Defining regions for locality health care planning: a multidimensional approach
Introduction
The competitive market driven health care system in the United States has yielded a wealth of research regarding the demarcation of health care regions (Erickson & Finkler, 1985; Garnick, Luft, Robinson, & Tetreault 1987; Phibbs & Robinson, 1993). Historically the organisational structure of the National Health Service (NHS) in the UK warranted little need for such specifically defined regions of responsibility. In more recent times, with trends towards decentralisation being influenced by wider political and social policies, the formal need in British health care policy to partition space for the purpose of health care delivery emphasises the role of space, and indeed the medical geographer in the categorisation and delimitation of areas. Administrative units, though not reflecting localities, may have been used in many cases to ease the retrieval of data for locality profiling with census and routinely available data related to morbidity and mortality, available at such a scale. Previous NHS policy has underestimated the enormity of the task involved in defining localities, with policy documents such as the ‘Cumberlege Nursing Review’ (DHSS, 1986), failing to recognise the extent of overlap between localities when stating that ‘we see no problem in identifying such neighbourhoods’ (DHSS, 1986).
Approaches to area definition of service delivery have included those carried out by the health authorities themselves and others carried out by geographers for the health authorities. The Exeter Health Authority in England led a pioneering attempt towards locality planning in the mid 1980s (King & Court, 1984). Local groups were formed comprising of both health care professionals and lay people, with the aim of better identifying need ‘to adapt services for the needs of communities’ (King & Court, 1984, p. 734). Spurred by the decentralisation approach in the NHS, and the continued discussion of the organisation of primary care at the local level, the Kings Fund, financially supported by the Department of Health, began three experimental projects delivering primary care at a locality level in inner London. The aim behind these was to ‘match services more closely to the needs’ of the population. The best developed of the three projects was the Riverside Health Authority, especially that of the Pimlico Patch Committee (Dunford & Hughes, 1988). The locality planning approach taken in West Sussex was developed based on a geographical perspective for the definition of spatial units for health care delivery. In an advancement on previous administratively led approaches to locality planning (Curtis & Taket, 1989), the process adopted in West Sussex used two approaches; ‘constrained’ and ‘unconstrained’, with the results of the ‘constrained’ approach being subject to the specific recommendations of the health authority (Bullen, Moon, & Jones, 1996).
In 1998 the purchaser/provider system of the NHS was re-emphasised following the decentralisation of service delivery and the introduction of locality planning in the form of Primary Care Groups (PCGs).2 This locality focus requires Local Authorities and groups of commissioning GPs to be responsible for health service provision within a geographically defined market area. As such there is now a greater need for health geographers in the UK to address the issue of regionalisation demonstrated by recent PCG related research (Naish et al., 1998; Moon, Mohan, Gibson, & Pollock. 2001).
In Northern Ireland, the definition of GP catchments was further supported in a government white paper outlining options for PCGs in the province. Outlining the geographical coverage of commissioning groups, it was stated that ‘GP practice populations would be a natural starting point for the development of these groups’ (DHSSPS, 2000, p. 27). In doing so it was considered that the groups would possess a strong local identity created on ‘natural geographical communities’. This paper explores the methodology used to define the GP catchments in the Western Health and Social Services Board of Northern Ireland prior to the introduction of the PCGs. This paper will present the development of a multidimensional approach to catchment definition that incorporates methodologies previously used in health care regionalisation to define robust areas for health care planning. The paper examines previous approaches in both health care regionalisation and functional regionalisation to assess their merits before proposing a methodology to define health service areas and presenting the results specific to one Health and Social Services Board in Northern Ireland.
Section snippets
Previous approaches to defining health care regions
Numerous studies (for example Joseph & Phillips, 1984; Curtis, 1982; Joseph, & Bantock, 1982; Shannon, Spurlock, Gladin, & Skinner, 1975) have illustrated that patient accessibility is directly related to the spatial location of health care facilities. The production of market areas for GP practices may ease the issue of accessibility providing territorial justice and enabling health service planners to allocate patients to their nearest facility with greater ease. In addition the organisation
Methodology
Functional regionalisation is the term given to techniques developed for the analysis of interaction data to define regions that are based on spatial interaction between areas. Regions are defined by minimising the between and maximising the within, region population flows, and tend to have the added advantage of not being restricted by the boundaries imposed through formal administrative division (Openshaw (1977), Openshaw (1995); Martin & Williams, 1992; Martin (1997), Martin (1998); Coombes,
The study area and core data sets
The study area, illustrated in Fig. 1, is the Western Health and Social Services Board (WHSSB), a rural Health Authority in Northern Ireland with a population of just over 300,000 patients served by 78 GPs working from 31 main surgery locations. The area was selected due to the mix of urban centres and rural areas allowing us to apply the methodology in both. Patient address and GP affiliation information for the population was obtained from the Central Services Agency (CSA). The key
Creating basic GP catchments
A selection of previous ‘one-stop’ methodologies were selected from the literature as being the eight most common methodologies used in studies of this nature and were employed to define eight separate GP catchment regionalisations. Although the selection is subjective, a review of previous research suggests that these methodologies represent the majority of single analytical approaches to date. The eight methodologies were: 75% and 85% Percentage Catchments, Market Share Catchments, Nearest
Creating a synthetic data matrix
In the next stage of the analysis the ED level patient to GP affiliation information from each of the eight methods was used as input to create the Synthetic Data Matrix (SDM). The matrix essentially represents a synthesis of the eight methodologies that allows us to measure the level of spatial connectivity between EDs. For every catchment model, each ED was allocated to one and only one GP location. Binary matrices were created for each model identifying the association between every pair of
Defining optimal catchments using the European Regionalisation Algorithm (ERA)
The SDM was analysed to create optimal catchments based on a synthesis of all eight methodologies using the ERA (after Coombes, 2000). As outlined earlier the two standard criteria used in the model were patient population size and level of self-containment (essentially a measure of separateness defined by a lack of linkage or flow across the boundaries of the proposed GP catchments). Self-containment is measured as a percentage, with a score of 100% indicating that no EDs within the boundary
Results
Increasing and decreasing the thresholds until selections of parameters were chosen for the final analysis allowed the ERA procedure to be optimised against the needs of this application. It was anticipated that 31 catchments would be created, or as close to this number as possible, reflecting the number of GP practice locations in the generation of the eight input regionalisations. Selections of the test parameters used in the process are outlined in Table 3 with the total number of output
Discussion
Whilst this research has been largely experimental, spanning the boundaries of two geography sub-disciplines, namely regionalisation and the geography of health care, it has strove to highlight the importance of research related to the territorial division of space in the geography of health care in addition to specific concentration on problems related to accessibility and utilisation of services.
Throughout this paper it has been highlighted that whilst the concept of defining GP catchments is
Conclusion
This paper has demonstrated that the issue of regionalisation within health care planning is more complex than first thought. Although the methodology presented provides a solution for overcoming the problem surrounding single-shot techniques, there are a number of limitations that must be recognised before concluding, which could be seen as starting points for further research.
An inherent problem within the initial analysis lay in the retrieval of the primary patient data used in the creation
Reference (49)
- et al.
Defining localities for health care planninga GIS approach
Social Science & Medicine
(1996) Spatial analysis of surgery locations in general practice
Social Science & Medicine
(1982)- et al.
Identifying medical regions using hierarchical clustering
Social Science & Medicine
(1980) - et al.
Measuring potential physical accessibility to general practitioners in rural areasa method and case study
Social Science & Medicine
(1982) - et al.
Measuring access to primary medical caresome examples of the use of geographical information systems
Health and Place
(1998) - Alvanides, S. (2000). Zone design methods for application in human geography. Unpublished Ph.D. thesis, School of...
- et al.
Zone design for planning and policy analysis
- et al.
Interactive spatial data analysis
(1995) - Bullen, N., Moon, G., & Jones, K. (1993). The definition of geographical communities for health planning: final report....
- et al.
Model building and the analysis of spatial patterns in human geography
Journal of the Royal Statistical Society
(1975)
Functional regions for the population census of Great Britain
An efficient algorithm to generate official statistical reporting areasthe case of the 1984 travel-to-work areas revision in Britain
Journal of Operational Research
Substantive issues in the definition of ‘localities’evidence from the sub-group local labour market areas in the west midlands
Regional Studies
GIS in community boundary definition
Mapping Awareness and GIS in Europe
Defining locality boundaries with synthetic data
Environment and Planning A
Contrasting Aproaches to Identifying ‘localities’ for research and public administration
Locality planning for health carea case study in east London
Area
Building the way forward in primary care
Pimlico patch committeean experiment in locality planning
Determinants of market share for a hospital's services
Medical Care
Geographic market areas for psychiatric and medical outpatient treatment
Journal of Behavioural Health Services and Research
The Accuracy of age—sex registers, practice medical records and family practitioner committee registers
Journal of the Royal College of General Practitioners
Appropriate measures of hospital market areas
Health Services Research
Cited by (48)
Stochastic joint homecare service and capacity planning with nested decomposition approaches
2021, European Journal of Operational ResearchCitation Excerpt :The studies on the home health care (HHC) and related health care problems are many, here we mainly review the streams of the HHC planning studies that considered demand uncertainty and that with service planning which are more related to our study. The studies on HHC planning and operations in the setting of given deterministic parameters can be found in staff assignment and routing (Begur, Miller, & Weaver, 1997; Braekers, Hartl, Parragh, & Tricoire, 2016; Ernst, Jiang, Krishnamoorthy, & Sier, 2004), location of HHC central facilities (Shortt, Moore, Coombes, & Wymer, 2005), urban districting (Blais, Lapierre, & Laporte, 2003), fleet assignment to urban districts, shift scheduling and staff allocation (Grenouilleau, Legrain, Lahrichi, & Rousseau, 2019; Hertz & Lahrichi, 2009) and inventory management (Chahed, Marcon, Sahin, Feillet, & Dallery, 2009). More studies on the HHC planning can be in the excellent survey papers of Gutiírrez and Vidal (2013) and (Kahraman & Topcu, 2018).
Inferring transit-based health seeking patterns from smart card data – A case study in Beijing, China
2020, Health and PlaceCitation Excerpt :Understanding people's health seeking behavior, such as their healthcare demands, hospital choices, and accessibility, are key to effective interventions to improve public health, narrow health disparities, and build sustainable cities (Wang and Luo, 2005). As a basis for characterizing people's trips for healthcare, the data collection or extraction are often faced with issues of covering a large and diverse population, as well as an extensive geographic area (Shortt et al., 2005). Although a variety of data sources and data extraction methods have been proposed, challenges remain to improve both the statistical and geographic representativeness.
Developing a flow-based spatial algorithm to delineate hospital service areas
2016, Applied GeographyCitation Excerpt :With more accurate travel distance (or time), the buffer zone within a certain travel distance (or a certain period of travel time) can be used to represent the service area of a provider, such as 15 miles (or 30 min) (Luo & Wang, 2003). One common disadvantage of both approaches above is that the specified number or percentage is subjective, varying by the analysts, such as 60% (Garnick, Luft, Robinson, & Tetreault, 1987), 75%, and 85% (Shortt, Moore, Coombes, & Wymer, 2005). More importantly, as customers living in one region may have different choices, multiple service areas may overlap with one another and not be mutually exclusive.
Optimal allocation of public service centres in the central places of functional regions
2015, IFAC-PapersOnLineDefining GP practice areas based on true service utilisation
2012, Health and PlaceCitation Excerpt :Other approaches have also been used to depict the area in which the majority of registered population reside. Such previous approaches to characterise GP practice catchment areas have varied in their complexity (Norris, 1952; Bailey, 1956; Wilson and Gibberd, 1990; Goody, 1993; Bullen et al., 1996; Lowe and Sen, 1996; Shortt et al., 2005; Jones et al., 2011) but most have relied on existing administrative boundaries. As such, they do not allow the degree of overlap between catchment areas to be considered and are only rough approximations of the true service utilisation areas.
- 1
Formerly of the School of Architechture and Planning, University of Newcastle upon Tyne, Britain.