PHN- 8720 THE FINANCING AND ECONOMICS OF HOSPITALS IN DEVELOPING COUNTRIES: KEY ISSUES AND RESEARCH QUESTIONS by Anne Mills December 1987 Population and Human Resources Department World Bank The World Bank does not accept responsibility for the views expressed herein which are those of the author(s) and should not be attributed to the World Bank or to its affiliated organizations. The findings, interpretations, and conclusions are the results of research supported by the Bank; they do not necessarily represent official policy of the Bank. The designations employed, the presentation of material, and any maps used in this document are solely for the convenience of the reader and do not imply the expression! of any opinion whatsoever on the part of the World Bank or its affiliat'=s concerning the legal status of any country, territory, city area, or of its authorities, or concerning the delimitations of its boundaries, or national affiliation. PHN Technical Note 87 20 THE FINANCING AND ECONOMICS OF HOSPITALS IN DEVELOPING COUNTRIES: KEY ISSUES AND RESEARCH QUESTIONS A B-S T R A C T In recent years, the attention of policy makers and researchers has been directed at primary health care, with little notice Laken of hospitls except to observe that they take a disproporLionate share of hai:n see resources. However, it is precisely because they are larga consumers ot resources that more attention needs to be paid to them by polic makeus aud resparchers. EL is conventional and tradi Lionl to describe develop ing country health systems as hospital-dominated, with the bulk of resources devoted to hospitals in urban centers. Casual investigation and observation certainly suggest this is the case. Yet the higher levels of a health system have an important supportive role in terms of supervision and referrals. Thus relevant questions are what mix of facilities is desirabic, how can they be most economically provided, and how should resources be shared between the different levels of facility. This paper has two main aims: to pull together what data are ava on the financing and economics of hospitals in developing countries; and to identify research questions and a possible research programme. The paper first discusses some of the problems of data availability and comparability. ft then reviews data on the hospital share of health sector resours, the functional breakdown of hospital expenditure, hospital factor mix, Lie extent to which hospital expenditure is distributed equitabLy, and wheLher the characteristics of hospital expenditure have been changing over LWme. Subsequent sections consider the burden of hospital recurrent costs and LNhe scope for increasing hospital efficiency through organizational and procedural changes. Patterns of hospital income are then considered, and particular attention paid to the potential for shifting the burden of hospital financing away from governments, through cost--recovery in the government sector or greater use of non--government or private sector services. In each section, a summary of the present situation is rliowed by I discussion of issues and research questions. The paper ends by outlining a possible research programme in the field of hospttal financing and economics. Prepared by: Anne Mills London School of Hygiene and Tropical Medicine December 1987 THE FINANCING AND.ECONOMICS OF HOSPITALS IN DEVELOPING COUNTRIES: KEY ISSUES AND RESEARCH QUESTIONS TABLE OF CONTENTS Page 1. Introduction 1 2. Sources and Definitions of Data on Hospital Financing 3 3. Patterns of Hospital Expenditure 5 3.1 Hospital Share of Health Sector Resource Use 5 3.2 Functional Analysis of Hospital Expenditure 13 3.3 Hospital Factor Mix 14 3.4 The Equitable Distribution of Hospital Expenditure 18 3.5 Longitudinal Analysis 21 3.6 Issues Arising and Research Questions 22 4. Hospital Recurrent Costs 25 4.1 The Recurrent Cost Burden 25 4.2 Recurrent Cost Issues and Research Questions 26 5. -Government Planning and Budgeting Mechanisms 29 5.1 Present Mechanisms 29 5.2 Planning and Budgeting Issues and Research Questions 30 6. The Sources of Income of Hospitals 37 6.1 Present Sources of Income 37 6.2 Issues Arising and Research Questions 43 7. Proposed Research Program 52 Tables 58 Reference 72 @ 4 LIST OF TABLES List of Tables Table Pa e 1 Hospital Expenditure as a Proportion of Total Health Sector Expenditure (Public and Private) 58 2 Hospital Expenditure as a Proportion of Total Current Health Sector Expenditure of Government 59 3 Hospital Share of Annual Development (Capital) Expenditure on Health - 60 4 Hospital Expenditure as a Proportion,of National Government Health Facility Expenditure 61 5 Hospital Expenditure as a Proportion of District Health Facility Expenditure 62 6 Ministry of Health Current Expenditure by Type of Hospital 63 7 Unit Costs by Type of Hospital 64 8 Hospital Current Expenditure by Function 65 9 Hospital Current Expenditure by Main Category of Input 66 10 Share of Hospital Current Expenditure on Salaries and Medical Supplies by Function 67 11 Distribution of Hospital Costs Between Capital Services and Recurrent Costs 68 12 Health Service Fees and Agricultural Wage Rates 69 13 Revenue from User Charges as a Share of Expenditure on Government Health Services 71 -．·，--. *&―떼，―」 THE FINANCING AND ECONOMICS OF HOSPITALS IN DEVELOPING COUNTRIES: KEY ISSUES AND RESEARCH QUESTIONS 1. INTRODUCTION The aim of this paper is to identify and explore the issues associated with the financing and economics of hospitals in developing countries in order to identify research questions and a possible research programme. It looks at questions such as what share of health sector resources do and should hospitals absorb? How should the share of resources for hospitals be shared between different types of hospitals? What is the factor mix of hospitals and how does this vary between hospitals and countries? What proportion of hospital resources do the various ho ;pital functions consume? Given the significance of hospitals within the health sector, how can government planning and budgeting methods be improved to increase control over hospital behavior and increase hospital efficiency? What are the sources of income of both public and private hospitals and what effect does their mode of finance and payment have on their behavior? These are all issues which have been the subject of debate for many years. It is conventional and traditional to describe developing country health systems as hospital-dominated., with the bulk of resources devoted to hospitals in urbdn centers. Casual investigation and observation certainly suggest this is the case. Yet it is increasingly recognized that the higher levels of a health system have an important supportive role in terms of supervision and referrals. Thus the question is what mix of facilities is desirable, and how should resources be shared between the different levels. 2 We have very little data to investigate this issue. Health service budgeting and accounting systems throughout the, world are ill-suited to investigating the allocation of resources to functions or purposes and few inter-country comparisons have been made of hospital financing. Therefore one aim of this paper has been to pull together what data are available in order to draw out issues and research questions concerning the financing and economics of hospitals in developing countries. Section 2 explains the data used in the various tables of this paper, and discusses some of the problems of comparability that need to be kept in mind when reviewing the tables. Section 3 then uses the data collected to review the hospital share of health sector resources; the functional breakdown of hospital expenditure; hospital factor mix; the extent to which hospital expenditure is distributed equitably; and whether the balance of expenditure has been changing over time. Section 4 investigates the burden of hospital recurrent costs and considers how countries might take into consideration the recurrent cost implications of hospital developments. Section 5 then discusses government planning and budgeting systems, to consider the scope for increasing hospital efficiency through organizational and procedural changes. In Section 6, attention turns to patterns of hospital income. The data base here is extremely poor, and the majority of the discussion concentrates on issues concerned with the potential for shifting the burden of hospital financing away from governments, either through cost-recovery in the government sector or greater use of non-government (NGO) services or the private sector. In each section, a summary of the present situation (in so far as it is known) is followed by a discussion of issues and research questions. Research 3 topics are brought together in Section 7, which outlines a possible research programme in the field of hospital financing and economics. 2. SOURCES AND DEFINITIONS OF DATA ON HOSPITAL FINANCING In order to obtain data relevant to questions of hospital financing, a brief search was done of published and unpublished literature. Time and access problems limited the search to English language sources and to the more readily available French sources. An important consequence was that the Latin American literature was neglected, except where available in English language reviews. A major problem then arose in interpreting and analyzing this data, since sources are often not explicit in defining exactly the costs quoted. In particular, there is confusion over: o the scope of the definition used for health sector expenditure: are non-Ministry of Health, health-related expenditures such as water and sanitation included? o the definition used for health service expenditure: are private as well as government expenditures included? o the definition used for government health expenditure: are the expenditures on health services of government agencies other than the Ministry of Health included? o the categories of expenditure included: are the value of services from capital included, or just operating-expenses? o is all resource use, whether it appears in budgets or not, included: is the value of resources supplied in kind omitted? 4 0 the category of hospital to which the data. refers: are they secondary or tertiary facilities? The data collected have been tabulated and appear at the end of the text. Where there are definitional problems, this is mentioned in the notes to each table. -The notes also provide data sources. If the data presented were available in that form in the original source, only the reference to the source is given; if the data had to be re-calculated, the notes refer to "data from" the reference. Where question marks are entered in the columns of tables, this indicates that that piece of information has not been separated out and is included in other columns. Where observations (e.g. cost per inpatient) are available for a number of hospitals and a mean is required, this is calculated from the figures for each hospital rather from aggregated data (i.e. total costs and total inpatients). The former is referred to as an "unweighted mean" since it reflects the variation in the measure of interest between hospitals, and is not weighted by the share of any particular hospital in the total activity. Countries differ considerably in the extent to which Ministries of Health offer complete coverage of the population, and in some countries (especially Africa) NGOs such as religious agencies provide complementary services. Since these are increasingly being linked into the network of government services (for instance through subsidies and organizational ties), the sum of Ministry of Health and mission expenditure has occasionally been used as a better indicator of total public health service resource availability than Ministry of Health expenditures alone. When categorization of expenditure by hospital type is required, the categories of "central", "general" and "district" have been used.' "Central" 5 refers to a hospital with a full range of specialists (and often teaching responsibilities). "General" refers to a hospital with a narrower range of specialists, but which may be a regional center for referrals from more local hospitals. "District" refers to a hospital which provides general medical and surgical inpatient services usually staffed by general medical officers not specialists, and receiving patients from and supervising community and basic health services. "Rural", used less frequently, is a smaller institution, usually without a medical officer in charge. 3. PATTERNS OF HOSPITAL EXPENDITURE The significance of hospital resource use can be assessed in a variety of ways. However, the data base is unfortunately limited, first because many countries cannot easily separate out hospital from other personal health service expenditure, and second because figures for the total against which hospital expenditure is to be compared are rarely available. The analysis below reflects these data constraints. 3.1 Hospital Share of Health Sector Resource Use The purpose of investigating the proportion of health sector resources that hospitals absorb is to answer the question "is the current share of financial resources allocated to hospitals and between different types of hospitals economically efficient?" Economic efficiency would require that resources be allocated between the various activities of the health sector in such a way that their effect on health is maximized. Unfortunately, there is no easy way of determining whether this is the case since the outputs of the health sector are not easily defined and measured unless the 6 analysis is concerned with a specific disease preventive or curative activity. In contrast, hospitals provide a complex mix of services, tackling a variety of different disease problems and producing outputs which vary in terms of the dimensions of morbidity, mortality, disability and quality of life. Research has therefore tended to concentrate on more limited but more measurable elements of hospital performance. One approach is to analyze the cost-effectiveness of particular hospital treatment patterns (see, for instance, Echeverri et.al. 1972 on the scope for increasing efficiency by replacing hospital care for certain conditions by a mix of hospital and home care). Another is to look at the characteristics of patients attending particular types of hospitals, to see whether they are being cared for at the "right" level (in terms of medical standards). Heller, for instance, suggested that for a health system to be efficient in the economic sense, no patient should be treated at or referred to a higher level if acceptable diagnosis and treatment is possible at a lower level (Heller, 1978). He proceeded to investigate the referral patterns, origins of patients and case-mix of different hospital levels of the Tunisian health system. This type of analysis is reviewed in other papers. Since the focus of this paper is on hospital financing, the section below concentrates on analysis of the quantity of resources flowing to hospitals (that is on inputs rather than on outputs) to see whether any common pattern is evident and whether there are some countries where hospitals are more or less dominant, suggesting that further research on the pattern of care lying behind this might be profitable and might provide guidance for other countries on the most efficient pattern of care. 7 Hospital expenditure as a proportion of total health sector expenditure. If the focus of interest is on alternative ways of distributing resources within the entire health system, including both health services and health- related activities, and both public and private agencies, then an appropriate analysis is to look at hospital expenditure as a proportion of total health sector expenditure. Unfortunately, few countries have produced this evidence and only two could be found that have recent information in an appropriate form (see Table 1). Because of the shortage of evidence, Table 1 includes the results from the international survey of health expenditure published in 1967 (Abel-Smith 1967) and from a survey of a number of developed countries published in 1981 (Maxwell 1981). There are considerable dangers in comparing data from different countries and periods because of the problems of identifying hospital expenditure and differences between countries in hospital functions. However, a rough comparison of the Abel-Smith and Maxwell data (especially for those countries common to both surveys) shows a clear tendency for the hospital share to rise over time. Thus while the Abel-Smith data is now 25 years old, it indicates the balance of expenditure at a time when developed country health services were less hospital and technology intensive, and perhaps closer to the pattern of health care at present prevailing in developing countries. Unfortunately, neither this survey nor more recent sources provided data on Latin America. All figures except the first exclude water and sanitation. Nutritional services are included in the Malawi data but excluded elsewhere. If water, sanitation and nutrition were to be included throughout, the share of hospitals would be reduced. None of the information on teaching and research is likely to be reliable. Despite'these data problems, it is possible to 8 conclude that in general, hospitals in developing countries may absorb around 40-50 percent of health sector expenditure. The data is far too patchy to identify countries that are out of line with the general pattern, and it would be useful to know whether Latin America conforms to this pattern. Table 1 does not show for developing countries the extreme hospital dominance that might be expected. An important reason for this is the inclusion in total sector expenditure of private expenditures, which in developing countries are usual-ly substantial (for instance 31.5 percent in Malawi) and the bulk of which are spent on non-hospital health services. A different pattern is shown by analyzing government health sector expenditure (see below). Hospital expenditure as a proportion of total current health sector expenditure of government. Table 2 shows a functional analysis of current government health sector expenditure. It should be noted that the figures for Indonesia and Colombia apply to specific geographical areas within those countries, and thus are likely to underestimate the hospital share since specialist referral hospitals will be excluded. In general, approximately 50-60 percent of current government health expenditure appears to be absorbed by hospitals. The richer the country the higher the share tends to be, as indicated by Malaysia, Colombia, Jamaica and Tunisia. Historical patterns of development are another influence: Zimbabwe, for instance, has inherited a sophisticated urban hospital system, and Papua New Guinea a relatively undeveloped system. The Nepal data suggests a third influence: 26 percent of expenditure is devoted to prevention (mainly malaria control). Countries such as Nepal, which have inherited a substantial 9 disease control programme, will devote a relatively smaller share of total resources to hospitals. Hospital share of annual development (capital) expenditure on health. Annual capital or development expenditure can be a misleading indicator of the proportion of investment funds absorbed by hospitals, since hospital construction is by its nature lengthy and often irregular. On this cautionary note, Table 3 presents recent data for four countries, and contrasts this with information from the 1967 survey. As might be expected, the hospital share varies enormously. It is interesting to contrast the relatively low shares of the countries for which recent data are available with the Abel-Smith survey. The latter probably reflects the large-scale hospital construction programmes that developed countries began in the 1960's; the former the relatively limited donor funds at present available for hospital construction and the priority given to extending coverage of primary health care. It should be noted that the Abel-Smith survey excluded capital expenditure on private physicians' offices, and thus overemphasizes the hospital share in those developed countries where primary health care is substantially provided by private physicians. However, if we look just at developing countries, a considerably higher share of capital expenditure was being spent on hospitals in the early 1960s than is being spent now (contrast, for example, the figures for Tanzania and Tanganyika). Hospital expenditure as a proportion of national government health facility expenditure. Hospital facilities and basic level health services (health centers, dispensaries, etc.) should in practice complement each other, with basic health services acting as a filter for hospitals. It is 10 therefore of interest to look at the distribution of expenditure between health facilities, excluding public health, prevention, training and overhead administrative expenses. Table 4 suggests considerable differences between countries, and is likely to be less affected by definitional problems than earlier tables (although the Brazil data represents a very approximate division between "hospital" and "primary"). Botswana and Malawi, and to a slightly lesser extent Tanzania, Zimbabwe and Malaysia, appear to have relatively well-developed primary-level facilities, in contrast to.Tunisia and Brazil. The domination of hospital expenditure is again shown, with 60-80 percent of facility expenditure devoted to hospitals. Hospital expenditure as a proportion of district health facility expenditure. The organization and distribution of health services in rural "districts" has received much more attention recently (Vaughan, Mills, Smith, 1984). A standard district usually contains a hospital with a network of peripheral facilities. Accounting systems, even if districts are given identifiable budgets, usually do not distinguish between hospital and non- hospital expenditure. The data in Table 5 were obtained from reports of special surveys. They confirm the picture from Table 4, and suggest that around 70 percent of district-level facility expenditure may be absorbed by the district hospital. The explanation for the different picture in Zimbabwe is suggested by Table 6 below which indicates that the district hospital level is poorly developed relative to general and central hospitals. Ministry of Health current expenditure by type of hospital. So far, hospitals have been treated as a single category. However, the type of hospital affects both its accessibility (central hospitals being few in number and concentrated in large cities) and cost (higher level hospitals 11 are likely to be more expensive). Table 6 reports data on the distribution of Ministry of Health current expenditure by type of hospital. These figures will be influenced by a number of factors. For instance, expenditure on a central hospital in a small country (such as Malawi) is likely to represent a larger share than in a larger country (such as Tanzania) since a certain size and staffing level is required whatever the size of the country. The level of urbanization is another influence, and a country with a number of large urban centers (e.g. Zimbabwe) is likely to have more higher level hospitals than a country with a more dispersed population. Table 6 provides evidence supporting the assertion that health sectors are dominated by large urban hospitals. With the notable exception of Tanzania, well under 50 percent of hospital expenditures are devoted to the more accessible district hospitals. Sixty to eighty percent are absorbed by central and general hospitals. Alailima and Mohideen (1984) comment in these terms on Sri Lanka: "The significance of tertiary-level services is evident in the high proportion of capital and recurrent expenditure allocated to these institutions -- 60 percent and 36 percent (of central government health expenditure) respectively. The Colombo Group utilized half of total tertiary-level health expenditure in 1979. One quarter of (government health) recurrent expenditure and only one tenth of (government health) capital spending went to secondary level curative institutions." The higher level hospitals not only absorb a larger.share of hospital resources than district hospitals, they also cost relatively more per unit of intermediate output. Table 7 expresses the cost per unit of intermediate output (per inpatient, inpatient day and outpatient visit) as a multiple of the unit cost in a district hospital. With the exception of Malawi (where 12 the "general" hospital is more like a large district hospital and is very overcrowded, reducing unit costs), general hospitals cost more per unit of output than district hospitals, and central hospitals more than general hospitals. Data are inadequate to draw any firm conclusions, but inpatient care in general hospitals can be up to twice as: expensive as district hospitals, and central hospitals between two and five times as expensive as district hospitals. The reasons for these cost differences are likely to vary and will be explored in other papers in greater detail. Costs will depend on size, the level of sophistication of services, the quality of services, the mix of services offered, case-mix and case-severity, occupancy rates etc. For example, those costing studies which disaggregate unit costs by speciality indicate the extent to which different specialities can give rise to different unit costs. In four regions in Peru, pediatric: hospitalizations were two to three times as expensive as gynecology, surgical hospitalizations three to four times as expensive, and medical hospitalizations two to six times as expensive (depending on the region; Robertson 1985). Within one hospital in Colombia, cost per inpatient day was approximately one and a half times as large for internal medicine cases, and twice as large for obstetric cases, than for pediatric cases (Robertson, 1985). The difference was much "less marked per discharge for obstetrics, presumably because obstetric patients had short lengths of stay. Of particular interest is whether ownership influences hospital unit costs, and in particular whether NGO or private hospitals provide similar services at a cost lower than that in government facilities. This is a largely unresearched topic in developing countries, with only isolated 13 observations. For example in Lesotho, the average cost per inpatient was reported to be almost three times higher in public than private hospitals, and similar differentials existed for hospital and clinic outpatients (Smith 1980). This pattern was not repeated in an investigation by the author of hospital costs in Malawi. The following range of costs was revealed.: Mission (Kwacha) Government (Kwacha) Hospital 1 7.10 Central hospital 1 12.43 Hospital 2 6.50 Central hospital 2 8.91 Hospital 3 4.10 General hospital 4.18 Hospital 4 4.00 District hospital 4.56 Hospital 5 2.60 Mission hospitals 1 and 2, in terms of services offered and case-mix, were probably similar to a general hospital, whereas mission hospitals 3, 4 and 5 were similar to district hospitals. 3.2 Functional Analysis of Hospital Expenditure Hospitals perform a range of different functions: for example, provision of inpatient treatment services within various medical specialties, specialist and general outpatient care, medical and paramedical support services, and other support services such as administration, cleaning etc. Given the cost differences described above and the possibilities for decentralization of certain hospital functions, it is important to know the balance of resources that are absorbed by the different functions. Unfortunately, very little functional analysis has been done of hospitals in developing countries, and 14 in Table 8, only a breakdown between inpatient and outpatient care could be presented. The table indicates the very substantial outpatient functions of hospitals. A common pattern appears to be for outpatient care to absorb approximately 20 percent of hospital current expenditure, although this proportion is much higher in the three Chinese examples. The patchy evidence suggests that the less specialized the hospital, the greater is its outpatients' role. 3.3 Hospital Factor Mix In the absence of competitive pressures forcing hospitals to adopt a minimum cost pattern of production, hospital factor mix is likely to depend on a number of influences including manpower training policies, the supply of manpower, and nationally acceptable medical practices. Prices of manpower, drugs, and supplies, etc. may or may not be a strong influence on choice of factor mix but will influence the relative shares of different types of input. In developed countries, it is common for a very high proportion of hospital current expenditure to be absorbed by wages and salaries, partly because of the highly trained manpower required and because wages and salaries tend to be high relative to non-labor inputs (Kravis, et.al. 1978). In contrast, in developing countries skilled labor in sufficient numbers may be unavailable (especially in Africa) and cheaper substitutes may be used, and wages and salaries are often relatively low. For example, many African countries pay (at least) world prices for drugs, but much less than developed countries for skilled and unskilled labor. It might be expected, therefore, that salaries and wages would consume a smaller share of total hospital expenditure in developing than in developed 15 countries. This is borne out by Table 9. In a number of African countries the share is 40-50 percent, with Sri Lanka and Thailand slightly higher and Papua New Guinea and Jamaica around 70 percent. Additional data from middle- income countries might indicate whether there is a general tendency for the share of wages and salaries to increase as national income increases from low to middle income levels. The Colombian data in the table may well be atypical since it relates to a psychiatric hospital. Of the other categories of expenditure, medical supplies would appear to consume around 20 percent, and utilities 5-10 percent. Other items of expenditure are more minor, except food when provided. China is a clear exception to the above generalizations. Its manpower share is unusually low, and medical supplies share exceptionally high. Without knowing more about the prices, functions and operational pattern of Chinese hospitals, it is difficult to explain this satisfactorily, though sources comment on widespread over-prescription of drugs (Jamison, et.al. 1984). The data are inadequate to look in any detail at variations in factor mix between hospitals of different types. There is a suggestion that the more specialized the hospital, the lower the share of salaries and wages, perhaps because buildings are more sophisticated, requiring relatively greater expenditure on support services and utilities. In many situations, however, this may well be offset by use of a more expensive manpower mix. In Sri Lanka, the factor mix of private hospitals (called nursing homes) is different than that of government. Compared to government curative services, private nursing homes have low wage and salary, drug and diet expenses (Alailima and Mohideen, 1984). No comparison is, however, presented 16 of the case-mix and case-severity of government and private facilities, so it is not known whether these are in fact comparable. Indeed, variations in the functions of hospitals may well explain some variations in factor mix. A few references present expenditure*shares separately for inpatients and outpatients (see Table 10). The picture, however, is very varied. Salaries and drugs together take up a higher share of outpatient than inpatient expenditure, presumably because of the absence of "hotel" costs and relatively fewer investigative procedures. In general, drug expenditure absorbs a higher share of outpatient than inpatient expenditure but a clear pattern is not discernable for expenditure on salaries and wages. Given this general pattern, the greater the outpatient function of a hospital, the greater is likely to be the share of drug expenditure, and the lower the share of nonsalary and drug expenditure. Since manpower usually absorbs the greatest share of hospital expenditure, it would be useful to disaggregate salary and wage expenditure by manpower type. Unfortunately, little information of this nature is available. In Botswana (all hospitals) doctors and dentists took up 12 percent, nurses 55 percent and technicians 7 percent of total hospital salary and wage expenditure (MOH Botswana, 1979). In Jamaica the shares were 7 percent for doctors and 43 percent for nurses for all hospitals, though for the main specialist MOH hospital the shares were 22 percent and 43 percent (Cumper, 1982). In Sri Lanka, there was a similar pattern with the higher level hospitals showing a greater skill intensity for both doctors and nurses. Also in Sri Lanka, Alailima and Mohideen (1984) comment that "the private sector, for a similar level of activity, uses scarce resources (physicians) more efficiently than 17 does government, albeit with a higher utilization of a less skilled type of labor (nurses)." Information on capital costs is notably lacking from most studies, so the preceding analysis has been conducted in terms of recurrent costs only. It is possible, however, to show the distribution of hospital costs between capital services and recurrent items for those cost studies which include an annuitized capital cost. Table 11 shows great disparity between countries in the share of hospital costs absorbed by capital services. At least part of this may well be due to differences in methodology: for instance, it is likely that some of the studies include only depreciation and not foregone capital earnings, though this is not always possible to tell from the studies. It is clear from the sources that the Uganda and Malaysia studies fully account for all capital costs. The difference between the share of capital estimated in each study can perhaps be accounted for by the nature of the buildings: in Uganda, these were of simple construction and style, whereas Malaysian hospitals are likely to have used more sophisticated materials and techniques. In many countries, the foreign exchange implications of government expenditure are likely to be an important consideration. Of all the cost studies reviewed only one, the Sri Lanka study using an input-output method of analysis, estimated the import effects of various types of health sector expenditure (Alailima and Mohideen, 1984). They comment that "health sector expenditure on Western medicine, both recurrent and capital, was characterized by its high import content and wide variability among health service levels. The most significant consumers were the specialized campaigns and the tertiary 18 level institutions in the public sector, recurrent expenditure on imports (mainly drugs) accounted for about 20 percent of the gross output at tertiary and secondary levels and less than 10 percent of the gross output at primary levels.....The import content of capital expenditure for the secondary and Ayurveda services was relatively low -- about one-quarter of all purchases; at all other levels it reached over 50 percent, because the domestic capital goods industry is not geared to provide the complex medical equipment and vehicles required by the health sector". 'Taking account of both direct import effects and the indirect impact on imports through domestic expenditure on goods and services, import intensities (ratio of total imports generated to gross output) were 0.26 to 0.28 for secondary level institutions and private nursing homes And 0.37 for the tertiary level. 3.4 The Equitable Distribution of Hospital Expenditure The equity implications of hospital expenditure patterns have already been touched on, in the sense that the distribution of expenditure between tertiary, secondary and primary levels has implications for the accessibility of health services. It is also possible to look more directly at equity, by analyzing firstly the geographical distribution of hospital expenditure and secondly the incidence of hospital expenditure, that is the relative consumption- levels -- in terms of cost of production -- of different income groups. Geographical distribution. In most developing countries, the distribution of hospital expenditure -- for instance in terms of expenditure per district - - is likely to be very unequal. Many countries have not, for instance, completed their network of district hospitals, so some districts will be without any hospital or with a very small one. Moreover, the smaller the 19 geographical unit of analysis the more unequal will be the distribution since a hospital may be planned to serve more than its immediate geographical area. Thus inequality can only be identified in the context of agreement on the appropriate catchment population of a hospital. Few countries know the extent of inequalities. Data on geographical distribution is hard to come by since many accounting systems do not disaggregate expenditure by district or region. The literature search revealed three country analyses. In Malaysia, hospital expenditure per capita of the district population showed very large differences (range M$ 0.00 to 50.15, mean 6.04 if districts with a general hospital are excluded; Heller, 1975). Taking the state as the base for analysis gives a more realistic picture since general hospitals are intended to serve the whole state. State expenditure per capita varied from M$ 7.00 to 24.88 (unweighted mean 12.81). Similar data from Papua New Guinea shows a mean of Kina 5.13 (range 1.43 to 33.49) for all provinces and means of Kina 6.35 (range 3.46 to 9.41) for provinces with a base hospital and Kina 3.38 (range 1.43 to 6.42) for provinces with a provincial hospital (Baker 1977). In contrast, per capita hospital expenditure for the province containing the capital city was Kina 33.49. In contrast to Papua New Guinea, Malawi shows less extreme variation (MOH, Republic of Malawi, 1984). MOH hospital expenditure per capita in one of the two districts with a central hospital was seven times greater, and in the other district three times greater, than the mean for seven districts with district hospitals (in contrast to a ten-fold difference in Papua New Guinea). District MOH hospital expenditure for the seven districts with a district hospital varied between KO.32 and 1.38, mean 0.77. It is interesting to note that mission hospital expenditure to some extent compensated for 20 this variation. Including mission expenditure gives a range of KO.68 to 1.41, mean 1.15, thus reducing the ratio of the highest to the lowest district from 4.3 to 2.1 None of the above analyses take any account of time and travel costs for users, which are likely to be greater where the area that the hospital covers is larger. The potential size of transport costs is indicated by data from a mission hospital in Uganda (King, 1966). Expenditure by patients took up 10 percent of the total cost (hospital plus patient) of inpatient care, and 75 percent of the total cost of outpatient care. Incidence. Hospitals may be used very unequally by different socioeconomic groups within the hospital catchment area. Thus, per capita expenditure per potential user gives only a rough indication of equity. Little research has been done of the socioeconomic characteristics of hospital users. However, Heller was able to put together his cost analysis of hospitals in Malaysia with data from a survey of socioeconomic characteristics of inpatient admissions in seventeen hospitals (Heller, 1975). Heller concluded that the rural poor were under-represented in the hospital user population, and that this underrepresentation was less in the urban population. Taking the population as a whole, only 20.9 percent of hospital inpatients were from households in the poverty bracket compared to 39 percent for the entire population. The principal groups overutilizing hospital inpatient services relative to their share in the population were upper income groups. Only in relatively urbanized regions was there adequate representation of the poor, suggesting that urbanization, probably a proxy .for accessibility, is a key determinant of inpatient service utilization. 21 Similar studies in other countries would be enlightening, especially if focussed on the extent to which the equity objectives of a nationalized health service are in practice achieved. Where government services are perceived to be of poor quality and alternative private services exist, the demand from higher income groups may be diverted to the private sector. For instance, in Sri Lanka, the per capita benefits accruing to low income groups from government Western health services far exceeded their expenditure on private health care, while the highest income group spent about double what it received from government services on private medical care (Alailima and Mohideen, 1984). 3.5 Longitudinal Analysis A cross-sectional analysis of expenditure shares or factor mix at one point in time is of value to researchers and policymakers who wish to determine appropriate patterns of expenditure. In terms of obtaining relevant information for country-level planning, it is often as important to know whether expenditure patterns are changing over time. For example, if countries establish policies, as many have done, to shift the balance of resource use away from hospitals and towards primary care, it is vital that an information system be in place to monitor changes. Studies that have examined changes in expenditure patterns are few in number, not least because there is always the temptation to improve definitions and methods in subsequent surveys, making it difficult to compare results with earlier work. The studies that have been done are enlightening. An analysis over three years in Papua New Guinea showed that hospital expenditure had increased by 31.6 percent when the overall Department of Health expenditure 22 had increased by only 10.4 percent (Baker, 1977). In Tanzania, hospital services took up about 75 percent of government recurrent health expenditure in 1970/71, but this had-fallen to 64 percent in 1978/79 (Ministry of Health, Tanzania, 1980). Greater emphasis on a standard methodological framework for analyzing health expenditure and collecting information would improve the quality of data available for monitoring policy implementation. 3.6 Issues Arising and.Research Questions Two main issues arise from this review of hospital expenditure patterns that require discussion and raise questions for research: the database at country level for policy analysis; and hospital expenditure patterns and unit costs. The database at country level for policy analysis. It is evident from this review that few countries have usable information on health sector expenditure patterns. The development of policy advice on hospitals and monitoring policy implementation requires an improved database. The recent attention given to health finance and expenditure surveys has investigated many of the methodological issues: the problems now are not so much methodological as they are problems of organizing and stimulating survey work in developing countries and improving routine information systems in the health sector. Hospital expenditure patterns. The aim of the preceding review of hospital expenditure patterns was to identify what research had already been done in order to see if any particular patterns emerged, if some countries could be identified as being out of line, and if this could lead to profitable speculation on possible reasons and to more focussed research. 23 The first conclusion is that we know very little about the allocation of resources to and within hospitals, and thus are in a very poor position to devise policies without further research. Given the poor database, it was possible in the analysis here to draw only rough conclusions. To summarize, it appears that: o hospitals may absorb approximately 40-50 percent of health sector expenditure; 0 hospitals absorb approximately 50-60 percent of current government health sector expenditure; o hospitals absorb approximately 60-80 percent of government national health facility expenditure and possibly 70 percent of district-level health facility expenditure; o around 60-80 percent of hospital.expenditure can be absorbed by central and general hospitals, the remainder going to district hospitals; o general hospitals can be twice as expensive to run per unit of intermediate output as district hospitals, and central hospitals between two and five times as expensive; o outpatient care often absorbs around 20 percent of total hospital recurrent expenditure; o salaries and wages account for between 40-70 percent of hospital recurrent expenditure, drugs and medical supplies around 20 percent, and utilities around 10 percent; o nurses absorb the largest share of manpower expenditure although the more sophisticated the hospital, the greater the physician share; o there is inadequate evidence to estimate the proportion of total hospital costs attributable to capital services. 24 Therefore, while some rough answers are available to the question of "what is", the data are not good enough to say "'what should be". Nonetheless, it is apparent that this type of analysis can shed light on appropriate patterns of resource use in the health sector, and is worth further research. There may well be further evidence in unpublished sources inaccessible to this review that could shed further light on expenditure patterns. However, because of the difficulties of ensuring that data from isolated studies are comparable, the analysis would best be improved by the systematic collection of data from a sample of countries using a common methodology. The analyses presented in section 3 provide a guide to the information required and its analysis. Other issues concerning hospital expenditure patterns that have arisen from this review and require further research include: o the relationship between hospital expenditure patterns and the level of income of a country; o the extent of variation in hospital unit costs and the reasons for them; o the relative efficiency of different levels of hospital and different ownership patterns; o the extent to which hospital functions can be decentralized at decreased unit cost, or should be centralized in one location; 0 the responsiveness of the factor mix of hcspitals to variation in factor prices and the extent to which increased efficiency could be obtained by changes in factor mix; o the socioeconomic characteristics of patients using hospitals of different types and ownership (especially public/private). 25 4. HOSPITAL RECURRENT COSTS 4.1 The Recurrent Cost Burden It is clear from the preceding section that hospital recurrent costs represent a significant burden to governments. The high proportion of Ministry of Health recurrent expenditure devoted to hospitals implies also that hospitals absorb the lion's share of scarce resources such as manpower and drugs and, equally important, absorb a high proportion of the scarce time and energy of policymakers, planners and managers. Hospitals, as complex institutipns employing a staff category of high social status and occupying a prominent position in the eyes of the public and politicians, tend to produce managerial problems which may preoccupy Ministry of Health senior officials to the virtual exclusion of other health activities. Another, often neglected, aspect of the hospital recurrent cost burden is that of maintenance of buildings and equipment. In government hospitals, this is often paid for by a separate government organization (for example, the public works department). It is thus of peripheral financial concern to Ministries of Health, though frequently of major relevance to hospital functioning when buildings are unrepaired and equipment breaks down. The skills and resources required to ensure the continuing operation of sophisticated medical technology are frequently lacking, resulting in the waste of resources tied up in capital items. The figures in the preceding section reported actual recurrent expenditure. They do not necessarily represent "correct" levels since recurrent costs may be either over or under financed. They may be overfinanced in the sense that the same mix of outputs could be produced at lower cost if cheaper 26 technologies were used -- for example a cheaper manpower mix, less use of costly drugs when cheaper alternatives are available, less use of diagnostic services when they do not assist diagnosis or treatment. They may be underfinanced when Ministries of Health face a relatively abundant supply of capital funds (e.g., from donors) but very restricted funds f6r recurrent expenditures. Further underfinancing may result from specific constraints such as the availability of skilled manpower, or restrictions on obtaining foreign exchange to import drugs. One way of expressing the recurrent cost burden is in terms of the ratio of the annual recurrent expenditure required for operating an institution to the total capital expenditure required to construct and equip it (called the recurrent cost coefficient; Heller, 1979). Heller has estimated the recurrent cost coefficients of district hospitals to lie between 0.11 and 0.30, and for general hospitals to be 0.18. This implies, for example, that within 4 to 5 years, the total recurrent expenditure of a general hospital would have exceeded its construction and equipping costs. Thomas estimated the recurrent cost coefficient for a district hospital in Botswana to be 0.33 (Thomas, 1982). 4.2 Recurrent Cost Issues and Research Questions The major issues arising from a consideration of hospital recurrent costs concern: first, information needs; second, ways of incorporating recurrent cost concerns into health service planning; and third, ways of minimizing recurrent cost implications. Information needs. More information is required of the recurrent cost implications of hospitals and their relationship to capital costs. Within 27 the scope of this review, it was not possible to check or add to estimates of recurrent cost coefficients, since the cost analyses reviewed presented inadequate evidence on replacement capital costs. Recurrent cost coefficients have been shown to be subject to considerable variation (Thomas, 1982; de Ferranti, 1983). Rather than refine these measures, it is preferable on the one hand to develop a greater understanding of the determinants of the relationship of capital to recurrent costs (see below) and on the other hand, to encourage countries to develop an information system that provides information on hospital recurrent costs. The information needs here tie in closely with the information required to identify and monitor hospitals' share of total expenditure, and one information system could serve both purposes. Incorporating recurrent cost concerns into health service planning. Providing information on the recurrent cost implications of hospitals is of limited use unless the information is incorporated in planning procedures in such a way that it can influence planning decisions. Most countries operate a development or capital budget and a recurrent budget. The recurrent budget usually has a time-horizon of one year, and neither the budgeting nor the planning process may require the forward projection (for instance for 2 or 3 years ahead) of recurrent expenditure. Moreover, each type of budget is usually the responsibility of different central government agencies, the planning agency managing the development budget, and the Treasury the recurrent budget. Thus many countries face both institutional and procedural difficulties in ensuring that proposals in the development budget are reflected in the recurrent budget. Such difficulties are often the focus of general government procedural reforms, although there are actions that the Ministry of Health can take to 28 improve its own coordination of capital and recurrent budgets. For example, any new hospital proposal (indeed any development proposal) should include an estimate of recurrent cost implications, and as the capital proposal is refined, the recurrent cost estimates can also be refined, to give estimates of increasing accuracy. In addition, the overall implications of the whole development programme.for the recurrent budget should be monitored and the total incremental recurrent expenditure requirements by year compared with estimates of the total funds likely to be available. Such changes do not require research, but are a matter of amending planning procedures once the necessary information is available. Minimizing recurrent cost implications. There is presently a lack of information on the determinants of the relationship between the capital and recurrent costs of hospitals. This is an area that requires further research. For instance, it is likely that hospital design has a considerable influence on recurrent costs. Ward layout influences staffing costs, choice of air- conditioning and heating systems influence fuel costs, building design and choice of equipment can influence maintenance costs. Stories abound of poor hospital designs that produce maintenance and operational problems. The poorer the country, the less likely it is to have indigenous experience of economical hospital design: architects may be used who follow slavishly developed country design standards, or who are ignorant of local conditions. Thus an important research issue is the extent to which hospitals can be designed to minimize recurrent cost implications. This is a particularly important consideration in those countries where recurrent funds are much more scarce than capital funds. 29 A final point in considering the issue of minimizing recurrent costs is the establishment of procedures within hospitals that encourage efficiency. This issue is considered at length in the section below. 5. GOVERNMENT 'PLANNING AND BUDGETING MECHANISMS Government planning and budgeting mechanisms are vital in two main respects: first, they provide the means whereby changes in resource allocation patterns are brought about; and second, they contain implicit or explicit incentives for hospitals to behave in particular ways, more or less conducive to achieving efficiency. 5.1 Present Mechanisms Many countries have poorly developed mechanisms for allocating funds to lower levels of the government health system and for monitoring resource use within hospitals. In some countries (for example Jordan) budgets are held centrally and resources supplied in kind to hospitals. In others budgeting may be decentralized to a geographical unit but not further, to institutions within that unit. For example, in Malawi district medical officers do act as budget holders, but the budgeting and accounting system does not plan or record the distribution of expenditure within the district, in particular between the district hospital and health centers. It is also common, as in Malawi, for salaries to be paid centrally, the control of only non-salary expenditure being decentralized. Moreover; decentralized budget holders are often not fully involved in the setting of budgets. They may be asked, as in Malawi, for their estimates of the next year's budget requirements, but the actual level results from negotiation between the Ministry of Health and 30 Treasury over the total budget by line item, and the budget finally allocated may bear little relationship to that originally requested. This process tends to reduce the responsibility that budget holders and local managers feel for managing their resources efficiently. This is aggravated by controls on their ability to manage budgets, for instance lack of discretion to transfer funds from one budget head to another, and by accounting systems that make it difficult to identify what resources are being used for. Budget holders may feel that their actions have little A effect on subsequent allocations, or even that saving resources may lead to future cuts in budgets, and so do not seek ways of using resources more efficiently or innovatively. Further problems may arise from contradictions between the responsibility hospitals feel for treating everyone, and the constraints imposed by limited budgets. The pressure due to demand can limit hospital control over outputs. This may be aggravated by the cost-creating actions of physicians which managers cannot easily control. Thus internal mechanisms for cost control and planned resource use may be weak. The result is often lack of responsiveness to cost control measures, expenditure levels that easily get out of control, and a concentration of resources in the hands of those physicians who wield greatest power in the hospital setting. 5.2 Planning and Budgeting Issues and Research Questions Health services in developdd countries have gradually improved their budgeting and accounting practices over the years in order to alter the pattern of incentives facing health service managers, encourage cost control and efficiency, and clarify the ways in which resources are being used in 31 order to plan changes in resource allocation patterns. Thus a fair amount is known about restructuring budgeting and accounting procedures. However, many of the difficulties that developing country health services face stem from deep-seated structural problems: acute uncertainty over the availability of public finance, vulnerability to external events and shortage of skilled accountants and managers. Thus the crucial question is what reforms are feasible in the context of particular developing countries, and can be maintained by the countries themselves rather than collapsing into disuse once external assistance ends. Data on successful health service budgeting and accounting innovations are severely lacking, and local accountants tend to be wary, often justifiably, of innovations that if they fail, may threaten the precarious control they have of health service expenditure. A list of possible improvements can be suggested, but much more experience is required of how they can work in practice. The need is not so much for academic research as for applied field research and case-studies of successful innovations. Decentralization of budgets. An important precondition for many budgeting and accounting reforms is the decentralization of budgeting in order to place responsibility for promoting efficiency at the institutional level. This raises three crucial issues. First, to what level should budgets be decentralized? In a national health service, the choice lies between decentralizing to (a) an agency with area-wide responsibilities (for example a state, province or district); or (b) direct to the institutional level (e.g. the hospital); or (c) to the area-wide agency which then decentralize to either institutions (e.g. hospitals) or to programmes (see below for a discussion on programme budgeting). The choice will depend at least in part 32 on the extent to which the national level wishes to control the local allocation of funds between institutions or programmes. The national earmarking of funds for a particular hospital may on the one hand provide a sense of responsibility to the hospital manager and help to control the resources used by the hospital, but on the other hand will limit the flexibility of an area-wide agency (such as a health authority) to switch resources between programmes or institutions to suit local circumstances. The second issue is how should funds be allocated from national level to local budget holders? The normal practice is for allocations to follow historical patterns, but a number-of countries are experimenting with resource allocation formulae that relate allocations to a geographical area to some measure of "need", usually population adjusted if possible by characteristics such as age, sex and mortality rates (Department of Health and Social Security, 1976). Direct allocations to hospitals could be based on the size of the population served or on expected workload, possibly using standard costs weighted if feasible by factors external to the hospital's control known to affect costs. The third issue is what degree of flexibility should budgetholders be given? For example, should they be able to transfer funds between budget heads, from one year to the next, or from recurrent to capital accounts? If they make savings, should they be permitted to keep all or a certain proportion to reallocate? If revenue is raised locally, should it be retained by the institution? Such flexibility can have an important influence on the incentives facing budgetholders. Functional budgeting and accounting. The first step in improving internal hospital budgeting and accounting is usually a functional budgeting 33 and accounting system. Hospital "functions" are enumerated: for example, direct treatment services, medical and paramedical support services and general "hotel" services. Each of these can be subdivided according to whether they are used by inpatients or outpatients. The approach can be purely an accounting one, or managers in charge of departments can be given their own budgets. Functional budgeting provides the basis for moving towards a routine system of performance budgeting where expenditures can be matched with measures of performance expressed in terms of activities. For example, unit costs could.be routinely produced, such as cost per x-ray procedure, per drug prescribed, per surgical inpatient or per outpatient visit. These can then be monitored over time in the same hospital and across hospitals, as a check on hospital efficiency. Functional budgeting and accounting lends itself both to a very simple system and to refinements that can build on the basic approach. For instance, Figure 1 shows the approach to functional analysis of hospitals in the National Health Service in the U.K. At the most basic level is a "subjective analysis" (by line item), then primary, secondary and tertiary analyses (the latter not yet routinely implemented). Such systems are now becoming much more feasible with the advent of microcomputers, but require research into their relevance elsewhere and careful translation to other settings. 34 Figure 1 FUNCTIONAL ACCOUNTING IN THE UK NHS SUBJECTIVE ANALYSIS (e.g. Salaries. Drugs, Provisions) 1 .I I Hospital Community Other Health Services Services Services I I . PRIMARY FUNCTIONAL ANALYSIS Direct Expenditure Indirect Expenditure (e.g. Patient Services) (General Services) * I Direct - Para-Medical General - Direct Treatment Services Services - Credits Services DEPARTMENTAL WORK UNITS producing MICRO UNIT COSTS SECONDARY ANALYSIS I II Hospitals Community Other Services Services In-patients Out-patients etc. Schools Domiciliary etc. Other units of Visiting production or service SECONDARY WORK UNITS producing MACRO UNIT COSTS e.g.- I I I Per I-P Day Per O-P Per case and Per 1.000 Population and Attendance & Per 1,000 Population Per I-P Case Per New O-P TERTIARY ANALYSIS Children Maternity Acutely III Elderly etc. (i.e. classes of beneficiary) Source: Rigden, 1983. 35 Programme budgeting. In programme budgeting, accounts and/or budgets and/or plans are expressed in terms of programmes which are output-related, that is specified in terms of the objectives of health expenditure (WHO, 1984). Objectives are usually specified in terms of improvement in the health status of various population groups, or reduction in particular diseases, rather than in terms of the institutions (e.g. hospitals) or activities (e.g. outpatient care) that merely serve those objectives. This "pure" view of programme budgeting is not easy to implement. A client or target group programme structure can be set up for monitoring or planning purposes, as in the U.K. (Lee and Mills, 1982), but if incorporated in routine budgeting and accounting procedures, would cut across institutional and organizational patterns. Indeed, hospitals present one of the most intractable problems for programme budgeting, since they serve many different population groups and are a major spender. For example, the U.K. programme budgeting structure adopted a compromise, allocating geriatric, pediatric and psychiatric hospital services to the elderly, child and mental illness programmes respectively, but retaining "general and acute hospital and maternity services" as a separate programme accounting for about 40 percent of total health and personal social services expenditure (Department of Health and Social Security, 1977). The programme analysis was produced by apportioning the costs of joint activities to different programmes and was used for planning and monitoring purposes, the .health service accounting system remaining unchanged. There still is, nonetheless, interest in more action-oriented client group budgeting in the U.K.: for example, experiments with a local programme budget for the elderly "purchasing" the services it requires for its clients 36 from a variety of health and local authority institutions (Maynard and Smith, 1983). The aim of such a system is to improve the signals flowing to institutions, to render them responsive to local needs rather than to their own organizational realities. However, such a system is likely to be beyond the capacity of most developing countries to set up. The question therefore remains, what is the most appropriate programme budget structure, and how can it be implemented. One alternative is to set up a programme structure as in the U.K. which is appropriate for planning and for monitoring the achievement of objectives but which is not carried through into the accounting system. Intermittent cost analyses are then necessary to apportion expenditure to programmes. The information needs here could be satisfied by an information system or surveys designed to provide information on hospital expenditure patterns. A second alternative is to orient the accounting sys-tem to programme budgeting. This will require some compromise to relate the programme structure to organizational patterns, and if a functional accounting system is in use in hospitals, may require little more than integrating it within a comprehensive programme structure for the health service as a whole. For example in the early 1970s, Malaysia began developing a "Programme and Performance Budgeting System" in the Ministry of Health. The prograrmes were administration, public health services, medical care services, dental services, health manpower and training, pharmacy and supplies, research, planning and development' and engineering services (Ministry of Health, Malaysia, 1977). Each programme was then subdivided by activity: for example, the medical care programme contained the activities of general patient care, special care psychiatry, special care leprosy, special care tuberculosis, and ministry and state 37 headquarters' administration. Within hospitals, cost centers were defined, such as the surgical department, with the eventual aim of allocating all expenditure to cost centers. At present, the programme structure is largely institution-based (for example keeping "general patient care" as a single category), but as the accounting system by cost centers within hospitals is improved, the structure can move towards a functional classification. Some of the benefits experienced so far have been a clearer idea of the major %consumers of resources within hospitals, leading to decentralized budgets (for instance for drugs) as a means of cost control (Mills, 1983). The example of Malaysia represents a gradual progression from state- based budgeting with no clear institutional breakdown, through institutional budgeting, to functional budgeting in hospitals as part of an overall programme budget structure. As yet it is an isolated report of innovations in hospital budgeting and accounting practices. More evidence is required from other countries to judge what innovations in budgeting practices are feasible, how they can best be implemented and what advantages they bring in terms of increased cost control and efficiency. 6. THE SOURCES OF INCOME OF HOSPITALS 6.1 Present Sources of Income The sources of hospital income have received even less attention than hospital expenditure. This section is, therefore, extremely short on evidence. It is convenient to structure the discussion in terms of hospitals in the public sector, the social insurance sector, the NGO (nonprofit) sector, and the private (for profit) sector. 38 The public sector. Sources of income for capital and recurrent hospital expenditures are likely to be quite different, especially in the least developed countries. Health is often not given high priority for local development funds, and capital construction is usually externally financed although where the finance takes the form of a loan, this implies a stream of repayments from local resources in the future. In richer countries, the sources of income for capital construction are likely to be more diverse. In Thailand, for example, public lotteries have been used to obtain funds to construct a hospital for the Navy. Despite the lack of evidence, it seems a safe generalization that the great majority of hospital income for recurrent expenditures in developing countries comes from government revenues. Virtually no income and expenditure accounts for government hospitals were found by the literature review, but evidence of the amount and extent of user fees for hospital services suggests that direct payments usually make up only a small proportion of hospital income. Table 12, taken from de Ferranti (1984), lists health service fees for selected countries and indicates their level by expressing them in terms of the daily agricultural wage. In general, with the notable exception of China, hospitalization fees are only a fraction of the daily agricultural wage. Moreover, certain categories of patient are likely to be exempt from these fees. For example, in Rwanda and Togo, fees are not levied on the "poor" (Laurent, 1982). The same is true in Zimbabwe (Government of the Republic of Zimbabwe, 1981) where there is a cut-off income of Z$150 per month, and Thailand, where a special Ministry of Health allocation pays for free medical care for those falling below a particular income level (Mills, 1980). 39 The small contribution that user fees make to hospital income is implied by the share of user charges in total government health service expenditure. Table 13, taken from de Ferranti (1984), shows that user charges often contribute well under 10 percent of health service expenditure. Jordan is an example of user charges appearing to be widespread but in fact contributing relatively little income (Walker, Longford and Mills, 1980). Fees are charged for outpatient and inpatient care, the latter being based on the length of stay and services received. The effect, however, is mitigated in a number of ways. Certain groups are exempt: civil servants (because they pay into a compulsory insurance fund), those defined as poor, school children, refugees (covered by a lump sum payment) and those suffering from certain conditions such as carcinomas. A sample survey at the main Ministry of Health hospital of attendances at the outpatient department and of inpatient discharges showed the following pattern: Category Outpatients (percent) Inpatients (percent) Civil servants 15.8 2.8 Dependents of civil servants 8.1 8.0 Poor 13.0 2.5 Refugees 23.2 School children 12.4 -- Medically exempt N/A 16.7 Others -- 0.8 Fee paying 50.8 46.0 40 Thus only approximately 50 percent of patients were classed as fee- payers. Moreover, a comparison of the fees charged for treatment for a common set of diseases with the likely actual cost of treatment indicated that fees covered only 15 to 20 percent of the actual recurrent cost. Thus fees are low relative to the cost of care. The social insurance sector. Patchy data is available for social insurance of the overall breakdown of income between employees, employers and government (International Social'Security Association, 1982). However, no information was found (though it may well exist) relating specifically to the income of hospitals caring for the insured or to the division of employee payment between premiums and co-payment charges. In general, it appears that co-payment requirements are not very common in social insurance schemes in developing countries, unless the patient opts for a different type or higher standard of care than that reimbursed by the insurance scheme. The only hospital-specific data found relates to the Philippines and may well be atypical. These data show that although when the "Medicare" scheme was introduced in 1972, 70 percent of hospital costs were paid by insurance and 30 percent directly by the patient, the proportions are now likely to be 30 percent insurance and 70 percent patient (Denoga, 1982). This change is at least in part due to benefits being specified in financial terms, and remaining static despite increases in hospital fees. From the balance of evidence available, it must be assumed that the great proportion of the income in hospitals serving a social insurance system originates from pre-payment mechanisms and government revenue (unless of course the hospital is not owned by the social insurance agency and has a substantial clientele of fee-paying private patients -- see below). However, 41 as important for the efficiency of the insurance sector is the means by which this income is channelled to hospitals. This aspect has received very little attention and no overall reviews of payment practices were found. For hospitals not owned by the insurance agency, the most common means of payment is probably fee-for-service, whereas insurance agency hospitals may receive budgets and employ salaried doctors. The NGO sector. Evidence relating to NGO hospitals comes mainly from mission agencies in Africa. A survey of all mission hospitals in Malawi revealed that 34 percent of their income came from patient fees, 33 percent from the government, 29 percent from overseas and 5 percent from other sources (Msukwa and Simkonda, 1982). Other analyses revealed two other interesting features. First, the proportion of income in the form of fees was lowest in hospitals, and increased as the health unit level decreased (being 40 percent in rural hospitals and health centers and 52 percent in subcenters). This presumably reflects the lower cost of care in peripheral units and thus the higher proportion of costs that can be recovered in fees. Second, there was considerable variation between different denominational groups. Unfortunately, no breakdown by type of facility is provided, but overall, CCAP and Anglican units obtained about 27 percent of their income from fees, Roman Catholics about 40 percent, Seventh Day Adventists 50 percent, and Seventh Day Baptists 70 percent. These proportions are influenced by a number of .factors: the extent of overseas income, the objectives of the services offered (for example part or all of a few facilities are clearly aimed at the more affluent section of the population rather than the rural poor), and the ability and willingness to pay of the local population. 42 A similar, varied pattern is evident from other parts of Africa. A mission hospital in Togo obtained 44 percent of its income from fees, and another 43 percent (Laurent, 1982). A review of the income of 50 African voluntary hospitals found that 12 hospitals got more than 70 percdnt of their income from fees, 17 hospitals more than 70 percent from government grants, and the remainder had varied sources of income (van Lemmen and van Amelsvoort, 1985). This information suggests that there is considerable scope for investigating the relationship between the level of fees charged and patterns of utilization, and this evidence may well be of use when considering the effect of levying fees in the government sector. The private sector. Little is known about any aspect of private (for profit) hospitals in developing countries, let alone their sources of income. Of considerable relevance to planning organizational changes in the health system would be data on sources of income, and especially the extent of direct payment by patients and the extent to which costs are covered by employer reimbursement or by social or private insurance arrangements. In Jordan, for example, where 20 percent of all hospital beds are in the private sector, it was apparent from discussions with owners of private hospitals that there were a substantial number of payment arrangements that limited the burden on the individuai patient (Walker, Longford and Mills, 1980). Little use was made of commercial insurance companies, but a number of hospitals had agreements with private companies, where the hospital either offered its own insurance scheme or debited companies for the cost of medical care for its employees. A similar pattern of employers paying substantial sums to the private health sector has been found in Indonesia (Abel-Smith, personal communication). 43 Thus it would be misleading to assume that private hospitals are substantially funded by out-of-pocket payments by patients. However, their actual mix of income is not known. Nor are the socioeconomic characteristics of users: for instance the extent to which private hospital care is used by poorer population groups. In Jordan, it appeared that private outpatient care was widely used even in low income urban areas but that private hospitals served mainly the more affluent population groups (Walker, Longford and Mills, 1980). 6.2 Issues Arising and Research Questions A large number of issues and research questions arise from this consideration of hospital sources of income, particularly those concerned with the potential for altering the mix of income sources of different types of hospital, the effect of user charges on hospital utilization, the effect of hospital and physician payment systems on hospital behavior and the effect of private sector expansion on MOH services. Perhaps the most dominant question presently facing Ministries of Health is how to develop the health system without excessive reliance on government revenue. Since hospitals are the major user of Ministry of Health resources, it is appropriate to focus the discussion here on issues concerned with the extent to which governments can be relieved of the burden of financing and/or providing hospitals. Four main options are available: increased cost recovery through user charges in government hospitals; increased cost recovery through insurance schemes which cover the cost of care at government hospitals; increased use of hospitals in NGO or private ownership with appropriate government financial transfers; and expansion of privately owned and financed risk-sharing schemes. 44 Each of these can be discussed in terms of research questions relating to its potential for substituting for or complementing government finance and its effects on utilization patterns and hospital and physician behavior. User charges in government hospitals. The above section made clear that there is at present relatively limited use of user fees in government hospitals. The economic justification for user fees is two-fold: first, that it offers an efficient means of raising money for hospital care; and second, that it increases the efficiency of resource use. In the case of the first justification, user fees as a method of revenue generation need to be compared with available alternatives, such as increasing income or sales taxes, and their relative advantages and disadvantages evaluated in terms of public finance criteria. In particular, research is badly needed on the cost of collection relative to yield. Nonetheless, government decisions will often be based on pragmatic criteria, namely that the Ministry of Health is given low priority for tax revenues, and if it wishes to increase its income, must seek ways of generating income itself. In this context, the second justification, that of increasing the efficiency of resource use, is of greatest relevance. The argument is that free services will encourage consumers to use services at a level where the marginal social cost exceeds the marginal social benefit. Levying fees will discourage at least some "unnecessary" use. However, it may also deter "tnecessary" use by those with low incomes. The discussion of the levying of user fees therefore involves balancing considerations of efficiency and equity, and requires evidence of the revenue potential of fees and their effect on consumers. The issue of user fees for health services and questions surrounding it have been systematically reviewed 45 by de Ferranti (1984). Here, the discussion will concentrate on four issues for research: the potential revenue contribution of user fees; the type of fees that should be levied; collection mechanisms; and the likely equity effects. The potential revenue contribution of fees will depend on whether they are considered nominal, designed to deter excess use, or large enough to achieve cost recovery. Hospital care can be expensive, and the amount of service received is only marginally controlled by the consumer. It therefore presents clear potential for risk-sharing, either through tax or.pre-payment mechanisms. Thus user fees in hospitals are unlikely to be set at a level to achieve cost recovery. Their actual revenue potential will depend on increasing our general understanding of the ability and willingness to pay of users, and on carefully assessing, in each country, what level of fees will produce sufficient revenue to make the policy worthwhile, but will not act as a major deterrent to use. To some extent this will depend on what mechanisms can be devised to exempt those unable to pay from fees, although the greater the exemptions the lower will be the revenue and the higher the cost of collection mechanisms relative to income received. Hospitals where fees are already charged (for example, mission hospitals) are likely to be a useful source of information for researchers. If fees are not set at a level to achieve cost recovery, there is little point in charging for every item of service. The type of fee will depend on factors such as its acceptability to the local population, effect on utilization and resource use, and ease of collection. Very little information is available on any of these factors though a few reports do exist (see Bekele and Lewis, 1986 on user fees in Sudan). More research is required on a number of issues. For example, arguments are frequently made for charging 46 an outpatient fee which differs according to whether the patient is referred or not. To what extent is this policy acceptable and feasible, considering that the hospital outpatients department usually also operates as the local health center? Inpatient fees can be per admission, and/or per day, plus some allowance for services provided (e.g. flat fee for x-ray or laboratory tests). A fee for admission is administratively simple, but has limited revenue potential because a high fee would unduly penalize those with short stays. A fee per day would probably need to have some upper limit, to avoid penalizing those who require long term treatment and discouraging patients from completing a course of treatment. Alternatives to a flat rate inpatient or outpatient fee are to charge for a particular item of service such as drugs or food, or to set up a revolving fund. To a considerable extent, the title of the charge is immaterial if non cost-covering fees are intended, and will largely be a matter of consumer acceptability and ease of collection unless the fee is intended to influence the use of a particular resource in a desirable direction (for example, economize on the use of drugs). These again are largely unresearched areas. Countries often have difficulty collecting existing fees, and there is little point introducing more general fees unless a collection system can be devised that is cheap and efficient. NGO hospitals may have useful experiences here that can be evaluated. In order .to satisfy equity concerns, any policy on user fees will need-a system for exempting from fees for those too poor to pay. Here again, more information is required on the feasibility of different approaches. Are informal systems, for example, leaving the decision to the hospital superintendent, most appropriate or can a formal system be set up with a minimum of cost and bureaucracy? The more complex the system, 47 the more likely is it that the poor will fail to take advantage of it. In Jordan, for example, a certificate from the Ministry of Social Affairs, granted after an investigation of the income of the applicant, was required to give exemption from fees (Walker, Longford and Mills, 1980). The number of certificates issued was very small in relation to the numbers likely to fall below the poverty line, suggesting that the procedure was a considerable deterrent. If fees are nominal, there should be no need for exemptions for illnesses that are particularly costly to treat. However, there may be illnesses (e.g. sexually transmitted diseases) where there should be no financial barrier to treatment. Thus certain exemptions for reasons other than poverty are likely to be required. Since a nationalized health service usually has explicit equity objectives, one of the most crucial research areas is to evaluate the likely equity effects of fees. Information is at present notably lacking on two crucial factors -- ability to pay and willingness to pay -- and on their determinants. An important point for government hospitals to keep in mind is that willingness to pay will depend not only on ability to pay but also on the consumer's evaluation of the service offered. Where services are seen as being of low quality, the levying of fees is likely to be a much greater deterrent to use than when quality is good. For example, in Nigeria increased fees plus widespread shortages of crucial hospital supplies have combined to produce considerable underutilization of hospital services (Attah, 1986). (Research on demand determinants and price elasticities is the subject of another paper in this series.) 48 Insurance coverage for treatment at government hospitals. Insurance schemes in developing countries have usually provided access to beneficiaries for either services owned by the insurance agency or private sector facilities. Relatively few have established agreements with government hospitals to provide care. Many governments are now interested in insurance as a way of increasing resources for the health system, but wish to benefit from the income themselves. Some already have small-scale schemes providing insurance coverage for civil servants, and wish to expand these to a larger proportion of the population. The crucial issue is how to offer those who can be brought into the health insurance net a service that compensates for the premium they are required to pay, but is integrated into the government health service system. Where services are already free or virtually free, requiring a premium to use these services, even if a nominal fee is waived, is unlikely to be attractive. Two solutions are possible. First, it is possible for insurance to be matched by a policy of greater cost recovery in the government sector, but then it is vital to have an efficient mechanism for those who cannot easily be covered by insurance (for example, the self-employed) and are too poor to pay fees. Alternatively, the insured could be given access to services at government hospitals which are more attractive or convenient than those available to the general public. Many hospitals already have side wards or private beds for which fees are charged, and these could be used by insured patients at no (or minimum) charge. Another suggestion is for the insured to have access to special sessions run outside normal working hours at government 49 health centers, with appointment systems and possibly doctors available (Abel-Smith, 1986). Much more research is required on existing schemes of this nature and on the potential for further expansion. For those countries unwilling to establish a separate health insurance sector but wishing to tap the income of the employed urban dwellers for government health services, such a scheme may well prove attractive. Increased use of hospitals in private or NGO ownership. Insurance coverage for treatment at government hospitals may suit those countries where a substantial network of government hospitals already exists and where for political or other reasons, expansion of NGO or private hospitals is not envisaged. Some governments, however, may wish to limit their involvement in the direct provision of medical care, but to continue to play a role in organizing financial support. This can be arranged via a social insurance system which finances the provision of care in the private sector, or alternatively, the government could use general tax revenues but could enter into contracts for care to be provided by nongovernment hospitals. A number of important issues arise in considering this option. First, there is the question of coverage. Could private providers deliver services throughout the country and on a long-term basis, or would the incentives available be inadequate to attract them out of the main urban centers? Second, could private hospitals provide an adequate range and quality of services? Patchy evidence suggests that private providers may concentrate on relatively low cost, acute services and maternity care, where 24 hour physician attendance is not required, nor sophisticated therapeutic services. 50 Research is required on what the private hospital sector presently does before formulating policies on its future role. But perhaps the most crucial issue is whether payment systems can be devised that ensure cost control and efficiency, The experience of social insurance agencies with financing care in the private sector has not been a happy one. Fee for service systems have provided incentives for hospitals to maximize their income by maximizing the number of services provided. For instance, a review of health care financing in Brazil pointed out that the numbers of x-rays and complementary tests have expanded at rates two or three times greater than that of total hospitalizations between 1970 and 1981 (McGreevey, 1982). Other criticisms of the practices encouraged by the payment system include the excessive use of high technology and drugs, unnecessary hospital admissions, and the overproduction and overbilling of services as a means to compensate for levels of remuneration thought by private physicians to be too low. Similar criticisms have been made of other countries with similar insurance schemes (International Social Security Association, 1982) although unfortunately, rarely backed by hard data. More information is required on the effects of alternative payment mechanisms on physician and hospital behavior. For example, Brazil has been experimenting with the use of medical auditors to limit entry to hospital care, and with payment for procedures, not detailed items of service (McGreevey, 1982). Capitation payments are commonly used as an alternative to fee for service for outpatient care, and may also have some potential for inpatient care. A more radical option is for the government or social insurance agency to invite private sector hospitals to tender for the supply of a certain package of health care., All 51 these possibilities need studying in terms of their effects on admission rates, lengths of stay, quality and quantity of services provided and costs. It is also important to review the wider repercussions of such schemes, especially where hea-lth manpower is in short supply and an expanded private sector may be at the expense of MOH services for the general public. Privately financed and organised risk-sharing schemes. The final option is for governments to encourage the development of health care schemes that require no government involvement at all, whether in provision or financing. Firms or groups of workers could come to an agreement with private sector hospitals to provide certain services. Such a scheme could be organised on the lines of a health maintenance organisation. If a national social insurance system is in operation, 'contracting out' could be acceptable so long as the health care scheme met certain minimum requirements. Such schemes already exist although they are little researched, and may have considerable potential for expansion amongst the more affluent urban dwellers. They are unlikely to be an affordable solution for poorer countries or population groups. It is interesting to note that in Chile the government's policy of encouraging the private health sector is reported to have had limited success (Viveros-Long, 1986). The objective of government policy was to provide greater incentives to private health sector investment in order to relieve the high demand for the public sector. A review of private sector spending in Santiago showed that there had been a large increase in ambulatory health care providers and investment in diagnostic equipment, but in the richer areas and without .any increase in private hospital beds. The conclusion was that the relative failure of the policy was due to lack of demand stemming from high unemployment and a skewed income distribution. 52 Much more investigation is required of the feasibility and behaviour of private sector schemes. Issues to be investigated should include: - the coverage of existing private sector arrangements; - hospitalization benefits provided; - payment mechanisms; - innovative forms of organisation such as HMOs; - the effect on MOH services, particularly on the availability of health manpower for MOH facilities; - the effect on cost levels for health care. Some countries in Africa and South East Asia (for instance Thailand and Malaysia) have experienced a considerable increase recently in private sector medical care, and could provide useful research sites for study of both the existing private hospital sector and innovative forms of organisation. 7. PROPOSED RESEARCH PROGRAM The issues and research questions identified in each of the preceding sections can be summarized and regrouped into five main research areas: hospital income and expenditure patterns; information needs for country- level planning; planning and budgeting methods; the potential for changes in patterns of hospital income; and the effect of physician and hospital payment systems on behavior. Hospital income and expenditure patterns. Research would be designed to produce information on the issues raised in section 3. The aims of the research would be to..identify: (1) a 'good practice' distribution of resources between hospitals and other types of health activity and between different types of hospital; 53 (2) the determinants of variations in hospital unit costs and the extent to which reduction in hospital unit costs is feasible through adjustments to factor mix and hospital design. Information would be sought on: (1) the allocation of health sector resources between hospitals and other components of the health sector, between different types of hospital, and between hospital functions; (2) the factor mix.and unit costs of hospitals class.ified by ownership and type and nature of services provided. This information could be obtained in two different ways. First, the - data presented in section 3 of this paper could be supplemented with similar information from any documents not available to the author and from health sector studies and reviews as they are carried out. This approach would suffer from the problems encountered here, namely the lack of comparability and other inadequacies of data obtained from secondary sources, and the limited range of countries for which data is available. A preferable method would therefore be to design a standard methodology for collecting the data, and use this in a sample of countries selected in order to display a variety of different patterns of hospital expenditure. Such a study would be a valuable addition to our knowledge about the role of hospitals in developing country health systems. Maxwell (1981) for developed countries and Abel-Smith (1967) provide precedents, but the proposed study would focus primarily on hospitals and their place in a health system. Information needs for country-level planning. Much of the earlier discussion indicated the need for improved information relating to hospitals. This is required for identifying resource allocation problems, planning 54 changes to health sector resource allocation patterns, monitoring changes, taking adequate account of the recurrent cost burden of hospitals in investment decisions and implementation of plans, and operating a programme budgeting system. This information can be produced by special one-off surveys, repeated when necessary, or by setting up a routine information system. The overall methodology for special surveys has been reasonably well- developed in connection with health finance and expenditure surveys and therefore needs little methodological research.- However, the focus of these surveys has been on a comprehensive enumeration of the entire health system, with relatively little attention given tp the design of methods to obtain specific items of information (e.g. private sector health expenditure) or to the design of surveys and sampling procedures that economize on the amount of research required. These two areas therefore require further methodological research. Once this has been done, the main need is for carrying out surveys in countries, that is for applied or operational research. Routine data collection methods have many advantages over special surveys if a country can introduce and maintain them. It is evident that existing information systems rarely produce the required information, and research is required into how routine systems can best be altered and added to. An improved system of health expenditure estimates under the aegis of a national accounts system has been proposed (Cumper, 1986) and its feasibility could be investigated. He suggests that an annual series of expenditure estimates could be maintained by combining the data available routinely each year (particularly government health expenditures) with those collected in detail at longer intervals but which can be extrapolated on an annual basis 55 according to some agreed procedure (e.g. estimates of private health expenditure derived from household expenditure surveys). A national accounts agency would be well-suited to undertake this work, guided by the Ministry of Health. Improving the information system on MOH health expenditure will depend to a considerable extent on what improvements in budgeting and accounting procedures are considered feasible (see below). At the local level, a particular research focus might be the extent to which micro-computers could be used for maintaining and analysing information on budgets, expenditure patterns and utilisation statistics. Planning and budgeting methods. This paper has identified a number of issues concerned with planning and budgeting methods: amending planning procedures to more fully account for the recurrent cost consequences of hospital development, devising resource allocation formulae in order to provide a basis for efficient and equitable distribution of funds from national to sub-national levels, amending budgeting procedures in order to decentralize budgeting and create incentives for efficiency at hospital level, and improving budgeting and accounting formats and practices by moving towards functional, performance and programme budgeting. These are not areas for conceptual or methodological research, but rather for action-oriented research where innovations are designed and introduced and their effects monitored. In the first instance, it would be appropriate to see if countries can be identified where innovations have already been tried, to review their experience. Following this, new experiments could be designed and introduced, and their success monitored with a view to identifying features relevant to the circumstances of other countries. 56 Changing the pattern of hospital income. The major research topics under this heading are: the potential for increased use of user fees and health insurance to cover the cost of hospitalization; and for shifting the balance of hospital financing and ownership away from government to the NGO and private sectors. These topics raise enumerable issues for research. They include: User fees: - the ability and willingness of users to pay fees, and the effect of fees on utilization patterns classified by socio-economic status, health condition etc.; - for which services should there be a charge and which should be free; - appropriate fee-levels and the proport.ion of hospital expenditure that can feasibly be recovered through user fees; - mechanisms for the collection of fees; - procedures for exempting those who cannot or should not pay fees. Insurance: - the potential for expanding the scope of insurance to cover hospital care for the self-employed and rural populations; - means of providing acceptable services for insured patients at government hospitals. Increased NGO and private hospital provision: - the types of hospital service which NGOs and the private sector are most suited to provide; - the extent to which NGO and private sector provision can offer a long-term, nation-wide hospital service; 57 - the scope for private sector patterns of health service organisation set up by agreements between employers or worker groups and private health service providers; - the best form of organisation for such private sector arrangements (e.g. HMOs); - the relative efficiency of public and private sector hospital services; - the effect of private sector expansion on the supply of MOH services; - the implication of private sector hospital expansion for the equity of the whole health system. Research could be carried out in three different ways: (1) by studying existing institutions - for instance the demand and supply of hospital care at different types of hospitals (e.g. fee- paying/non fee- paying, public/private etc.) within one country; (2) by studying hospitals before and after a policy change, such as the introduction of user fees; (3) by comparing a sample of countries with different patterns of hospital ownership and hospital income. Effect of physician and hospital payment systems on hospital behaviour. There are many comments on the effect that physician and hospital payment systems - especially fee-for-service systems - have on behaviour but few hard facts. Given the present interest in promoting non-government forms of financing and ownership, this is a crucial research area. As with the previous topic, research is required first to document the existing situatioh, and second to monitor the effects of policy changes. Of particular interest are the effects of changing payment practices on admission rates, lengths of stay, quality and quantity of services provided and costs. Table 1 Hospital expenditure as a proportion of total health sector expenditure (public and private) Country Personal health services Pub. hith Teaching/ Other Total Date Hospitals Non-hospital services Research Malawi (1) 28.1 25.3 27.9 2.1 16.6 100,0 1980/1 (2) 33.8 30.5 13.2 2.5 - 20.0 100.0 1980/1 Sri LankA (3) 43.5 47.4 9.0 ? 0.0 99.9 1979 Australia (4) 43.7 52.9 1.9 1.5 0.0 100.0 1960/1 Canada (5) 42.7 54.5 1.0 1.8 0.0 100.0 1961 Ceylon (5). 50.0 43.9 4.5 1.7 0.0 100.1 1957/8 Czeckoslovak(4) 41.6 49.8 2.4 6.3 0.0 100.1 1961 Rhodesia/ (4) 18.6 1.4 0.0 0.0 100.0 1961/2 Nyasaland (4) Finland (4) 59.9 35.9 2.5 1.7 0.0 100.0 1961 France * (4) 40.7 56.0 1.7 1.6 0.0 100.0 1963 Israel (4) 45.3 50.4 1.6 2.7 0.0 100.0 1961/2 Kenya (4) 88.5 8.3 3.2 0.0 100,0 1961/2 Netherlands (4) 92.1 2.4 5.5 0.0 100.0 1963 -Poland (4) 36.4 56.7 2.8 4.1 0.0 100.0 1961 Sweden (4) 52.9 42.4 1.2 3.5 0.0 100.0 1962 Tanganyika (4) 44.8 50.2 4.3 0.8 0.0 100.1 1961/2 UK (4) 51.4 44.7 1.9 2.1 0.0 100.1 1961/2 USA (5) 36.2 59.0 0.3 4.6 0.0 100.1 1961/2 Yugoslavia (4) 40.7 52.5 3.9 3.0 0.0 100.1 1961 Sweden (6) 71.1 28.1 0.4 0.4 100.0 1975 UK (6) 62.8 24.5 6.2 1.5 5.0 100.0 1974/5 Canada (6) 59.0 32.1 3.2 1.1 4.6 100.0 1975 Italy (6) 48.0 40.5 2.2 ? 9.3 100.0 1975 Switzerland (7) 44.9 46.0 3.8 5.3 100.0 1975 Australia (8) 56.9 36.0 3.1 1.9 2.1 100.0 1974/5 Netherlands (9) 52.6 41.9 1.1 ? 4.4 100.0 1974 US (8) 50.4 40.0 2.4 2.5 4.7 100.0 1974/5 Notes and references (1) 'Public health' includes water, sanitation and nutrition. 'Other' includes administration. Data from MON Republic of Malawi 1984. (2) Mater and sanitation excluded. Data from MON Republic of Malawi 1984. (3) Central government other than MON excluded. Alailisa and Mohideen 1984. (4) Recurrent costs only. Abel-Saith 1967 (5) Recurrent costs only. A certain amount of expenditure on hospitals is included in 'non-hospital'. Abel-Stith 1967. (6) Self-medication included in 'non-hospital'; administration included in 'other'. Data.from Maxwell 1981. (7) Self-sedication included in 'non-hospital'; administration expenditure not available. Data from Maxwell 1981. (8) Excludes fees of private doctors attending hospital patients. Self-tedication included in 'non-hospital'; - administration included in 'other'. Data from Maxwell 1981. (9) Excludes fees of private doctors attending hospital patients. Self-medication expenditure not available; administration included in 'other'. Data from Maxwell 1981. - 58 Table 2 Hospital expenditure as a proportion of total current health sector expenditure of government Country Admin. Hospital Primary Prevention Training Other Total Date Botswana (1) 4.4 42.0 21.5 13.3 .7.7 11.2 100.1 1978/9 Malawi (2) 8.0 57.6 15.5 5.6 2.8 10.5 100.0 1983/4 43) 8.3 62.3 19.6 5.0 2.5 2.3 100.0 1983/4 (4) 22.6 51.9 14.4 5.6 1.0 4.5 100.0 1983 Tanzania (5) 0.7 64.0 17.1 11.3 5.2 1.6 99.9 1978/9 Zimbabwe (6) ? 69.0 ? ? ? 31.0 100.0 1980/1 Indonesia (7) 12.8 37.8 30.7 5.3 3.2 10.2 100.0 1980/1 Malaysia (8) 4.4 61.2 18.9 8.8 6.8 - 100.1 1973 Nepal (9) 2.5 23.2 40.1 25.8 ? 8.5 100.1 1981/2 PN (10) 9.5 46.6 25.0 9.7 5.8 3.4 100.0 1976/7 Tunisia (11) ? 64.7 11.4 ? ? 23.9 100.0 1971 Jamaica (12) 7.1 60.9 18.1 8.7 1.9 3.3 100.0 1980 Colombia (13) ? 65.2 26.5 2.2 1.6 4.4 99.9 197314 Notes and references (1) Includes central and local government expenditures. Data from MOH Botswana 1979. (2) MOH expenditure only. *Other' includes grants to mission facilities including hospitals. Author's estimates. (3) MOH and Mission expenditures. Author's estimates. (4) Total government expenditures including transfers to missions. Data from MO Republic of Malawi 1984. (5) Data from MON Tanzania 1980 (6) MOH expenditures only. Data from Government of Republic of Zimbabwe 1981. (7) 2 provinces. Data from Wheeler and Volpatti 1981. (8) Peninsular Malaysia. 'Primary' includes small % of environmental health. Data from Heller 1975. (9) Division very approximate. Both development and recurrent. 'Primary' and 'prevention' include substantial central administrative costs. Data from WHO 1982. (10) MOH only. 'Hospital' includes urban health centres. Baker 1977. (11) 'Other' unspecified in source. Data from Heller 1978. (12) Data from Cusper 1986. (13) Area of Candelaria. Capital services included. Proportions estimated on the basis of information given in Robertson 1985. 59 Table 3 Hospital share of annual development (capital) expenditure on health Country Hospitals Other Total Date Botswana (1) 7.5 92.5 100.0 1978/9 Malawi (2) 19.3 80.7 100.0 1984/5 Tanzania (3) 31.7 68.3 100.0 1978/9 Indonesia (4) 38.7 61.3 100.0 1980/1 Canada (5) 99.7 0.3 100.0 1961 Chile (5) 45.8 54.2 100.0 1961 Czeckoslovak(5) 52.6 47.4 100.0 1961 France (5) 74.8 25.2 100.0 1963 Israel (5) 80.2 19.8 100.0 1961/2 Poland (5) 74.0 26.0 100.0 1961 Sweden (5) 90.3 9.7 100.0 1961 Tanganyika (5) 80.0 20.0 100.0 1961/2 UK (5) 76.4 23.6 100.0 1961/2 Venezuela (5) 88.5 11.5 100.0 1962 Yugoslavia (5) 48.9 51.1 100.0 1961 Notes and references (1) Data from MOH Botswana 1979. (2) Estimated not actual expenditure. Author's estimates. (3) Data from MOH Tanzania 1980. (4) Water and sanitation excluded for comparability with other figures. Data from Wheeler and Volpatti 1981. (5) Data from Abel-Smith 1967. 60 Table 4 Hospital expenditure as a proportion of national government health facility expenditure Country Hospital Primary Other Total Date Botswana (1) 66.2 33.8 0.0 100.0 1978/9 Malawi (2) 56.8 43.2 0.0 100.0 1983/4 (3) 59.0 41.0 0.0 100.0 1983/4 (4) 62.5 57.5 0.0 100.0 1980/1 . Tanzania (5) 57.8 28.5 13.7 100.0 1978/9 Zimbabwe (6) 75.5 24.5 0.0 100.0 1980/1 Malaysia (7) 71.9 28.1 0.0 100.0 1973 Tunisia (8) 85.0 15.0 0.0 100.0 1971 Brazil (9) 82.2 17.8 0.0 100.0 1981 Notes and references (1) Central and local government expenditures. MOH Botswana 1979. (2) Ministry of Health expenditures only. Author's estimates. (3) Ministry of Health and mission expenditures. Author's estimates. (4) 9 districts including 2 urban. Expenditure of central-government and missions. MOH Republic of Malawi 1984 (5) All districts. 'Other' is 'hospital preventive services'. Data from MOH Tanzania 1980. (6) All districts. 'Primary' includes provincial offices. Data from Government of Republic of Zimbabwe 1981. (7) Peninsular Malaysia. Data from Heller 1975. (8) All districts. Data from Heller 1978. (9) Whole country. Expenditures of Ministerio de Previdencia e Assistencia Social counted as 'hospital' and of MOH as 'primary'. Mesa-Lago 1983. 61 Table '5 Hospital expenditure as a proportion of district health facility expenditure Country Hospital Primary Other Total Date Shana (1) 67.3 32.7 0.0 100.0 1975/6 Malawi (2) 69.7 30.3 0.0 100.0 1973/4 Zimbabwe (3)- 45.6 54.4 0.0 100.0 19130/1 Notes and references (1) 2 districts. Government plus mission. Data from IDS Health 6roup 1978. (2) 2 districts. Data from MOH Malawi 1975. (3) All rural districts. Primary includes provincial offices. Data fron Sovernment of Republic of Zimbabwe 1981. 62 Table 6 Ministry of Health current expenditure by type of-hospital Country Central General District Special Total Date Malawi (1) 54.0 7.0 35.4 3.6 100.0 1983/4 (2) 44.7 5.8 45.4 4.1 100.0 1963/4 Tanzania (3) 23.0 - 70.0 6.0 99.0 1978/9 Zimbabwe (4) 60.0 21.0 13.0 6.0 100.0 1981 Malaysia (5) 16.8 44.5 38.7 - 100.0 .1973 Sri Lanka (6) 63.8 - 36.2 - 100.0 197? Tunisia (7) 68.7 22.1 9.2 - 100.0 1971 Colombia (8) 35.6 31.3 27.2 5.9 100.0 1979 Notes and references (1) MOH only. Author's estimates. (2) MOH plus missions. Author's estimates. (3) 'Central'=teaching and zonal hospitals; 'district'=regional and district hospitals. Data from MOH Tanzania 1980, (4) 'District'. includes 3Z rural hospital expenditure. Data from Government of Republic of Zimbabwe 1981. (5) Peninsular Malaysia. Data from Heller 1975. (6) Alailima and-Mohideen 1984. (7) 'Central'= Institutes plus general hospitals; 'General'= regional hospitals. Data from Heller 1978. (8) 'Central'= University, 'General'= regional, 'District= local. Data from Robertson 1985 p.128. 63 Table 7 Unit costs by type of hospital (district hospital cost = 1) Country Cost unit Central Seneral District Rural Date Malawi (1) /IP day 2.4 0.9 1.0 - 1983/4 (2) /IP 3.5 - 1.0 1982 (3) /OP visit 1.2 - 1.0 - 1982 (4) 10P visit 5.6 - 1.0 - 1982 Zimbabwe (5) /bed 4.8 2.4 1.0 0.3 19B1 (6) /IP day 5.5 2.0 1.0 0.2 1979 Malaysia (7) /IP day - 1.6 1.0 - 1974 (8) l0P visit - 1.0 1.0 - 1974 (9) /IP day 1.6 1.3 1.0 - 1974 (10) /IP 2.4 1.9 1.0 - 1974 (11) /OP visit 2.1 1.3 1.0 - 1974 PNO (12) /IP day 2.0 1.2 1.0 - 1976/7 (13) locc bed/day 2.0 1.1 1.0 - Nay 1978 (14) /BP visit 1.5 1.1 1.0 - May 1978 Thailand (15) /IP day - 1.2 1.0 - 1981 (15) /OP visit - 1.2 1.0 - 1981 Tunisia (16) /IP day - 1.8 1.0 - 1971 (16) /OP visit - 1.9 1.0 - 1971 Columbia (17) /IP 1.2 1.1 1.0 - 1979 Notes and references (1) All hospitals. Author's estimates. (2) Unweighted-sean for 7 district and 2 central hospitals. Data fro& MOH Republic of Malawi 1984. (3) Unweighted mean for 7 district hospitals and Lilongwe central hospital. Data from MOH Republic of Malawi 1984. (4) Unweighted mean for 7 district hospitals and Queen Elizabeth central hospital. Data from nOH Republic of Malawi 1984. (5) Government of Republic of Zimbabwe 1981. (6) Unweighted mean of 4 central, 1 general, 7 district and 1 rural hospitals. For both central and district hospitals the range is sore than 3 fold within each category. Data from Robertson 1985 p. 122. (7) Mean from aggregated costs and outputs of 4 general and 7 district hospitals. Heller 1975. (8) Mean from aggregated costs and outputs of 4 general and 7 district hospitals. Heller 1975. (9) Unweighted mean of 3 general and 7 district hospitals treating Kuala Lumpur as 'central'. Data from Heller 1975. (10) Unweighted mean of 3 general and 7 district hospitals treating Kuala Lumpur as 'central'. Data from Heller 1975, (11) Unweighted mean of 4 general and 7 district hospitals treating Kuala Lumpur as 'central'. Data from Heller 1975 (12) Unweighted mean of I general (central), 3 base (general) and 15 provincial (district) hospitals. Data from Baker 1977. (13) Unweighted mean of 1 general (central), 3 base (general) and 4 provincial (district) hospitals. Data from Baker 1978. (14) Unweighted mean of 1 general (central), 3 base (general) and 4 provincial (district) hospitals. Data from Baker 1978. (15) 'Seneral'=provincial. Provincial hospital cost is mean of cost quoted for 4 specialties. District hospital cost is mean of costs for 3 different sizes of hospital. Data from MOPH Thailand 1983. (16) Heller 1975. (17) 12 University (central), 101 regional (general) and 453 local (district) hospitals. Data from Robertson 1985 p.128. 64 Table 8 Hospital current expenditure by function Country Hosp. type Adain Inpatients Outpatients Other Total Date Botswana (1) All MHH 14.3 58.8 18.9 8.0 100.0 1978/9 (2) Mission 18.4 48.3 21.2 12.1 100.0 1978/9 Malawi (3) -District ? 78.3 21.7 0.0 100.0 1973/4 China (4) County hosp. ? 42.6 57.4 0.0 100.0 1980 (5) Cosaune hosp ? - 23.0 64.4 12.6 100.0 1980 (6) Commune hosp - ? 10.8 72.8 16.5 100.1 1980 Malaysia (7) Central ? 83.2 16.8 0.0 100.0 1974 (8) Seneral ? 81.3 18.7 0.0. 100.0 1974 (9) District ? 74.1 25.9 0.0 100.0 1974 Tunisia (10) Seneral ? 91.1 8.9 0.0 100.0 1971 (11) Regional ? 86.9 13.1 0.0 100.0 1971 (12) District ? 77.6 22.4 0.0 100.0 1971 Coluibia (13) Psychiatric 7 73,8 13.8 12.4 100.0 1973/4 Notes and references (1) Data from MO Botswana 1979. (2) Data from HH Botswana 1979. (3) 2 districts. Data from MOH Malawi 1975. (4) Data from Hinman and Parker 1982. (5) Data. from Hinman and Parker 1982. (6) Data from Hinman and Parker 1982. (7) Kuala Lumpur. Data from Heller 1975, (8) Unweighted mean for 4 general hospitals. Data from Heller 1975. (9) Unweighted mean for 7 district hospitals. Data from Heller 1975. (10) Data from Heller 1978. (11) Data from Heller 1978. (12) Data from Heller 1978. (13) 'Other'= teaching & research. Includes capital services. Data trom Robertson 1985. 65 Table 9 Hospital current expenditure by main category of input Country Hospital Salaries Medical Food Transport Util- Maint- Other Ital Date type & wages Supplies ities enance Africa (1) Mission 47.6 18.1 5.8 ? ? 4.7 23.8 100.0 ? Africa (2) Mission 47.0 19.1 4.4 ? ? 4.2 25.2 99.9 1983 Botswana (3) All MOM 64.1 17.4 7.9 1.2 ? ? 9.3 99.9 197819 Malawi (4) Mission 41.6 19.1 0.0 5.4 10.1 8.0 15.8 100.0 1981 (5) Mission 45.0 20.6 0.0 4.7 8.8 7.0 13.8 99.9 1981 (6) District 52.0 17.2 8.0 7.1 11.9 0.0 4.0 100.2 1973/4 (7) District 42.0 29.4 11.5 8.2 4.3 0.0 4.6 100.0 1973/4 (8) Central 30.1 28.5 ? 12.6 ? ? 28.9 100.1 1980/1 (9) District 30.4 28.6 ? 24.7 ? ? 16.4 100.1 1980/1 Tanzania (10) Mission 46.0 18.0 11.0 7.0 5.0 ? 14.0 101.0 1981/2 (11) Reg., Dist. 45.0 25.0 13.0 ? ? ? 17.0 100.0 1978/9 PNG (12) All MOM 67.7 20.2 0.0 3.7 7.6. 0.0 0.8 100.0 1976/7 (13) Central 60.0 20.2 0.0 1.2 9.0 4.6 -4.0 99.0 1978 (14) General 63.2 21.3 0.0 3.2 5.8 5.7 0.8 100.0 1978 (15) Provincial 66.1 17.0 0.0 3.2 9.4 3.7 0.8 100.2 1978 Sri Lanka(16) Tertiary 51.0 ? ? ? ? ? ? 51.0 1979 (17) Secondary 52.0 ? ? ? ? ? ? 52.0 1979 (18) Private 33.0 25.0 ? ? ? ? 20.0 78.0 1979 Thailand (19) District 55.4 ? ? ? ? ? 44.8 100.2 1977 China (20) Commune HC 36.1 53.3 ? ? ? ? 10.5 99.9. c.1982 (20) Commune HC 22.0 49.9 ? ? ? 28.1 100.0 c.1982 (20) Commune HC 20.4 58.0 ? ? ? 21.6 100.0 c.1982 (21) County hosp 24.3 41.6 3.5 7 .? ? 30.5 99.9 1980 (21) Commune hosp 40.5 45.5 3.5 ? 8.2 ? 2.4 100.1 1980 (21) Commune hosp 35.9 54.5 0.8 ? 7.1 ? 1.6 99.9 1980 Jamaica (22) University 61.9 9.4 ? ? ? - 28.7 100.0 1980/1. (22) All MOM 73.3 . 10.8 ? ? ? - 15.8 99.9 1980/1 Colombia (23) Psychiatric 82.2 9.1 ? ? ? ? 8.8 100.1 1973/4 Notes and references (1) 50 Voluntary hospitals. van Lemen and van Amelsvoort 1985. (2) 23 voluntary hospitals. van Lessen and van Aaelsvoort 1985. (3) Data from MOM Botswana 1979. (4) Contributions in kind excluded. Msukwa and Siskonda 1982. (5) Contributions in kind included. Author's adjustments to Msukwa and Siskonda 1982. (6) Rumphi district. MOM Malawi 1975. (7) Mulanje district. MOM Malawi 1975. (8) 2 central hospitals. Data from MOH Republic of Malawi 1983. (9) 6 district hospitals. Data from MOM Republic of Malawi 1983. (10) Designated district hospitals. van Lessen and van Aselsvoort 1985. (11) MO Tanzania 1980. (12) Data from Baker 1977. (13) 1 central hospital. Data from Baker 1978. (14) 3 'base' hospitals. Data from Baker 1978. (15) 4 'general' hospitals. Data from Baker 1978. (16) Other categories of expenditure not given. Alailima and Mohideen 1984. (17) Other categories of expenditure not given. Alailima and Mohideen 1984. (18) 'Other'= profits and depreciation. Figures therefore not strictly comparable to rest of table. Alailisa and Mohideen 1984. (19) District Hospital in Lampang. Data from Robertson 1985. (20) Data from Van der Saag 1983. (21) Data from Hinman and Parker 1982. Table 10 Share of hospital current expenditure on salaries and medical supplies by function Country Hosp. type Z of total current expenditure by function on Salaries and med. suppi. share Salaries Medical Supplies as % of tot, functional expend. Inpatients Outpatients Inpatients Outpatients Inpatients Outpatients Malawi (1) District 32.7 79.4 32.0 18.9 64.7 98.3 (1) District 53.2 48.0 10.3 38.9 63.5 86.9 Malaysia (2) All N/A 68.8 N/A 18.9 N/A 87.7 China (3) County 22.4 25.7 30.4 49.2 52.8 74.9 (3) Commune 31.1 33.5 28.0 60.0 .59.1 93.5 (3) Commune 37.5 25.5 30.1 69.8 67.6 95.3 Notes and references (1) Data from NOH Malawi 1975. (2) Data from Heller 1975. (3) Data from Hinman and Parker 1982. 67 Table 11 Distribution of hospital costs between capital services and recurrent costs Country Hosp. type Capital Recurrent Total Vate Uganda (1) District 19.0 81.0 100.0 1966 PNG (2) Central 14.3 85.7 100,0 1978 (3) Seneral 11.4 88.6 100.0 1978 (4) District 15.7 84.3 100.0 1978 Malaysia (5) Gen/Distr IP 43.8 56.2 100.0 1573/4 (6) Sen/Distr OP 37.5 62.5 100.0 1973/4 Thailand (7) District 10.9 89.1 100.0 -1977 China (8) County 6.1 93.9 100.0 1980 (9) Commune 6.6 93.4 100.0 1980 (10) Commune 4.5 95.5 100.0 1980 Colombia (11) Psychiatric 10.9 89.1 100.0 1973/4 Notes and references (1) 1 hospital. Data from King 1966. (2) 1 hospital. Baker 1978. (3) Unweighted mean for 3 'base.' hospitals. Data from Baker.1978. (4) Unweighted mean for 4 'general' hospitals. Data from Baker 1978. (5) Meerman 1979. (6) Meerman 1979. (7) 1 hospital in Lampang. Pre-project figure. Data from Robertson 1985. (8) 1 hospital. Data from Hinman and Parker 1982. (9) 1 hospital. Data from Hinman and Parker 1982. (10) 1 hospital. Data from Hinman and Parker 1982. (11) 1 hospital. Robertson 1985. 68 TABLE 12. HEALTH SERVICE FEES AND AGRICULTURAL WAGE RATES Country, FEESa RATIO OF FEE TO DAILY AGRICULTURAL WAGE Year Outpatient visit In-hospital Daily agricultural wagec Outpatient visit 'In-hospit Adult Child Hospitalization delivery Adult Child Hospitalization delive Botswana, $.30 free ... $.37 Pula 7.4 $8.78 .034 0 ... .041: 1917 Barundi, $.22 $.22 $2.20 ... F 100 $1.11 .198 .198 1.98 ... 1993J Cameroon, CFA 600.- CFA 600.- CFA 500.-g CFA 5000 CPA 600 $1.83 1.0 1.0 .83 8.33: 1983 China, I 1.5-3d V 1.5-3 d V 27-46d ... V2/$1.17 $1.17 .75-1.50 .75-1.50 13.5-23.0 1982 Indonesia, $.36 $.36 ... ... Rp 400-450 $.60-.68 .529-.6 .53-.6 1983 Lesotho, $1.20 $.60 $.60h ... R 2.8 $3.60 .33 .167 .167 1980 Malawi, free free free K 1-3 K.59 $0.52 0 0 0 1.695-5.0; 1982 k Mali MF 0-2500e MF 0-2500 MF 200 MF 0-800 MF 500-1200 $.74-1.77 .0-5.0 .0-5.0 .167-.4 0-1.f 1982 Niger, free free free ... CFA 712 $2.17 0 0 .416 ... 1981 Pakistan, Rs 1.- Rs 1.- Rs 5.- ... Rs 12.02 $1.01 .083 .083 0 1982k Philippines, Pesos .25-5 Pesos .25-5 free ... Pesos 18.53 $2.35 .013-.270 .01-.270 0 ... 1982m Rwanda RwF 20.- RwF 20.- RwF 10-100 free RwF 92 $1 .22 .22 .11-1.09 0 1977 Togo, ... ... S .50-.70h $1.40-2.33 CSA 337 $1.67 ... ... .299-.410 .838-1.3 1979 . Burkina Paso, free free CFA 75.- CFA 100 CFA350-500 $1.03-1.18 0 0- .15-.214 .20- 19821 Source: For fees, World Bank Sector Reports and Econanic Memoranda For wages, World Bank country economists and project staff, except in the case of Malawi and Pakistan, for which ILO data were used. a Official fees for public facilities, except as noted. Actual charges may differ depending on enforcement and local practice. b First consultation only; fees for follow-up consultations may differ. c Estimate of actual minimm wage for agricultural labor, except as noted. Wide intra-country variations exist. d Average patient charge in selected comunme health centres in Shandong Province. e Consultations at public facilities are free of charge; fees are levied by traditional health workers. Weekly fee, which includes all services. 9 Registration fee. h Daily fee. Estimate based on wages paid to workers in the Padat Kenya Program. 3 Wage data are for 1982. k Wage data are for 1980. 1 Wage data are for 1983. m Wage data are for 1981. 70 TABLE 13. REVENUE FROM USER CHARGES AS A SHARE OF EPENDI1RE CN GOVERNMENT HEALTH SERVICES- Country, Year % of total % of recurrent Notesc expenditure expenditure Botswana, 1978 2.5 2.8 Burundi, 1982 3.3 4.0 For Health Ministry only. Colombia, 1980 17.3 28.4 Ghana, 1976-1977 ... 3.0 Total health as a percent of recurrent expenditure. Down from 5% in 1966-1967. Indonesia, 1982-3 12.9 15.5 All levels of government, excluding government emp- loyees' insurance scheme. Jordan, 1982 10.9 13.2 Excludes Royal Medical Service sponsored by Defence Ministry, due to lack of data. Lesotho, 1980-1 5.2 6.0 Down from 16% in 1974-1975. Malawi, 1982 2.8 3.0 Pakistan, 1980-1 .1.5 2.5 Peru, 1981. 7.2 8.0 Percentage of total (recur- rent plus capital) expendi- ture. Down from 12%. Philippines, 1981 6.4 6.8 Health Ministry only. Down from 14% in 1978. Rwanda, 1982 . 5.7 7.0 Sri Lanka, 1982 0.6 0.7 Down from 3.0% in 1974. Sudan, 1980-1 0.9 1.4 Central government only. Togo, 1979 ... 6.0 Tunisia, 1982-3 1.8 2.0 Zimbabwe, 1980-1 2.0 2.2 All levels of government, excluding Parirenyatwa Hospital. Down fran 10% in 1974-1975. Source: de Ferranti (1984) * a For several countries not listed, insufficient data exist to compute a percentage, but other evidence implies that the figure must be either (i) zero because no fees are charged or collection is not enforced, or (ii) very small (eg under 2%) because fees are minimal or, again collection is poor. Countries in this category include Angola, Bangladesh, Bolivia, Cameroon, Egypt, Gabon, Guatamala, Honduras, Jamaica, Liberia, Libyan Arab Jamahiriya, Mali, Morocco, Nigeria, St. Lucia, Democratic Yemen and Zambia. b In China, relatively high levels were found in reviews of the following selected areas: Shanghai County (26% for 1980); Yexian County, County Hospital (74% for 1981). c Figures exclude quasi-public institutions (eg social insurance schemes). 71 References Abel-Smith, B. (1967). An International Study of Health Expenditure. Public Health Papers No. 32. WHO, Geneva. Abel-Smith, B. (1986). "Funding Health For All - Is Insurance the Answer?" World Health Forum 7 (1) 3-32. Alailima, P. and F. Mohideen (1984). "Health Sector Expenditure Flows in Sri Lanka." World Health Statistics Quarterly 37 (4) 403-420. Attah, E.B. (1986). Underutilization of Public Sector Health Facilities in Imo State, Nigeria. PHN Technical Note 86-1. World Bank, Washington, D.C. Baker, W.G. (1977). In-depth Study on Hospital Expenditure. Assignment Report. WHO Regional Office for the Western Pacific, Manila. Baker, W.G. (1978). In-depth Study on Hospital Expenditure (phase II). Assignment Report. WHO Regional Office for the Western Pacific, Manila. Bekele, A. and M.A. Lewis (1986). "Financing Health Care in the Sudan: Some Recent Experiments in the Central Region." International Journal of Health Planning and Management 1(2) 111-128. Cumper, G.E. (1982). Report of Consultancy on Economic Analysis of the Health Sector in Jamaica. Evaluation and Planning Centre, London School of Hygiene and Tropical Medicine. Cumper, G.E. (1986). Health Sector Financing: Estimating Health Expenditure in Developing Countries. EPC Publication No. 9, London School of Hygiene and Tropical Medicine. de Ferranti, D. (1983). Some Current Methodological Issues in Health Sector and Project Analysis. PHN Technical Reports GEN 24. World Bank, Washington, D.C. de Ferranti, D. (1984). "Strategies for Paying for Health Services in Developing Countries." World Health Statistics Quarterly 37 (4) 428-450. Denoga (1982). Health Insurance in the Philippines. In International Social Security Association (1982). Department of Health and Social Security (DHSS) (1976). Sharing Resources for Health in England. HMSO, London. Department of Health and Social Security (DHSS) (1977). The Way Forward. HMSO, London. Echeverri, 0. et al (1972). "Postoperative Care: In Hospital or at Home?" International Journal of Health Services 2 (1) 101-110. 72 Government of Republic of Zimbabwe (1981). Planning for Equity in Health. Unpublished. Heller, P.S. (1975). Issues in the Costing of Public Sector Outputs: The Public Medical Services of Malaysia. World Bank Staff Working Paper 207. Washington, D.C. Heller, P.S. (1978). "Issues in the Allocation of Resources in the Medical Sector of Developing Countries: The Tunisian Case." Economic Development and Cultural Change 27 (1) 121-144. Heller, P.S. (1979). "Underfinancing of Recurrent Development Costs." Finance and Development 16 (1) 38-41. Hinman, A.R. and R.L. Parker (1982). "Costs of Care." American Journal of Public Health 72 (supplement) 83-88. IDS Health Group (1978). Health Needs and Health Services in Rural Ghana. Institute of Development Studies, Brighton. International Social Security Association (1982). Medical Care Under Social Security in Developing Countries. Studies and Research No. 18. ISSA, Geneva. Jamison, D.T. et al (1984). China. The Health Sector. World Bank, Washington, D.C. King, M. (1966). Medical Care in Developing Countries. Oxford University Press. Kravis, I.B. et al (1978). International Comparison of Real Product and Purchasing Power. Johns Hopkins University Press. Laurent, A. (1982). Health Financing and Expenditures. Rwanda and Togo. Third World Series: Sandoz Institute for Health and Socio-economic Studies, Geneva. Lee, K. and A. Mills (1982). Policy-Making and Planning in the Health Sector. Croom Helm, London. Lee, K. and A. Mills (1983).- The Economics of Health in Developing Countries. Oxford University Press. Maxwell, R.J. (1981). Health and Wealth. Lexington Books. Maynard, A. and J.C.C. Smith (1983). The Elderly. Who Cares? Who Pays? Nuffield Provincial Hospitals Trust, London. McGreevey, W. (1982). Brazilian Health Care Financing and Health Policy: An International Perspective. PHN Technical Notes GEN 6. World Bank, Washington, D.C. Meerman, J. (1979). Public Expenditure in Malaysia. Oxford University Press. 73 Mesa-Lago, C. (1983). Financing Health Care in Latin America and the Caribbean with a Special Study of Costa Rica. PHN Technical Notes GEN 8. World Bank, Washington, D.C. Mills, A. (1980). "Health Services for Low Income Groups - Access to Free Medical Care." Health Planning Division, Ministry of Public Health, Thailand, mimeo. Mills, A. (1983). Report on a Visit to Malaysia, Thailand and Geneva. EPC, London School of Hygiene and Tropical Medicine. Ministry of Health, Botswana (1979). The Financing of Health Services and Activities in Botswana. Republic of Botswana. Ministry of Health, Malawi (1975). Results of a Study of Expenditure on Health Care in Malawi. Unpublished. Ministry of Health, Republic of Malawi (1983). Report on Health Financing Survey. Ministry of Health, Republic of Malawi (1984). "A Survey of Health Sector Costs and Financing in Malawi." World Health Statistics Quarterly 37 (4) 375-386. Ministry of Health, Malaysia (1977). Programme and Performance Budgeting System Implementation Manual. Ministry of Health, Tanzania (1980). Evaluation of the Health Sector. United Republic of Tanzania. Ministry of Public Health, Thailand (1983). Unit Service Costs in Provincial Hospital Services in Thailand. And: Cost of Rural Health Facilities in Thailand. Health Planning Division. Unpublished. Msukwa, L.A.H. and H.P.M. Simkonda (1982). Report of the Evaluation of the Private Hospital Association of Malawi. University of Malawi, Centre for Social Research. Rigden, M.S. (1983). Health Service Finance and Accounting. Heinemann, London. Robertson, R.L. (1985). Review of Literature on Costs of Health Services in Developing Countries. PHN Technical Note 85-21. World Bank, Washington, D.C. Smith, J. (1980). A Review of the Resources and Service Area of the Health Facilities in the Kingdom of Lesotho. Ministry of Health, Maseru. Quoted in Prescott N, Warford J, Economic Appraisal in the Health Sector. In eds Lee K, Mills A (1983). Thomas, M. (1982). "Current Issues Surrounding Recurrent Costs in Health Services of Developing Countries." Strengthening Health Services Division, WHO, Geneva, Draft. 74 van der Gaag, J. (1983). Commune Health Care in Rural China. PHN Technical Notes GEN 20. World Bank, Washington, D.C. van Lemmen, F. and V. van Avelsvoort (1985). "How do African Voluntary Agency Hospitals Spend their Money?" Unpublished. Vaughan, P., A. Mills, and D. Smith (1984). "The Importance of Decentralized Management." World Health Forum 5 (1) 27-29. Viveros-Long, A. (1986). "Changes in Health Financing: the Chilean Experience." Social Science and Medicine 22 (3) 379-385. Walker, G., M. Longford, and A. Mills (1980). Health Insurance in Jordan. Report to the Overseas Development Administration, London. Mimeo. Wheeler, M. and J.B. Volpatti (1980). Health Service Finance. Departement Keuangan Djarkarta, Indonesia. World Health Organization (1982). Country Resource Utilization Review, Nepal. WHO, Geneva. World Health Organization (1984). Programme Budgeting as a Part of the Managerial Process for National Health Development. WHO, Geneva. 75