Travel patterns for government emergency dental care in Australia: a new approach using GIS tools


Professor Estie Kruger
Centre for Rural and Remote Oral Health
The University of Western Australia
35 Stirling Highway
Crawley WA 6009


Background:  Government subsidized dental care is provided as a community safety-net to complement the private dental sector. The aim of this study was to detail the geographic catchment characteristics of three outer metropolitan government dental clinics.

Methods:  Three outer metropolitan dental clinics with the greatest number of geocoded triage patients were selected for the study. In total these three facilities had 5742 patients over the 12-week period with 2010 at clinic A, 1278 at clinic B and 2454 at clinic C. Cumulative proportions of patients’ residential address locations at distances were calculated; there was close correlation between the three clinics. A best fit curve with a correlation coefficient of 0.998 was developed.

Results:  In summary, approximately 50% of patients were within 6 km of the clinic and 75% were within 10 km.

Conclusions:  This study has critical outcomes for the planning of future services in developing a network model for care. The data presented will assist in the development of more evidence-based approaches to planning new service network structures.

Abbreviations and acronyms:

Geographic Information Systems


Socio-economic Index for Areas


Government subsidized dental care in Australia is provided as a community safety-net to complement the private sector (which provide more that 80% of total care).1 As with most health safety-net approaches, public patient access is restricted to the needy through income derived access limitations. In most jurisdictions in Australia this is managed using a health care card that provides access to a suite of government subsidized services, including dentistry. The provision of government subsidized dental care in Australia is becoming far more complex. Significant workforce shortages are having a major impact on sustaining services. These workforce shortages are predicted to continue for at least another decade.2 As part of strategic planning in this difficult service environment, a strong understanding of the nature, demographic and access drivers for care are needed. Patient mobility and in particular lower socio-economic patient access are critical factors in determining the shape, size and distribution of service networks. The aim of the study was to detail the geographic catchment characteristics of three typical outer metropolitan government dental clinics and to develop modern Geographic Information Systems (GIS) based techniques for understanding clinic catchment zones.

Methods and Statistical Analysis

Triage data for a 12-week period in 2005 were collected from all government dental clinics in the State of Victoria, Australia. (This period was chosen to closely reflect the time the population census data were recorded.) Patients were triaged using a computer based universally accessible system (that is in all government dental clinics in the State) either over the phone (more that 80%) or in person. A total of 34 646 episodes of triage occurred over the period. Of these, there were 4015 that were second (or more) events for the same person which were removed from this study’s sample set. Each triage event was recorded in detail and the data for the events have been reported previously.3 After removing data for the single major centralized dental hospital (that is also the primary specialist and teaching service centre) and data anomalies that occurred during recording, a total of 22 601 records remained in the jurisdiction wide frameset (65%). The teaching hospital data were removed as this posed some complications regarding specialist and student care (in terms of analysis). All these records were geocoded to attain the longitude and latitude of the residential address using Victorian health based geocoding services. In addition, all clinics were similarly geocoded using Google earth™. GIS consists of a computer based system for the input, storage, maintenance, management, retrieval, analysis and output of geographic or location based information.4 By illustrating juxtaposed multiple layers of information, GIS is emerging as an important novel tool in health care planning.5,6

Population data were extracted from the most recently available online census data cube available from the Australian Bureau of Statistics at the highest acuity level available in Australia (census district).7 A Socio-economic Index for Areas (SEIFA) category was used to indicate the socio-economic disadvantage of each census district. SEIFA is produced from the Australian Bureau of Statistics 2001 national census, and is a set of four indexes: The Index of Relative Socio-economic Disadvantage; The Index of Relative Socio-economic Advantage and Disadvantage; The Index of Education and Occupation; and The Index of Economic Resources. There are five SEIFA categories: 1 being most disadvantaged, 2 being above average disadvantaged, 3 being average disadvantage, 4 being below average disadvantaged and 5 being least disadvantaged.8

All data were collated in Excel (version 2003 from Microsoft Corporation) and geographic integration of the data were completed using ArcGIS (version 9.3 from ESRI Corporation).

Three outer metropolitan dental clinics with the greatest number of geocoded triage patients were selected for the study. These clinics were selected not only on the basis of a rich data sample, but also because they were in similar socio-economically deprived communities and therefore typified the regions where government safety net dentistry is focused. The spacing of areas around each clinic also had minimal overlap, and lastly, an even distribution of surrounding roadways. The clinics will be referred to as clinic A, B and C to ensure anonymity. In total, these three facilities had 5742 adult patients over the 12-week period with 2010 at clinic A, 1278 at clinic B and 2454 at clinic C.

Catchment zones for each clinic were developed by taking the 95% of triage events closest to the clinic and those outside the ‘closeness’ criteria where more than a single individual was within a census district. These asymmetric zones were then used as the basis of the analysis of populations.

Development of an innovative summary method

As part of the research, new methods of simply defining catchment characteristics were examined. A simplified single diagrammatic approach was found to convey the complex underlying data clearly. In summary, the north, south, east and west maximal extensions of each catchment area (relative to the clinic) were measured and recorded. In addition, the angulation degrees from north (i.e. compass bearing) were measured for the maximal density of events (in this study, triage events) and added to the diagram as a dotted arrow. Each diagram had a 5 km (wide and high) square overlayed for scale.


The application of concentric circles around each clinic provided an overall distance from the clinic that people associated as their ‘local’ clinic (Fig 1). Cumulative proportions of patients as the circles got wider found that there was close correlation between the three clinics. A best fit curve with a R-squared value of 0.998 was developed. In summary, approximately 50% of patients were within 6 km of the clinic and 75% were within 10 km (Fig 2). The best fit curve will allow managers to make an estimate of the number of patients that will be expected to require emergency care based on the socio-economics of the surrounding district. The formula can be used by inserting the distance (as x) into the function and this will then calculate the proportion of patients that will require emergency care in a given period (the y variable).

Figure 1.

 The triage events for one clinic overlayed with 1, 2, 4 and 8 km radius circles.

Figure 2.

 The cumulative percent of patients’ distance from dental clinics A, B and C. The R squared (R2) of 0.998 shows that the calculated formula (where x is the distance from the clinic and y will be the percentage of patients) explains a large percentage of variance to the empirical data from this study. Distance was measured from the patient’s residential address. A best fit curve using the complete dataset (with a R squared (R2) above 0.99) was constructed and documented.

Closer examination of the distribution of patients around each clinic found a far more asymmetric arrangement (Fig 3). This appears to be driven by differences in the distribution of socio-economics of populations as well as probably by transport and access factors. Catchment zones that represented over 85% of all patients but were representative of the asymmetric distribution were developed based on census districts. The overlaying of SEIFA deciles found a visual association between low socio-economic status and clinic usage (Fig 4). Although this is not unexpected against a backdrop of the clinics being a community safety net with access limited by income.

Figure 3.

 Every triage event (green, purple and blue dots) associated with the three clinics A, B and C (red dots).

Figure 4.

 The location of triage events (green dots) for clinic A overlayed on the associated census districts. Light tones are census districts that have SEIFA scores of 4 to 10 while dark tones have SEIFA scores of 1 to 3 and are therefore the lowest (30%) socio-economic areas of the region.

Each catchment zone consisted of between 200 and 500 census districts. Each census district is of a different population size but when summated the total population catchment zone was not significantly different (Table 1). It ranged between 190 000 and 270 000 adults.

Table 1.   Total number of adult (older than 15 years) males and females that reside within each catchment zone of the clinics (A, B and C)
Males125 51690 810134 580
Females130 28996 628140 039
Total255 805187 438274 619

Simplified catchment depictions using the new methodology were developed for each clinic (Fig 5). It can be quite clearly seen that clinic C was placed at the extreme east of elongated catchment zone while clinics A and B were more centrally located. In addition, it can be seen that in all three cases the maximal density of users was towards the north-west of each clinic, C being much closer to east than A or B.

Figure 5.

 An innovative method of simplified depiction of the catchment zones for each clinic A, B and C. The four lines radiating from each central clinic depict the maximum extent of the catchment area (in kilometres) while the dotted arrow represents the compass bearing of the maximal density of triage events. The square around each clinic is 5 km wide and high and acts as a scale bar.


The results of this study have found that about 75% of emergency dental patients from the sampled outer-metropolitan clinics reside within 10 km of the clinic. A formula with a strong correlation to the data was developed that can be applied in the future for estimating demand. For example, the number of patients drawn to a clinic from a similar type of region could be estimated using the formula. However, one would need to be careful to not apply it universally to all clinics as relevant aspects that might differ need to be considered. Importantly, the study highlighted the asymmetric nature of catchment zones and provided a metrology for analysis of this asymmetry. The basic concept of understanding ‘catchment zones’ has been applied elsewhere in dental public health research (e.g. fluoride water access9) but not to the service provision side of the equation where significant capital resource allocations are made.

Although the data used in the study were based on emergency demand, the aim of this research was not to look at these factors but to use the dataset to define geographic boundaries. This approach is not unreasonable as emergency dental care is approximately 50% of all care provided while general dental care usually follows (on the same patient) from the emergency event and thus the geographic detail will remain valid.

Recent studies of health access using GIS have used travel time as a measure of access. However, in this Australian first study distance by line-of-site has been used as a near proxy. In Australia, access to cars by metropolitan residents is some of the highest access in the world and a highly developed road network with little congestion allows for relatively easy movement. In this study line-of-site distances were used as an indicator of a patient’s reasonable measure of access to care. Since the distances measured were in the order of 10 km maximum, there is little risk of this approach not being a reasonable approach to the issue.

The application of census data and GIS tools to target care at areas of greatest need is not a new concept in health care.10 Dental disease is a very strongly socio-economically linked disease cluster. The socio-economically disadvantaged suffer far greater burdens of disease than the more affluent members of society (this is true across many societies). By applying disadvantage based mapping and service targeting access to safety net care can be significantly enhanced and efficiency gains made.

To our knowledge, no published study in Australia has examined the public’s distance from clinic approach to accessing dental care. This study has critical outcomes for the planning of future services in developing a network model for care. Further analysis of the effects of variables such as relative household wealth in the public’s approach to distance to access dental care needs to be undertaken. However, with the data presented in this research health planners can develop a more evidence based approach to planning new service network structures.


The authors would like to thank Dr Christophe Duigou for his early input into this research project.