65. The District Hospital

Economics of District Hospitals: A Summary of Reported Experience

The previous sections outline the suggested functions and extended role of a district hospital. Although some countries have adopted the principle of essential packages of services and defined detailed norms and standards for care at this level as part of long-term health sector strategies, many countries lack any specific hospital strategy (WHO 1994). Even where a well-articulated strategy exists, decades of different political, social, economic, and historical influences on health system development result in great variability of district hospitals, both between and within developing countries. Thus, some district hospitals of 500 beds have a full complement of specialist consultants and access to a wide range of diagnostic and therapeutic services, while other hospitals of as few as 30 beds, but more often 80 to 150 beds, are run almost entirely by medical assistants and nurses, sometimes lack reliable power or water supplies, and often offer few or no high-quality modern diagnostic services. This variability makes it daunting to extrapolate findings from one setting to another and may seriously undermine the value of attempts to provide useful general descriptions of hospitals. In particular, when interpreting calculated costs of care at a national or individual level, we must remember several critical points:

• Relevant data may often be missing or inadequately defined at a country level.

• Because a number of accepted ways of calculating costs exist, particularly at the level of individual interventions, different methods are likely to lead to different estimates. The particular design used to estimate costs should be considered when interpreting any results.

• In particular, a central feature of the hospital is that many of its resources are used for more than one activity, so unit cost estimates depend crucially on how the costs of these resources are allocated among activities.

• The relative prices of inputs can vary substantially between regions and countries.

• In the majority of cases, only the cost of care is reported without reference to outcomes so that the cost per unit of health benefit (however defined) is unknown.

• Calculated costs usually reflect the care offered; it may not be the same as the care that is necessary, of an acceptable quality, or most effective.

• Cost estimates cannot indicate the extent of unmet need or other sources of inequity.

• The costs of care will depend to some extent on the severity of illness of the patients and, for average costs per bed day, on the variety and relative proportions of different illnesses (the case mix). These areas are rarely commented on or adjusted for.

Levels of Provision of Hospital Care

Data on the levels of service provision for many developing countries are crude. In the absence of any more meaningful data, the number of beds is most often used as a (poor) substitute. Bearing this weakness in mind, sources estimate the average number of total hospital beds to be 1.3 per 1,000 population in developing countries (World Bank 2002), a figure probably declining in many developing countries (Hensher and others 1999), with varying estimates of the average number of doctors from 0.5 per 1,000 population in low-income countries generally (World Bank 2002) to 0.09 doctors per 1,000 population in Sub-Saharan Africa (Peters and others 2000). These estimates are considerably lower than the averages for beds and doctors of 7.2 per 1,000 and 2.9 per 1,000, respectively, in high-income countries (World Bank 2002). Although these estimates provide some indication of the major disparities in service provision between rich and poor countries, their value is limited. Lack of information on the relative distribution of beds and staff by geographic zone, or between district and higher referral levels of care in a single country, and the fact that bed and staff numbers are probably a poor reflection of activity make these figures a poor substitute for data on patient throughput and outcomes, statistics rarely available for district hospitals. Furthermore, with the concentration on provision of service, the demand for services may often be ignored. It is still true in many countries that most deaths, presumably many preventable, occur at home and that many chronic diseases are inadequately treated. The need for hospital care is largely undetermined, but some have argued that the lack of provision of district hospital care, in Sub-Saharan Africa at least, is a significant impediment to improving overall health status (Van Leberghe, de Bethune, and de Brouwere 1997).

What Do District Hospitals Cost at a National Level?

Although it has been argued for some time that hospitals consume too much of health sector budgets, thereby depriving primary care of adequate resources, it is surprisingly difficult to identify how much hospitals cost in low- and middle-income countries. Even where data exist on health expenditure, such data are often at a highly aggregated national level and the functions that are included (clean water and sanitation, for example) are not always clear (World Bank 2002). Furthermore, whether private or nongovernmental expenditure, capital expenditure, or the value of noncash inputs—such as donations of equipment or volunteers' time—are included is rarely apparent. Add to this ambiguity the nearly impossible problem of separating what is spent at different levels of the health or hospital system—for example, to distinguish between district and referral hospitals—and it should be clear that we currently have only a crude understanding of the costs of district hospitals as a unit of service provision (Mills 1990a).

If just government health expenditure is considered, the available data suggest that hospitals at every level taken together consume 50 to 60 percent of recurrent national health budgets, with the proportion appearing to increase as countries become richer (Barnum and Kutzin 1993). If private expenditure on health care (insurance and out of pocket) is included, the proportion of total health expenditure consumed by all hospitals falls to 30 to 50 percent of the total in developing countries (excluding South America) (Mills 1990a). Whereas these figures reflect total hospital sector expenditure, the limited data available suggest that district hospitals may receive less than 50 percent of this total in many countries, consuming fewer resources than secondary and tertiary referral facilities (Mills 1990a).

The Nongovernmental and Private Sectors

In many countries (especially in Africa) nongovernmental institutions, often religious organizations, are major health service providers, and private physicians are often as numerous as those in the public sector. In Kenya, for instance, the number of private and nongovernmental hospitals is equal to the number of public hospitals (Government of Kenya 2001), while in Indonesia, 32 percent of hospital beds are private (Gani 1996). This potentially important contribution to the hospital sector may also be underrecognized, particularly in urban settings, where multiple, small facilities may operate without registration, resulting in inaccurate local, regional, and national data on levels of overall service provision. Although few data exist on the effectiveness and quality of these hospitals, the belief is widespread that they may be more efficient than public sector hospitals. This belief is not necessarily borne out by the limited data available (Bitran 1996), and concerns exist about the quality of care provided by private as well as public providers (Brugha and Zwi 1998).

District Hospital Efficiency

Data on hospital efficiency in developing countries are scant. Considerable variability has been observed in the technical efficiency with which surgical services were provided in a small number of Indian hospitals, with differences in total salary costs being the main explanatory variable (Purohit and Rai 1992). Also in India, some evidence has been provided that nongovernmental hospitals may be more efficient, on average, than public hospitals, although considerable variability existed within both groups (Bhat, Verma, and Reuben 2001). In Kenya, public hospitals were found to have an average inefficiency level of 30 percent (that is, the same resources could have achieved a 30 percent increase in output) with significant contributing factors including shortage of appropriate professional staff members, poor combinations of inputs (resources), nonfunctioning theaters and laboratories, lack of transportation, irregular distribution of drugs and supplies, and frequent breakdowns in medical equipment (Owino and Korir 1997). All these data highlight the critical role of human resources, often a hospital's principal recurrent input cost (see the next section). Underinvestment in or absence of staff or inadequate flexibility in reallocating roles between different health worker groups may prevent hospitals from functioning efficiently (Hensher 2001).

What Are the Costs of Providing Care in District Hospitals?

In a detailed review of actual hospital expenditure, Mills (1990b) identified two input categories that together accounted for two-thirds or more of recurrent expenditure in almost all settings. Salaries varied between 20 and 80 percent and medical supplies between 15 and 58 percent of reported hospital expenditure. These and other data also suggest that, in many countries, costs of referral hospital care are often more than double the cost of equivalent care at district hospitals, although without knowledge on case mix or illness severity such data are hard to interpret (Barnum and Kutzin 1993; Mills 1990b). More recent data collected from seven church-supported hospitals in Tanzania also demonstrate considerable variability in the proportion of costs attributable to salaries and supplies even within a single organization in the same country (Flessa 1998). The strong dependence of hospital costs on salaries particularly cautions against generalizations across countries.

In the following analysis, all original U.S. dollar costs have been adjusted to represent the U.S. dollar cost in 2004. The Tanzanian nongovernmental hospital data indicate that the average cost per inpatient day derived from 1995 reports (including expenditure on maintenance and expatriate salaries) would equate now to approximately US$3.60 (range US$2.60 to US$6.00) in district hospitals (Flessa 1998). However, if care had actually been provided according to the standards defined by the provider (including recommended staffing levels, building maintenance, and equipment), the estimated cost per day would have risen to the equivalent of US$11.60 (range US$9.20 to US$15.90) (Flessa 1998). This cost compares with costs reported in Kenya in 1993-94 (Kirigia, Fox-Rushby, and Mills 1998), adjusted to 2004 prices of actual inpatient costs per day from two district hospitals of US$8.30 to US$10.10, and adjusted 1995 data from a district hospital in Bangladesh of US$15.90 (McCord and Chowdury 2003). In a middle-income country, South Africa, the cost per inpatient day calculated between 1996 and 1998 and adjusted to 2004 prices in five district hospitals ranged from US$37.80 to US$96.30 (Daviaud and others 2000). These data do not necessarily reflect the cost of optimal care, and the Tanzanian study demonstrates that even in externally supported hospitals actual expenditure may be insufficient to provide good-quality care and cover essential maintenance, resulting in steady deterioration of capital stock and worsening efficiency in the long term.

Data describing costs of treating some specific conditions in district hospitals are summarized in table 65.1. Given the difficulties in extrapolating data across contexts and the potentially significant effect of exchange rate fluctuations, great caution should be used in interpreting these data, which, it should be noted, derive in all cases from specific research rather than routine sources.

Measuring the Effect and Cost-Effectiveness of District Hospitals

In the previous section, some limited data on the costs associated with provision of care at the district hospital were presented. What of a hospital's cost-effectiveness? Ideally we would like to know the aggregate health output of a hospital in terms of improved health compared with a situation in which there is no hospital. Such data do not exist, even from industrial countries, where the hospital has been the subject of intense academic study.

However, some attempts have been made to estimate the effect of a hospital by comparing the observed outcome of illness treated with hospital care to consensus expert opinion on the likely outcome of illness in the absence of hospital care. Using this approach in Kenya, Snow and others (1994) estimated that a well-functioning rural district hospital might reduce all-cause child mortality by 44 percent in a population with reasonable access to the hospital (see table 65.2). Extending this approach, researchers in a small rural hospital in Bangladesh calculated the benefit of hospital admission for patients of all ages suffering from life-threatening conditions using a slightly modified DALY (McCord and Chowdury 2003). Over a three-month period, the total costs (including all staff, capital, and hotel costs) of running the hospital were calculated and divided by the estimated total number of DALYs gained attributable to inpatient care over the same three months. The authors report an average cost per DALY of approximately US$11.00 in 1995, or US$13.30 in 2004 dollars (McCord and Chowdury 2003; see table 65.3). This figure compares favorably with costs per DALY of many primary care interventions regarded as highly cost-effective (World Bank 1993). To what extent these results depend on the quality of primary care, the referral system, the inpatient care, the hospital administration, and the commitment of health personnel working for a small independent nongovernmental organization will remain uncertain until more such data become available.