Geospatial analysis and decision support for health rvices planning in Uganda
Shuaib Lwasa
Urban Harvest, International Potato Center, Kampala, Uganda
Abstract.As the utilization of geospatial techniques continues to surge, spatial information has become an integral part of decision-making. In Uganda, the u of geospatial techniques in provision of health rvices planning has gained momentum after a comprehensive survey of health units and the development of a national health rvices geodata-ba. Planning for the provision of health infrastructure rvices requires quality information to rationalize the loca-tion, and allocation, of rvices in relation to the population. Health rvice planners are always faced with a question of where to locate rvices in relation to need and how such distribution would be affected by resources to meet the requirements of the population. Becau resources are scarce, prioritization is indispensable and thorough analysis becomes important in the planning process. This paper analyzes access to health facilities using the population gridding approach, coupled with location of health infrastructure facilities for decision support in health rvices planning. Keywords: population gridding, location-bad rvices, health planning, Uganda.
空王
Introduction
管理育人The role of geospatial technologies in planning and management of location-bad rvices is underscored by the veral studies related to social rvices provision. The health infrastructure in Uganda (and elwhere) depends on the physical structure and supporting equipment established for provision of health rvices. It usually involves facilities for different health rvice needs, equip-ment such as cold-chain facilities for storage and management tools for the distribution of health rvices to the population (MoH, 2002). Like in other developing countries of sub-Saharan Africa, health systems in Uganda are increasingly facing challenges in ensuring health care to its population (Leonard and Masatu, 2007; Mullan, 2007). Challenges are pod by a number of factors,including population growth, uneven population distribution in relation to the natural resource bas, low access by some areas due to limited transporta-tion networks, availability of human resources and financial requirements for managing and provision of the health rvices. There is evidence that poor access to health care can be counter productive to growth and development and that rural communi-ties are particularly vulnerable to the conquences of inaccessibility to health rvices (Leonard and Masatu, 2007). Health rvice provision is one of the many basic location-bad social rvices that need to be provided in-tandem with the spatial dis-tribution of the population. Planning for the pr
ovi-sion of health infrastructure therefore requires qual-ity information on location of rvices, capacity of facilities, population and catchment distance (the distance traveled by the furthest accessing patient to a health facility). Although a country’s population requires proximity of health infrastructure and health care, some areas are rved by distant facili-ties becau of resource constraints. Prioritization of health rvice location therefore considers veral factors, need and gaps being important among them (Katz et al., 2006). According to the national health
Corresponding author:
Shuaib Lwasa
Urban Harvest, International Potato Center Naguru Hill, Katarima Road, Plot 106
P.O. Box 22274
水菠萝Kampala, Uganda
Tel. +256 41 4287571; Fax +256 77 2461727
E-mail: s.lwasa@cgiar; lwasa_s@arts.mak.ac.ug Geospatial Health 2(1), 2007, pp. 29-40
S. Lwasa- Geospatial Health 2(1), 2007, pp. 29-40 30
policy (MoH, 2002), health infrastructures, sup-porting populations in rural as well as in urban areas, are to be established in various administrative units within 2 km reach. The Uganda Bureau of Statistics (UBOS, 2002) conducted a socio-econom-ic survey on health rvice access and the result indi-cates that, within 5 km radius, the national average of health facility access is 73.2% with 69.6% for the rural population and 95.8% for the urban popula-tion. This was considered an improvement from the staggering proportion of the population living at the 10 km limit in 1991. Access to health rvices is clo to the policy target of 2 km radius for urban areas but not for rural areas. This rural-urban qual-ity divide can be explained by focusing on veral issues, but it ems that the number of people rved by a health facility and the distance to the nearest health facility are important indicators (Abel-Smith and Rawal, 1992; Shrestha, 2000; Leonard and Masatu, 2007).
This paper analyzes access to health facilities cou-pling population and location of health infrastructure facilities for decision support (Clancy and Cronin, 2005). The analysis is intended to enhance the under-standing of location-bad rvice analysis and provi-sion of information for planning of health rvices.
Decision-making in health rvices planning
Decision-making is a process of solving a problem which is said to exist if “someone is in doubt as to which choice is best to remove his dissatisfaction with his prent state” (Clancy and Cronin, 2005; WHO, 2006; O’Connor et al., 2007). Such a per-son, or in this context government, can identify three aspects related to the choices, namely:
(i)one or more outcomes that he/she desires; (ii)two or more unequally efficient or effective cours of action; and
(iii)environment-containing factors that affect the outcomes (WHO, 2006).
In decision-making there is an ideal of behaving objectively and rationally in which optimal cours-es of action are found and relevant information for the decision are assumed to be readily available. But as Clancy and Cronin (2005) obrved, infor-mation may not be readily available depending on the level and type of decision to be made. In addi-tion, O’Connor et al. (2007) pointed out the importance of decision quality and supportive role of rearch coupled with decision-specific instru-ments for decision making. In practical terms deci-sion-makers usually do not have all the relevant information when making decisions becau of the time and cost constraints in gathering such infor-mation (Tunis et al., 2007). A decision-maker will stop gathering once some information is available o
n the basis of which a decision can be reached. But more information, either directly gathered or analyzed from existing data, would probably yield better decisions as pointed out in the context of health rvices planning (Einberg, 2002; WHO, 2006). Thus decisions tend to focus on procedures that lead to a solution that may not necessarily be optimal.
The path usually followed in decision-making is the procedural rationality in which the cour of action involves arching for a satisfying rather than an optimal alternative (Katz et al., 2006; Yates et al., 2006; Health Services Rearch, 2007). Information arch and its evaluation are very criti-cal and, when spatial data is considered, as depicted in Figure 1, available techniques for evaluation pro-vide a wide range of possibilities for manipulation to ensure better decisions bad on existing infor-mation (ILRI and CBS, 2002). For policy analysis, part of the decision-making process is how spatial information can be captured and what methods would be ud to evaluate such information. Decision-makers usually operate within a tight time frame with inadequate resources and informa-tion (WHO, 2006; Yates et al., 2006). As worded by Mullan (2007), they are influenced by special-inter-ests, bureaucratic imperatives, and political forces who visions extend no further than the next elec-tion cycle. The current health rvices planning sys-tem in Uganda utilizes population data linked to administrative units with different spatial attributes
S. Lwasa- Geospatial Health 2(1), 2007, pp. 29-4031
and thus do not “spatiallize” population for appro-priate allocation.
Having discusd the process and importance of decision-making, it is prudent to also highlight its application in health rvices planning. Like other location-bad rvices, health rvices are critically tied to the space in which populations live. As point-ed out by Shrestha (2000), it is a challenge to provide health rvices acceptable when the population is unevenly distributed. Although Gardner et al. (2007) do not explicitly highlight the importance of geospa-tial techniques, the focus on information technology indicates that all possible cours of action in health rvices planning, location inclusive, need to be explored. The geospatial analytical techniques are invaluable in analyzing and visualizing rvice gaps and needs and their introduction reprent a crucial step towards allocation for rvice delivery and improvement of health care (Yates et al., 2006).
Overall purpo and objectives
Location-bad rvices for health delivery prent a rearch and policy challenge in dealing with the spatial distribution of, and the relation of such rv-ices to, other variables including the population.怎么炸酥肉
Geospatial analytical techniques have been applied for the analysis of information much needed in decision-making. The data analysis can range from visualization, exploration through spatial sta-tistics to spatial econometrics (Bivand, 1998; Jeong and Gluck, 2002; Davis, 2003). The underlying spa-tial relationships notwithstanding, exploration and visualization of data is very significant for decision-making becau it clarifies where the need is or where hotspots and gaps exist. Such information provides a basis for evidence-bad planning and management of rvices (Clancy and Cronin, 2005). Spatial obrvations upon which analysis can be undertaken include fields or surfaces, point patterns and lattice obrvations where attribute values relate to a grid, e.g. an administrative unit (Bivand, 1998). The prent study address the question how enhanced information regarding the relationships between rvices, capacity and catchment distance on one hand, and the population distribution on the other, can improve decision-making in the planning of health rvices. This question emerged from the fact that current rvices planning mechanisms are largely bad on econometric models which could possibly be enhanced if coupled with spatial analyti-cal techniques. In this way, the historical explana-tions of locating rvices could not only be explored but pointers for policy on what actions to be taken would also be provided (Walsh et al., 1993). The study employs triangulation of spatial techniques, including population gridding and spatial aggrega-tion of health rvices using spatial statistics to gen-erate information for improved d
ecision-making. The specific objectives were to spatially analyze health facility access by relating location and popu-lation, including:
(i)application of location-bad analyzes of access
to health facilities in Uganda;倒垃圾英语
(ii)generation of spatial information for visualiza-tion and support planning and delivery of health rvices; and
(iii)demonstration of the u of geospatial informa-tion and techniques in provision of information
required for planning and health rvices delivery. Fig. 1. The place of spatial data in the decision-making process.
S. Lwasa- Geospatial Health 2(1), 2007, pp. 29-40 32
Materials and methods
Population data
Population data utilized in the analysis were derived from the results of the Population and Housing Census of Uganda (UBOS, 2002). The data were captured as aggregate data for administrative units of parishes, the cond-level administrative unit in Uganda, and detailed enough for the national-level type of analysis. The choice of parish-level pop-ulation data was determined by the unavailability of readily usable spatial data layer for the lowest unit, i.e. the village. This population data is available in tabular form with linked data and parish labels. With the shapefile (NF A, 1996) of the national administrative layer, the data was entered into the geodataba for further processing. Processing of the data included adjustment of the population with the annual growth rate to compute estimates for the year 2006 (UBOS, 2005). To that end, the population esti-mates ud in the analysis were adjusted using the annual growth rate of 3.2%. Although this prents a problem of uni
社会保障卡是医保卡吗fying the growth rates for all adminis-trative units, the resultant total estimate was consid-ered accurate enough for the purpo of this study. Administrative boundary data and h ealth facilities GPS point data
Spatially-explicit data were acquired from the updated national administrative geodataba (NFA, 1996). Although some boundary changes were noted, especially at the sub-county and district lev-els, the parish-level boundaries had not significantly changed. Geospatial analytical techniques were uti-lized to process the data for the analysis. A grid tool providing the opportunity to determine spatial reso-lution enabled generation of a regular square spatial layer that was geo-referenced and linked to the administrative boundary of Uganda. The processing of the spatial distribution of the population involved veral steps, for example, excluding water bodies from the administrative layer, to enable a more accurate estimate of the population density. The resultant map layer was overlayed with the grid layer and population estimates calculated for each area unit and later aggregated to each grid. Becau a grid is a square cell, the process produced area units which were dicted by the administrative boundary layers during overlay to enable calcula-tion of population of each unit area bad on the density and area of the disaggregated areal units of the grids (Diechmann and Balk, 2001).
In this study a 5 km spatial resolution, as described in de By (2004), was utilized and the result of this process was a spatial layer with a population esti-mate for each grid (Fig. 2). This technique is more robust than administrative-bad summarizes which are not spatially-populated and where the popula-tion densities include the water bodies. In addition, the health infrastructure databa was acquired from the geodataba created by global positioning system (GPS) data which had been captured from 2000 to 2003 and continuously updated. Additional data available from the health facilities geodataba include health facility grades and type of ownership, catchment distance, and bed capacity which was also collected during the GPS point survey. Geostatistical analysis (Anlin, 2004) and ArcGIS were utilized to spatially explore the data, visualize and statistically analyze relationships between the key variables of population, number of health facilities, average catchment distance and bed capacity. The data were explored in terms of loca-tion randomness within the grids. The study utilized the regression model and Moran’s I statistic for rela-tionships between variables generating significance maps which were visualized to enhance understand-ing of access to health by distance and the spatial distribution of the facilities.
Results and discussion
Exploration of population data
To establish where people are generally concen-trated, the population was “spatiallized” using a
S. Lwasa - Geospatial Health 2(1), 2007, pp. 29-4033
掎角之势
gridding approach (Diechmann and Balk, 2001;Davis, 2003) regarded as fairly accurate. The grid-ded datat is utilized to visualize the spatial distri-bution of population in Uganda before relating it to provision of rvices. Decision-making requires robust data on location of rvices and population which provides the demand of rvices. Exploration of population distribution in Uganda indicates con-centration of population in areas around the major water bodies of Lake Victoria and Lake Kioga, and the mountainous areas although there are a few out-liers, especially in the areas of conflict in the North and North-East. As shown in F igure 3, the explo-ration also revealed smaller concentrations in areas with populations less than the mean of the grids.The implication of this analysis is that location of rvices needs to be nsitive and responsive to pop-ulation distribution. A further analysis of the data indicates that 251 grids have populations in the upper outlier implying high concentration of popu-lation in relatively small areas which are the areas of current conflict. This distribution has implications for rvices location and planning and, later in the paper, is utilized to analyze the relationship between population and rvices to generate information uful for decision-making.
Health infrastructure development in Uganda The Ministry of Health (MoH) in Uganda has a programme for health infrastructure construction. It equips the health centers (HC) across the country which are designated as HC II, HC III, HC IV and hospital 1according to the grade of sophistication.This programme is a respon to the need for improving access to basic need of health recognizing the inadequacy of health care facilities (UNDP ,
2005). In addition to the quality of staff associated
Fig. 2. Grided population of Uganda at 5 x 5 km resolution.
吴国章Fig. 3. Gridded population of Uganda showing areas of con-centration.
1HC II, health center II rves an administrative unit which is a parish; HC III, health center III rves an administrative unit which is a sub-county; HC IV , health center IV , also termed health sub-district headquarter, rves constituency; hospital, rves as district or can be a regional referral health unit.
S. Lwasa- Geospatial Health 2(1), 2007, pp. 29-40 34
with each grade, all health facilities in Uganda are graded bad on the health rvices offered and the administrative or political unit rviced. The HC II are managed by nursing officers and the rvices include treatment of non-complicated ailments that require clinical rvices, bed rests for some patients and they act as points for near-neighbourhood advi on health care (Shrestha, 2000). The HC III rvices a sub-county and are managed by a clinical officer and their activities include maternal health care, a labour ward, non-complicated surgery rv-ices and general treatment. An HC IV is managed by a qualified doctor, has a labour ward, surgery the-atre, offers general health-care besides supervising all the lower level health centers in a constituency, and provides outreach programmes and guidance. The hospital reprents the highest level available. It offers all health care rvices, can have veral qual-ified doctors as well as lower level health work
ers, manages outreach programmes, and some rve as regional referrals for complicated ailments. This health rvices tier is linked to the administrative units and the higher administrative levels in which management responsibility is also shared between local governments such as districts, sub-counties or parishes. The highest authority in the medical area rests with MoH and the Government. An implicit assumption in health rvices provision is that the less the distance to the health facility, the more access to health rvices.
According to UBOS (2005) the access to health facilities and its rvices has improved from 49% coverage in 1992 to 69.9% in 2005 for the popula-tion living within 5 km of a health rvice unit. However, this is bad on social surveys conducted during the census which may not accurately provide for the distance factor. The challenge is that the dis-tance measurements are interpreted differently by the many rural communities in Uganda and, con-quently, lf-reported distance data may not always be correctly reprented in the surveys. Obviously, the distance influence the level of access to the health rvices since the transportation of sick peo-ple, and the time it takes, have a bearing on the respon and action taken. Rural communities are particularly affected becau there are still marked variations in access to the health facilities both with-in and between districts (Harrison and Verhoef, 2002; Barber et al., 2007; Leonard and Masatu, 2007). Beyond physical acce
ss, many of the health facilities do not provide the full range of esntial primary health care rvices. Country wide, there is a total of 2314 health facilities of all grades and their distribution by district is shown in F igure 2. The distribution of health centers by grade is ana-lyzed in Table 1. The HC II distribution appears sat-isfactory but that of HC III, HC IV and hospitals still requires improvement, especially in relation to the population rved and the rvices offered.
Analysis of access to health facilities bad on dis-tance
Physical access to health facilities is critical in health rvices planning since the distance to a health facility is significantly associated with morbidity. Although the literature indicates improvement in health rvice access (NEMA, 2000/01; UBOS, 2005) and there is continued investment in health-related infrastructures, especially in rural areas, there are still questions regarding access to health rvices in relation to population distribution. The level of access to rvices has been analyzed previously on the basis of a survey conducted country-wide (UBOS, 2005). In my opinion, however, it does not adequately portray the access levels. Given the short-comings of lf-reported data and the many interpre-tations of distance by the various ethnic groups in the country, there is a need to verify the reported sta-
Table 1. Descriptive statistics of health facilities in Uganda, stratified by grade.
