Document(s) 7 of 19

Introduction

Health care systems in sub-Saharan Africa face increasingly diverse and complex health problems, rapidly growing populations, and severe resource constraints. Rational allocation of scarce resources is difficult and is dependent on the size of catchment populations. Expensive hospital-based health care systems are protected by strong vested interests, reorientation is mainly rhetorical, and primary health care is making only slow progress.

Health care management and the use of health information at the local level are restrained by highly centralized decision-making processes. Several other weaknesses further restrict the usefulness of the health information system in Kenya. Some highly desirable information such as population-based epidemiology, service quality data, and sociocultural information is not being collected.

Problems also exist in the flow of information from the field, including delays, nonreporting, nonresponse, and a generally unsatisfactory quality of generated data. Moreover, current reporting is largely restricted to acute and brief illness episodes in people fit and affluent enough to seek care at a health facility; those who live far away, are too sick to travel, are worried about cost implications, or are chronically sick and disabled and have little to benefit from a visit remain "invisible," neglected by the service system and overlooked by planners. Much of this is known by the local community, but unknown to the health care system and its staff.

The Kenyan population is heterogeneous with ethnic, religious, and socioeconomic differences influencing illness concepts and demands for health care. However, the only available information on the local population is the Census of Population, normally conducted every 10 years, which does not coincide with the actual catchment populations served by health facilities. The health information system should reflect these circumstances, but available data are almost exclusively about care-seeking clients and their service utilization. A need exists for a more precise and complete description of the catchment population and health situation. It is important to generate this information at village, community, and division levels.

Three interrelated problems are prominent. First, health care planning and programing are highly centralized and largely incremental in nature. Second, most staff are poorly prepared for management and planning responsibilities, particularly at and below the district level. Third, existing health information systems are unsatisfactory both as tools for planning and as a basis for local community development and action. The use of geographic information systems for determining catchment populations may help to alleviate some of these inherent problems.

Geographic Information Systems (GIS)

Applications
A geographic information system (GIS) is designed to work with spatially referenced data (Maxwell 1976; EPA 1987). In other words, a GIS has both attribute and spatial databases with specific capabilities for manipulating spatially-referenced data, as well as a set of operations for working with the attribute data; in a sense, a GIS manipulates spatial and nonspatial data.

The essential components of a GIS are the same as those of any other information system, namely:

• Data acquisition;

• Preprocessing;

• Data management;

• Data manipulation and analysis; and,

• Product generation.

The main emphasis in data analysis should be on the production of maps of demographic data. This activity relies on the collection of point data from known locations, and the global positioning system (GPS) is be used to record precise locations where sample data are located. At the end of the data collection activity, an associated count data (point data) will exist for each sample location. In the collection of data, efforts should be made to locate those areas from which no participants are coming. This will aid in the interpolation, by ensuring that areas with zero count (no patients) are taken into consideration.

This point data will be input into the GIS. To facilitate data input, the data should be in ARC/INFO ungenerated point format, that is: Value, X-coordinate, Y-coordinate, with value being the observed count for the location identified by the X-, Y-coordinates.

The next step is to generate a points coverage (digital version of a map) in GIS and to build a correct topology with related feature attributes. Additional information can be added to the attribute table, for example counts by sex. This coverage will form the input to the triangulated irregular network (TIN) module.

Using the ARC/INFO TIN routine, a TIN coverage will be generated. From this coverage, an isoline coverage will be produced. The final coverage will show those areas of equal numbers through the use of contours.

The determination of health facility catchment areas is more involved, as it requires the active participation of the local communities. Homesteads/villages are identified with a unique number. As patients visit a health facility, they will be required to also provide their homestead/village particulars. In this manner, for a maximum of about 2 months, the spatial coverage for a facility serving a typical village (600-1000 people) can be identified, and the specific areas where patients originate mapped.

The frequency of visits will then be tabulated or graphed. Interpretation of this data must take into consideration a number of factors, such as transportation costs, condition of roads, availability of local medicinemen, and so on. These factors will affect the size of the population visiting a health facility, frequency of visits, types and conditions of illnesses involved, and so on.

For each health facility, a plot showing where patients come from in a given period will be produced in the form of a dot map. All patients visiting health facility n will also be recorded as n, X-Coordinate, Y-Coordinate, where X-, Y-Coordinate is the homestead location. Isoline maps produced from this data will show the geographic reach of health facility n. Areas of overlap between catchment areas will be determined by overlaying maps from health facility n and n+1.

Health planners are confronted with the problem of selecting the optimum location for a health facility in relation to the spatial distribution of the catchment population and infrastructural facilities already in place. Although this is not an easy exercise, the application of GIS may aid in decision-making. A study to find the best location for a health facility would start by identifying homesteads, road networks, and health facilities in the study area. For the allocation of health facilities, a number of variables such as road conditions and types and travel times will have to be incorporated into the database before running the allocation function. The various solutions generated will have to be evaluated against the criteria specified for siting the health facilities. Such criteria may be to locate a facility where travel costs are minimized.

Software
Although there are several pieces of software on the market, the PC ARC/INFO system is preferred for its extra spatial analytical capabilities. However, it needs a number of modules to facilitate work in the project described above. These are:

• A TIN module, for generating triangulated irregular networks;

• A NETWORK module, for resource location/allocation and selecting the shortest path between a given set of points; and

• NEXPERT, an expert decision system (EDS).

Conclusion

The application of the GIS technology for determining catchment populations is feasible. The data required from health facilities serving a typical African village would take a maximum of 2 months to collect. The incorporation of remote sensing is also needed for handling migratory populations, such as the nomadic communities found in the semi-arid and arid lands in eastern and northern Kenya.

References

• Cohen J.M.; Hook R.M. 1987. Decentralized planning in Kenya. Public Administration and Development, 7, 77-93.

• EPA (Environmental Protection Agency). 1987. US cancer mortality rates and trends, 1950-1979. US Environmental Protection Agency, Research Triangle Park, NC, USA. EPA 600/1-83/015.

• GOK (Government of Kenya). 1984. District focus for rural development. Office of the President, Nairobi, Kenya.

• Mason, T.J.; McKay, F.W.; Hoover, R.; Blot, W.J.; Fraumeni, J.F. Jr. 1975. Atlas of cancer mortality for US counties: 1950-1970. Government Printing Office, Washington, DC, USA. DHEW Publication No. (NIH) 75-780.

• Maxwell, E.L. 1976. Multivariate system of analysis of multispecral imagery. Photogrammetric Engineering and Remote Sensing, September, 22-24.

• Ministry of Health. 1986. National guidelines for the implementation of primary health care in Kenya. Ministry of Health, Nairobi, Kenya.

• Shlien, S.; Smith 1977. A rapid method to generate special theme classification of LANDSAT imagery. Remote Sensing of Environment, 4, 67-77.

• Wetmore, S.P.; Townsend, G.H. 1975. A geographical mode for storage identification: Analysis of Ecological data. KREMU, Nairobi, Kenya. Technical report no. 1.

Document(s) 7 of 19