determining cause/effect relationships. 38 The scientific- medical information subsystem. This is used to store bib- liographical data. On request, it issues information on specific topics and pro- vides for the publication of bibliographical reference books. The subsystem of standard reference information. This is used to store and, on request, provide information and reference material. Regional level At regional level there is a considerable number of the above-mentioned subsystems and in a number of regions steps are being taken to build integrated medical information systems for those regions ; these will include a large num- ber of subsystems, both specific and nonspecific. An experimental register of mental patients is also being developed at regional level. In a number of large cities, work has started on integrated systems or subsystems of emergency hospitalization and intensive therapy. Institutional level A number of information subsystems are being developed, for example: 1. The information system of a scientific research clinical institute per- forms a wide range of tasks: automated archiving of case histories , automated analysis of biomedical data, statistical processing of the results of scientific research projects, and mathematical modelling of pathological processes. At the same time, it performs management functions such as the management of bed utilization, pharmacies, book-keeping accounts, and salaries. 2. An information system is being elaborated for a large multipurpose hospital. To a great extent, this system repeats the tasks mentioned above. 3. An information system is being elaborated for a polyclinic, its main objective being to build a patient follow-up system depending on the different medical measures employed. AU the above-mentioned information systems (subsystems) are being developed and are becoming gradually more complicated and refined. Problems may be resolved in different ways, which makes it possible to select the most successful variations and recommend them for wider use. CONCLUSION Fig. 1 shows how the health information system is being developed in the USSR. Construction of the health information system has already gone beyond the stage of discussing whether it is needed ; this question no longer 39 Institution V, Q) ~ :J 0 V, ~ "' u C .c Q) u Q) u .., C -0 Q) u C .., ,._ C V, 0 C Q) "' ·;:; Q) 3:: "iii E "' -~ C 0 ,._ Q) .., C. Q) -0 > "' C .., Q) Q) Q) "' "' ~ ,._ .= ~ ~ 0.. FUNCTIONS Republic Regional "iii u C'l 0 .., C Q) E Q) C'l "' C "' ~ National u E Q) ~ C. Q) ·;:; C "' I ~ "' .., Fig. 1. Scheme of health information systems in the USSR . arises. Because of the size of the country and the great number of people involved, the Ministry of Public Health has the overall responsibility for de- veloping the system, but allowance is made for a certain amount of duplication and this flexibility makes it possible to study the effectiveness of different approaches to specific problems. Some systems are developed quickly, others more slowly ; nevertheless , progress in this field is gradually overcoming the problem of integrating the subsystems into a single, unified health informa- tion system. 40 5 THE STRATEGY IN SCOTLAND M.A. Heasman° The National Health Service in Scotland comprises 15 health boards covering geographical areas whose population varies from 17 000 in the Shet- land Isles to one million in Greater Glasgow. Each board is totally responsible for the provision of health services within its area. In addition, a Common Services Agency (CSA) provides services common to all health boards and, where applicab le, to the Scottish Home and Health Department (SHHD), which is the government department responsible for health in Scotland. With certain minor exceptions, each health board receives an annual block grant for its area and, provided national salary agreements are carried out, boards are free to use these grants as they wish. The present health information system in Scotland has slowly evolved from a situation where vital statistics and limited health services data were mainly required for central government purposes , and as a matter of historical record , towards a system by which information for monitoring, evaluation, planning, and management of health services is more readily available in proper form. Strategic planning of the Scottish health information system is based mainly on a report by Bodenham & Wellman.b The Information Services Division (ISD) of the CSA has the task of providing a service for production of information used in the planning, manage- ment , and administration of the health service. It provides computer advisory and support services, and has personnel with statistical and research expertise for collecting, analysing , and interpreting statistics. Other important areas of a total health information system, such as information on research, drug moni- toring, and bibliographical services , are not dealt with by ISD but by other parts of either the central government or the National Health Service. National policy requires that each health board should be responsible for the collection and primary processing of data required for managing the health service in its area. Their work is coordinated by ISO and, when necessary, standardized in order to achieve an efficient local service which, at the same a Director, Information Services Division , Scottish Health Service, Common Services Agency , Edinburgh, Scotland . b Bodenham, K. & Wellman, F. Foundations for health service management. London, Oxford University Press, I 972 . 41 time, forms part of a nationally integrated system for handling the large amount of information required centrally . Because the health boards vary in size there is a tendency to form consortia to provide the necessary expertise and computer back-up facilities for these services. In Scotland , the general policy is for statis- tical and computing services to come under the joint control of information services as a whole , and for information services to be a health service responsi- bility rather than a responsibility of central government or of a central govern- ment statistical service. Main policy issues are , however, decided by SHHD. PRES ENT STATUS ISD consists of 5 statistical branches headed by 3 community medicine specialists and 2 statisticians. These branches are responsible for analysing all health service statistics (except vital statistics) and for coordinating their collection from the health boards. A small research group of social scientists is available to undertake ad hoc studies and assist in the utilization of statistical data. There are also a computer advisory branch , to advise on all health service installations, and a small computer unit. At health boards or consortia with developed information services, there is usually a community medicine specialist who, together with an information services officer , is responsible for the collection and processing of statistical and other data at that level. There is usually a computer installation, and statistical and records skills are available . The routine collection of health statistical data in Scotland can be con- veniently divided into four parts. (a) National vital statistics continue to be collected and processed by the Registrar General for Scotland , which is an independent government depart- ment and is , therefore, not part of ISD, although there are close formal and informal links between the two bodies . (b) Hospital inpatient statistics are collected by means of statistical case abstracts completed on the discharge of a patient. Initial processing takes place at some health boards or consortia and it is hoped to extend this to all health boards or consortia as soon as possible. Computer tapes of data for individual patients are then transmitted centrally. Other individual case abstracts are col- lected for cancer registration , abortion notifications , school medical examina- tions, handicap register , and dental treatment. All these are processed centrally at the present time. (c) Manpower data are collected for all grades of manpower in the Scot- tish Health Service , but are particularly detailed in the case of medical and den- tal personnel. As far as possible, these data are related to that collected for payroll purposes. (d) Other statistics are collected mainly in summary form at health board level and transmitted centrally for collation. They cover areas such as 42 immunization and vaccination status, summary statistics of hospital work including outpatient data, notification of infectious diseases , work in the field of ophthalmology, work of public health nurses , etc. In general, the system works best where satisfactory primary processing has been achieved at health board level, enabling the system to be responsive to local demands. Closer contact with the provider of statistics also facilitates the production of accurate data. The central system is used increasingly in providing data for planning and management purposes, and the response time to requests is satisfactory for data that already exist. On the other hand, it is extremely difficult to introduce new schemes or to undertake ad hoc studies quickly to fill gaps in the system. A general criticism of the system is that there are in- sufficient interpretive skills available to present the data in an optimal manner. Although routine tabulations form the principal source of data analyses, an important part of the service is to re-analyse the data in an ad hoc fashion. FUTURE STEPS Several developments are taking place in information services in Scotland . (a) Some pioneer work has been done on the tabulation of existing data for the use of physicians in the hospital service , enabling them to review their work. These tabulations have not been very effective, however , and work is now in progress to examine alternative systems of providing information that is both meaningful and useful at the level of the individual physician or division.a (b) A system enabling record linkage of data from the individual case abstract scheme described above is now in existence and is used for ad hoc research studies, e.g., in the detection of long-term toxic effects and in the prospective follow-up of individuals involved in cardiothoracic screening pro- grammes. Further amendment of the methodology of the scheme is in progress and it is hoped that it will soon be available to produce longitudinal morbidity statistics as a routine . (c) Work is in progress within ISD together with other interested parties to develop personnel records that can form the basis of more detailed man- power statistical information for use in manpower planning services. (d) At one health board (Tayside) a master patient index is under develop- ment ; this will carry details of a large number of different health service con- tacts and will also be used for payment of general practitioners. The basic concepts involved are fully described by Bodenham & Wellman .b a A group of physicians having the sam e or similar specialties. b Bodenham, K. & Wellman, F. Foundations for health service management. London, Oxford University Press, 1972. 43 (e) Perhaps the most important future development involves the design of a system able to collect and process data on ambulatory care for hospital outpatients and in general practice. Difficulties exist here, however, both in the magnitude of the problem and as far as primary care is concerned, because of the nature of the contract of employment with family practitioners. It is almost inevitable that some form of sampling procedure will need to be introduced but economic circumstances prohibit any active work in this field at present. The last few years have seen active growth in the use of information in Scotland. Nevertheless, the number of individuals who are able to appreciate the potential and imperfections of the system is still limited, and too many decisions are still taken without recourse to information that is available. As indicated above, the health information system as a whole is still not responsive enough to changes in the health system itself, and certainly the lack of data on ambulatory care is an important obstacle to the full development of statistical health information services. Again, as indicated above, health boards that have developed information services are making the most rapid progress but there are still some health boards where such development is rudimentary, despite the fact that some of these have the biggest health problems. Here also, the present economic difficulties are hindering future development . 44 6 THE GENERAL SITUATION IN EUROPE IN 1977 A.A. Weber 0 This paper summarizes the answers received to a "questionnaire on in- formation systems in the health services", which was sent by the WHO Regional Office for Europe to the 32 Member States of the European Region in the spring of 1977. The questionnaire covered the following aspects : definition of the national health information system {NHIS); the existence of an overall development plan for the NHIS ; national procedures for reviewing user requirements and for making in- formation available to users ; aspects of the NHIS that were not satisfactory and the type of improve- ments required , especially those that justify greater attention at the European level. Replies were received from 25 Member States.b In most cases the answer was prepared after consultations between experts from various sectors of the health services such as the central health statistical department , the central health computing services, and the general directorates of the central health administration , as well as from scientific and research institutes , universities, and departments outside the central health services such as the central sta- tistical services. The following sections summarize the answers received. SCOPE OF THE NATIONAL HEALTH INFORMATION SYSTEM The following definition of a NHIS was proposed to the Member States : "An organization of people , machines , and methods interacting together to a Director, Health Info rmation, WHO Regional Office for Europe, Copenhagen, Denmark. b Algeria , Austria , Belgium, Bulgaria , Czechoslovakia , Denmark , Finland , France, German Democratic Republic, Germany (Federal Republic of), Greece, Iceland, Ireland, Monaco, Morocco, Netherlands, Norway , Poland , Portugal, Romania, Spain, Sweden, Switzerland, United Kingdom, and Yugoslavia . 45 provide the necessary data and infonnation on the health situation of a nation in support of the planning and management of its health services". Most replies agreed with this definition. However, several modifications, clarifications, or additions were proposed . Eight countries considered the term "health situation of a nation" too res- trictive and suggested that the definition of the NHIS explicitly mention the factors (social, economic, geographic , etc.) influencing the health situation, the health resources available, and the consumption of services. Others considered the term "machines" too restrictive and preferred "facilities". Some thought it important to indicate that the NHIS is a subsystem of the national socio- economic information system. Based on these comments, a definition that would be generally acceptable would be as follows: "An organization of people , facilities, and methods interacting together as a subsystem of the national socioeconomic information system to provide the necessary data and information on the health situation of a nation and the factors (social , economic, geographic, etc .) influencing it, on the deployment of the health resources available, and on the utilization of those resources in support of the planning and management of the health services." It was made clear that a NHIS as defined above is to be considered the ideal, ultimate goal, but for practical reasons it is necessary to consider a more limited, medium-term objective. PLACEMENT OF THE HEALTH INFORMATION SERVICES In practice there is no single service responsible for the collection, pro- cessing, storage, retrieval, and analysis of "health information" in its broadest sense. In most countries the information services operate on four levels: national, subnational or regional, local, and institutional. In most countries, the information or statistical department within the national health administration deals with most of the information that is cen- trally collected and processed. This department coordinates the activities of other health and health-related information services, usually in collaboration with the national statistical service. Frequently, the main subdivisions of the central health services, such as the central hospital administration and the central personnel administration, collect and process their own information, though not always in a fully coordinated way. Universities and research and scientific centres often collect health information, mainly on an ad hoc basis and often through research programmes. Social security or health insurance schemes are also involved in the pro- duction of health information, as are other services in different ministries. In countries with a large private health sector, information concerning the activity of such private services is usually collected direct by these services. In coun- tries with a federal structure , the health information services are usually well 46 developed at the regional level. In other countries, the importance of health information services at the regional level (as well as at the local level) depends to a large extent on the responsibilities delegated by the national administra- tion. Whereas the work and responsibility of the information services attached to regional or local administrations is usually more or less standardized , the services and procedures for the collection, processing, and analysis of basic in- formation at the institutional level are not standardized or formalized in the majority of countries. EXISTENCE OF A DEVELOPMENT PLAN OR "MASTER PLAN" FOR THE NHIS There were important differences from country to country in the meaning given to the term "master plan". Slightly more than half of the countries indicated the existence of a development plan for their NHIS that had been approved or endorsed by the national authorities or was being drafted or prepared for approval . This included all the East European socialist republics and most of the Nordic countries. PROCEDURES f-OR REVIEWING INf-ORMA TION REQUIREMENTS Evaluation procedures to review information gaps or to stop or modify the production of statistical series that have become obsolete were recognized as an important but still neglected aspect of the work of information services. Most of the countries that answered this question have some mechan- ism for ascertaining the information requirements of categories of users as well as for assessing the adequacy of the information available to meet those require- ments. These mechanisms are usually informal and ad hoc. Several countries mentioned informal discussions between users and pro- ducers of information . Eight countries reported sending questionnaires or a draft data collection and tabulation plan to producers and to known users , often on a sample basis to see if this met their possibilities and requirements. PROCEDURES f-OR DISSEMINATING !Nf-ORMATION TO USERS Most of the routine information generated locally as a by-product of the activities of the health services is used locally for the management of those services. Although greater use of such information could be made by local managers , dissemination of information to them is usually straightforward. The 47 problem is more complicated when information is to be reported to the regional and central levels, and especially when such information has to be fed back to local users in a form directly useful to them and in a reasonably short time. Almost all central health information services, and to a lesser degree regional and local services, regularly publish the information collected in the fonn of statistical yearbooks, annual reports, etc. These publications usually have a rather fixed content and are meant to satisfy the basic requirements of a large variety of users. Most of these services also provide basic tables or other statistical presentations for reports on the functioning of health services presented by the chief medical officer to the legislative bodies. Several, if not all, of these information services also attempt to present special reports, mono- graphs or studies on a specific subject, either in one periodical series or in in- dividual papers issued as internal, informal documents. Most of the national information services produce lists or catalogues of their routine and ad hoc publications and make them widely available. It is now recognized that a greater effort must be made to arrange the data processed and stored in information services in appropriate data bases or data banks so that questions asked by a broad variety of ad hoc users can be answered rapidly. The participation of the staff of the information services in task forces, committees or groups concerned with the planning, monitoring, or evaluation of a health programme, has been reported in several instances to be an ex- cellent procedure for making statistical services aware of information require- ments and, conversely, users aware of existing information of direct value and relevance to their problems. AREAS FOR IMPROVEMENT AND COORDINATED ACTION The answers to the questionnaire indicated a large number of aspects of national health information systems that required improvement. Many of these were also mentioned during the technical discussions of the WHO Regional Committee for Europe in 1977. Procedures and organization Insufficient data linkage and integration has been one of the most fre- quently mentioned topics. It goes from difficulties of linking information related to the same person by the use of a unique national identification num- ber to the problems of adequately integrating different series related to a single topic such as mother and child hygiene, especially when the various series are collected by different services. The difficulty of relating information on de- mand for care and on provision of services, lack of compatibility betweeen hospital service statistics and information on the health status of the com- munity, and the impossibility of relating results of ad hoc studies with routine statistics were all mentioned as typical examples. 48 The importance of feedback in informing producers of data and thereby increasing their interest and improving the quality of data collected was stressed , as was the inadequate identification of user requirements in many answers, such as the fact that statistics are too input-oriented. Special technologies Insufficient relevance , timeliness , completeness and quality of data col- lected were mentioned by several countries as one of the main drawbacks of many statistical series. The use of proxy measures and information on their value and limitations, the improvement of diagnostic facilities and uni- fication of diagnostic criteria, the development of more appropriate methods of data validation and analysis, more frequent use of sampling, and improve- ment or better utilization of diagnostic facilities and laboratories were all mentioned as examples of techniques or methods that would improve the relevance and quality of data. The increased use of computers to handle large amounts of medical re- cords has made the medical profession and the public even more conscious of the important issues of privacy, safety and protection. It is generally agreed that computerized record systems could be at least as safe as traditional record handling methods, but the large amount of data involved as well as the facility of linking information are creating new problems. However, many research workers were concerned that recent legislation on data protection prevented linkage of personal data and therefore made epidemiological research very cumbersome, if not impossible. Much of the experience so far in computer-assisted preventive, thera- peutic or diagnostic procedures has not provided conclusive results, and the cost-effectiveness of such procedures is very difficult to assess . Interesting possibilities are created by the availability of cheap, "small" computing sys- tems with on-line data acquisition and validation as well as file interroga- tion through interactive terminals , which could be placed close to the source of data or to the decision-makers. Nevertheless much remains to be done in the use of systems analysis in the medical decision process and in the stan- dardization of definitions, criteria and classification. Specific health problems and programmes The absence of valid information on the health status of population groups, as well as on the needs for health care and on population require- ments, call for special morbidity and attitude surveys to supplement routinely available statistics on utilization of services. In most countries the utilization of primary care services is very poorly documented as are most outpatient serv- ices. This was mentioned by several countries as one of the largest gaps in current health statistical systems. Whereas currently available hospital statistics provide in many coun- tries a reasonable picture of hospital utilization , the functioning of the hos- pital is almost everywhere much less satisfactorily documented. The work 49 and functioning of emergency and casualty services, radiological diagnosis departments and laboratories as well as the types of patient using such facili- ties, are poorly documented. Information for health planning on items such as the cost of health services by source of funds, by type of beneficiary or by health delivery programme is very incomplete and such data need to be collected, for example, within the System of National Accounts (SNA) . Many countries reported incomplete information on active physicians and even less accurate or complete information on nurses and paramedical personnel. Although the in1portance of reliable manpower projection for rational planning was mentioned, many countries complained about the quality of available information for meaningful projection, as well as the absence of manageable models. There were comments on the insufficient data available on pharmaceutical products, in relation both to consumption and to side effects. Similar remarks were made on the insufficiency of information on the consumption of toxic substances as well as on their degree of toxicity. Finally, several answers referred to the lack of satisfactory information on the quality of the environment, such as indicators of pollution and the great pau- city of statistical series linking pollution and disease prevalence or incidence. so 7 REPORT ON THE TECHNICAL DISCUSSIONS Munich, 6-10 September 1977 Most health administrators now feel the need for information to assist them in their decision-making and monitoring functions but few, if any, would say that their needs are adequately met by the health information systems that currently exist to service them. Some of the ineffectiveness of health informa- tion systems results from communication failures involving physicians , admin- istrators and statisticians. The first two groups have very different needs from each other, and both have difficulty in communicating their needs to the health statistician ; the statistician, for his part, is aware of the shortcomings of his data in providing answers to questions, even when such questions are not vague and imperfectly fom1ulated. Moreover, data collection itself can be dauntingly difficult. Some dissatisfaction arises from the unfulfilled promise of computers. The early years of health computing gave rise to high hopes that they would facilitate the use of information in decision-making , yet health services per- sonnel and the computer manufacturers have, together, failed to produce solutions that come up to these expectations. This has led to disenchantment and frustration . It was around these problems of communication and of the inappropriate- ness of the infom1ation provided that much of the technical discussions centred. No catch-all solution was found; indeed, no such solution exists. Nevertheless, hope was expressed that awareness of the problems and mutual understanding of each other's difficulties will, in th emselves, go some way towards solving the difficulties. TYPES OF HEAL TH INFORMATION Health information usually starts with the patient-doctor contact but may begin , for example, with the employment of an individual in a health authority. At this level , and in other analogous situations in health services, th e information collected is detailed, ofte n imprecise , and poorly structured, ye t the face-to-face contact allows sufficient information to be collected and pro- cessed for the immediate objectives of that contact to be attained. Information of this nature then has to be ftltered, compressed and standardized ifit is to be 51 of use to the administration of the health facility in which the original contact took place. At this level the detail required is considerably Jess than for the face-to-face contact , yet some facts may be required on each. As information moves towards the central health administration, Uuough district and region, the detail required on the individual patient becomes Jess, but the number of patients or staff members on which information may be required is larger. This introduces the need for standardization of terminology and of recording methods so that the data can be easily collated and analysed. Such concepts as standardization are not readily seen as relevant to the phys- ician and others in the front line of health care, even though better information for decision-making should eventually result in better facilities for such people. The connexions between collection of information, decision-making and im- proved facilities are too remote. Much effort, therefore', needs to be devoted to ensuring greater understanding of health information services and their relevance to modern health care. At the same time, collection and analysis of routine information should be restricted to those items that are necessary, or potentially necessary . Collection of information that might simply be "interesting" should be ruthlessly discarded. In general, routine information will be of greater value in answering factual questions and for monitoring purposes and this should be its primary function. Although routine data can also be of great use in the testing of hypotheses, ad hoc investigations involving separate data collection will be necessary in most cases. Health information systems should be made as adaptable as possible. Flexi- bility can only be achieved if data are stored in a basic form so that recombina- tion can be made as appropriate, whether this be for estimating trends, for forming part of a strategic model, or for providing indicators of the state of health of a group or community. Grouped and summarized data are usually too in flexible for many of the further analyses that may be required. A health information system should be an all-embracing functional sys- tem. Many still consider health information as being primarily concerned with statistics and it was to this aspect that much of the discussion was directed. Nevertheless, while this will always remain an important component, a total health information system is much wider and includes scientific and tech- nical infonnation as well as "soft" and even anecdotal data. Most countries of the European Region have important components of a health information system but, at the same time, there was considerable agreement between them as to the shortcomings of these components. CURRENT DIFFICULTIES ASSOCIATED WITH HEALTH INFORMATION SYSTEMS The primary use of a health information system is in management. Yet where good data exist feedback, even to management, is often insufficient, difficult to interpret and greatly delayed. The output often comprises ex- tensive statistical data difficult to understand, particularly by busy people 52 untrained in statistical techniques. What are required are easily understood analyses, together with succinct and appropriate commentary that should not sllirk from explaining the important shortcomings of the data. The trend towards such analyses is increasing, and where this is so there is increasing respect for, and understanding of, the capabilities of a good health information system. Yet from many countries came complaints that little information is made available beyond that of routine publication. It is apparent that some governmental statis- tical organizations make routine statistics very widely available and yet, at the same time, produce little in the way of interpretive general commentaries and are reluctant or unable to produce further analyses in response toad hoc requests dealing with particular problems. These are required for management but they are also required increasingly for epidemiological purposes, ranging from com- plicated tabulation and analysis to a study of individual case records. Many gaps exist in information systems. The most commonly expressed need was for data in the field of primary and/or ambulatory care. Several delegates expressed considerable concern over the lack of data on health per- sonnel. Another area in which infom1ation is generally lacking is that relating to the environment and, in particular , to data required for the detection and control of environmental hazards . Concern was ex pressed over the problems being created for health informa- tion systems by the need for confidentiality, particularly recent and impending legislation on privacy. Increasingly, epidemiological and health services research depends for its basic data on the ability to link different events occurring at different times in an individual's life ; knowledge of the detailed workings of a health service depends on an ability to link records relating to different parts of a health system. Several speakers referred to the need to reach compromises acceptable both to those concerned with privacy and to those who require to use the information. In general, decentralization of information improves privacy but this is not always a blessing, for many of the linkages required for research purposes can only be achieved regionally or centrally. There is a need to inform and educate all health personnel about the need for and use of information in decision-making at all levels. Another problem discussed was the lack of facilities for training in health statistics ; there is a need for special training in health matters for graduate statisticians and econ- omists but more importantly for training those responsible for the collection of basic information , so that they not only understand the importance of their task but are also able to undertake simple analyses. Closely related to training and educational needs is the requirement for more and better consultation between the users and producers of information. No effort should be spared in improving communication between the user and the producer of health information , for it is only in this way that an exact identification of require- ments can be made and the limitations of health information understood. These efforts should take place at all levels of the health services and should occur both frequently and regularly. Some of the discussion focused on the relationship between the health system and the computer. Many hopes have been raised by computer enthusi- asts, including the manufacturers , only to be dashed when information was found to be difficult to collect or to analyse. Despite the undoubted flexibility 53 of the computer, it has not been as successful in many health service situations as had been hoped. There has been talk of "computerizing a hospital"; one person suggested that a better term might be to "hospitalize the computer". Tltis suggests that even more needs to be done by computer people to make their hardware and software acceptable to the physician and health administrator. This may not be easy to achieve. Many computer personnel still ca rry an idealized picture of a health service with clear objectives and relatively simple problems to solve; those in health services know that there are few organizations more complex. Many individuals in health services have idiosyncratic ideas that do not readily lend themselves to the discipline imposed by computers. The optimism of the past has led to the pessin1ism of the present. From pessimism, with good leadership, a new spirit of realism can arise that can result in the computer finclingitsproperplace in the health services of the developed countries of the world in the next decade . MAJOR DEVELOPMENTS REQUIRED The view was generally expressed that too much effort has been concen- trated on hospital information , although even with hospital data very little is availab le to measure either the cost or the effectiveness of treatment policies. Information to answer many of the major strategic problems of the modern health service is nowhere available, and it is obviously very difficult, if not impossible, to obtain. For example, while it is obviously important to find an optimum balance between expenditure on prevention and that on treatment , that in formation only exists in very restricted areas of study. Even where mor- bidity, resource and costing data are available, effectiveness has proved impos- sible to measure. This is at least partly the result of lack of agreement on measures of outcome. Indeed , the only readily available measure of outcome of health care is death or survival. The design of an easily used and widely ap- plicable measure of "quality of life" is one of the most important research questions for health information systems today yet, even though considerable effort has been expended, usable solutions seem as far away as ever. Related in some ways to the need for outcome measures is the need for health status indicators yet, as with the former, lit tie progress has been made ex- cept in a theoretical sense. Perhaps too much has been expected of both of these . Measurement of the health status of a population is more difficult now that mor- tality in the young and middle-aged has been so much reduced, for in all health systems solutions to the problems of morbidity measurement have proved dif- ficult to find. It was suggested that mo re progress might be made if a clearer distinction were made between the measurement of health at a population and at an individual level. There should be recogn ition of the fact that measurement of total health status is an impossible object ive, and effort should be concentrated first on measures that are both easy to make and objec tive in character. At least some of the gaps in current health information might be filled by a concentration of effort on hea lth interview surveys. The suggestion was made that information collected on major causes of morbidity not likely to be treated 54 in hospital might prove to be a significant factor in determinin g health policies. In addition to the possibility of data collection on morbidity, health interview surveys can also be a source of much-needed information on public opinion on health matters, for it is increasingly appreciated that health planning must take such opinion into account, even though objective measurement of the perception of health needs may be difficult to make. It was suggested that some of the problems associated with the collection of data on ambulatory health care might be solved by the establishment of mobile statistical units ; these could visit health facilities such as polyclinics in order to obviate the extra load that data collection imposes on busy staff. One of the ways in which computers can provide an understanding of health problems is their use in simulation by means of mathematical models. Model building is still in its infancy as a product of health information systems, yet it may be that the intelligent use of micro- and macromodels of health systems will do as much as anything else to convince administrators of the value of health information, because it will enable them to understand the possible results of alternative decisions before the implementation of any one of them. Models offer an exciting use of information that can immediately be seen to be relevant, thus doing much to improve communication. The full development of national health information systems will take many years, particularly in the wider context covered in these technical dis- cussions. Yet the message to those present at the technical discussions was clear enough: despite the difficulties the view was expressed that countries should begin to develop outline health information systems. To a great extent each country must decide which health information system is most suited to its needs. Although there will be a large common element, the political, economic, social and health service structures are so diverse that it would be in1possible and unwise to develop a standard system. Each country should make full use of such subsystems as already exist, adapting them as needs become more clearly defined ; to create new sources of data is to add a whole new dimension of difficulty. Full development will take many years and the final shape (if, indeed, this is ever reached) may differ greatly from the initial conception. At the same time it should be remembered that it is remarkably difficult to measure ilie effectiveness of a health information system either in whole or in part. An efficient system should produce a high return fo r low investment and this can be partially achieved by using information for several purposes, such as using payroll information as a basis for manpower statistics. Flexibility of response is an importan t factor in the production of meaningful analyses from routine data or the collection of ad hoc material. Nevertheless , an efficient information system must also hold information against possible need . CONCLUSION A health information system is a functional en tity within the framework of ilie health service as a whole. It cannot exist by itself; if it is to be of any 55 value at all it must be used as a tool by others, whether this be management, research workers, pressure groups or the public. In other health fields the acquisition and use of expensive equipment is actively promoted by both physicians and administrators, even when there is little evidence of its effectiveness. The same people, however, are not suf- ficiently active in promoting the development of a health information system, even though they may appreciate the need for one . The reason for this lies partly in the complexity of the problem, partly in the communication block already mentioned, and partly in the urgency of decision-making, which often cannot await the arrival of barely relevant information. Much lies in a failure to find out what information is really needed, even though, as one speaker put it, "one of the most difficult things is to know what should be known". The technical discussions were useful. They identified problems and, even if few solutions were found, the realization that most of the problems were common to most of the countries represented should itself act as a spur to their solution. The increasing cost of health services and the need to achieve the most efficient use of expensive resources demands that decision- making should rely more on objective information and less on subjective opinion. 56 - - - --- - PART II SOME TECHNICAL AND OTHER PROBLEMS 8 DATA SECURITY AND DATA PROTECTION IN HEALTH INFORMATION SYSTEMS C. Griesser0 The term "health information system" covers a wide ran ge of electronic data processing of health care information. Data made anonymous, aggregated, prepared and issued for health statistics cannot endanger the individual's right to privacy when they are used in a proper and legal way, and this discussion will therefore concentrate mainly on identified data relating to health care. Also , no mention will be made of statistical data, even though these are derived from originally identified data . Identified health care data are those groups of data that include personal information linked by a common social identifier, such as a person's name, his insurance number, his social security number, or any other unique or typical identifier recognizable and/o r decodable by per- sons other than the subject himself. A health information system in the broadest sense may be defmed as an organization that operates in the health care environment. It is composed of (a) hardware , including transmission devices : (b) personnel such as or- ganizers, planners , designers, managers , programmers, operators and users ; (c) software (operating systems and applicat ion programs); (d) organiza- tional rules influencing human behaviour ; and (e) health care infonnation gained from patients. In a narrower sense health information systems are man - computer systems with numerous interfaces between health care personnel and hardware com- ponents such as terminals. This type of system follows the principle of integration in order lo reduce the bulk of the data acquired. Such an integrated system makes health care information available to many users for many purposes (multi- user /multipurpose system). According to their extent , structure and means of data exchange (directly by on-line transmission or by indirect off-line exchange), such systems may affect different numbers of individuals (patients or insured persons), particularly when a common data base(/) exists for one hospital, for a group of hospitals or medical practices, or for some socia l security institutions. The common factor is that identified health care information concerning each individual involved may be processed and stored. Since identified health care data principally relate to the results of physician -patient encounters, there is a need for well-established data protectio n measures. a Director, Department of Medical Statistics and Documentation . Hospital of the University of Kie l, Fed eral Republi c of Germany. 57 The construction and operation of a health information system is justi- fied only when the outlay in terms of money, personnel and organization is worth while in relation to the benefits achieved. Such benefits must be available not only to the health care organization itself by rationalization of work, but above all to the patient, by the provision of better medical and social services. Conflict exists, however , between the patient 's right to privacy and the needs of the health care organization. Therefore, the pos- sibly contradictory interests of patients concerned to preserve their privacy, of physicians in observance of their professional discretion, and of users of health care information must be taken into account by safeguarding the health information system against interference or encroachment from out- side (2) and by clearly defined channelling of information between the insti- tutions and users involved in health care and social security (3). Further- more, the possible threat to privacy depends on the scope of the health in- formation system, i.e., whether it operates in a single hospital or part of a hospital only; at the local level; at the regional level or for a group of institu- tions; or at the national level. DEr-lNITION Of' TERMS IN DATA SECURITY In discussions on threats to privacy and in the relevant regulations and laws - existing or planned - various terms are often used synonymously. It may be necessary, therefore, to settle on definitions based on previous con- siderations and proposals (4-6). Privacy The term "privacy" may be understood as the right of an individual to determine the degree to which he is willing to share with others infor- mation about himself. There is at present the possibility of exchange of such information among other individuals or organizations (5), but this should not be done without the consent of the individual concerned (7). The com- plement to this is the patient's right to secrecy and the protection of his personal data against misuse or unjustified publication inside or outside the health care field (8). Confidentiality The term "confidentiality" may be defined in health information pro- cessing as the professional duty of physicians, nurses, paramedical and cleri- cal personnel, computer personnel and nonmedical research workers in insti- tutions of health care, as well as of social security personnel acting as "third parties" (7), to safeguard the secrecy of information processed by the health information system. 58 Data security Data security is the result of implementing measures to protect data against unauthorized events leading to unintentional or intentional modifica- tion, loss, destruction , or disclosure of personal health care data. Data security is directed to two main targets: safeguarding data integrity and assuring the privacy of patients and physicians. Safeguarding data integrity Data integrity may be understood as the state that exists when computerized data taken from source documents have not been exposed to accidental or malicious alteration, loss or destruction. The objective, therefore , of the main- tenance of data integrity is the availability of complete, valid and reliable data. This means the safeguarding of acquired, stored , transmitted or retrieved data against (a) destruction by hardware failures, software deficiencies, opera- tional mistakes and external physical damage (fire, water) ; (b) theft of recording media ; and (c) falsification. Assuring privacy This may be defined as follows : (a) safeguarding confidential personal health care information; (b) controlling access to synoptic information from collected and stored data, from which the patient's identity has been removed , and which can be reidentified by "sniffing techniques" (9) ; (c) preventing deliberate or unintentional misuse of those data ; and (d) preventing unauthorized copying of data stored or processed in a computer-aided health information system. Data protection This term comprises all measures taken to safeguard health care data from occurrences that intentionally or unintentionally lead to modification , des- truction , loss or disclosure of those data . TIIE MENACE TO PRIVACY AND DATA INTEGRITY FROM HEALTH INf-ORMATION SYSTEMS The safeguarding of privacy is a leading ethical principle and a major pre- requisite for a trusting physician-patient relationship . Nevertheless, privacy in health care and data security have not been absolute and without their prob- lems in the last hundred years or so. Originally, the relationship between the 59 7 ! patient and his physician was one to one (JO) . In the course of the develop- ment of modern medicine one of the first threats to privacy was caused hy one or more persons assisting the physician in his practice or in hospital. Certainly those persons now working in health care - the nurse-secretary in the doctor's surgery, the hospital nurse , medical technicians , hospital medical staff and, less directly, administrative clerks and other hospital employees - were and are under a legal and ethical obligation to keep secret all facts gathered in the per- formance of their duties (J 1, 12). Initially this obligation was strictly observed, but things have since changed and in reality there is not infrequent contraven- tion of professional discretion. Another factor contributing to the expansion of the circle of those from whom absolute confidentiality is expected is the duty of physicians to make notes on their patients. Consequently, since medical records are now stored in ftling cabinets or in hospital libraries, it is not so difficult for an interested but unauthorized person to gain access to such infomiation. The crucial change took place, however , with the introduction of systems of social security, and especially of health insurance run by statutory or private health insurance institutions. These systems act inter aha as payers in place of the patient , and represent a third party. Therefore, a conflict arose between the individual's constitutional right to privacy and the need for access to cer- tain facts about the individual by official, semi-official or private institutions. These health insurance institutions, particularly the statutory ones, operate as managing, controlling and planning authorities. They have to look after the interests of their insured members, and need to know medical facts that originate during the physician-patient encounter in order to function efficiently. This extension of access to private information to third parties is covered either by legal and regulatory restraints or by common agreement. Nevertheless, both the patient as the source and "prime owner" of the information and the physician as the originator and "secondary owner" of the medical data have lost absolute control over that information. Generally speaking the original one-to-one relationship has been changed into a triangular one, with some often uncontrolled channels leading outside the system, as shown by Westin (7). Nevertheless, these conditions have generally been accepted by patients and physicians. Occasional unintentional violations of privacy have not always been recognized; sometimes they have even been taken for granted and excused as human error. Normally only deliberate violations result in prosecution under the laws governing the unauthorized transmission of confidential information. The impact of the computer The introduction of computer-supported health information systems with their extended capability of storing personal health care information in directly accessible data bases, together with the possibility of linking it to other data base(s) storing identified data concerning the same individual , has led to some anxiety outside as well as inside medical circles (8). Indeed, the danger of the concentration of health care information by computer-supported health infor- mation systems cannot be ignored (J 3) and some of the widespread fears are 60 justified. On the other hand, it would be wrong to assume that the advent of the computer is responsible for all the evils possible from automated data pro- cessing. As has been shown , privacy has long been threatened by alterations in health care procedures , and any loopholes in conventional health care informa- tion systems likely to lead to violations of privacy must be discovered and plugged before computerization of such a system is contemplated. Violations of privacy in computerized health information systems have been caused by (a) uncontrolled adoption of structures , functions and human behaviour in conventional systems and/or (b) blind confidence in the supposed technical possibilities and advantages of the automation of administrative pro- cedures, without sufficient attention being paid to the human factor. 1 t may be said that since man's discovery of fire every technological advance has carried its inherent dangers (14), but such dangers must be appreciated and prevented in time to avoid excessive harm. Just as he has learned to overcome the danger of fire, man may obviate the threat to privacy in health care by ap- propriate data protection . Since the relationship of conventional information processing in health care to computer-supported health information systems is Hke that of a "friendly log-fire to a nuclear power station" (J), it is highly advis- able, indeed imperative from both the social and medical points of view, to intro- duce measures to prevent violations of the citizen's right to privacy. Computer-aided versus conventional systems Whereas in conventional health information systems medical record Hnk- age is limited and the transmission of health care data is controlled by the phy- sician(s) originating them, in an automated system there is less, if any , such control. In these circumstances the possibility of misuse of data must be con- sidered; if access to confidential information is not properly supervised its unauthorized use is possible. This may also happen in conventionally structured health information systems if uncontrolled channels of communication exist from the health care system into other information areas or systems . The drawbacks of computer-aided health information systems with regard to privacy should not be excused by arguing that it is more feasible to channel information and to protect health care data against misuse than in conventional systems. On the other hand, it cannot be denied that the introduction of com- puters in to health care has made the medical profession more conscious than before of its professional duty in respect of confidentiality. Concrete threats to privacy Accord ing to Peterson & Turn (15) there are two types of disclosure of confidential information: accidental (unintentional) and deliberate . Unintentional violations of privacy Accidental disclosures include (a) hardware failures and (b) software deficiencies due either to inadequately designed or "debugged" programs or to simple operational errors. These types of error will inevitably occur, but 61 good memory protection in the central processing-unit of the computer and hardware reliability can minimize such dangers. Often accidental disclosure of confidential data is due to carelessness and ineptness on the part of those handling the data, because they fail to realize the importance of proper safe- guards (JI) . This is true not only for the computer centre itself but for the interfaces between computer and man at every stage . Deliberate encroachment on privacy Deliberate violations of privacy are more malicious, and for this reason protection against them is more difficult. Some of these threats are of a passive infiltration nature and include predominantly technical procedures, such as electromagnetic pick-up; wire-tapping ; concealed transmission from the central processing unit, input/output devices or communication lines; and periodical inspection of the contents of waste baskets in computer centre areas, in rooms equipped with remote job entry devices or (printing) terminals, or in phy- sicians' offices. Other threats are of an active infiltration nature and include: browsing, i.e., searching through storage to locate or acquire information without necessarily knowing of the existence or the nature of the informa- tion being sought (5); gaining access to a (protected) system by posing as an authorized user; unauthorized access gained via another user's legitimate connexion; hardware or software loopholes created intentionally, providing unauth- orized access for the purpose of collecting, altering or destroying data; infiltration through active communication channels, e.g., by intercepting and possibly modifying messages; physical means, e.g., theft by an employee of recording media or un- authorized copying of their contents. These many threats to data security may suggest that a computerized health information system is unprotected. However, in most cases the infiltra- tion of any information system using technical devices requires an intimate knowledge of the system in question, time and opportunity to work undistu rbed, and an accomplice on the staff. Therefore the personal integrity of all persons concerned with the health information system cannot be overemphasized. PRINCIPLES OF DATA PROTECTION IN HEAL TH INf-ORMATION SYSTEMS It is possible to make two generalizations: I. Data protection depends on (a) the laws governing data security or legal regulations relating to a particular health information system, and (b) the structure of a particular system and the various specific tasks to be solved . 62 2. There is as yet no impregnable information system. It is important, therefore , to make the security mechanisms very difficult and expensive for an intruder to bypass. It must, however, be acknowledged that these can go too far and may be unjustified if there is a marked loss of flexibility or the cost becomes prohibitive. Accordingly, the methods used in data security must be reasonable, legally and technically practicable , and psychologically acceptable. Whereas , by means of strict management rules, it is relatively easy to establish effective data protection in (off-line) batch processing, problems of data security become more difficult in a complex multi-user/multipurpose interactive system with shared access to data bases. Additional and different aspects must be taken into consideration when the data processing applies not only at the local level, such as the hospital, but at the regional or central level (J 6). Basic guidelines Data protection in health information systems and its organization, realization and control must not be regarded as just a hobby. It requires the ability to be conscious of all the risks that threaten data security. Imaginative thinking is also necessary to foresee those risks in order to trace and counter weaknesses in the system. Above all, it must be recognized that data security and data protection measures are not necessarily confined to the computer centre itself. It is not the intention in this chapter to discuss all the possible measures that can be taken against violations of privacy and confidentiality. It is , how- ever, possible to outline certain basic guidelines for data protection in health information systems that have been developed by several authors or groups ( 4, 6, 7, 12, 16- 18), such as: J. The definition of: health care data and their structure (J 9); the sensitivity of health care data, especially those stored in a com- mon data base . This is specially important if the health information system covers more than one medical service or includes institutions of social security such as health insurance funds; the requirements of the various users of health care information. Other than in conventional systems it is feasible in a computer-aided system to divide the person-related infomrntion into two or more parts according to the needs of the user. There are at least two dif- ferent components: administrative and medical. The administrative clerk, for instance , cannot have access to medical data ; rea l time demands that may determine the mode of access ; the right of use rs to access to data; responsibility for the reliability of health ca re data , i.e ., the phy- sician as originator; 63 integrity, i.e., the responsibility of the systems manager as the keeper and trustee of the health care information . 2 . The complete disconnexion of a health infonnation system from any other general or special information system (2) . This particularly applies when hard- ware is shared. A health information system has to be regarded as a self-contained entity with its own hardware and computer staff, and therefore the assumed advantages of integration with other infonnation systems should not be over- estimated . The confidentiality of highly sensitive data and the patient's right to privacy prohibit a mixed system in which data becomes unprotected. The principle of confidentiality is inviolable (16), and such confidentiality should not be confined to criminal investigation or revenue information systems, but must also be applied to health information systems. 3. The avoidance of processing health care data externally by commercial or official common service bureaux (J). 4. The separation of functions in the computer centre. This means special security measures in the computer rooms to prevent unauthorized entry. 5. The management and supervision of the health infom1ation system by an experienced physician (3). In this way the non medical computer staff (opera- tors, programmers, systems analysts) become colleagues of the physician and subject to observance of the same professional discretion as other paramedical personnel. 6. The appointment of an independent commissioner for the internal auditing of data security (2), including the education of all persons working within the health in formation system and the observance of a code of ethics (J 2). 7. The prohibition of derived (secret) data base files that could be withdrawn from an established control more easily than the "official" files. 8. The provision of periodical security audits, either internal or external. Realization of data protection Generally speaking there are three approaches that must be integrated to prevent the possibility of breaches in a data protection system (2): hard- ware precautions; software techniques in operating systems as well as in app(jca- tion programs; and organizational measures and rules. Since hardware pre- cautions and software techniques must always be applied to an operational health in formation system, it is the responsibility of planners and designers of such a system to prepare and set up clear and feasible organizational regula- tions that guarantee data security. This involves the definition of: (a) categories of data accurding to sensitivity. availability and accessi- bility (on-line/off-line): 64 (b) categories of users of health care infomiation according to their in- formation needs, depending on their functions and responsibilities within the health care system. This confers on them a right of access to certain categories of data according to their real time demands, together with opportunities to modify stored health care information; (c) categories of application program in a teleprocessing system, con- sidering their inherent potential in manipulating health care data and producing new information; (d) categories of hardware (terminals , remote job entry subsystems) con- cerning access to categorized health care information and/or application programs; (e) possible threats to and deficiencies in data security that must rely on an analysis carefully carried out in all parts of the health care system ; and (f) channels transmitting health care information from one part of a (complex) health information system to another. The categories of data , user, application program and hardware can be described by means of several matrices, such as data/user, data/application program, etc. These definitions result in multidimensional system coordinates (2) that describe and represent the health care system in its various dimen- sions, such as: patients, represented by their data; institutions, e.g., care unit, department , hospital, doctor's surgery, health insurance fund; users; hardware; software (methods and rules). Authorization is the leading principle. Security can be accomplished inter alia by passwords, which enable one to verify the authenticity of the user asking for access to the information. The importance of organizational measures ("orgware '') It would not be wise here to discuss the technical aspects of data protec- tion that are described elsewhere (J 6- 18). However, since they have to follow precondit ions established in the planning and designing phase, it might be worth whi le to consider some aspects of organization. A beginning may be made with "orgware", which in effect concerns the design of the information system and the description of the organization of the institution(s) to be serviced. Tasks originally assigned to "orgware" can laier be transferred to soft- ware applications but arc a useful first step towards an automated solution of data protection . It is a sensible principle to build control and supervision into the computer. Nevertheless, human control of the access to and the usage of the in formation is always necessary. It must therefore be the task of key supervisory personnel to control the proper use of passwords. 65 In any system of data security, the human factor cannot be neglected. Since to a great extent organizational measures consist of fairly strong and sometimes rather rigid commands and/or prohibitions restricting individuals' activities, they may provoke psychological reactions even to reasonable regula- tions. It must also be appreciated that despite initial attentiveness and aware- ness of the need for data protection, in the course of time a certain slackness may creep in. These considerations apply to the medical as well as to the non- medical staff of the health information system. It must be remembered that the majority of the computer staff are not physicians, nurses or medical tech- nicians, but mathematicians, information scientists , programmers, operators and medical record librarians. Nonmedical people working in the health care environment must be motivated to observe the same professional confidentiality as that exercised by medical personnel. REFERENCES I. Schneider, W. & Bengtsson, S., ed. The application of computer techniques in health care, with special regard to hospitals. Computer programs in bio- medicine, 5: 171- 249 (I 97 5). 2. Griesser, G. Data protection by organizational means in a hospital in- formation system. In : Proceedings of the IF/P-WG 4.2 Working Con- ference, Kiel, June 23-25, 1976. Amsterdam, North-Holland, 1976, pp. 123- I 32. 3. Steinmuller, W. Datenschutz in R.isikosystemen . Nachrichten fiir Doku- mentation, 28: 74-83 (I 977). 4. IFIP-Working Group 4.2. Considerations on the subject "data protection". Computer programs in biomedicine, 5: 85-86 (I 975). 5. Federal Information Processing Standards Task Group 15: Computer Sys- tems Security. Glossary of terminology for computer system security. 197 5. 6 . Data security - threats and deficiencies in computer operations. A report on a completed study. IBM Svenska Publication G 320-5646. 7. Westin, A.F. Computer, health record and citizen rights. USA National Bureau of Standards Monograph 157 , I 976. 8. Griesser, G. Scope and purpose . In: Proceedings of the IFIP-WG 4.2 Working Conference, Kiel, June 23-25, 1976. Amsterdam, North-Holland, 1976, pp. 1-6. 9. Schlorer, J. Identification of medical records and retrieval of dossiers from a statistical data bank. Methods of information in medicine, 14: 7- 13 ( I 975). IO. Witt, J.L. Peop le in confidence , the expanding circle . In: Acheson, E.D., ed. Record linkage in medicine. Edinburgh, Livingstone , 1968, pp. 333-338. l I. Vallbona , C. & Beggs-Baker, S. Data protection in a community medi- cine environment. In : Proceedings of the IFJP-WG 4.2 Working Con- ference, Kiel, June 23-25, 1976. Amsterdam, North-Holland , 1970, pp. 45-54. 66 12 . Gabrieli, E.R. Ethics of medical computing. In: Shires, D.B. & Wolf, H., ed. MEDINFO 77. Proceedings of the Second World Conference on Medi- cal Informatics, Toronto, August 8-12, 1977_. Amsterdam , North-Holland, 1977 , pp. 729- 730. 13. Wagner, G. Medical record linkage. In: Anderson , J. & Forsythe, J.M., ed. Information processing of medical records. Amsterdam, North-Holland, 1970, pp. 41-54. 14. Acheson, E.D. Linkage of medical records. British medical bulletin, 24: 206-209 (I 968). 15. Peterson, H.E. & Tum, R. System implications of information priv- acy. In: Proceedings of the Spring Joint Computer Conference. AFIPS, Vol. 30, 1967. 16. Griesser, G. et al. Summary and conclusions. In: Proceedings of the IFIP- WG 4.2 Working Conference, Kiel, June 23-25, 1976. Amsterdam, North- Holland, 1970, pp. 180- 195 . 17. Griesser, G., ed. Realization of data protection in health information sys- tems. In: Proceedings of the IFIP-WG 4.2 Working Conference, Kiel, June 23-25, 1976. Amsterdam , North-Holland, 1977. 18. Griesser, G. Technical aspects of data protection in health information systems. In: Shires, D.B. & Wolf, H., ed. MED INFO 77. Proceedings of the Second World Conference on Medical Informatics, Toronto, August 8- 12, 1977. Amsterdam, North-Holland, I 977, pp. 723- 728. 19. Sauter, K. et al. A data structure model for a health information system. Computer programs in biomedicine, 6: 171 - 177 ( 1976). 67 9 ADV ANT AGES AND DISADVANTAGES OF CENTRALIZED AND DECENTRALIZED DATA STORAGE AND PROCESSING A.S. Kiseleva The development of health inforn1ation systems in various countries over the past decade has been accompanied by a trend towards specializa- tion according to the level at which they function, the principal levels being (a) physician-patient , (b) medical institution , (c) area, and (d) regional and national. At the same time, there is also a tendency for th e first two of these levels to be amalgamat ed within the framework of a single medical institution information system . This kind of specialization and separation is governed primarily by the diversity of aims and objectives of management, by the data used, and also by the "product" of the information system. The physician - patient level These systems are intended for the storage in a data bank of machine- produced case histories, results of laboratory analyses, and in format ion con- cerning the dynamics of the patient's condition as affected by treatment. The final aim is to provide assistance to the physician in establish ing a diag- nosis and managing the treatment process . Ideally , the product of the system is machine-produced diagnosis , prognosis, and recommendatio ns regarding the choice of treatment tactics . The medical institution level These systems store administrative data about personnel , indices of staff workload (number of patients per physician, number of visits, etc.), figures re lating to the functioning of institutions (number of inpatient beds occupied, bed turnover, etc.), financial information, and so on. At the same time, they store data on patient fl ow in an institutio n (statistics). The aim of such systems is to help directors of institutions to run those institutions as efficiently as possible. The end product consists of various statistical and financial summaries for the institution's administra tion, and possibly also optimizational and modelling tasks. a Director, Statistical Cen tre for Psychiatry, Moscow, USSR . 69 The area level At the area level, the health information system is concerned with the management of all the health care institutions within that area. There is a great similarity, therefore , between the area level and the institution level in the sense that items of administrative information (statistical summaries, reports, financial and other documents) are either constructed according to a single principle or are sufficiently similar to ensure continuity of flow from the insti- tution to those in charge of the national health service. One way in which the area-level system differs from that at the institution level is the need for directors of area health services to make an evaluation ( even if it is not very detailed) of the health status of the population in their areas. On the basis of this evaluation, decisions are taken regarding changes in priorities for development of the various sectors of the health service. The end-product of area-level information systems is, on the one hand , a statistical picture of the functioning of the health service and, on the other, a statistical picture of the health of the population in the area. Since directors of area services are faced with tasks relating to changes in the emphasis to be given to the develop- ment of various sectors of the health service , the area-level system should, ideally, be capable of dealing with various optimizational, forecasting and modelling tasks, so as to assist the directors of the health service. The national level The chief task of the health information system at the national level is to produce a statistical assessment of the health of the population, both in the country as a whole and in specific areas, and also to detect those factors that exert a favourable or unfavourable influence on the population 's health. On the basis of this information , the directors of the national health service take decisions relating to changes in long-term strategy by shifting the priorities given to the various sectors of the service, i.e., by redistributing funds . Such decision- making at the national level is extremely complex, calling for a detailed know- ledge of health care resources and of the requirements and resources in respect of each new programme. For this reason, the national health information system must also collect and store data on health service resources. Such a system has as its product, not only a wide-ranging and detailed statistical picture of the health of the population and a statistical picture of the distribu- tion of health resources (both throughout the country and in individual areas) , but also a study of the processes by which changes in resources affect health. This is investigated either by comparing the health pictures resulting from different resource allocations in different areas , or by comparing population health pictures at different periods in the history of the health service. These data are absolutely essential for planning and forecasting. Ideally , the national health information system should be capable of assisting national health leaders to take strategic decisions by presenting various alternative strategies by the simulated model technique. At the present time no major country has set up a national health informa- tion system , but work in this direction is under way in many countries. In 70 the USSR, such a system is being set up , based initially on the existing health statistics system. In addition, for a number of years a national information system has existed to collect primary registry information concerning every mentally ill person entered on or taken off the register, and also data on resource provision for psychiatric institutions. The effect of computerization Modern computers and data transmission methods impose restraints on the volume of data that can be taken up by information systems. At the physician-patient level the entry document is the formal case history, containing many hundreds of items, but the number of such entry documents does not exceed several thousand (more rarely, tens of thousands) annually. At the medical institution level, in addition to the case history, use is made of a number of administrative documents that reflect patient flow, the work.load of the personnel , resource expenditure, etc. At the area level, serving a population ranging from several hundred thousand to several million, there is no longer a place for such voluminous entry documents as the case history, with its several hundred items. Thus, personal registration documents (several dozen items) are mainly used to characterize patients. The number of types of administrative document is increasing rapidly. At the level of a large country (tens or hundreds of millions of in- habitants) the entry documents used at present are the documents obtained at lower levels. The depth of processing of data and the degree of mathematical pro- cessing also differ substantially according to the level of the information system. This is accounted for not only by the increasing complexity of tasks at the higher levels but also by purely pragmatic considerations; it is easier to bring together qualified personnel in a centralized national information system than to attempt to obtain sufficient personnel for a large number of decentrali ze d systems. The choice between centralized and decentralized health information sys- tems is affected by a variety of factors, such as the size and economic potential of the country , the degree of development and centralization of the health service, and the degree and characteristics of development of the health sta- tistics service. In the USSR it is accepted that there is a need for health infonnation systems at all levels, both centralized (national) and decentralized . This pre- supposes a need for compatibility of information between the systems at the various levels, and the formation of an extensive network of compu- ter centres. 71 INTERVI EW SU RVEYS FOR DECISION -MAKERS t F. Sawickia !Nf'ORMATJON SYSTEMS FOR HEALTH ADM! ISTRATORS A health information system may be defined as "a mechanism for the collection, processing, ana lysis, and transmission of information required for organizing and operating health services, and also for research and training" (J). However, it seems impossible at the moment to build in any country such a comprehensive, multipurpose and multiobjective information system. An in- formation system designed for health administrators, managers and planners should be limited only to the information that is really re levan t to their needs, which are d ifferent at each level of management; information that is oriented to scientific research or to serving an inquiring mind has only "curiosity value", as opposed to the "decision value" of information that serves some practical purpose, especially in relation to the decision-making process. Ty pes of data collected The amount and kind of data collected differs from one country to an- other depending on the political and social system, on the size of the country and the size and structure of its population, on the organization and structure of the health services, and on the stage of development of the statistical and computing services. With.in each country. the amount and kind of in formation needed for the management and planning of the health services also differs at the national, provincial (regional and district), and local levels. Nevertheless, it is possible to specify the kind of data that may be useful for health administrators in any country or region. These may be grouped into the following categories: (a) data on demographic events ; (b) health data, including information on objective and subjective aspects of morbidity and disability; 0 !l ead. Department of Medical Statistics, Nat iona l Institute of ll y!!ienc. Warsaw. Poland . 73 (c) data on the network of health institutions and facilities, and on their staff, activities (prophylactic, therapeutic and rehabilitation) and equipment; (cl) data on drug production and consumption; (e) data on physical and chemical environmental factors related to health; and (j) data on health-related social, economic and occupational issues, in- cluding sociomedical information needed for the evaluation of health services and health activities and programmes. From these data, the minimum infom1ation necessary for health administra- tion should be selected for further collection, processing and analysis. Sources of data Three main sources of the above data may be distinguished, namely re- porting systems, data banks and population surveys. The inflow of data to the information system from reporting systems and from data banks (if such exist) is usually continuous. Population surveys most often provide the information systems with data related to specific problems in a given time, although in some countries continuous population surveys are performed. Reporting systems Existing reporting mechanisms, such as vital statistics or notifiable disease surveillance systems, have to cover all events all the time, but the complete- ness of reporting is often variable, both between and within countries. The great advantage of such data is that it relates to an entire population, and if the under-reporting rates arc relatively low, one can continuously 'provide comparable statistics for various periods of time and various regions. The disadvantage of routine reporting systems is that the reports usually contain aggregated data. This limits to a great extent the possibility of per- forming more sophisticated analysis and makes it difficult or even impossible to link health data with other information , such as environmental, social or demographic data. Another disadvantage of current reporting systems, especially those that provide information on the activities of health institutions, is that in- formation is usually rela1ed to the number of cases only, for example to the number of persons discharf!cd from hospital or to the number of consulta- tions in outpatient clinics, a11d not to the number of persons treated in those institutions. The validity ol da ta collected by such reporting systems is often affected by the lack of uniformity in reporting particular phenomena, par- ticularly those relating to health. Data banks (2) Data banks, which at present are very few, will in the future play an im- portant role in supplying iiL·,ilth and health-related data to information sys- tems , especially at the l,iwc1 leve ls of management. 74 Population surveys ( 3, 4) Population surveys are no t usually organized on a continuous basis but as ad hoc surveys. Their great advantage is that they provide various types of in- formation related to the same individual , and that information on an individual's health problems, his other characteristics, and his opinions, behaviour or atti- tudes can be linked automatically (5). Population surveys in which the number of persons interviewed is limited allow more attention to be paid to standard- ization of the information . Surveys provide an excellent method of studying specific topics that need not necessarily be investigated continuously over a long period of time. POPULATION SURVEYS Types of survey There a re severa l types of survey, some of which are particularly useful for hea lth administrators. Descriptive surveys are normally used, but in some cir- cumstances analytical or even experimental surveys (intervention studies) are performed to cover the needs of health managers. Descriptive surveys Descriptive cross-sectional surveys provide information concerning the distribution (incidence or prevalence) of diseases , disability and/or sociomedical events at a given time in a specific population. Longitud inal studies provide data concerning more than one point in time; these may provide infonnation on seasonal or secular changes in disease patterns, or on changes in a disease with time, when each subject is observed on more than one occasion. Evaluative surveys Evaluative (cross-sectional or longitudinal) surveys, i.e., those designed to assess the value of a health care system , a health institution or service, or a health activity o r programme , are of particular value to heal th administrators. According to their objectives, evaluative surveys may be classified as descriptive or analytical. When an evaluative survey takes the form of an experiment it may be termed an intervention survey. Evaluative surveys form a part of opera- tional research , which has been defined by Bailey (6) as "the application of scientifi c methods of investigation to the sort of problems that confront executive authorities". Measurement methods Different methods of data collection may be employed: exami nations, varying from simple visual observations to clinical examinations ; interview 75 and questionnaires; and documentary sources, such as medical and other (e.g., social security) records. The usefulness of the various measurement methods Interviews are often employed in health population surveys, and the use- fulness and adequacy of the results thus obtained may be compared with those obtained from a health examination survey or from health service records (5, 7). A great advantage of popu lation-based data gathered both by interview and by health examination survey is that they refer to an unselected popu- lation, whereas institution-based data, i.e ., service statistics, refer only to the users of such services. In addition, service statistics, as compared with data from interview surveys, usually contain very limited data on the various health-related characteristics of individuals. In interview surveys it is possible to gather health and health-related information referring to both the respondent and to other members of his family or of his household, and this facilitates analysis of health problems as related to various environmental, social and economic factors. When information on the health status of individuals is needed, health exam- ination surveys generally provide more valid information than the interview sur- veys. However, health examination surveys are more expensive ;moreover, when the subjective experience of the individual, such as perception of disease symp- toms, motivation and opinions, must be considered, an interview must also be carried out. This is why, apart from interview surveys, interviewing is often used together with other medical and/or laboratory measurements. Furthermore in some countries, especially the developing ones, where information may be quite scarce, an additional advantage of interview surveys is that they facilitate the collection of information not only on health but also on the basic demographic and social characteristics of the population. How- ever, even in countries with comprehensive and continuous reporting systems, interview surveys are conducted to obtain additional information that is not made available by existing routine statistical means . Choice of measurement method The choice of an adequate survey measurement method is a difficult and responsible task . When a particular method is under consideration, the purpose and objectives of the survey must be considered in relation to, on the one hand the resources needed for conducting such a survey and the validity of the resulting information and, on the other hand the resources needed and the validity of information obtained or obtainable by other methods. The interview as a measurement method Interviewing is an invaluable method of measuring subjective phenomena, such as: the occurrence of symptoms of disease, either at the time of interview or in the past ; perceived morbidity, disability and impairment; opinions; beliefs; attitudes; and some behavioural characteristics (4, 7, 8). 76 Interviews and self-administered questionnaires An alternative method of measuring subjective phenomena is the self- administered questionnaire, i.e., a questionnaire without an interviewer. The use of self-administered questionnaires is simpler and cheaper, and they may be sent, for example by mail, to many persons simultaneously. Never- theless, in general interview surveys, despite their expense, are preferable. This is particularly true when a certain level of education, literacy and skill is expected from the respondent, when there is a large number of questions in a questionnaire, and when the questions are related not only to facts but to attitudes, opinions and beliefs. Advantages of interviews Interviews have many other advantages. Well trained, skilled interviewers can stimulate and motivate the respondents. A good interviewer can create an atmosphere conducive to answering the questions, can explain the reasons for conducting the interview, can repeat questions that are not understood, and can ask so-called "follow-up" or "probing" questions to clarify or amplify a response. When embarrassing or otherwise awkward questions need to be asked, the interviewer can present them in the correct way, for example in the case of embarrassing questions he may use special techniques such as the "randomized response" method. Health interview surveys Health population interview surveys were initiated in Great Britain at the end of the J 9th century, in London and York . In the same country in 1912, Bowley was the first to take a sample instead of studying the entire population. The first methodical health interview survey, covering a sample of households, was organized in I 921-24 in Hagerstown, USA. In the last 30 years a great number of such studies have been performed in many coun- tries, covering both entire nations and populations living in particular areas (9, JO) . Continuous interview surveys Continuous health interview systems are in operation in only a few coun- tries. The continuous Survey on Sickness was conducted in the United King- dom from I 943 to J 952. Continuous health interview surveys have been in operation in Japan since 1953 and in the United States of America since 1957, and in the United Kingdom a new survey started in 1971 (the health section of the General Household Survey). Health survey systems are cur- rently planned in many other countries (5). In continuous surveys some data may be collected year by year, and some occasionally after various intervals of time. For example, the questionnaire used in the Health Interview Survey in the USA includes basic information 77 collected on a continuous basis {this part is occasionally revised) and from time to time various additional items are included (J 1). International surveys Interviewing techniques have also been employed in international studies on the perceived needs of populations, on health resources and on the utiliza- tion of health services. In 1968-69 the WHO/International Collaborative Study of Medical Care Utilization was performed in 12 areas in 7 countries (12). In each area a sample of a noninstitutionalized population was inter- viewed. The standardized questionnaires, carefully translated from English into 4 other languages, in each area were administered by interviewers who had all received the same training. Comparison of results obtained in dif- ferent areas in different countries should be helpful in the development of new conceptual models and new methods of analysing national and local problems related to the organization, efficiency and effectiveness of health care systems(/ 3). Specific problems Interview surveys may be designed to answer a number of questions, but such multiobjective surveys are more expensive, laborious, and time-consuming. More often the objectives of health surveys are limited to specific problems, especially those that are not available from existing reporting systems. As an example, in recent years a number of surveys have been designed to study dis- ability. Such information is not obtainable by means of existing reporting sys- tems, and an increasing proportion of handicapped persons creates important health , social and economic problems. Such studies have been performed in Denmark , the Federal Republic of Germany, the Netherlands and the United Kingdom (/ 4-17). Data on disability are collected on a continuous basis by means of the interview surveys in the United Kingdom and the USA. In Poland, questions on disability were included in the questionnaire used in the micro- census in 1974, and the questionnaire to be used in the forthcoming national population census will include two specific questions related to disability and impairment.0 ORGANIZATION OF INTERVIEW SURVEYS The principles of survey organization are the same irrespective of the measurement technique used. a J. Bejnarowicz, unpublished data , 1974 and personal communication, J 978 . 78 Stages of a survey The stages of a study may be listed in logical sequence, in which each phase is dependent on the preceding one ( 4, 18, 19) . 1. Plan of the study 1. I Formulation of the objectives of the study 1 .2 Selection of the study population 1 .3 Selection of variables to be measured 1 .4 Determination of measurement methods 1.5 Planning the organization of a field study 1.6 Planning the analysis 1.7 Organizing supply and maintenance of the necessary equipment 1.8 Estimation of costs and budget plan 2. Training of staff 3. Pilot study, pretests , etc. 4. Correction of the preliminary programme according to the results obtained in the pilot study 5. Main field study - data collection 6. Data processing 7. Analysis of data 8. Final report Of the above problems only those explicitly related to the interview surveys will be briefly discussed . Objectives of the study Study objectives should reflect the purpose of the study. The purpose of the study should be decided by a heal th administrator, if necessary in consulta- tion with epidemiologists and /or health statisticians ; he should specify in operational terms the nature of the problem and the types of information that would help him to make a decision (20). The study objectives determine the whole future planning and perform- ance of a study. Before the study objectives are finally established, the availabil- ity of local resources should be taken into consideration and other avail- able sources of data should be explored. The expected resulting information should be considered as well as its suitability for improving the process of 79 choosing among alternative ways of decision-making with regard to previously specified purposes and needs. When an intervention study is designed, at this stage of planning the hypotheses to be tested must be stated. Study population The study population and its size must be suited to the attainment of the objectives of the study. The study population and its elements (units of inquiry) should be precisely defined . It is necessary to decide whether an entire popula- tion will be included in the study, or only a part of this population. When the latter is selected the method of drawing a sample (purposive or random) should be chosen. Sampling ( 19, 21) In most cases, results of acceptable precision may be obtained by random sampling, with a substantial saving in cost, time and labour as compared with what would be required in studying the entire population. However, in some cases, such as in intervention studies, purposive sampling may better meet the study objectives. Owing to the high cost of interview surveys, these are usually performed on a sample basis; most often a random (probability) sample is drawn from the parent population. Sampling design ( 19, 21) The sampling design is an integral part of the survey design. When the sampling is planned it is necessary to include some additional items in the overall design of the survey. It is most important to establish the precision of estiniates that will be sufficient to make an appreciable contribution to decisions, the sample size and the method of sample selection. Precision of estimates and sample size ( 19, 21) The precision of the sampling estimates depends on the variability of the examined characteristics and on the sample size. The sample size may be established when the following problems are taken into consideration: (a) desired precision of the estimates - the smaller the sample the lower the precision that can be expected; (b) expected frequency of events to be studied - the smaller the sample the less that can be learned about low-frequency events; (c) nonsampling errors, such as noncoverage (incomplete sampling frames, missing subjects, etc.), nonresponse , and observational and processing errors - the larger the sample the greater the difficulties in avoiding these errors; (d) costs, which increase with the size of a sample. 80 In practice, when the precision is fixed, the method of sample selection and the sample size should be determined at the lowest possible cost; when costs are fixed, sample selection and sample size should aim to obtain maxi- mum precision. A sample is economical if the precision per unit cost is high. One has to remember that precision depends only on the size of the sample and neither on total population size nor on the size of the sampling fraction. Selection methods ( 19, 21) There are several methods of sample selection, such as simple random, systematic, stratified or nonstratified, and single-, double- or multiphase selection. Sampling units are the basic units of sample selection. In element sampling the units of inquiry are the only sampling units, whereas cluster sampling in- volves the selection of groups, i.e., clusters of elements, as sampling units. One has to remember that some of the methods of sample selection may increase or decrease the variance of the characteristic under study and there- by the size of the sampling error. Increasing the variance decreases the pre- cision of estimates and vice versa. Generally , stratification may increase the precision of estimates, whereas cluster sampling and multiphase selection may reduce precision, sometimes to a great extent. Control groups When intervention studies are designed, the choice of an adequate con- trol group should be done at this stage of planning. Choice of variables At the next stage of planning the list of characteristics (variables) that are to be measured should be completed. In health interview surveys various variables may be measured, namely: (a) occurrence of specific acute and/or chronic conditions; (b) occurrence of selected symptoms or groups of symptoms; (c) perceived disability (short- and long-term) and impairment; (d) demographic and social characteris tics of the respondents; (e) behavioural characteristics related to health (smoking, drinking, drug consumption) ; (j) employment and economic status of the respondents ; (g) utilization of health services; (h) expenditure related to the use of medical care, other health services, medicines, etc.; (i) needs and demands for medical care; and (j) level of satisfaction of health service users. 81 It is obvious that, according to specific needs , objectives and resources, only some of the above problems will be investigated in individual interview studies performed at different levels of management. Measurement methods Interviewing (8) In health interview surveys the measurements are performed by means of interviewing. An interview is an interaction between the interviewer and the respondent, in which both participate. If this interaction is handled properly the interview becomes a powerful technique, but if it is not the results may be biased and invalid. The questionnaire is the instrument that should facilitate communication between the interviewer and the respondent. In order to obtain valid results it is necessary to prepare a uniform, valid, and useful questionnaire and to train interviewers carefully. Design of the questionnaire (8) The problems that should be taken into account when a questionnaire is designed cover the construction and wording of particular questions, the use of "open" or "closed" and short or long questions, the choice between a direct and an indirect approach to specific problems, the use of simple or multiple questions for measuring specific variables, the sequence of questions within a particular topic, and the sequence of topics within the questionnaire. In general, the function of questions is to translate the interview objectives into language familiar to the respondent and to assist the interviewer to achieve a high level of motivation in the respondent. The words used in the question- naire should be understandable to respondents at all levels of literacy, educa- tion and skill. The language must be "shared" by both the respondent and the interviewer, and must be appropriate to the role of the interviewer. Bias in interviews Invalid questionnaires and unskilled interviewers may be an important source of bias. This bias may be diminished by careful designing of the ques- tionnaire, by training the interviewers and by controlling their field work. The respondent is an additional source of bias, forgetfulness, attaching little importance to the events, and threat or embarrassment being the main reasons. The possibility of checking the validity of answers is limited to factual events; for example the reported occurrence of chronic conditions can be checked against the physician's records. It is impossible to check the validity of answers related to the respondent's opinions, behaviour or attitudes. Plan of the field study The detailed plan of the field study should include all organizational procedures related to interviewing, to the (if possible random) allocation of interviewers to selected sampling units, and to the methods of supervision of their field work. All additional documents and forms needed in the study as well as detailed instructions for both interviewers and supervisors should be prepared . Budget preparation When the cost of the study is planned, and when sampling is included in the survey design , it is necessary first of all, as mentioned above, to establish a proper balance between the sample size and the allowed costs on the one hand and the desired precision of the estimates on the other. All expected expenses should then be taken into consideration. Kish (21) has proposed the following formula for expressing the total cost T of a survey : T = K + Kv + nc + ncv = K + Kv + n(c + Cv), where K represents the constant cost factors , such as the design of the study, the construction of the questionnaire, the analysis of results and the survey report. These factors are affected neither by the sample size nor by the type of sample design used. K v represents the cost factors that vary with changes in the sample design , such as computation of estimates, weighting of sample results and addi- tional staff training. These factors are not affected by the sample size. n denotes the number of elements in the sample. c is an average cost per element of the factors not affected by changes in sample design, such as making and recording of observations, coding and punching. cv denotes the average cost per element of the factors that vary with changes in the sample design, such as cost offrame(s),selection procedures, location of sampling units and transportation costs. The above formula may be helpful when precision of estimates, sample design , selection technique and costs of a survey are considered. Training of interviewers The training of interviewers should be mainly practical . One of the best ways of training is based on so-called "role playing", when one interviewer after another plays alternately the role of the respondent and that of the inter- viewer, using the same questionnaire to be used in the field study . Interviewers should also be taught about interviewing techniques and dynamics, and about the role and significance of their behaviour, language, even their attire . They have to learn that it is necessary to follow the fixed order of questions in the questionnaire, how to deal with inadequate responses, when they may ask additional questions, how to behave in expected or unexpected difficult situa- tions during the interview , and how to motivate respondents in an appropriate 83 way. The interviewer must be able to explain the objectives of the study to the respondent and to convince him of the importance of his role in that particular survey. When the whole group of interviewers follows all these instructions and when they understand the dynamics of the interaction process of which they are to be a part, they are equipped to collect information in a way that assures the validity and comparability of the answers obtained. Pilot study (4, 8, 18) Before the main field study is started, it is usually necessary to test all the methods, procedures and equipment (including questionnaire and interviewers) . In addition, pilot studies allow one to establish the sample size and to assess the size of the sampling error. All organizational field procedures should be tested as well. Prior to the field pilot study, the questionnaire is normally pretested several times on small groups of respondents, in order to determine whether the questions elicit responses that meet the objectives of the study, whether the questionnaire promotes an appropriate relationship with the respondent, and whether the individual questions are understood without additional explanations o r rewording. When difficult or inadequate questions are idenfi- fied, they are changed before the pilot study. The final changes in the question- naire are made when the pilot study is completed. The pilot study also provides an opportunity of observing the work of interviewers and of checking the validity of their activities. REFERENCES I. WHO Regional Office for Europe. Health information systems: report on a Conference. Copenhagen, 1973 (EURO 4914). 2. WHO Regional Office for Europe. Health data banks: report on a Joint IFIP/WHO Working Group. Copenhagen, I 977 (ICP/DHS 002). 3. Witts, L.J ., ed. Medical surveys and clinical trials, 2nd ed. London, New York, Toronto, Oxford University Press, I 964. 4. Abramson, J .H. Survey methods in community medicine. Edinburgh & London, Churchill Livingstone, 1974 . 5. Armitage, P. National health su rvey systems in the European Community. International journal of epidemiology, 5: 321-326 {1976). 6. Bailey , N.T.J . Operational research. In : Witts, L.J ., ed . Medical surveys and clinical trials, 2nd ed. London, New York , Toronto, Oxford University Press, 1964. 7. Kalimo, E. Use of health interview surveys. In : Pflanz, M. & Schach, E., ed. Crossnational sociomedical research: a seminar convened by the Akademie fiir Sozialmedizin Hannover. Stuttgart, Georg Thieme , 1976, pp. IO I - I 07. 84 8. Kahn, R.L. & Cannell, C.F. The dynamics of interviewing, 9th ed. New York, London, Sydney, John Wiley & Sons, 1965. 9. Kessler, I.I. & Levin, M.L., ed. The community as an epidemiologic labora- tory. Baltimore & London, Johns Hopkins Press, I 970. 10. Lindhardt , M. In: The Sickness Survey of Denmark. The Morbidity Survey of /950. Copenhagen, Munksgaard, 1960. 11. Health interview survey procedure 1957-/ 974. Vital and health statistics. Series 1, Number 11. Rockville, National Center for Health Statistics, 1975 (DHEW Publication No. (HRA) 75-1341). 12. Kohn, R. & White, K.L., ed. Health care. An international study. London, New York, Toronto , Oxford University Press, 1976. 13 . White, K.L. et al. Health services: concepts and information for national planning and management. Geneva, World Health Organization, 1977 (Public Health Papers, No . 67) . 14. Andersen, B.R. /The physically handicapped in Denmark/. Copenhagen, 1964, Vol. 2 (in Danish). 15. Korperbehinderte. Bevolkerung und Kultur. Fachserie A , Reihe 7. Gesund- heitswesen, Sonderbetrag, Ergebnisse der Mikrocensus 1966. Wiesbaden , Statistiches Bundesamt, 1969. 16. The physically handicapped. The Hague, Staatsuitgeverij, 1976. 17. Harris, A.J. Handicapped and impaired in Great Britain. London , Office of Population Censuses and Surveys, 1971. 18. Sawicki, F. / Elements of statistics for physicians/. Warsaw, PZWL, 1974 (in Polish). 19. WHO Technical Report Series , No. 336, 1966 (Sampling methods in mor- bidity surveys and public health investigations: ten th report of the WHO Expert Committee on Health Statistics). 20. Cassel, J. In : White, K.L. & Henderson, M., ed. Epidemiology as a funda- mental science. New York, Oxford University Press, I 976. 21. Kish, L. Survey sampling. New York , London, Sydney , John Wiley & Sons, 1965 . 85 11 MODELS OF THE HEALTH SYSTEM AS A BASIS FOR DATA COLLECTION E.N. Shigan° A health care system is a large and complex dynamic system consisting of interrelated subsystems that are very closely related to external systems and are joined by a common aim. It comprises several hierarchical levels - global, regional, national, district, etc. The development of systems methodology re- flects the necessity to describe such a complicated system from the point of view of different disciplines in order to define problems and to suggest to decision-makers approaches to their solution. The main focus in systems analysis is on the comprehensive, interdisciplinary, analytical investigation of the system and its development. In this chapter I shall explain how the aims of the system correspond to its structure, functions, and organization. Matching the aims to statistical indices used to estimate the achievement of those aims shows that for the achievement of some there are many statistical indices in "routine" reports whereas for others there are either few or none. For example, it is very import- ant to have indices that can help in estimating the relationship and inter- action between, for example, different medical departments; unfortunately such indices - the most important from the systems point of view - do not exist in "routine" statistics, and in order to find these interacting indices it is necessary to conduct special studies. Such an approach will help to define problems to which solutions are needed , although sometimes the decision-maker knows of problems without having to conduct special preliminary studies. Nonmedical specialists such as mathematicians, economists and engineers need such a "diagnosis" of the system not only for problem formulation (health managers can easily tell them about problems), but to understand better the behaviour and develop- ment of the health care systems being investigated. Nearly all the management problems that the decision-maker is continually facing may be divided into three groups: operative, short-term and long-term. These may be further subdivided into more detailed categories of problem, and so on. Finally, a complex of problems may be built up having different a Head, Modelling Health Care Systems Task, Human Settlements and Services Area, International Institute for Applied Systems Analysis, Laxenburg , Austria. 87 priorities and levels of difficulty. For the solution of these problems the decision-maker uses different methods : his own knowledge and experience; "routine" and comprehensive study data ; and expert evaluation . For the solution of some particularly strategic problems the decision- maker needs to conduct special management experiments with real objects (hospitals, district health care services, etc.). This involves the participa- tion of many institutions, much time and money, and taking decisions over several years. Despite the large number of important tools , such as mathematical models, available to the decision-maker , there are relatively few examples of their application in health care. There are several reasons for this, but the problem is mainly due to the gap that exists between the decision-maker and the model builder in understanding health care systems and their problems, behaviour and trends. Even if they select the same problem, their understanding of the health system, its development and ways of solving a particular problem (this is re- ferred to as the "mental model") could be very different. This is why, in order to improve the usefulness of mathematical models, the participation of the decision-maker is necessary during all stages of the modelling process, be- ginning with problem formulation and ending with the implementation of the model in health care practice (Fig. I). Participating in this process a decision-maker, on the basis of his own mental model , defines the system, its boundaries, subsystems, levels of ag- gregation, criteria , the relevant factors and variables , and their influence and interaction. In order to improve mutual understanding and to help com- municate a clear representation of the mental model, it is very useful to pre- sent it graphically (block-scheme, causal loop diagram, etc.). A very import- ant role of the decision-maker involves gathering information about sub- systems, variables and criteria. He knows the sources of the necessary in- formation ("routine" statistics, special studies , literature) and he can esti- mate the reliability and importance of these data. A decision-maker can deter- mine which variables are controllable by himself and which must be dealt with at a higher level. Proceeding from this consideration, a model builder selects methods for mathematical description and computer programming and subsequently creates mathematical and then computer models. During this phase the model builder can return to previous stages ( data collection, causal loop diagrams, etc.), and sometimes he needs to consult health managers. The processes of running and tuning the computer model are simpli- fied by the help of terminal equipment. In the course of testing the model a model builder can fix different deviations that should be corrected. For tuning the computer model several different approaches can be used , be- ginning with official statistical data and results of special surveys. A decision- maker can help in selecting the most reliable data. A comparison of com- puterized data about population structure , medical staff, beds and other 88 I I I I I I I I I I I I I I I I I I I ,,,✓✓,,,,,'~ I I I problem formulation conceptual framework graphical presentation informational analysis method data collection mathematical modelling computer programming running and tuning the model evaluation implementation Fig. 1. The model-building process, showing the mutual participation of the decision-maker and the model builder . 89 precise indices with real data taken from "routine" reports will help to esti- mate deviations, which can then be adjusted in the next run. Secondly, it is possible to use the opinion of the decision-maker concerning expected results, data, alternatives. etc. Thirdlv. a combination of both approaches is possible, viz. the use of reliable statistical data in combination with expert opinion. The participation of the decision-maker ~ ~!i_e tunin~ process is important because the ultimate objective of the computer model is to help hin1 in managing his health care problems. After the tuning of the computer model comes the last stage of the model- ling process - implementation. This final stage depends mainly on the actions of the decision-maker. The model should be adapted whenever new data are available. This description of the standard process of model building shows the need for participation by the decision-maker at almost every stage. His knowledge, practical experience and intuition help the model builder to create adequate computer models for the solution of health care problems. The degree to which different sources of information are used may vary. Sometimes it is enough to use only "routine" statistics, and some data may be found from comprehensive studies. If the necessary data do not exist in na- tional publications, it is possible to use information from other, similar coun- tries. Very often, however, a model builder lacks some items of informa- tion, such as data on the relationship between different subsystems of the health care system and, especially, about the interaction of the health care system with external systems. In this case the decision-maker can help to some degree in the creation of the fust version of the health care model. It is impossible to describe here all the examples of modelling for health care systems, since each model is oriented to the solution of specific problems. At the International Institute for Applied Systems Analysis (IIASA) in Laxenburg, Austria, a team of specialists in different fields from Austria, Japan, the United Kingdom and the USSR, in close contact with some national centres and WHO, is elaborating a national health care system (NHCS) model. The creation of this model will help the decision-maker at the national level in simulating the activity of the health care system, and in testing different policy alternatives. In particular the health manager needs, in order to allocate health care resources properly , forecasts of population development, as well as of the dynamics of the health status of the population, taking into consideration the strong dependence of the health care system on socioeconomic, environmental and other systems. The mental model, elaborated by Venediktov in 1974 (1), reflects the concept of the health care system. A graphical representation of this mental model, in the form of a functional chart, is shown in Fig. 2. The problems, con- ceptual framework and structure of the model have been discussed at several conferences organized by the IIASA Health Care System Modelling Group, together with national centres and WHO (2-4). IIASA's dynamic simulation model of the national health care system con- sists of several interrelated submodels, which are elaborated in the following sequence: population; health; resource demand; resource supply ; resource al- location; external systems. The selection of a mathematical model to describe each of the above subsystems depends on the problems involved. 90 I t Socioeconomic condit ions Assessment of Population :1 Analysis of individual reQJJirements I & community health Industry I w Training I Research on major ~- r+--Agr iculture I problems I ii i r7 Transport I Recommendations for Placement ij I I Collection of data on I practice :, 1 I population health Science & Technology :, '!. ,, I & medical care l Organization ,, I National income 1-iealth education - of work ,, I Evaluation of Nutrit ion Preventive health 11z 1 information 11 o I surveillance Further ,' 3 I ;i: Standard of training 11""1 Management .. living Demographic Routine health -0 11~ I decis ion-mak ing :, surveillance ~ ,, a.1 i indices -0 Sphere of 0 ::r ~ ,1~1 Planning & ~ services .., Epidemic control < ,15· 1 Incidence and C: .; ◄-------- - ----------------- 11~. I implementation ~ ~ )( Education prevalence of ;;;· / 2.1 o· Immunization ,,,, diseases :, Monitoring & ::r ,, I Culture " I Preventive screen ing ,, I decision-making .. I - - :,' I I I Natural conditions Disab ility Preventive care Assessment of I I Links with other .: requirements I I 5 I LJ systems Cl imate Physi cal Case find ing I I Landscape status I Primary & emergency Financing -I i I care ::Jl Air -0 : Diagnosis & assessment i Buildings & ~ 2. fac il ities 0 Water s 1 of work capacity I~ ~ o· I Provision of Soil :, I Care & rehabilitat ion instruments & I General r~com- , I Fiora I Prevention of relapse equ ipment mendat1ons I Drugs, & restrictions Fauna I Other funct ions preparations, Natural foci of disease etc. I '° Fig. 2. Functional chart of a public health system. Submodels have already been built at IIASA covering population, pre- valence estimation, resource requirements and resource allocation. An example is "AMER", the aggregate model for estimating resource requirements. Resource requirements are determined on the basis of estimations of population trends, morbidity rates and desired standards. The block diagram of this model is shown in Fig. 3. Each block in its tum may be described in quantitative parameters, equations and systems of equations. Initial data for the model were: (a) population structure, mortality rate , fertility rate and cause of death, obtained from WHO publications; and (b) standards obtained from "routine" reports and special surveys in the United Kingdom and the USSR . These initial data, together with hypotheses and information from studies, were used to calculate intermediate data - current population structure, general morbidity, and substitution coefficients describing the relationship between outpatient and inpatient services at the same horizontal level. It is clear that the model could be broken down further by estimating other resources, their structure , etc. For the mathematical description a set of equations was selected and translated into one of the computer languages {FORTRAN). This computer model will help the decision-maker working with the model through terminal equipment, not only to forecast changes in population structure and morbidity trends, but also to test different plan- ning alternatives, such as how many hospital beds would be needed if the average length of stay in hospital were reduced. Such a reduction is possible if the majority of inpatients undergo laboratory tests in outpatient centres before admission. After building a computer model it is possible to test diffe rent combina- tions, hypotheses and assumptions and to estimate, in quantitative terms , the results of such simulation. If the results of such testing do not satisfy the decision-maker he can return to the mental model , to data collection, etc . This simple model is based mainly on "routine" data and literature and does not take into consideration the effect of external influences, but there is no question that all components of this model depend to a large degree on the socioeconomic situation. All demographic indices and processes differ greatly in countries with different socioeconomic systems . The model builder also needs data about the interaction between the environment and the health care system. Unfortunately, these data do not exist in official publications and have to be specially produced. Without this information it is impossible to build a computer model to satisfy the decision-maker. It should be mentioned that many different kinds of interaction exist between subsystems of the health care system, and these can be measured as quantitative indices (interaction coefficients). The interaction coefficient can reflect, step by step, the situation in the treatment or prevention process (for example, how many patients were sent from one medical establishment to an- other). It can also reflect the substitution effect, as in AMER between out- patient and inpatient services (substitution coefficients), or the duplication effect when , for example, laboratory tests performed on outpatients are 92 I I I I I I I MORB IOITY BLOCK POPULATION BLOCK rate , given DEAIII Initial prevalence rate, given PREVI I) Risk ca lcula- tion RISKO) Current all ca uses death rate I I I I I I I I I I I I I I Initial popula• ti o n age structure, given POP(I) Current fert il i ty rate Ini t ia l fe rtility rate BRTHIII 12 I--------- -- -- ----------- - --------- -' 1 ____________________________ t ------------ -------------- ----------------------------------------- I I I I I I OUTPUT : Resource requirements I------------- - I 16 Bed turnover interval (days) BTI 1 STANDARDS (Control vanables) Inpatient ~ 1--------' ~~~\o;e~:~;:: ments OR IN inpatient doctcr equiv • alent DPERB Actual value of COEPII) of consul t ations per episode COEP10l l l Workload (consu ltations per year) WL 1 I-------------------- ---------------------- --- ------------------ ---- Fig. 3. Block diagram : resource requirement estimation model of a health care system. 93 repeated on inpatients. An interaction coefficient can also reflect the rela- tionship between subsystems by means of different medical resources (drugs, beds, equipment, money, etc.). Besides interactions between subsystems at the same horizontal level (horizontal interaction) there is also vertical interaction, for example the relationship between the central hospital and a local hospital in the treatment of people in that locality . These very important coefficients concerning the interaction between horizontal and vertical levels are absent in "routine" statistics and must be taken from specially organized studies. Without the interaction effects and their coefficients it is impossible to conduct quantitative analyses of any sys- tem and to build mathematical models . Modelling makes additional demands on the reliability of medicostatistical information, because it is very difficult to build adequate mathematical models on the basis of poor information. Taking into account that the building and tuning of the model is based first of all on "routine" information, it is necessary to have a set of coefficients that reflect the existing deviation between official statistics and special studies. With these corrective coefficients it is possible to build a model on the basis of "routine" information (5). The development of the health management process on the basis of models leads to some other related problems, viz, the need to concentrate "routine" and special study data in a single national centre to make it easier to find informa- tion needed by health managers and model builders; and the need to teach the decision-maker about systems analysis methodology and the model builder about health management problems. .REFERENCES 1. Venedictov [Venediktov ], D.D. Systems analysis of health services. In: Bailey, N.T J. & Thompson, M., ed. Systems aspects of health planning. Proceedings of the I/ASA Conference, Baden, Austria, 20-22 August, 1973. Amsterdam, North-Holland, 1975, pp. 19-30. 2. Bailey, N.T.J. & Thompson, M., ed. Systems aspects of health planning. Proceedings of the I/ASA Conference, Baden, Austria, 20-22 August, 1973. Amsterdam, North-Holland, 1975. 3. Venedictov [Venediktov ], D.D., ed. Health system modeling and the in- formation system for the coordination of research in oncology. Proceed- ings of the I/ASA Biomedical Conference, Moscow, USSR, and Laxen- burg, Austria, 8-12 December 1975. Lax.enburg, International Institute for Applied Systems Analysis, I 977. 4. Shigan, E.N. & Gibbs, RJ., ed. Modeling health care systems. Proceedings of a I/ASA Workshop, Laxenburg, Austria, 28-29 March, 1977. Lax.en- burg, International Institute for Applied Systems Analysis, 1977. 5. Shigan, E.N. Alternative analysis of different methods for estimating prevalence rate. Laxenburg, International Institute for Applied Systems Analysis, I 977. 94 12 CRITERIA FOR THE EFFECTIVE PLANNING AND ESTABLISHMENT OF COMPUTERIZED HEAL TH INFORMATION SYSTEMS P. le Beux 0 & F. Gremy b Until quite recently the financial investment and human effort needed to choose and set up a computerized information system were so great that it was difficult - at least in theory - for an authority or a manager to make any decision unless it was firmly based on precise economic or sound technical criteria (1, 2). Moreover, the introduction of computerized information systems in any organization involved changes in operation that went counter to previous habits and methods of work. This meant that, in addition to the technical and economic difficulties of setting up information systems, there were human problems due to the misgivings of the existing staff and the users, who often expected better services than could be offered to them. This was especially so in the field of medical and hospital services, where any technical innovation not of direct benefit to the patient is considered unnecessary and where, moreover, potential users of the computerized information system had very little time to define their needs and accept the constraints associated with the development and use of any computing system. As a result, the introduction of computers in medical and hospital services was often restricted to applications of a purely administrative nature, such as invoicing, accountancy and salaries. In other cases more ambitious projects proved ineffective, too expensive, or ill-suited to requirements. This led to the notion that the introduction of computerized information systems was a weighty matter, requiring preliminary studies with precise planning and ob- jective criteria (3, 4). This is still the case with large computer systems. In many cases, however , owing to the technological developments of the last IO years , the introduction of a dedicated computer configuration nowadays involves no more problems than the introduction of a new piece of medical equipment in a specialized unit. In the discussion that follows, the applications of computers in the health services will be considered to be limited to the processing of medical or public health data ; applications in the fields of administration, accountancy, invoicing, education and basic research will be disregarded. Nonetheless, the development 0 Departement MAI, Universite technologique de Compiegne, France. b Departement de Biophysique et de Biomathematiques, Faculte de Medecine Pitie- Salpetriere, Universite Paris 6, Paris, France. 95 of computerized information systems for medical purposes should no longer be disorderly or irrational. Information science is now an area of technology that is well understood, and outside the field of basic research computer systems (both hardware and software) are seen as industrial products that should not be set up without due regard both to requirements determined in accordance with objectives and quantifiable criteria and to available resources (5). As a general rule, the acquisition of any new facility in the medical sec- tor must be justified by preliminary studies to evaluate its suitability for the purpose intended. In the acquisition of computerized information systems this is relatively easy owing to the wide variety of equipment now available, ranging from systems based on microprocessors to the single centralized sys- tem ( 6). In many cases it has been found that past projects failed because they tackJed problems that were too difficult in order to justify the acquisi- tion of a computer or the development of an information system. Nowadays it is possible to set up systems that are immediately ready for operation, at a reasonable cost. In particular, data can be processed locally in distributed modular and interconnected systems, and after sifting and aggregation these data, already verified and reliable, can be sent to a higher level for further processing. In the following sections an attempt will be made to indicate the nature of these various processing levels, each of which may have very different criteria. To begin with, the techniques and potential applications now available for medical computing will be reviewed. This review will then serve as the basis for a definition of the technical, economic and cultural criteria relevant to the selection, planning and implementation of computer-assisted information sys- tems. Finally, recommendations will be made in relation to these criteria. HOSPITAL COMPUTING SYSTEMS: STATE OF THE ART A number of possible applications of computerized information systems in medical and hospital services may be distinguished, excluding the purely administrative applications mentioned above. Some systems are already in operation, some are at the design stage or in course of implementation, and others are the subject of research. Automation of laboratories There are many applications in this field, the first in which the intro- duction of computer systems has been fully accepted; indeed some such systems have been in operation for some time . This is due essentially to the automation of analytical instruments (autoanalysers), which can be controlled by minicomputers and the results presented in processed fom1. Moreover, the need to minimize the manipulation and transcription of numerical results justifies the complete computerization of the processing chain. Such systems can also carry out real-time checks and even ensure the quality controls needed 96 for assessing the reliability of results. These systems are now routinely used in most laboratories. The advent of the microcomputer will lead to a further expansion of these applications . Surveillance of patients with acute disease This application is also very common, but results have not always been as convincing as was hoped. The essential problem in this type of application is data acquisition; this requires precise sensors operating with minimal deviations so that they give the alarm at just the right time. Another difficulty is that several patients often have to be monitored simultaneously, because the tech- niques used are so expensive. This increases the problem of reliability and it has been difficult to adapt the system to individual cases. Here also, technical ad- vances and reductions in costs have opened up prospects for autonomous sys- tems that watch over a single patient. As a result it should be possible to con- struct compact monitoring systems that can be adjusted according to the characteristics of the individual patient. Systems for management of medical records Despite the impressive number of applications attempted in this field, very few systems are in fact in operation. This has to do with the nature of medical language. Any clinical observation is essentially qualitative and depends on descriptions in ordinary language; on ly the results of biological examinations can be quantified. Furthermore, the prerequisite for the organ- ization of a computerized medical records system is that the records can be processed, but at present the text-processing of ordinary language is im- possible owing to its grammatical complexity and especially its an1biguity. Nevertheless, the attempts made in thjs field have been of some benefit in showing that there is no such thing as a complete and universal medical rec- ords system, or at least that such a system cannot be fully computerized. Various types of medical records system, intended for different types of processing, can in fact be distinguished. First, there are "survey" records, intended primarily for sta tistical pro- cessing, for wruch it is essential to collect precise and reliable data from a significant sample. This type of appli cation, in which the physician must as- sume a research role for epidemiological or long-tem1 studies on a precise prob- lem, necessitates the use of scientific and statistical methods. The second type of application is in the establishment of systems containing computerized reports or summary medical records for documentary purposes, which can be used by the physician in his role as the organizer of health care. A third type of application is in the storage of data needed for day-to-day inpatient care: requests for examjnations, observations, medical prescriptions, results of examinations, radiological reports, etc. In this case the essential problem is to establish a man-machine communication interface adapted to the medical and hospital environment. Here also, new techniques are now available, and these applications have been facilitated by the development of sensitive interactive systems that are easy to use ( 7). 97 Integrated hospital information systems A number o f projects · of thj s type have been launched , notably in the United States of America (8-10) and in Europe (11) . The cost of develop- ing and operating such systems , however , has always been underestimated, and in some cases the system has not met the expectations of physicians. This demonstrates just how important it is to determjne the objectives at the outset. It should be pointed out , however , that the techrucal options avajlable when the systems were designed were much more Jimjted , and very often a centralized system had to be selected for econ omic or even tech- nical reasons. In fact, systems o f trus type can now be set up at much lower cost by using only two levels o f processin g: one at the urut or department level and another at the hospital level. This ensures communjcation between the various elements of the first level and sifts and processes the data needed for the running of the hospital . Such an approach al lows greater flexibility in meeting the needs of each unit , and it also permits a modular type of development (8) starting with the units that are easiest to computerize {laboratories, pharmacy ( 7) , radiology, radiotherapy, etc.) . Moreover , owing to the reduced cost of equipment , sim- plified versions of these first-level systems can be set up in numerous units of the same type , thus reducing the development costs of the software and the system. Epidemiological studies Up to now, the collection and processing of epidemiological data have been carried on in parallel with, but independently from, all the informa- tion sys tems set up in the medical and hospital services. Indeed, epidemjo. logical stu d ies are often considered to be outside the realm of medical com- puting. In actual fact, epidemiological studies based on scientific methods are always very meticulous and use data collected manually. Other studies , more pragmatic in their approach, confine themselves to the collection of general data , such as mortality statistics. It should be noted that in many developed countries there are no complete and reliable data on morbidity , even though only such data can provide the basis for true planning of health care. Another point is that overall epidemiological studies on a scientific basis, covering a whole town , region or country, are very rare because of the difficulty of obtairung data and the cost of setting up the study. This situation appears to be an anachronism , in the sense that the major prob- lem is the collection of data mostly contained in existing medical records. The problem of collecting epidemjological data is, in fact , on a level addi- tional to the two levels mentioned above: all the morbidity data must be filtered through to the community level. Not only would reliable data thus become available , but the cost of data collection for epidemio logical studies would be greatly reduced . 98 CRITERIA FOR PLANNING AND SETTING UP COMPUTERIZED INFORMATION SYSTEMS Technical criteria From the technical point of view the criteria for planning and setting up a computerized information system are of three types : I . Criteria relating to the performance and reliability of the hardware and the structure of the system. 2. Criteria relating to specifications and development of software. 3. Criteria relating to maintenance and operation. Criteria relating to hardware It is essential that the capacity of the hardware should be app reciated and that equipment should be selected in accordance with req uiremen ts. It must always be borne in mind that the bottlenecks in an information system often occur at the input/output level. It is therefore important at the outset to make an approximate assessment of the volume of input data to be pro- cessed and the volume of output results , in addition to specifying the volume of permanent and temporary files and the mean number of accesses to these files. The next stage is to specify the processing volume. Here also it is im- portant to distinguish clearly between the different types of processing ; scienti- fi c calculation does not require the same constraints as the processing of non- numerical data. A central unit suitable for the type of processing envisaged must therefore be selected; for scientific calculation hard-wired instructions for floatin g point calculation are needed , whereas character processing requires instructions for the manipulation of characters , stacks, lists , etc. With rega rd to the structure of the system, a choice must be made be- tween interac tive o r real-time processing and batch processing. In the case of interac tive processing, effective memory protection devices will be needed , as we ll as potential memory (paging) mechanisms. In the case of batch pro- cessing, a mem ory separation system with segmentation may suffice. Finally, the reliability of the hardware must be evaluated, with par- ticu lar rega rd to the operating temperature ran ge, electrical environment (current regulation and emerge ncy battery), volatility o f the memo ries, back- up procedures, e tc. Criteria relating to software These c rit e ri a also depend on the type of processing desired. In the case of interact ive processing many tools arc needed in the form of utility programs (for ed itin g. file manipulation. execution, etc.). As regards the monit o r, i.e .. the so ftware that controls the entire system , a distinction sho uld be made 99 between multiprocessing systems (for real-time opera tion) , multiprogramming systems (for batch processing) and time-sharing systems (for interactive pro- cessing). The most importan t factor in rel ation to the basic software is the relia- bility of the monitor and the presence of clear, precise diagnostic messages. Another important cri terion for software is the presence of efficient com- pilers for scientific calculations (FORTRAN , PL/ I , ALGOL and other lan- guages) and for business data processing (COBOL, PL/ I). In interac tive sys- tems, interpretative languages such as BASIC or APL may be used because they are easy to modify in the execution phase. In the case of systems intended for commercial ope ration, however, this type of language gives less efficient programs than compiled languages. Mention should also be made of the specialized medical systems and lan - guages, such as MUMPS (8). This type of system is highly suit ab le for auto- nomous applications such as labo ratory systems. It s major disadvantage , how- ever, is its incompatibility with a wide range of hardware. It is increasingly common to find basic data processing software that obviates the need to write specialized programs for handling large se ts of data. In choosing a system, the only ge nuinely objec tive tests on software are bench- mark tests on standard programs in different systems; by these tests the perfor- mance of the entire hardware and software can be evaluated on programs that will be routinely used. Criteria relating to maintenance and operation These criteria relate to the ease of operating the system and the speed with which a malfunction can be detected and correc ted . These problems become increasingly difficult and expensive to solve as the system increases in size. With modern intergra ted technology, however, it is possible to produce highly compact systems that are easy to maintain ; in many cases there are automatic error detection procedures, and degraded procedures can be introduced. At the software level, data protection and retrieval procedures are needed to keep the files and data bases intact. The most fragile parts of a system, how- ever, are the periphera l unit s; in certain cases those that are most used have to be duplica ted to guarantee continuous operation. Economic criteria It is obvious that economic criteria are always very in1portant in choosing a computerized information system. As stressed in the introduction to this chapter , however , the costs of hardware have great ly diminished with the ..id- vent of integra ted technology. and this in turn reduces software costs because of the possibilities fo r wider application. The large r the hardware system, the fewer will be the systems on the mar- ket among which the software costs are to be divided. Moreover. if the hard- ware system is big, it will be desirable to use it for many different applications. which will increase the difficulties - and therefore the costs - of producing the software. Conversely if, by virtue of its low price. a hardware system can be used for a decentralized applica tion, a large number of systems ca n be se t up 100 and the marginal cost of the software (which will also be simpler) will be re- duced. In short, the greater the degree of decentraliza tion in data collection, the lower the cost of the software. This point, however, is ofte n not appreciated by planners. There are already on the market monoterminal systems incor- pora ting microprocessors and backing stores (cassettes o r floppy disks) no more expensive than a typewriter! Thus, first-level sys tems fo r decentralized data acquisition and processing can now be developed at reasonab le cos t. At the second level (medium-size institutions) it is now possible to obtain powe rful hardware and software (high-performance minicomputers) that cost less to install and operate than a telephone exchange. Th e staff costs associated with the operation of such systems can also be reduced as the only require- ments are an operation supervisor, an operator and a few analyst-programmers for the development of applications. At the th ird level (regional or national leve l) large computing centres of the trad itional type are still needed. Cultural and human criteria Account must also be taken of o ther criteria, more difficult to evaluate, which are grouped under the heading of cultural criteria . Many medical projects have failed because they did not take into account the habit s and constraints peculiar to medical personnel. When computerized information procedures are introduced it is important that the staff should not be overloaded and also that the system should feed back information likely to render their work easier and more interesting. This demands cooperation from staff, training, and the develop- ment of man- machine interfaces that are simple, reliable and effective. Consideration must also be given to the special characteristics of medical thinking. The use of computers in this field has been very limited (J 2), al- though advances in the fundamental life sciences confront the physician with a mass of information th at he is often unable to digest or use. The consequences are twofold : on the one hand , a trend towards extreme specialization, associated with major increases in health expenditure ; on the other, the severe difficulties encountered in passing on the knowledge needed to bridge the gap between theory and practice and adap t the tools available to the daily routine with- out an excessive increase in the numb er of supplementary examinations, which are expensive and often distressing for the patient. The introduct ion of computers in medicine should therefo re be used to further research into the methods on which medical activity is based, with particular rega rd to the development of more precise language, the definition of objectives, the analysis of decision-making processes in diagnosis, prognosis and therapy, and the evaluation of these processes and their consequences. Fina Uy, an important criterion in re lation to medical systems is to ensure that data are protected both agains t malfunctions of software o r hardware and (in the case of confidential medical data) against misuse. It is therefore im- portant to provide protective mechanisms and to control access to highly con- fidential data . Here also, decentralization of data acquisition and processing gives the attending physician the same direct control over his own re cords as he would have over manual records. IOI / RECOMMENDATIONS On the basis of the criteria and ideas brie0y expounded above, a number of recommendations can be put forward for the planning and implementation of computerized information systems in health services. The first recommendation, at the technical level, is that the objectives should be defined with extreme precision : the possibility may then be con- sidered of using modular systems, and the extent to which the system meets the objectives can be evaluated at each stage of development. A second technical recommendation is that systems should be developed in which the functions of data acquisition and processing are performed in the most peripheral location possible. This obviates all the transcription stages that make information processing techniques so cumbersome and unsuitable for medical purposes . It also solves the problem of security and protection of medical data, as the system is accessible to relatively few people (those in- volved in producing and/or processing the data). Only such information as is deemed to be of value can then be passed on to the higher level , under the supervision of the head of the unit or department. This ensures, moreover, that the data passed on have already been verified and are reliable . This recom- mendation is closely associated with the concept of hierarchical systems, in which each level processes data originating from the level below. Here also, technically feasible solutions are available, based on the technology of inter- computer communication networks and data base systems. At the planning level it is better to develop autonomous systems and evaluate their performance and use before trying to interconnect them at a higher level. With regard to the criteria for selection of hardware and software, it can be said that computer technology has now grown up as an industry and the hard- ware and software likely to meet required specifications can be evaluated fairly objectively . One re commendation at this level , therefore, is that free rein be given to competition, if possible , and that a system be selected on the basis of essentially techno logical criteria. At the level of operation and maintenance it is important to lay down ex tremely stringent requirements for the functioning and repair of the system , ot herwise any malfunction may have dire consequences. Here also, a decentral- ized approach has the advantage that complete stoppage of the en tire system is avoided in the event of a breakdown. From the economic point of view, the main recommendation is that no computerized infom1ation project should be embarked on without first assessing in detail all the software develo pment costs. The problem of hardware costs is becoming less cr itical , since the range of choice among small and medium- sized sys tems is now fairly wide and performance / price ratios often vary little among manufacturers. The choice of hardware can therefore be dictated by purely technical criteria in accordance with the applications envisaged. For software , o n the o ther hand. development costs may differ considerably, de- pending on wheth er a service company is engaged or the software is developed int ernall y. If the problem is clea rly defined and of an industrial nature (e.g., 102 laboratory admi nistration), the first of these alternatives will be preferab le - the essential aim being to obtain a final product whose operation and reliabil ity are guaranteed in the contract with the designer. In cases where the prob- lem is not adequately defined, however, it will first be essential to carry out feasibility studies, and perhaps produce prototypes, before deciding to set up the sys tem. CONCLUSION In conclusion it may be said that in developing and set ting up computerized in fo rmation systems the first thing to be done is to define the problems pre- cisely and to look at solutions in which the data are gath ered as near as possible to the point where they are created. As a first step, decentralized and virtually autonomous systems could conceivab ly be developed, adapted to one specific problem that can be reproduced time and time again. In this way, the costs of develo ping software can be reduced. At the second level, the systems must be developed and selected in ac- cordance with technological and economic criteria , but here also the installa- tion and operating cos ts are low relative to th e benefits that can hen ceforth be expected from low- or medium-power systems. At the final level , the important criteria continue to be administrative, and these depend on the health policy advocated at the regional or national level. In this situation the choice of system is de termined solely by the ob- jec tives and the financial resources available . REFERENCES I . Yaiiniinen, I. Decision making and resource allocation in the fi eld of health ca re experiences of Helsinki University Central Hospital Region. In : de Dombal, F.T . & Gremy, F., ed. Decision making and medical care: can information science help ? Amsterdam, North-Holland, 1976, pp. 42-60. ' Forsythe , M.J . Decision making in resource al loca ti on. In : de Dombal, F.T. & Gremy, F. , ed. Decision making and medical care: can info rmation science help ? Amsterdam, Nor th-Holland , 1976, pp. 63- 73. 3. Haro , A.S. Meth ods of determining commu nity health needs. In : de Dom- bal, F.T . & Gremy, F., ed. Decision making and medical care: can info rma- tion science help ? Amsterdam, North-Holland, 1976, pp. 311-3 I 9. 4. Fagnani , F. Mathematical mode ls in health planning. In: de Dombal , F.T. & Grcmy, F., ed. Decision 111aking and medical care: ca11 information science help ? Amsterdam, North-Holland , 1976, pp. 323-332. 5 . Ackoff, R.L. Towards a system of systems concep ts . Ma11agemen1 science. 17: 661 (1971). 103 6. Spencer, W. The health information system. Computer programs in bio- medicine, 5: 171-188(1976). 7. Le Beux, P. Frames selections systems and languages for medical applica- tions. Ph.D . Thesis , University of California, San Francisco, 1974. 8. Barnett, G .O. The modular hospital information system. In: Stacey, R.W. & Waxman , B., ed. Computers in biomedical research, Vol. 4 . New York & London , Academic Press, 1974. 9. Collen, M.F ., ed . Hospital computer systems. New York , Wiley, I 974. I 0. Lamson, B. Present possibilities for realizing a HIS. Computer programs in biomedicine, 5 : 2 I 5 - 245 ( I 976). 11. Reichertz , P.L. Implementing hospital computer systems: Medica l School of Hannove r Hospit al computer system. In: Collen, M.F. , ed . Hospital computer systems. New York , Wiley, 1974. 12. Gremy, F. Avenir de l'informatique medicate. Congres de la Societe des Electriciens. des Electroniciens et des Radio-electriciens. Grenoble, 20- 24 septembre 19 77. 104 13 TRU E MORBIDITY AND THE USE OF HEALTH CARE SERVICES 0 A. Fenton l ewisb Morbidi ty exists when a person is in a state other than that of pe rfect health . Such morbidity , however, is not necessarily perceived by the individual concerned , and in that case in formation may only be derived from population sc reening or other he alth surveys by examination. When morbidity is per- ce ived the individual will usually seek heal th care or ad minister se lf-care. Only the former gives rise to personal in fo rmation sui tab le for planning and manage- ment, since we have surprisingly little knowledge of the extent of self-care. Thus in a national hea lth service we have information on the use of various components of the service but such information relates only to a part of the total morbidity that must exist (Fig. I). { ot screened Not perceived creened ---------~ Morbidity Perceived - __ __. ~btain health care e ek health care ail to obtain health care i o not seek health care Self -care Fig. 1. Response pathways in morbidity. In a health service that is freely available, but limited by normal economic constrain ts in its capacity to provide , there will always be a level of unfulfilled 0 Th e views expressed in this chapter arc those of the a uth or and not necessar il y those of the Department of Health and Social Security . h Senior Medical Officer, Departmen t of Hea lth and Social Sec urit y, London, Un it ed Kingdom. 105 demand. Regional variation in this level is no indication, in itself, that true morbidity or need for health care also vary. This situation will continue , for mankind is adept at using available resources and the effective availability of health care cannot really be judged in the absence of adequate measures of effectiveness. We know that population sparsity makes equality of provision an ideal that we can seldom attain, and that the overall cost of provision is heavily weighted by the excess cost of looking after those in remote areas. On the other hand, high-density urban populations also render p .-ovision more expensive but perhaps on a more uniform basis. How do we balance these separate difficulties? Rural populations characteristically have a more in- dependent attitude to the natural hardships of life, and perhaps also to minor illnesses, but this does not justify inequality of provision. In this respect measures of self-perceived illness that do not necessarily lead to formal health care are of value, but we must look to the distortion that arises within these figures by variation in standards of education, wealth, age and family support. Self-perceived illness does provide a measure that may be considered independent of the level of health care locally available and concomitant measures of socioeconomic status allow us to identify other relevant factors. The results are interesting. Wales, for exan1ple, is characterized by a relatively high use of general practitioner services and this is reflected in a level of drug prescribing that is approxinlately 30% higher than in Eng- land. Against this, the urban areas of Wales have a standardized mortality ratio showing a I 0% excess whereas measures of self-perceived illness (from the General Household Survey) show a 23% excess for acute illness and 28% for chronic illness. We may confidently state, therefore , that there is a genu- ine excess of provision over need. The blemishes in this argument are many, but they may be diminished by examining the relative patterns of excess within different age and socioeconomic groups. For example, Forster (J) has revealed a relatively greater use of health care services in the higher social classes when the data are adjusted for differences in self-perceived morbidity within each class ; this is despite the much lower overall use of health services by the upper classes. It is not possible to interpret this finding simply but it provides a thought-provoking contribution to planning. The relationship between health care use and self-perceived illness in different social classes is but one aspect of a relatively complex relationship between need and provision, since differences also exist between age groups, the sexes, occupations, and rural and urban communities. In England and Wales (population about 50 million), for every hospital admission there are 10 outpatient attendances and 30 consultations at the general practitioner level, the latter occurring at a frequency of three per person per annum (2). There is, however , a large difference in balance between new and repeat at- tendances: in primary care 60% of all attendances represent a fresh episode of illness, whereas only 30% of outpatient attendances are new in the same sense. Much the same finding emerges from studies linking hospital attend- ances with general practitioner referrals. Here, approximately half those attending hospital have done so following a consultation with a general prac- titioner. The balance must largely be made up of referrals between hospital departments and it is difficult to escape the conclusion that many of these 106 conditions could be handled within primary care. There is, on average, one outpatient attendance per year for every member of the population. THE GENERAL HOUSEHOLD SURVEY The General Household Survey is sponsored by the Central Statistical Office as a central government tool and is conducted by the Social Survey Division of the Office of Population Censuses and Surveys (OPCS) (J). It aims to cover a sample of 10000 households, amounting to some 30000 individuals in Great Britain, and has been mounted annually since 1971. The sampling frame is based on the nine economic regions of England and Wales, as well as Scotland. The survey is conducted by an interviewer using a questionnaire (with a special version for children), and the main parts deal with family popu- lation and income, housing, employment, education and health. The health section succeeds the Survey on Sickness that covered the period 1943-1952, and it is now possible to relate all the information obtained in the health section to five main variables: age and sex ; marital status; employment status; socio- economic group; and educational qualifications and background. The health section has several main themes that do not vary greatly from year to year and these cover: (a) chronic sickness, with and without limitation of activity; (b) acute sickness ( over a 2-week period), with and without restricted activity; (c) general practice consultations (over 2 weeks), identifying the place and type of contact, the number of people consulting, and the number of con- sultations made per person; (d) use of health and welfare services ( over one calendar month), covering: heal th visitor community nurse chiropody services home help meals on wheels cent res for the elderly child health clinic/child welfare clinic social worker/child care officer/welfare officer; (c) inpatient treatment (over 3 months), covering the means of referral , the type of treatment and whether a National Health Service (NHS) or private patient; (f) outpatient attendance (over 3 months), covering the means of referral , the main symptoms or complaints, and the time on a waiting list; 107 (g) medicine takers (over 7 days, adults only) , covering the type of drug and whether on prescription or self-prescribed ; (h) smokers (adults only, but a self-completed questionnaire was available for those aged 16 and 17 years), covering the type of product, the number of individuals smoking, and the quantity per smoker; and (i) family size (for married women aged 34 years and under), including the age of children by the present marriage and the intended final family size. Questions on symptoms and disability have also been included successfuJly, although such results are not readily compatible with the International Classi- fication of Diseases. The health questionnaire has now probably reached the largest size that is likely to be effective, and the insertion of additional questions requires the re- moval of others. Morbidity trends are, however, relatively moderate and there is seldom need to repeat questions every year. Thus careful planning enables the health section to be used to cover a wide range of subjects on an intermittent basis. Although the study has no longitudinal element, there is good evidence of consistency from year to year despite the unreliability of such self-reported data . Already information obtained from this study has been used to supple- ment NHS data for planning purposes (for example, age-specific use of outpatient departments, since NHS-derived statistics do not include the characteristics of the user) . In primary care these self-reported data have shown good compara- bility with workload information obtained from general practitioners, and are free from the element of bias described below. The total annual cost of the survey is £600 000 (at 1978 prices) and the cost of the health section is about £50 000 per annum. Thus we have established in the United Kingdom a health survey by interview conducted on 0.5% of the population every year ; there is no real prospect of this survey being extended to include an examination component without a great increase in total cost. Annual reports are published by OPCS on this survey although, of neces- sity, these reports contain only limited tabulations of the data available . The published material from the health section amounts to some 40 pages and this leaves much valuable source data to be explored, especially by cross-tabulation with material from other parts of the survey. The Social Survey Division of OPCS has its own capacity to explore these data and, in addition, both the Department of Health and Social Security and academic units may obtain com- puter tapes for their own studies. It will be some time before we have assessed the full usefulness of this survey in health care. THE SECOND NATIONAL MORBIDITY SURVEY This second survey (the first National Morbidity Survey was made in 1954/ 55) covered 2 years beginning in November 1970. Data were collected continuously from 115 general practitioners in 53 practices caring for a popula- tion of nearly 300 000. The focus of the study was the Birmingham Research 108 Unit of the Royal College of General Practitioners (supported by the Depart- ment of Health and Social Security), which had pioneered the first National Morbidity Survey. Data processing was carried out by OPCS, which published the results of the first year of the Survey in 1974 (2) and those of the second year in 1979. An indexing system was used that identified separately, for each disease, first and subsequent consultations as well as referral to hospital for opin ion, admission or investigation. Although coded identifiers were used for trans- mission of data from the practices , it has been possible to make further retro- spective enquiries via the practice records. The survey enabled measures of disease incidence and prevalence to be calculated, as well as the number of patients consulting, the number of consu ltations or episodes, and the referral rates to hospital for each disease or group of diseases. This study provided an opportunity to examine the range of variability in both consultation rates and referrals to hospital, and confirmed that the largest single source of variability in hospital use was the differential referral rate from general practice. These data now exist as a means of investigating this variation, and to this effect the data may be Linked to provide a complete file of morbidity for each patient, which may further be combined with socio- economic data from the 1971 census. An elaborate procedure was adopted to achieve this link with the census by using a coded identifier ( the Hogben num- ber) based on the name and date of birth of the subject. This preserved the anonymity of the patient while being reasonably adequate for local linkage within the census enumeration districts covering any one practice. The survey made considerable demands on the participating general practitioners and it was inevitable that they should form a biased , self-selected group. For this reason no adequate regional sampling frame existed and it is reasonable to suppose that variability within this group may well be much less than that between all general practitioners. A smaller study has continued since 1972, providing a continuous population-based data base as well as weekly (and, if necessary, daily) returns of infectious disease. These latter returns are published regularly by OPCS and are available to the Department of Health and Social Security as a monitoring device. OBJECTIVES FOR THE THIRD NATIONAL MORBIDITY SURVEY IN 1981 The first two surveys confirmed that consultation patterns do not change dramatically with time and we now have, in add ition, self-reported information on both primary and secondary care from the annual General Household Survey. Thus planning for the third survey is directed to those aspects of health care that have not previously been exp lored and to developing an understanding of primary health care based on the earlie r studies. No country has resources so great that it does not need to appraise care- fully its health problems in orde r to use those resources to maximum effect. 109 The NHS assumed in 1948 responsibility for existing buildings, services and practitioners. Revenue allocations based on this historical disposition of facilities perpetuated regional and other geographical inequalities of provision, and it has taken many years to put resource allocation on a per capita basis unadulterated by the need to maintain existing, unbalanced services. The recognition of per capita need has introduced its own problems : (a) the need to take into account the relative cost of providing a unit of health care at different population densities (the sparsity of rural areas and the high density of urban areas both increasing this cost); and (b) the different needs of populations in different environments (such as industrial areas traditionally associated with high mortality and presumably high morbidity). Although it cannot be assumed that the use of local hospital services is a measure of relative need, general practitioner services do not suffer from such an obvious limitation of supply , and the general practitioner alone is able to assess the health of his patient as opposed to the morbidity that caused him to seek treatment. Although 67% of all patients attend their general practitioner at least once a year, little is known about the health of the remainder. Do they have a high threshold of discomfort or do they resort successfully to self-medication? The practitioner has the opportunity to study the health of this interesting group and to assess the extent of any illness that did not lead to a call on his services. We therefore see the survey in 198 I as an opportunity to study the health of individuals as opposed to their tendency to consult their general practitioner. Furthermore, it should be possible to devise a sampling frame that takes into account the 14 regional health authorities responsible for the administration of the NHS in England. We would also like to explore the consultation patterns of those general practitioners who are not self-selected by motivation to participate in such large data-collecting exercises. Although it is tempting to conclude that a good practitioner also maintains good records, 90% of all episodes treated by general practitioners probably require little more than a basic record stating the date, the problem and the action taken. Such a rudimentary but nonetheless adequate record should perhaps be seen as the basis for part of the survey. Otherwise the motivation demanded of the contributing practitioners may be diminished by concentrating on the I 0% or so of episodes that may be described as serious or significant. These episodes may be indexed using a limited list of diagnostic labels and yet provide much of the important information that arises within gene ral practice from the handling of non-trivial illness. CIIILD HEALTH A dual system of recording births has existed in the United Kingdom for many years. As with marriages and deaths, births are required to be registered 110 but there exists in addition a mechanism for notifying the Area Medical Officer of the birth, along with any relevant medical details that could affect the health of the child . At the local level there has always been an exchange of information between the two for checking purposes , but we are now attempting to transfer information on birth weight to the registrar so that this important risk factor may be made available as a national vital statistic. Fortunately this is an item of medical information that does not raise a confidentiality issue , but thought is also being given to the more difficult question of transferring data on the length of gestation and the parity of the mother. With the development of computerized information systems, notification of a birth to the Area Medical Officer now initiates an all-embracing record of child health care and is used, for example, to control the immunization pro- gramme. Notification is made by the (usually medical or nursing) attendant at the birth and this system provides nearly l 00% enumeration of births (most of which occur in hospital or nursing homes). Cross-linkage with the registration process ensures complete coverage and thus the computerized child care sys- tems suffer no potential losses. This fact alone worries those who value their individual right to privacy, but this is an example of a system that functions for the overall good of the community at small cost to personal privacy. At present this system is maintained until the child enters school, where the record will be used to initiate the school health record system. There is still some com- plexity in that the various preventive medical procedures (such as immuniza- tion) may be offered both by the general practitioner and by health authority clinics. The patient has freedom of choice and some information systems are managing to provide a service for this dual arrangement. CONFIDENTIALITY OF HEALTH CARE DATA Information about health care may be divided into that which is purely ad- ministrative and that which is clinical. The confidential nature of clinical informa- tion is self-evident, but attendance for treatment may in itself be a matter of some concern to the patient. In some cases the place of treatment is a good guide to the diagnosis and thus all this information needs to be treated under an um- brella of medical confidentiality. In the National Health Service a principle of ex- tended confidence is adopted, in that information about the patient is freely exchanged between all those health care professionals who may become respon- sible for him. This point is not always understood, although such transfer is quite essential to avoid duplication of effort at public expense. The information expert wishes to apply the data he collects to every pos- sible useful purpose but medical information is collected solely for the treat- ment and care of the patient. Clearly there is some difference of view here, and careful decisions have to be made on the balance between service to th e community and the privacy of the individual. Successive linking of even ts for any patient builds a more complete picture of his health care but also creates a more sensitive and confidential file. Even when the identity of the patient 11 1 is removed from the linked data , it may still contain unique features that ren- der identification possible. We adhere to the ge neral principle that medical data is not linked to any other information system and is not normally used for any purpose other than the health care of the patient. There are exceptions to this principle , usually in the interests of research, but such exceptions are made with care, always with the agreement of the treating physician, and if possible with the agreement of the patient himself. PLANNING AND RESOURCE ALLOCATION In planning a unified health service we have service-derived statistics of usage supplemented by less reliable data from surveys such as the General Household Survey and from studies within primary care . Data used for planning become a matter of public interest, and it is essential that such data should be proof against manipulation to influence planning. Thus we have the strongest argument in fa- vour of using mortality data to measure relative need or at least as a relative out- come measure . Imagine the impact of advice provided by the popular press to those interviewed in the General Household Survey if the regional results could influence that region's share o f the national resources! T o improve the value of data on extramural care they need to be linked to data on hospital care. Basic data on inpatient care is collected by regional health authorities in a system called hospital activity analysis (HAA). In most cases these data are rendered anony mous before they leave the hospital but some auth- ori ties collect identified data to permit longitudinal linkage. In neither case , how- ever, are these da ta made available to central govern me nt , and the 14 regional health authorities of England process their own HAA (an uniden ti fi ed 10% sam- ple of this information is used for natio nal statistical purposes). It is doub tful whether present resources will permit extension of HAA to cover ou tpatient at- tendances, as this would increase the magnitude of the task by a factor of IO and, in general, outpatient records are of doubtful adequacy for such a task. However, a link between the referring general practitioner and HAA is feasible and likely to be productive. The known variation (4) between referral rates for general practi- tioners and the influence that this variation has on hospital usage is a valuable contribution to the study of the balance of care. This would provide the essential link between community and hospital care and shed some light on the change of balance that occurs between age groups and between social classes. CONCLUSION Information systems grow like Topsy - we are for ever adding to them but seldom discarding that wh ich is redundant. For every human activity we must judge the value of the information system that we use in its manage- ment, because every sum spent on health information is a sum withdrawn 112 from immediate patient care. Therefore such systems must be developed in the knowledge that they will increase cost-effectiveness. Data gathering is only justified by data use and extramural care is an important field in which to apply this premise. REFERENCES I. Forster, D.P. Social class differences in sickness and general practitioner consultations. Health trends, 8: 29-32 (1976). 2. Office of Population Censuses and Surveys. Morbidity statistics from general practice. Second national study 1970- 71. London, HMSO , 1974. 3. Office of Population Censuses and Surveys. The General Household Sur- vey: introductory report. London, HMSO, 1973. 4. Crombie, D.L. Information for and from general practice. Proceedings of the Royal Society of Medicine, 70: 407-410 (I 977). 11 3 14 LIAISON OF ENVIRONMENTAL EXPOSURE DATA WITH MORBIDITY DATA0 A. Sjostrom b & P. Westerholm C It is well recognized that the interaction between environment and health - with both concepts used in a wide sense - is extremely complex. Studies on these relationships are being carried out in increasing numbers and their quality is improving. Results of descriptive and analytical scien- tific work in this field have demonstrated how disease and injuries are dis- tributed in populations in a way that is subject to both random and non- random influences. It is reasonable to assume that many of the differences and peculiar- ities in these distributions are in some way related to the environment. The term "environment" as used here embraces not only well-known and identi- fied hazardous agents but also other factors or groups of factors - except those of a genetic or constitutional nature - that may directly or indirectly influence health. Thus the environment also includes factors of a general nature such as air, water, food (dietary habits) and the working environ- ment. Personal lifestyle is also an important health determinant ; it consti- tutes the outcome of multiple interacting processes: socioeconomic, cultural, familial and individual. It is a well-known fact that income and education are related to health. All these circumstances have practical implications, one being that dis- eases shown to be associated with the environment in one way or another should not be regarded as immediately preventable. It is obvious, however, that an improved understanding of disease in populations constitutes a basis for a better understanding of the underlying mechanisms and for various types of policy options in planning for medical services and for setting re- search priorities. ° From the National Board of Health and Welfare, Stockholm, Sweden. b Head, Division of Statistics. c Consultant Epidemiologist. I I 5 EXPOSURE A first step in studying a relationship between exposure and disease is to determine whether their occurrence correlates with time between com- munities or other population groups available for study. A prerequisite for epidemiological research is access to sufficient, relevant and accurate data on both exposure ( or correlates of exposure) and the health endpoints or health indicators under study. This immediately involves the investigator in problems of dosimetry ; both exposure and disease must be reliably measured. This also applies to factors other than that being studied, since these often play a role in disease etiology. The interpretation of any association between a particular type of exposure and a disease is fraught with difficulty if these other factors are not properly taken into account. However , this may prove difficult; such factors may be unknown or, if they are known, their distribu- tion may be hard to ascertain in the population under study. As regards de- fining or describing the type of exposure, the problems usualJy encountered are of four types. First, one particular type of exposure may occur in many different subgroups of the population. Second , the exposure under study may coincide with the presence of many risk factors and risk modifying factors in the study population. Third , in population groups there may be large or small proportions of persons who are not exposed or who have no significant exposure. Fourth, adequate measurement methods for exposure may not be available. All these problems may be encountered in various types of descriptive epidemiological study, both where preformed popula- tion groups are studied and where selection of the study population is based on the same exposure criterion. The problem of dosimetry is generally most directly encountered when attempting to measure some defined type of exposure or agent in the environ- ment. For some environmental factors or risk modifying factors methods of sampling and analysis may be available and used to provide values for exposure on a continuous numerical scale. In other situations a noncontinuous scale may be used for correlates of exposure (occupation, domicile , socioeconomic group, etc.). These other chosen parameters may be more or less valid or relevant as correlates for the exposure under study, and this imposes a need to exercise caution in interpreting findings. Where analytical methods of measurement are available it is theoretically possible to achieve a high degree of precision in describing an exposure situa- tion. If a suitable formula is found for a measurement strategy with appropriate selection of points of measurement in space and time, the results can - at least in theory - be transferred to give a representative account of the exposure of individuals. Such methods are, however, usually costly and difficult. It should be remembered that situations with a constant exposure are extremely rare; for most types of environmental factor, exposure is highly variable. In fact, seen in an epidemiological context, exposure even to a weU-known agent that is available for detection and measurement is often highly complex. Exposure intensity (in terms of concentration), exposure duration and exposure frequency (in instances of intermittent exposure) all are important parameters. Thought 116 should also be given to the latency period that may be present for some groups of diseases, i.e., the time between exposure and clinical manifestation of the associated disease. MORBIDITY DATA Many of the studies on environmentally induced disease carried out so far have been based on mortality and related mortality statistics. There is an increasing awareness of the limitations of such an approach. In principle, the use of morbidity data and data from clinically oriented research opens up the possibility of using more sophisticated health endpoints or even physiological parameters as tools in environmental epidemiological research. For some speci- fic kinds of condition, such as cancer and births of congenitally malformed children, systematic information systems - in the form of registries - have been established in many places. For other illnesses, including many important chronic conditions such as arteriosclerotic vascular disease, routine data are scanty. The difficulties encountered in describing the morbidity or physio- logical parameters in this area are related to variability in the criteria used in diagnosis and problems associated with the establishment and implementation of standardized and uniform nomenclatures. Furthermore, various types of selection phenomena may cause problems; for example access to hospital data only may result in severe limitations in data collection as regards condi- tions or phenomena dealt with on an outpatient basis. As regards monitoring the health status of populations to identify high- risk groups or clustering of disease in subgroups of the population, one pos- sibility is to improve the standard of morbidity statistics. Furthermore, it is necessary to standardize nomenclature , criteria and definitions, thus making the morbidity data more reliable. The beginning of a life-long personal health information record could be the construction of a medical birth registry, since the first exposure to environmental risks takes place during pregnancy and delivery. It is then possible to gain further information from inpatient and outpatient statistics, special morbidity registers, health surveys, etc., provided the collected data can be linked to the individual. However, while we are improving sources of data about the environment and illnesses that may have a known or suspected association with it, we must make the best possible use of the limited data currently available. A MODEL f-OR EPIDEMIOLOGICAL SCREENING BASED ON INPATI ENT STATISTICS The introduction of the use of computers in administration has been of great importance in collecting data, including those for statistical use. The computer has also made it possible to apply statistical methods to large quantities of data. This development has not, however, been without problems for the user of 117 those statistics. TechnicaJ progress has made so much primary data available that , even after the statistician has structured and reduced the material, it is o ften still too bulky to be of immediate use. Another difficulty is that sta- tistics are largely processed on the assumption that the data are of relatively high quality - that is, they are reliable and reflect the true circumstances. Unfortunately this is not aJways the case. The material can of course be cor- rected , but correcting large quantities of data involves a great deaJ of time and often results in the user of such information suspecting its vaJ ue. Inpatient health statistics on episodes of illness are often of this kind - large quantities of primary data sometimes of doubtful quality. In particular, data on diagnoses may be poor - underreported, not satisfactorily coded and so on - so that the research worker may hesitate to use it. Moreover, although at the national and regional levels statistical means and yearly trends may be generated from basic data of this kind , it is very difficult to use such data at the area level, for example for monitoring and surveillance of health-related problems. Even in situations in which there are plenty of potentially useful data this aspect is difficult to handle . Method and test material A model is presented below on how to use dubious data in order to obtain valuable environmentally-oriented information of epidemiological interest. The method has shown promise and makes area-related surveillance possible . Since large quantities of data are available one can apply the same technique as used in mass health surveys. By using a screening method at the beginning of processing, the more important data can be selected and made more manageable . In 1964 the National Board of Health and Welfare in Sweden introduced statistics based on inpatient episodes, reported individually from the hospitals in one of the seven hospital regions, the Uppsala region , which contains about I 5% of the Swedish population. Today , statistics of this kind are produced by most Swedish hospitals and cover more than 900 000 discharges per year. The purpose of these statistics was primarily to provide a basis for better planning but, since such statistics include information about diagnoses , they are also used to a certain degree to obtain morbidity data. These statistics have been of great vaJue in the planning of the health serv- ices but have not been adopted extensively as a source of morbidity informa- tion. One of the reasons has been scepticism as to the quality of the data among scientists, who are used to working with small series of carefully col- lected , high quality data. Inpatient statistics from the Uppsala region for the years 1973 and 1974 were used in testing the model. The number of diagnoses reported during that period was 572 000. By means of the unique identity number given to each inhabitant of Sweden it was usually possible to ascertain whether a particul ar diagnosis was the reason for hospitaliza tion more than once during the period . Reduced to diagnoses per individual (a diagnosis was counted only once even if it was the reason for several episodes of hospitalization during the period) , the num- ber of "cases" in the test amounted to 4 29 000. Thus one o f the essential requirements for the model was present - a large quantity of data. II 8 The other requirement - debatable quality - was also at hand. The material was biased in several ways: people residing in the region may be treated in hos- pitals outside the region; the principles of reporting vary between different hospital units - some only reporting the most important diagnoses, others giving a more complete picture of the patient's troubles; the criteria of the diagnosis may differ; and the diagnosis may of course sometimes be wrong. The data processing also includes possibilities of bias, such as errors in punching. We wanted the highest possible differentiation by geographical area and cliagnosis consistent with the need to have a sufficient number of cases for analysis. To fulfil this purpose each geographical unit had to have a common distribution of other variables, which could be used as exposure backgrounds. The Uppsala region was divided into 4 7 local areas on the basis of the 100 or so registered local insurance areas. In the statistics the diagnoses were coded accorcling to a five-digit version of the Eighth Revision of the Interna- tional Classification of Diseases, adapted for hospital use. However, in trying to use this more detailed code, difficulties arose because many of the diagnoses were reported in nonspecified groups. It was assumed that the population of every local area was at the same risk of being sick and had the same access to hospital treatment. If this were so, the frequency of hospital cases with a certain diagnosis would be the same in each area. By comparing the mean frequency of diagnosis in the region as a whole with the frequency in any one area it would thus be possible to find differences based on chance or other factors. The influence of chance must be eliminated as far as possible by the use of statistical methods. A magnetic tape was produced containing the necessary data, inducting hospital, clinic, diagnosis (five-digit), sex, identifying number, month of dis- charge and area of residence. For the main data processing the items needed were diagnosis (three-digit), sex, age at latest discharge and area of residence. The rest of the data was produced in order to identify the patient in the hos- pital records, thus providing a gateway to further research. We know that the need for hospital care increases with age, especially after the age of retirement. The age distribution of the populations in the involved areas was not the same. In order to reduce these discrepancies the data were age-adjusted using five age groups : ,,;;19 , 20-64, 65-74, 75- 79, and ~80 years. The estimated number of cases (£) was calculated by taking the frequency of a certain diagnosis within a given age group in the region as a whole, projecting that frequency to the local area , and summing up. As there are differences in sickness by sex , calculations were made for both sexes and the total. The observed (reported) number of cases (0) was then compared with the estimated cases.a The difference between O and £maybe a Sg = to tal population of the region in the age group g Ilg = number of cases in the age group g of the region Pga = population in the age group g of the local area a Ea = estimated number of cases in the local area a = ~ ng. Pga Ea ,_, S g= I g 11 9 due to chance. In order to obtain a significant difference - with only one in a thousand ( or less) being caused by chance - the chi-square test was used and only data from differences that passed this test were listed for further ob- servation .0 Evaluation On finding an interesting divergence, the first step is to produce personal data from the first magnetic tape. The patient's case record is then studied and the patient himself or his physician may be interviewed. Unfortunately, case records often include very few data on, for example, occupation, working place and other environmental factors. Although hitherto considered more or less im- possible, it is necessary to invent and press in to use some kind of social case record for use in research on the influence of environmental factors. To obtain a better overall view of the material, a skeleton map of the local areas was constructed with lines drawn between connecting areas (Fig. I). Such maps were used to indicate the areas of interest for each iliagnosis. Using such maps drawn on transparent material to indicate areas with different ex- posures, correlations can be easily recognized. As exposure variables one may use , for example , the distribution of certain industries, climatic conilitions, availability of health resources, and urbanization. Results Published results are not yet available but the preliminary exercises seem to be promising. Most of the interest has, however , been focused on the func- tioning of the model. Certain diagnoses were found to depend on the hospital admission area and the model also reflected differences in admission criteria. Acute myocardial infarction was heavily over-represented in one area. The reason for this was an unusual use of the ICD caused by a lack of codes. This is an example of using the model to check the validity of the statistical data. Complicated iliagnoses tended to appear more frequently in the area where the regional hospital is situated. This showed that the model was working 0 The regular chi-square formula may in this case be simplified to: x 2 = (O-E)' (ldf) E .. In order to limit the probability of chance influence to 0 .001 or less , x2 has to be ;;, I 0.827. The formula will stand if £ ;;, 5 . However, this last condition may be an obstacle under certain circumstances. As an example, let us assume that there is a regional total of 20 cases with a certain diagnosis, all of them in the same local area. As E will be less than 5, this highly suspicious fact will not be registered. l11erefore another , and perhaps a little un- orthodox , condition was added: if I ,;;, £ < 5 and x 2 ;;, 16 the data were accepted for listing. 120 N q Fig. 1. Schematic map of the Uppsala region, with the areas represented by rectangles. A I ine between two rectangles means that the areas represented by those rectangles have a common boundary. properly ; this hospital is the most specialized in the region and consequently the population in the neighbourhood has the best access to specialists capable of making this kind of diagnosis. Another example of the usefulness of the model, which better illustrates the possibility of epidemiological follow-up, was in an area with many injuries as fractures, dislocations and contusions. Inquiries made at the hospital in the area showed a certain industry to be responsible for a high proportion of the injuries , the model thereby indicating the advisability of further research. Findings from this type of study will not normally allow any immediate conclusions to be drawn about causal associations between the environment and health endpoints. They must be followed by further studies, which should include a critical scrutiny of the validity of the data. 122 15 A METHOD OF MEASURING HOSPITAL ACTIVITY IN RELATION TO THE PATIENT B. Marielle 0 THE DEVELOPMENT OF HOSPITAL MANAGEMENT IN FRANCE When the activities of a hospital and the financial conditions under which it is managed are examined, one feature springs immediately to mind: the lack of any documentation providing a systematic link between the budget figures and the true medical significance of the hospital's work from the point of view of its patients. Improvement of budgetary and financial management Only a few decades ago, the typical French hospital was a charitable insti- tution financed by collections, gifts and special funds. Under these circum- stances hospital management was rudimentary, in terms both of financing and of economic management proper. Within the space of a few years, however, the hospital has evolved into a complex body governed by precise laws and regulations, especially as regards finances, facilities and organization. The Hos- pital Reform Act of 31 December 1970 gave substance and further impetus to this radical transformation. Among the various factors that have contributed to the marked improve- ment in the financial management of hospitals, two should be emphasized. l. Every public hospital is a legally accountable entity, with financial auton- omy, and as such is responsible for its own financial management and situation. 2. The system by which running the hospital and paying the salaries of medi- caJ staff is financed has a twofold basis: the charges for each day spent in the hospi taJ and the medicaJ fees. Hospital revenue thus depends on the hospital's activity, measu red in terms of the number of days patients spend in the hos- pital and of the medical procedures performed there. Hospitals are therefore burdened by severe financial constraints, which de- mand specific responses from hospital management in two forms : knowledge of a Co nsc iller technique, Bureau des Etudes et du Plan , Ministere de la Sante publiqu e, Paris, France. 123 the leve l of activity achieved (statistics on the hospital's activity); and control of the expenditure incurred in each centre of activity (analytical accounting). Gradually, in collaboration with the authorities responsible for their super- vision, French public hospitals have developed statistical and accounting tools with which hospital management can be greatly improved, at least from the point of view of costs and financing. The other facet of hospital management, however, which has the basic aim of contributing to a better economic view of the hospital, has not made the same advances. The need for a genuine concept of hospital economics At present no economic tool is available with which management of re- sources can be properly related to the definition of hospital objectives. Con- sequently, the concept of hospital activity is interpreted in different ways, depending on whether it is seen from the medical standpoint or from the so-called economic (but in reality , financial) point of view. In medical terms it is regarded as a set of diagnostic or therapeutic decisions on individual patients, but without any organized relation to the hospital resources needed as a result of those decisions . In financial terms it is considered to be a means of creating revenue. Accordingly, in social security terms it is conce ived as "production", the products being the days spent by patients in the hospital and the medical procedures performed. From this point of view, hospital act ivity is evaluated without regard to the patients and their special characteristics. Moreover , the charges per day in hospital and the medical fees do not reOect the real costs of hospitalization per patient, and so no significant improvement can be made in management of resources. , How then can harmony be achieved between the resources used and ob- jectives aimed at by hospitals if it is impossible (a) to describe the combina- tion of resources needed to undertake diagnostic , therapeutic and monitoring procedures, and (b) to relate hospital activity to the patient, and thus to the morbidity observed in the hospital? This difference in views on hospital activity, togeth er with the dichotomy between the hospital that provides medical (and perhaps social) services and the hospital that "produces hospitalization" makes it difficult to define a true concept of hospital economics, based on the relationship between resources and objectives. It may also cause serious misunderstanding between medical personnel and those involved with admin istrative and financial matters in the hospital system. The present state of information systems in French hospitals reOects both the progress ach ieved in hospital management and the difficulties still en- countered in efforts to give it its true economic dimension. Thus, quite apart from the question of the evaluation of hospital care and hospital activity in general, there is no relationship between morbidity data used for national statistics and the costs of hospital services , as measured by analytical accounting and by means of the documents used for breaking down expenditure within each hospital. Nevertheless, some work has been carried out, with the aim of proving that there is, or is not, a good correlation between hospital morbidity and the costs it generates. 124 It is true that many factors other than the state of health of the popula- tion affect hospital costs. The "cost of disease", expressed as an absolute value, can therefore only be the cost of particular treatments given to particular pa- tients in a clearly specified public health system. This gives rise to difficulties of method, which should be resolved before any radical reform of hospital management is attempted. These were the considerations that led the Minister of Health to embark on a unique experiment in which hospital activity was measured as the relation- ship between input and output. THE HOSPITAL ECONOMICS APPROACH : AN EXPERIMENT WITH ACTIVlTY CHARTS AT THE GUSTAVE-ROUSSY INSTITUTE, VILLEJUIF This experiment was carried out in 1976/77 by the General Directorate of Public Health, in collaboration with the Cancer Control Centre of the Gustave- Roussy Institute, Villejuif, an establishment that is highly specialized in terms of its aims (research, treatment and teaching), the techniques it employs ( cobalt therapy , X-ray therapy, chemotherapy, etc.), and the pathological conditions it is endeavouring to control (benign and malignant tumours, which are classified with high precision). By presenting the objectives, methods and main results of the experiment, we hope to give a concrete idea of what is meant by the hospital economics approach mentioned earlier and to show that this type of economic assessment is feasible in France. The aim of the experiment The aim was to develop management tools that would reflect the true operation of the hospital and enable resources (personnel, equipment and buildings) to be fully utilized and adapted to the needs of the population served. These indicators of true hospital operation would then serve as a basis for determining hospital activity at any moment in time, in relation to the number and characteristics of the patients , as well as for evaluating hospital activity in terms of the progress made towards good utilization of resources. As a direct consequence it would then be possible to simplify financial management ; for exan1ple, the hospital could receive a block alloca- tion from the authorities responsible for social security and social assistance (block budget system). The new management tools were to be based primarily on a nomen- clature of the activities performed by hospital departments, activity being understood in the sense of a coherent combination of resources aimed at achieving a certain objective or at least fulfilling certain functions. Several types of indicator would be related to these activities, namely : indicators of physical resources (man-hours and degree of utilization of equipment and 125 premises), which could later bet ran slated into financial resources; and indicators of performance, i.e., an evaluation of the services "offered". A record of activities, related to the individual patient, would provide a valid basis for establislii11g a relationship with the morbidity observed in the hospital and would be a first step towards an evaluation of the services "ren- dered" by the hospital, or its impact on the demand for care in its own particu- lar area. Thus, in its aims the study was essentially a macroeconomic investiga- tion, with the aim of adjusting supply to demand in the field of health care. Methods The backbone of the work done at the Gustave-Roussy Institute is the establishment of a new nomenclature of activities, which goes beyond the nomenclature of medical procedures on which payment for medical services is currently based; this con .:ept of "activity" provides the link between data on morbidity and data on h-ispital resources and thus on costs. In addition to the t:stablishment of a nomenclature of activities adapted to the needs of the Instil ute, a study was made of the con tents, frequency and future use of activity charts provided for the various departments and for the directorate of the hospital. The establishment of a new nomenclature of activities When the Ministry of Labour and Health launched the "programme budget" project in 1974 , the aim was tc, re111edy the deficiencies of the traditional budget by breaking down expenditurr in relation to the aims pursued, rather than by types of expendit urc (pcrsonn ·I. equipment, subsidies for operation or facilities, etc.). The programme concept was therefore fundamental, being defined as a coherent combination of resources entrusted, over a number of years, to the leader of a project or projects designed to achieve specific objectives. The list of programmes arranged in order, covering all the budgetary allocations, the administrative departments and their activities. without omission or duplication, constitutes the programme structure of the central administration. It appeared that this method, or at least the essentials of the method, could be used for analysing hospital activity and for working out a new manage- ment system for a hospital. There were two reasons for this view. I. The hospital, like the administration, ei11ploys various means for the attain- ment of objectives, by a complex process in volving the laws of technology and science as well as the laws created b_v all organizations . .., By bringing together all the parties concerned in the establishment of the programme structure, this method allows the conditions to be created for a genuine dialogue between medical personnel and administrators. The approach adopted at the central administration involved a progression fro111 the general (groups of programmes) to the particular (individual pro- grammes and then their constituent clements). In this experi111ent, however, 126 the reverse approach was chosen, starting from the elementary activities within each unit and ending with the laying down of overall programmes (or treat- ment plans). The new nomenclature of activities is based on analysis of the "elementary activities" of the units. These elementary activities are combinations of re- sources that are easy to identify , and that are not specific to the disease or to the patient,a e.g., radiography, taking of samples, or consultations. These elementary activities are grouped into "complex activities", which are carried out specifically in relation to a particular disease and patient, e.g., cobalt therapy or hysterectomy. At the inter-unit or departmental level, these com- plex activities are themselves grouped into "phases of treatment" of the pa- tient, in accordance with a "programme of therapy", e.g., hysterectomy com- bined with postoperative hospitalization, breast radiotherapy, etc. The phases of treatment are the constituent elements of a "complete treatment" of the patient at the Gustave-Roussy Institute, in accordance with a "therapy proto- col or plan" drawn up by a medical committee consisting of a surgeon, a radio- therapist and a chemotherapist. There are 15 committees at the Institute, corresponding to the various possible sites of primary cancer in the human body (breast, face, etc.). The aim is therefore to establish a "programme structure" for each committee; these structures, superimposed on each other and aggregated, make up the total activity of the Institute. It should be noted that the existence of systematic and well-defined treatment protocols for malignant tumours contributed greatly to the suc- cess of the work. The method was relatively easy to apply in technical units, such as sur- gical and radiotherapy units, where the organization of work is centred on surgical operations and irradiation procedures that can rapidly be identified. In inpatient units, however, the analysis of hospital activities presents major methodological problems. The multiplicity of the elementary procedures involved in patient care, catering, and dietetic and secretarial services makes it difficult to establish a "production function" in these sectors, owing to the diversity of the factors (medical, psychological, cultural and social) affecting the nature and extent of the services offered. An experiment carried out at the St Justine Hospital, Montreal, Canada, in which the anticipated workJoad of nursing and auxiliary staff was evaluated daily in relation to the care provided, suggested that days spent in hospital could be broken down in accordance with a classification based on the various combinations of elementary procedures performed by the nursing and auxiliary staff. A number of operational studies conducted in the United Kingdom have reached a similar conclusion. Five broad classes were thus distinguished, based on the average time devoted by nursing and auxiliary staff each day to the direct care of each inpatient. By 0 The distinction between "disease" and "pat ient" here takes account of the fact that the patient has specia l cha racteristics (age, sex, etc.) in addi•ion to his pathological condition . 127 means of such a classification, a "hospitalization prof11e" showing the daily magnitude of direct nursing care can be plotted for each type of pathological condition and each type of treatment. Similarly, the establishment of a nomen- clature of activities in which the phases of treatment and plan of treatment are defined opens the way for analysis of the "t reatment prof11e" of each unit , each department and the hospital as a whole. The analyses of treatment profiles and of the way they develop, taking into account both the characteristics of the patients and the hospital resources actually employed, will be submitted periodically to the persons responsible, in the fom1 of "activity charts". Activity charts in hospital management The structure of the activity charts is as follows: (a) a segment headed "Patients and Activities" contains information on the population prof11e , i.e. , the special characteristics of the patients , the treat- ments they have received, and the medical and paramedical activities performed in relation to them ; (b) a segment headed "Resources" describes the resources available to the unit (personnel , facilities, premises) as well as the day-to-day use of medical resources and laboratory services within the unit ; (c) a segment headed "Analysis of Deviations" is intended to present a dynamic view of hospital activities. From one quarter to the next , deviations will in fact become apparent in the perfom1ance of any one unit; there will be changes in the size and struc- ture of the population treated, in the number and nature of the treatments performed, and in the quantity and quality of the resources employed. In addition, the deviations between what was forecast and what was actually achieved will be quantified and studied annually. By analysing the deviations from past performance and the deviations from annual forecasts it will be possible to evaluate the unit in terms of its own objectives. In this way various phenomena can be monitored : I . The effec t of variations in patient characteristics on the types of treat- ment specified for patients . Thus, in a medical unit, an increase in the num- ber of brain cancer patients will cause increases in the number of radiothera- peutic procedures performed , in the average length of hospitalization , and in the number of days of Types 2 and 3 (intensive care by nursing and aux- iliary staff). This may ultimately result in greater mobilization of paramedi- cal personnel. 2. The effect of changes in the treatment protocol on the leve l of utiliza- tion of resources. Many therapeutic trials are carried o ut at the Gustave-Roussy Institute, espec ially in the field of chemotherapy. In consequence, patients with the same medical characteristics (same tumour site, same stage of the disease, etc.) do not necessarily receive the same treatment in one period as in 128 another. This may be illustrated by the example of a therapeutic trial initiated in the field of surgery , with the aim of replacing the combination "tumour ex- cision + cobalt therapy" by a simple surgical operation - mastectomy. As a result , there will be a tendency to increase the number of surgical operations necessitating slightly longer postoperative hospitalization, together with a ten- dency to reduce slightly activities in the field of cobalt therapy. The effects of this change on the resources to be employed in the surgical block , the inpatient department and the cobalt therapy unit can be measured and presented in the charts of the units and departments concerned. 3. The effect of variations in staffing and resources in general on the intake and treatment of patients. The resources available may influence the policies adopted with regard to the admission and transfer of patients in the various units. They act as a constraint when the demand for care exceeds the capacity of the unit ; when the reverse is the case, however, they may induce an artificial increase in the unit's activity. For example , there may be a relative short- age of manpower during holiday periods. If the characteristics of the patients and their treatment remain unchanged, the unit will tend to retain patients longer so that the care needed in excess of its capacity can be spread out over a period of time. This will be shown clearly by the changes in the number of standard inpatient days. As a result, the admission of new patients will be delayed, and the unit's objectives may not be achieved if this problem is not anticipated. 4 . Effects on other units. In the example given above, it can be foreseen that the surgical units and radiotherapy units may soon experience a decline in activity because, for reasons beyond their control, some of their resources are not being used. As laboratories provide services for all other medical units, it seems essential that the use oflaboratory services, blood products and medical sundries should be recorded in the charts. The difficulties in such an analysis of deviations arise from the fact that several phenomena, sometimes with opposite effects, may occur simultaneously; in an inpatient unit, for example, any increase in the gravity of a patient's condition will have an effect on the intensity of direct care, but this effect may be offset by a lengthening of the average period of hospitalization, owing to a reduction in paramedical personnel. INTRODUCTION OF TH E NEW INFORMATION SYSTEM AT THE GUSTAVE-ROUSSY INSTITUTE Data collection and processing In French hospitals today there are many types of information system that have developed in response to various needs, notably: the need for individual invoicing of medical services and hospital services in general ; the need to keep track of a patient throughout his period in hospital (patient record card, records 129 of examinations and treatment given by hospital units, temperature chart, medical files, etc.); the demand for a continuous supply of data to ensure that treatment is unified and properly carried out; and the importance of moni- toring the activity of individual units by means of periodical statistics. It is understandable, therefore , that the introduction of a further data collection and processing system for better analysis of hospital activities was greeted with some initial reservation by all staff involved. To allow for these problems, ways had to be sought of "grafting" the new system for measuring hospital activity onto existing documents and systems, while preserving the essential criteria for good management analysis, namely: reliability and completeness of the data collected; and the possibility of col- lating and perhaps aggregating all the data on a patient, irrespective of the period of hospitalization. The method tried out at the Gustave-Roussy Institute was based on three main ideas: (a) adapting the existing data files of the "medical record" type, so as to record the activity of surgical blocks, radiotherapy units and pathology lab- oratories; (b) establishing a simple system for recording the activity of inpatient units and devising a computer program for processing the data obtained; and (c) studying the link-up and compatibility problems posed by the existence of several different data files. These ideas were examined in close collaboration with the medical and para- medical personnel, as well as the administrative personnel and computer specialists at the hospital. The solutions arrived at, based on compromises between all the parties involved in hospital activity, demonstrated that it was possible to bridge the traditional gulf between "medical information" and "accounting infor- mation". This is illustrated by the way data on the activity of inpatient units are collected, using a slip that is stapled to the patient's temperature chart and sent to the management analysis unit at the end of the patient's stay in the hospital. Use of the activity charts By means of these charts , management analysts can provide decision- makers with a whole range of quantitative and objective information as a valuable aid to decisions on policies concerning admission and transfer of patients within the hospital , taking into account the objectives it has been set, and the organization and development of services. In fact, the periodical results should serve two purposes. First , they should facilitate annual and medium-term planning of the resources available to the hospital. In this respect , the activity charts may be used in the discussion of the budget (in talks between the heads of units and the hospital management, and between the latter and the super- visory administration). Second, they should reveal specific problems in organiza- tion and ultra-short-term planning of hospital resources. 130 The value of management tools of this kind, however , goes beyond the presentation of quantitative results. Making the medical profession sensitive to the financial implications of its decisions, and the initiation of administrators in the day-to-day problems of the medical profession should open the way for a fruitful dialogue bringing new ideas on hospital policy. On the other hand, without the cooperation of the medical profession and administrators, the charts might be nothing but empty, useless frames. CONCLUSION A method based on that used for drawing up programme budgets in public administrations has been successfully adapted in the development of a system of activity measurement and management analysis for a hospital. The method presupposes, and indeed makes possible, a system of hospital management based on definition of objectives at all levels of responsibility, and on know- ledge and evaluation of concrete achievements and the resources needed for these achievements. The underlying principle of the method is that the way to achieve proper utilization of hospital resources, with due regard to the characteristics of pa- tients, is to analyse the deviations observed between forecasts and performance, or even between performance in different periods. It thus opens up possibilities for a new concept of relations between hos- pital units and the hospital management, and also between the hospital and its supervisory bodies: "progress agreements" can then be made between the l)eads of units and the hospital management, laying down the objectives for each party, taking into account both the foreseeable demand for care and the existing level of resources (staff, available beds , technical facilities, etc.). Phy- sicians thus have a direct means of becoming involved in the management of their units, as the presentation of their "treatment profiles" may clarify their own assessment of their activity . In public administrations, programme budgets provide the framework for "programme analyses". In the same way, this method of measuring hospital activity allows alternative methods of treatment to be studied for the same pathological condition. The experiment at Villejuif reveals the process by which expenditure is generated in relation to the objectives of hospital care. It gives reason to hope that one day a true system of economic analysis may be set up in all hospitals, so that the relationship between hospital resources and their social efficacy can be properly understood. Any attempt to apply such an experiment to all French hospitals would encounter insurmountable obstacles, both financial and methodological. Can a programme structure be laid down for every hospital? How can such proce- dures be harmonized with existing statistical , information and management analysis ~ystems? These problems, which make it so difficult to grasp and analyse 131 in a coherent way all the phenomena acting on and within the hospital, ex- plain the diversity of approaches in relation to the hospital system. The value of the experiment at the Gustave-Roussy Institute lies in the fact that it widens the range of ideas on this subject and thus makes a modest but original contribution to the current debate . This is especially so with regard to hospital charges and their reform; the Gustave-Roussy Institute is taking part in trials of the so-called "total budget" concept, which is designed to en- sure regular and complete financing of the hospital. 132 WHO publications may be obtained, direct or through booksellers, from: ALGERIA: Soc1ctc Natoonale d'Ed111on et de Diffusion, 3 bd Zirout Youccr. AtGlfRS ARGE1'TINA: Carlo. Hirsch SRL. Flonda 165, Galerias Guemes. Escntono 453/465. Bl ENOS AIRES Alll>'TRAI.IA: Ma,/ Order Sales · Au51rahan Government Pubhshing Scn.·1ce, P.O._ Box 84. CANBER~ A.C.T. 2600; or m·er ,h, counter from Australian Go..,·crnmcnl Pubhshmg ServlCC Boo~shops at: 70 Alinga Street. CA~IERRA Crn A.CT. 2600; 294 Adelaide Street. BatSBANE, Queensland 4000; J47 Swanston Street, MEIIO< RNE, VI{' 3000; 309 Pitt Street. Svo~EY. N.S.W. 2000: Mt Newman House 200 St George\ Terrace, PFRTH, WA 6000; lndus1ry House. 12 Pi~ic Street: ADELAIDE, SA 5000. 156-162 Macquane Street . HoBAu, TAS 7000- Huntcr Pubhcatoons, 58A G1pps Street. C1xuNGwooo. VIC 3066 AUSfRIA: Gerold & Co .. Graben 31, 101 I V1FNNA I BANGLADFSH: The WHO Programme Coordinator, G.P.0. Box 250, 0Ac-< A 5 - The Association of Voluntary Agencies, P.O. Box 5045, DAc-<A 5 BELGIUM: Office 1nlcrnational de Librairic, 30 avenue Marn1Jt, 1050 BRl'SSELS - Suhs,·np11ons 10 World Health onlr • Jean de Lannoy. 202 avenue du Roi, 1060 8Rl'SSELS BRAZIL: BibhOleca Regional de Med1cina OMS/OPS. Unidade de Venda de Publ1caQOCS. Caixa Postal 20.381. Vila Clemcntino. 04023 SAo PAlH>. S.P BURMA: .,., India. WHO Regional Offoce CA.NADA: Smxlt> and hull,.; copies of ,ndfridua/ puhlmwons ( not su/>. S(-np11on.r~: . Canadian Public Health Associa1ion. 1335 Carling A,enuc. Suite 210. OTTAWA, Ont KIZ 8N8. Suhsmpwms · Suhscnp- tum ordt·n. arcompamed hr c·hl'qul' madt' ()UI w the Royal Bank of Canada. OTTAWA, Account World Hcallh Organ1z.at1on, should ~ u·m 10 tht World Heahh Orgamzalion. P.O. Box 1800, Postal Stallon 8. OTTAWA, Ont. KIP SRS. Com·spondenre wnummx suhsc-r'lption.s tlwuld hf' addrts .. u-d 10 tht World Heallh Organization. Distribution and S.tle\, 1211 GE!',,ifVA 27, Switzerland CHINA: China National Publications lmpon Corporation. P.O. Box 88. BEUINC, (PEKING) C'OLOMBIA: D151nlibros Ltd. Pio Alfonso Garcia. Carrera 4a. Nos Jl>-119. CARIAG>NA CYPRl:S: Publisher,,· D1stnbutors Cyprus. JO Dcmocratia,, Ave Ay10> Dhomctoou,. PO Box 4165. NICOSIA CZECHOSLOVAKIA: Ania. Ve Smcckach 30. 11127 PkAGl'' I DENMARK: Munksgaard Ltd .. Norrcgadc 6. 1165 Cllt'ENHAUN K ECUADOR: L1brcria Cientil1ca SA. PO Box 362, Luque 223, Gl'AYAOlll EGYPT: Nabaa El Fikr Bookshop. 55 Saad Zaghloul Street. Al.l:.XAl''ol>RIA EL SALVADOR: Lobrcna Estud1an11I. Ed1l1cio Comcrcoal B No 3. A\.emda Lihrnad. SA" SALVAIX>R t'IJI: The WllO Programme Coordinator. PO Box 113. Scv. FINLA.!'1D: Akatccmincn K1rjakauppa. Ke\ku,katu 2. 00101 HEUl""KI 10 f"RANCE: L1brJirie Arneue. 2 rue C'asimir-Dclavigne. 75006 PARIS GERMAN DEMOCRATIC REPUBLIC: Buchhau, Lc1pllg. Po51fach 140. 701 LEIPZIG Gt:RMANY, FEDERAL REPUBLIC OF: Govo-Vcrlag GmbH. Gonn- hc1mcrmassc 20. Postlach 5360. 6236 Esrn-N - W. E. Saarbach. Postfach 101610. Follerstrassc 2. 5000 KOLN I - Alex . Horn. Sp1cgcl- gassc 9, Postlach 3340. 6200 W1lSBADlN GHANA: fide, Entcrpme,. P.O. Box 1628. ACC-RA GREECE: G. C. Eldtheroudakis S.A .. Librainc m1crna1ionale. rue Nik" 4. A lltEM (T 126) HAm: MJx Bouchcreau. Librairic .. A la Caravelle''. Boite pos1ale I 11- 8. PtMtT•,\ll•PRINCE HONC KONC: Hong Kong Go\.ernmcnt lnforma11on Service\, Bea- consfield Hou5C, 6th Floor. Queen\ Road. C'en1ral. VK'TORIA HUNGARY: Kultura. P.0.B. 149. BUDAPEST 62 - Akadcm1ai Kony,,,,._ bolt. Vaci utca 22. Bt·OAPEST V ICELAND: SnaebJorn Jonsson & Co .. P 0 . Box 1131, Hafnarnraeti 9. RE't'~AVl"- INDIA: WHO Regional Office for South-Ea.\! Asia. Wor1d Health House. lndraprastha Estate. Rong Road. NEW DUHi I 10002 - Oxford Book & Statoonery Co .. Sc1nd1a llousc. NEW DELHI I 10001; 17 Park Street, CALU'TTA 700016 (S11/,.aKe'11) INDONFSIA: M/s Kalman Book Service Ltd. Jin Cik1ni Raya No. 63. p O Box 3105/Jkt .. JAKARTA IRAN: Iranian Amalgamated D1s1ribu1ion Agency. 151 Khiaban Soraya, TEHlRAN IRAQ_: Mm1,1ry of Information. National House for Publishing. Distri- hulmg and Advert1Mng. 8AGHDAll IRELAND: The Statoonery Office. D,·BuN 4 ISRAEL: Hc,hgcr & Co .. 3 Nathan Strauss Street. JERl'SAI u, ITALY: F.d111oni Minerva Med1ca. Corv.i Bramante 83-85. 10126 Tl!RJN; Via Lamarmora 3. 20100 M1uN JAPAN: Maru,en Co. Ltd. P.O Box SOSO. ToomJ lntcrnatoonal. 100-31 KOREA, REPUBLIC OF: The WHO Programme Coordinator, Cen- tral P 0. Box 540. SEOUL KUWAIT: The Kuwait Bookshops Co. Ltd. Thunayan AI-Ghanem Bldg. PO Box 2942. KLWAIT LAO PEOPLE'S DEMOCRATIC REPUBLIC: The WHO Programme Coorchnalor. P.O. Bo-. 343. VIENTIANE LEBANON:~ Levant Di51ributors Co. S.A.R.L. Box 1181, Ma•da"1 Street, Hanna Bld11, BEIRlT LUXEMBOURG:. Libra inc du Centre, 49 bd Royal. L, XEMBO< R!i MALAWI: Malawi Book Scn11ce, P.O. Box 30044. Chichoti, 8LANTYRt 3 MALAYSIA:~ WHO Programme Coordinator. Room 1004. Fit,pa- trick Building. Jalan RaJa Chulan. Kl ALA L, MPl R 0~2 - Jubilee (Book) Store Ltd, 97 Jalan Tuanku Abdul Rahman, P.O. Box 629. Kl:At.A LCMP\IR 01-08 - Parry's Book Cenlcr. K. L. Hillon H,>tel. Jin. Trcachcr. P.O. Box 960. Kl'ALA Ll MPl'R MEXICO: La Prcnsa Medica Mciicana. Ediciones Cienlificas. Pasco de las Facultades 26. Apt . Postal 20--413. MtxKl> CITY 20. D.F. MONGOLIA:"' India. WHO Regional Office MOROCCO: Editions La Pone, 281 avenue Mohammed V, RABAT MOZAMBIQUE: INLD. Caixa Postal 4030. MAPl·To NEPAL: "' India. WHO Regional Offoce NE111ERLANDS: N. V. Maninus Nijhoffs Bockhandel en Uitgc>e" Maatschapp1J. Lange Voorhout 9. TH, HAU E 2000 NEW ZEALAND: Governmcn1 Prinling Office, Mulgrave SI.reel. Pri- vale Bag. Wn.UN<..iTON I. - Go1·emmt•n1 &x.,kJhops at: Ru1land Street. P.O. 5344. AtntANll; 130 Oxford Terrace, P.O. Box 1721 , CHRJST<"Hl RC·H: Alma Sl.rect, P,O. Bo-. 8.57. H4MILT01'; Princes Slrcct, PO Box I 104. Dt ""'N - R. Hill & Son. Ltd. Ideal House. Cnr Gil- lies Avenue & Eden SI., Newmarket. Al'('Kl.AND 1 NIGERIA: Uni>enlty Bookshop Nigeria Ltd. Uni>en1ty of Ibadan. IBADA~ - G O Odatuwa Publishers & Booksclle" Co .. 9 Benin Road, Okirigwc Junction. SAPfl £, BE"'-DH STATE NORWAY: J. G. Tanum A/S. P.O. Box 1177 Scntrum, O.w I PAKISTAN: Mirza Book Agency, 65 Shahrah-E-Quaid-E-Azam. P 0 Box 729. LAHORf 3 PAPUA NEW GUINEA: The WHO Programme Coordinator. PO. Box 5896. BOROKO PHILIPPINES: World Health Organi1.a11on. Regional Office for the Western Paciloc. P.O. Box 2932. MANllA - The Modern Book Com- pany Inc., P.0 Box 632. 926 Rizal Avenue. MANILA POLAND: Skladn101 Ksi~garska. ul Mazowiccka 9. 00052 WAMSAW {t.tupt {H'nod,calsJ - BKWZ Ruch, ul Wronia 23. 00840 WARSAW ( ~ric"1irals onl) ') PORTIJGAL: Livraria Rodrigue\, 186 Rua do Ouro. Llsao""' 2 SIE.RRA LEONE: Njala Univer,uy College Bookshop (Uni,ers1ty of S.crra Leone). Pnvale Mail 8a(l. FRUTOW""' SINGAPORE: The WHO Programme Coordina1or, 144 Moulmem Road. G.P.0. Bo• 3457. St-.c,APO .. I -Select lll••h (Piel LJd. 215 Tanglin Shopping Centre. 2/f-'. 19 Tanglm Road. Sl'C.Af'Otu: 10 SOlJlll AFRICA: Van Scha1k\ Bookstore (Ply) Ltd. P 0. Ill» 724. 2611 Church Street, Pltnoa1A 0001 SPAIN: Comcrcial Atheneum SA. Con,cJo de C1cn10 UO-IJ6. BARn• LO""'A 15; General MoscardO 29. MADRID 20 - L1hrena 01.u de Santos. Lagasca 95. MADRID 6: Balmcs 417 y 419. BAanwNA 6 SRI I.ANKA: su India. WHO Regional Office SWEDEN: Akt,ebolagct C. E. frit,es Kungl. Hovbokhandel. Rcgcnngs- gatan 12. I0327 SronHOtM SWTTZERI.AND: Medizinischcr Verlag Han, Huber. U,ngga." Stra,,c 76. 3012 BERN 9 SYRIAN ARAB REPUBLIC: M. Farras Kckhia. P.O. Box No. 5221 , ALEPPO lliAILAND: .,., India. WHO Regional Office ru:u~~:A: SocictC Tunisicnnc de Diffusion. S a\.·enue de Carthage. nJRKEY: Hasct K1tapcvi. 469 l\11k.lal Caddcso. Bcyoglu. ISTANBl'I UNITTD KINGDOM: H.M . Statooncry Office : 49 High Holborn, LoNDO~ WCIV 6HB: 13a Castle Street, ED4Nat,aGH EH2 JAR: 41 The Hayes. CARDlfl Cfl IJW: 80 Chic~•• Street. BELFAST IITI 4JY; Brazennose Street. MANC'HESTER M60 SAS; 258 Broad Street. BoaMINCiHAM Bl 2HE: Southey House. Wine Street, BatsroL IISI 2BQ All ma,/ ord,r, .,hould l>t· .wnt ,,, P.O. Box 569. I.ONDON S1::1 9NH UNITTD STAn:s OF AMERICA:. Sing/, and hu/4 wpi,., of 1nd1- r1dual puhli<·a11onJ (nul sult.t(:nptwn.s): WHO PuhhcalK.ms Centre USA. 49 Shendan Avenue. ALBANY, N.Y. 12210. Suhsmp111m.< Sul,. S<:nptwn ()rdl'n, aCTompamed hy rht'rA madt out w tht• Chemical Bank. New York. Account World Health Organizatior. . . ,hou/d I>,, .'il'nt 10th, World Health Organization. P.O. Box 5284. Church Street Sta- tion. NEW YOltK, N.Y. 10249. C1Jrrtsp<mdl'nu conctmlng Juh.~·riptwn.t should 1w addrr.urd ta tht World Health Organiza1ion. Distrihulion and Sales. 1211 GENEVA 27, Switzerland. Puhliratwm an, also a,·ail• ah/, from th, United Nations Bookshop, NEW YORK, N.Y. 10017 (n,tail onh) USSR: For rradrrs ,n thl' USSR rrquirinK Russian rduionJ.· KomM> molskij prospckt 18, Medicinskaja Kniga. Moscow - For rtaden <mts1dt' 1hr USSR rtquirlng Russian t'dition.s.· Kuzneckij mOSI 18. McJ:dunarodnaja Kniga. Mose-ow G-200 VENEZUELA: Editorial lnteramcricana de Venezuela C.A.. Apanado S0.785. CARACAS 105 - L1brcria del Este. Apartado 60..137. CARAC·AS 106- L.ibrcria Medica Pam. Apanado 60.681, CARAUS 106 YUGOSLAVIA: Jugo.lo•enska Knjiga, Terazije 27/11. 11000 BELGRADE ZAIRE: Librairie uni\'ersitairc. avenue de la Pa1i N° 167. 8.P. 1682. KINSHASA I Special terms for developing countries are obtainable on application to the WHO Progrsmme Coordinators or WHO Regional Offices hsted above or to the World Health Organization, Distribution and Sales Service, 1211 Geneva 27, Switzer• land. Orders from_ countries whe~e sales agents have not yet been appointed may also be sent to the Geneva address, but must be paid for In pounds sterhng, US dollars, or Swiss francs. Price: Sw. fr. 11.- Prices are subject to change without notice. C/1/fJJ i f