To date, there appears to have been a desire by policy-makers and others for a single all-purpose classification to guide the distribution of health care resources to rural communities, without significant debate about whether the defining variable is the degree of 'rurality' or 'remoteness' or some other aspect of accessibility, disadvantage or contextual factor that underpins the problems associated with health care provision in these eligible areas. Numerous authors have debated "what is rural" and sought definitions based on characteristics such as low population density and small population centres, isolated populations and large distances, as well as observed environmental, agricultural and other economic activities [29
]. In reality, the related concepts of rurality and remoteness are multi-faceted, thus precluding agreement on one universally accepted classification [8
]. Nonetheless, governments continue to seek some agreed objective measure or classification on which to base their resource allocation decisions. This search is not limited to Australia, with a number of alternate classifications in existence [35
], but generally these too only capture similar elements of rurality and so they offer no significant design alternative to the Australian classifications.
Table provides a summary of important characteristics and related required decisions associated with any geographical classification. The first and most important distinction is to be clear about its purpose; that is, what is the classification designed to measure. For example, the ASGC-RA classification was unambiguously designed to measure geographical remoteness of populations. On the other hand, the RRMA classification captures some elements of 'rurality' including population size and density. Within each classification method, subjective decisions are required that determine its outcomes and the consequent degree of acceptance by users. These decision points include the choice of algorithm, the number of groupings and how they are determined, as well as the size of spatial units. The RRMA classification, despite its inherent weaknesses, is still preferred by many groups over the ASGC-RA classification [43
], chiefly because of its ability to discriminate between areas at a finer geographical scale, thereby giving somewhat more homogenous groupings. Rather than continue to search for a single solution that suits all applications, it is more appropriate to develop classifications closely aligned with a specific defined purpose. A number of examples illustrate what can be achieved:
Summary of decisions required regarding important characteristics of geographical classifications
i. The Griffith Service Access Frame (GSAF) is one classification that measures access disadvantage in relation to education services, and is intentionally designed as a tool for resource allocation [48
]. The GSAF is characterised by measuring access only to the nearest service option, and has been adopted by many Australian states in the distribution of rural (education) resources. Such an option may be an appropriate method for measuring access to hospital and specialist care services in the field of health.
ii. More recently, McGrail's new index of rural access, tested in Victoria, has been developed as a more appropriate measure of access to primary care services in rural areas [50
]. This index is specifically designed to include the key elements of access to GPs (availability, proximity, mobility and health needs), utilise more appropriate advanced accessibility methods (modified two-step floating catchment areas [52
]) and use the smallest feasible geographical units (collection districts).
iii. The GPARIA classification, a modified version of the ARIA classification, was specifically developed for the purpose of distributing Rural Retention Program grants to GPs working in rural and remote communities. GPARIA measures both remoteness and isolation by incorporating proximity to nearby GPs of both the population and GPs in its construction. How well this classification adequately differentiates all aspects of factors affecting retention decisions across rural and remote communities is a moot point.
iv. The District of Workforce Shortage (DWS) status is a simple yes or no stratification for all Statistical Local Areas (SLAs), which has been regularly updated every quarter for over 10 years using Medicare data [54
]. An area's DWS status reflects whether the ratio between population size and the number of services provided within an SLA is below the national average. It should be noted, however, that its value is questionable because population-provider ratios are a poor measure of access, particularly for 'small' rural areas [52
] and its dichotomous definition does not allow small areal variations to be detected. However, new methods such as McGrail's index of access can improve its application.
In addition to forming the basis for resource allocation decisions, geographical classifications are often used as statistical tools to guide rural health research, and, in particular, the presentation of results such as health outcome measures in evaluating the effectiveness and quality of health care [56
], or through measuring service utilisation rates as an indicator of need for services [22
]. A few other examples include the association between cancer survival rates and ARIA [58
], the association between primary care management of chronic heart failure and RRMA [59
], the association between Attention Deficit Hyperactivity Disorder (ADHD) treatment and ASGC-RA [60
], or the association between mental health status and RRMA [61
] or ARIA [62
]. Measures of the extent to which services and interventions are resulting in improvements in the health status of rural Australians are contingent upon how rural is defined.
The use of inappropriate classifications can serve to mask or average-out important health inequities that characterise rural communities. Many authors have failed to fully appreciate the significance of rural delimitation. Simply bundling together places of diversity (heterogeneous) into convenient (presumed homogenous) categories often obscures the inherent variations within rural areas [63
] and seriously affects the resultant pattern of health status and differentiation [64
]. Many possible covariates, such as differing demographics, socio-economic status, access to health services and health behaviour, are frequently not included within statistical reports that are broken down by geography, despite their possible influence on the extent to which apparent associations with rurality are significant. In short, while significant associations between geographical classifications and various health and health service outcomes are interesting, they often conceal the true effect within rural populations.