Introduction
Approximately 56000 new strokes occur each year in Australia and this is expected to double by 2050 (Stroke Foundation, 2017). Over one-third of stroke survivors have a stroke-related disability that interferes with their everyday life (Australian Institute of Health and Welfare (AIHW), 2013). These disabilities can span across all areas of functioning including physical activity, communication, learning and psychological wellbeing (AIHW, 2013), and have negative effects on the ability to return to work and on daily life post-stroke (Hofgren, Björkdahl, Esbjörnsson & Stibrant-Sunnerhagen, Reference Hofgren, Björkdahl, Esbjörnsson and Stibrant-Sunnerhagen2007). Despite these disabilities, only 50% of patients with stroke discharged from hospital in Australia have a discharge plan (AIHW, 2013), leaving the remainder to self-manage their recovery and access primary healthcare services to meet their needs. For these individuals, their outcomes depend on the extent to which primary healthcare services are properly distributed to meet needs.
In Australia, a recent and substantial focus on access to stroke care in acute settings has diverted attention away from access to primary care for stroke survivor services. Looking overseas, research has suggested that access to primary care and community-based stroke services may be limited. A focus group study in the United Kingdom found that there was a dearth of information provided to patients and their carers about access to local services (Hare, Rogers, Lester, McManus & Mant, Reference Hare, Rogers, Lester, McManus and Mant2005). Furthermore, participants were largely unsure about which organisations to access for assistance (Hare et al., Reference Hare, Rogers, Lester, McManus and Mant2005) and relied heavily on primary healthcare services, including general practitioners (GPs), as the point of contact for services (Hare et al., Reference Hare, Rogers, Lester, McManus and Mant2005). This limited research suggests that there is currently insufficient attention on access to primary healthcare services, despite them playing a critical role in health outcomes. There is also only limited evidence about factors that impede access to services for stroke survivors. In the United States of America, a survey study (N = 83) found that the most common barriers reported by stroke survivors to participating in physical activity, which is a key risk factor for stroke, included a lack of awareness of fitness centres in the area (57% of respondents), a lack of transport to a fitness centre (57%) and a lack of knowledge about where to exercise (44%; Rimmer, Wang & Smith, Reference Rimmer, Wang and Smith2008). Distribution of services and geographic accessibility for the stroke population is rarely considered despite evidence suggesting that the proper distribution of services has a large bearing on population health outcomes (Guagliardo, Reference Guagliardo2004).
As the first step in understanding accessibility to healthcare services for Australian stroke survivors, this study applied Geographic Information System (GIS) techniques to map and then examine the spatial distribution of primary healthcare services in one Primary Health Network (PHN) in South-East Queensland to determine whether these services are located in geographically accessible areas relative to the discharge location of patients with stroke. Internationally, GIS technology was recently used in Canada to determine the drive time for accessibility to stroke centres and emergency medical services for hyperacute stroke management (Eswaradass, Swartz, Rosen, Hill & Lindsay, Reference Eswaradass, Swartz, Rosen, Hill and Lindsay2017), with the researchers concluding that most Canadians live within 6-h road access of stroke centres and emergency medical services. A second study exploring stroke prevention services (services that provide assessment and intervention to minimise the risk of subsequent stroke) in Canada found that just over 87% of Canadians can access a stroke prevention service within a 1-hour drive time (Jewett et al., Reference Jewett, Harroud, Hill, Côté, Wein, Smith and Lindsay2018). As this research focused on the general population (rather than patients with stroke specifically), there remains a need to determine access to primary healthcare services for the stroke population, specifically in Australia. Therefore, the study sought to answer the following question: How accessible are primary healthcare services for patients with stroke recently discharged from hospital? Specifically, we aimed to (1) map the locations of a recently discharged stroke cohort from a single hospital; (2) map the distribution of GP, pharmacy and allied health providers (AHPs) in the PHN associated with the hospital and (3) examine accessibility based on travel time over the local road network.
Methods
Study group
Ethical approval was obtained from the relevant university (The University of Queensland Human Research Ethics Committee) and health service district (Metro South Human Research Ethics Committee) before beginning the project. Both the inpatient discharge data and the service location data within the Brisbane South PHN were collected. A 10-km buffer zone was applied around the border of the Brisbane South PHN with both inpatient and service location data to account for ‘edge effects’ when measuring spatial accessibility (Luo & Wang, Reference Luo and Wang2003, p. 867). Medical inpatient discharge data were supplied by a large metropolitan hospital in Brisbane, Queensland (which follows the same boundaries as the Brisbane South PHN). The diagnosis of stroke was based on the ICD-10-AM stroke codes (International Statistical Classification of Diseases and Related Health Problems, Australian Modification, ICD-10-AM codes I60 – I64 inclusive).
Data collection and analysis
Data included the discharge location (street address, suburb name and postcode) and characteristics (e.g., admission date, discharge date, discharge code, health cover type, principal diagnosis, discharge destination, sex and age) of each individual patient. Data on the locations of AHP, pharmacy and GP service clinics were collected from the Brisbane South PHN. Data in the 10-km buffer zone area were purchased from the Medical Directory of Australia through its online data portal, with each record containing the name and street address of the service. Both the patient data and the healthcare service data were geo coded using Google and Bing Application Program Interface and mapped in GIS. Patients who died in hospital and those whose discharge address was outside the 10-km buffer zone of the Brisbane South PHN were excluded from subsequent analysis.
Road network data were used to measure travel time from each patient’s discharge location to all service locations (GP, pharmacy and AHP) through the network analysis tool in GIS. Travel time is estimated using road distance, as this accommodates the variation in road types with varying speed limits across the study region. For instance, highway/motor ways would have a higher speed limit of 100 km h−1 compared to a local road with a speed limit of 40–50 km h−1. This is in line with the literature when measuring spatial access to healthcare services in a GIS environment (Luo & Wang, Reference Luo and Wang2003; Martin, Wrigley, Barnett & Roderick, Reference Martin, Wrigley, Barnett and Roderick2002; McGrail & Humphreys, Reference McGrail and Humphreys2009). Travel time was computed based on the designated road distance and speeds; however, it does not account for real-time traffic conditions nor time for parking and walking to the service site, and is thus likely to underestimate the actual travel time. In the absence of complete data on the capacity of health service provision at each service location, which may also vary over time, we measured accessibility as the total number of services (GP, pharmacy and AHP) located within a threshold travel time from each patient’s home location. Various travel time or distance thresholds have been used in the literature. For instance, Liu et al. (Reference Liu, Wong and Jin2009) used three distance thresholds at 3, 5 and 10 km (which corresponds to 3-, 5- and 10-min travel time at an average road speed of 60 km h−1) to measure spatial access to primary healthcare services for baby boomers in Singapore. Luo and Wang (Reference Luo and Wang2003) used 30 min as the initial threshold to measure spatial accessibility using a two-step floating catchment area (2SFCA) method; they tested the sensitivity of the accessibility measure using different travel time ranging from 20 to 50 min. McGrail and Humphreys (Reference McGrail and Humphreys2009) improved the 2SFCA method with an initial 10-min travel time to define an area with no discernible barrier to healthcare services and 60 min as the maximum limit to travel to use a service, with a distance decay effect in areas between 10 and 60 min to measure access to primary healthcare services in rural areas in Victoria, Australia. Considering that our research subjects are patients with specific heathcare needs, and the measure of travel time did not take into account real-time traffic conditions, walking/parking time, or the fact that some patients may rely on public transport, we used relatively shorter travel time thresholds for three levels of proximity:(1) within a 5-min drive from home, (2) within a 10-min drive from home and (3) within a 15-min drive time from home.
Measures of accessibility at the individual scale were aggregated to statistical area level 2 (SA2) for the purpose of mapping the spatial patterns of accessibility within the region. SA2 s are a general-purpose, medium-sized area built from whole Statistical Area Level 1 designed by the Australian Bureau of Statistics (ABS) with the aim of representing a community that interacts together socially and economically. The average population of a SA2 is about 10000, but can range from 3000 to 25000 persons (ABS, 2011). The average number of services available within each defined level of travel threshold is used to measure accessibility to GP, pharmacy and allied health services at the level of the SA2.
We categorised the accessibility results using standard deviation (SD) to indicate how much the accessibility measure in each SA2 differs from the mean value across the entire Brisbane South PHN area. Other classificatory methods such as Jenks Natural Breaks could also be used to classify the accessibility measure, but the resulting output would arguably be less able to illustrate the deviation of the measure from overall accessibility across the region.
Results
A total of 2817 patients with a total of 2979 admission records were available in the inpatient discharge database from the hospital for the approximately 6-year period spanning January 2011–January 2017. Amongst these, 440 patients died in hospital at first admission; 727 patients were discharged to a location outside the South Brisbane PHN but within Australia; and 55 patients were from overseas. For the study population of 1595 patients residing within the South Brisbane PHN, 1492 patients were admitted to the hospital only once during the six-year study period, and 103 patients had multiple admissions (i.e., 90 had two admissions, 10 had three, 2 had four and 1 had six admissions).
On the service provision side, there were a total of 321 GP clinics, 255 pharmacies and 802 AHPs within the Brisbane South PHN. Figure 1 illustrates the spatial distribution of the GP clinics, pharmacy and AHPs, with the number of patients with stroke per 10000 population (from 2016 census) in each SA2 illustrated in six categories ranging from less than −1.5 SD to over 2.5 SD across the whole PHN (Figure 1).
The study area, spatial distribution of health services [general practitioner (GP), pharmacy and allied health providers] and number of patients with stroke per 10000 population by statistical area level 2. Areas shown in darker blueish colour have higher patient-to-population ratio than those in yellow to brownish colours.
Figure 1 shows an uneven spatial distribution of patients who were discharged with stroke relative to the total population across SA2 regions. In particular, it is notable that certain areas shown in darker blueish colour have a much higher patient-to-population ratio which is over 1.5 SD from the mean in the region. In these areas with a larger proportion of patients with stroke, individuals may be more likely to experience a discrepancy between availability of healthcare services and realised access to services such as GP clinics, pharmacies and AHPs.
Figure 2 illustrates the number of healthcare services that are accessible to patients within a 5-min drive from the residence of the patients with stroke. It shows that SA2 s with a larger number of patients with stroke (represented by larger black dot symbols in Figure 2) in the north-western and north-eastern regions of the Brisbane South PHN have lower numbers of healthcare services within the specified drive time (i.e., less than −0.5 SD).
Number of health services per 1000 population that is accessible by patients with stroke within a 5-min drive from home. Areas shown in darker blueish colour are those with higher number of general practitioners (GPs)/pharmacy/allied health providers (AHPs)per 1000 population than those in yellow to brownish colours. SA2, statistical area level 2. Part a) represents data pertaining to GP clinics, Part b) represents data pertaining to pharmacies, and Part c) represents data pertaining to AHP services.
When the travel time increases to 10 (Figure 3) or 15 min (Figure 4), the number of services available to patients with stroke increases. However, the spatial variation of access to the services across all SA2 areas remain the same. For instance, those areas on the northwest and northeast of the region with a larger number of patients with stroke remain at lower level of access under all travel thresholds, as do the larger SA2 areas in the south of the PHN region.
Number of health services per 1000 population that is accessible by patients with stroke within a 10-min drive from home. Areas shown in darker blueish colour are those with higher number of general practitioners (GPs)/pharmacy/allied health providers (AHPs)per 1000 population than those in yellow to brownish colours. SA2, statistical area level 2. Part a) represents data pertaining to GP clinics, Part b) represents data pertaining to pharmacies, and Part c) represents data pertaining to AHP services.
Number of health services per 1000 population that is accessible by stroke patients within a 15-min drive from home. Areas shown in darker blueish colour are those with higher number of general practitioners (GPs)/pharmacy/allied health providers (AHPs) per 1000 population than those in yellow to brownish colours. SA2, statistical area level 2. Part a) represents data pertaining to GP clinics, Part b) represents data pertaining to pharmacies, and Part c) represents data pertaining to AHP services.
Discussion
Stroke is a long-term chronic condition that, in many cases, requires ongoing management following discharge from acute hospital (AIHW, 2013). Present data suggest that access to primary healthcare services is variable across the region. Furthermore, access is poor for a substantial minority of patients with stroke. This is a concern given that mobility is frequently impaired following stroke (Hill, Ellis, Bernhardt, Maggs & Hull, Reference Hill, Ellis, Bernhardt, Maggs and Hull1997) and driving is often prohibited, further compounding stroke patients’ ability to access primary healthcare services if they are not local and well connected by public transport. In the light of the expected increase in stroke over the coming decades (Stroke Foundation, 2017), this study suggests that there is an urgent need for PHNs to strategically locate future services in a way that maximises access for patients with stroke.
This study identified that certain areas with a larger proportion of patients with stroke did not necessarily have close proximity to GP clinics, pharmacy and AHPs, particularly in the north-western and north-eastern regions. When the allowable travel time was increased, the number of services accessible to patients also increased. However, the spatial variation of access to the services across the PHN as a whole remained largely unchanged. The mismatch in areas with relatively high numbers of patients with stroke and low numbers of services suggests low potential access to primary healthcare services for patients living in these areas, which might also lead to low realised access. This finding requires further exploration in future research, as access to stroke services at all stages of the stroke management continuum is an important focus of the World Stroke Organization’s Global Stroke Services guidelines and action plan (Lindsay, Furie, Davis, Donnan & Norrving, Reference Lindsay, Furie, Davis, Donnan and Norrving2014). It was also notable in this study that the larger southernmost SA2 s were also under-serviced although they had fewer stroke patients overall. This issue also requires further exploration, beyond the scope of the study, to determine whether patients living in these areas do have low realised access to stroke services, and if this is the case, to develop mechanisms to improve access to stroke services in these geographical regions.
The data revealed differences in the accessibility of the three types of services. Specifically, there was a bigger variance between the areas with respect to GP clinic access than the other two services. For instance, at the 5-min level, the difference in access to GP clinics ranged from −0.5 to 2.5 SD compared to a range of −0.5 to 1.5 SD for AHPs and pharmacies. Given that GPs are typically the point of contact for services (Hare et al., Reference Hare, Rogers, Lester, McManus and Mant2005), this variation in GP access is of particular concern. Furthermore, in Australia, the current primary health context places GPs as the managers/coordinators of care and, in some cases, requires the GP to optimise coordinated access to team-based care (e.g., under GP management plans which set up access to AHPs in a coordinated way). Therefore, the current project can inform the planning of new GP clinics which can be located in a way that maximises accessibility for patients with stroke living in areas of low access to GP clinics. It must be noted, however, that the pattern of results with respect to discrepancies in travel time may not be clinically significant. Therefore, when planning new GP clinics, geographical accessibility would be one factor amida range of other factors, including health-related ones.
Intriguingly, among all patients with stroke who were discharged from the hospital, one third were from outside the Brisbane South PHN area, including 2.3% from overseas. It is possible that this may have been at least partially due to the target hospital being a quaternary hospital with the ability to provide specialised acute stroke interventions not available at all hospitals, leading paramedic services and families to go specifically to the hospital instead of other more local hospitals that do not provide the same specialised stroke services. It is also possible that the finding may have partially reflected the highly mobile nature of the population with patients being away from their homes (such as at work in Central Business District (CBD)) at the time of their stroke.
Limitations and Future Research
The current project mapped the location of discharged patients with stroke in relation to primary healthcare services. However, it is not known to which degree patients were aware of and actually accessed these services. Further research could explore actual use of, or realised access to, primary healthcare services. Concomitantly, while proximity to service locations is one dimension of access, service capacity is an equally important factor to consider, as this would allow more accurate mapping of the demand and supply of services and how these aspects interact and impact accessibility. Future research could explore the availability of health providers within each of the clinics. Another limitation in this current study was that the data, while from an approximately 6-year period and encompassing a large number of cases, were only from a single healthcare service. It is unknown whether similar observations would occur in other healthcare services and regions. However, the GIS mapping approach used in this study could be used by other healthcare services and with other patient populations to explore this issue further. It must also be noted that the healthcare service in this study encompassed both metropolitan and non-metropolitan areas with substantial variation in socio-economic status. It should also be noted that some of the AHPs were mobile and were unable to be geocoded, as they could not be located to a single site. In addition, the accuracy of the geospatial analysis is dependent on the accuracy of the pre-collected data by the services involved. Furthermore, as the study did not explore actual use of services, it is possible that individuals living near the border of other geographical regions may have travelled to these other regions and may have, therefore, had different access to services than captured in the present analysis. As the AHP data comprised AHPs as a group rather than information about the location of specific disciplines (e.g., the location of physiotherapy practices and speech pathology practices were not mapped individually). Acknowledging the diversity of services provided by AHPs, future research could explore access to each of these disciplines separately. Future research could also link patient functioning at post-stroke follow-up and services access to the location of the three types of primary healthcare services. Finally, the hospital that provided the stroke patient data for this project is one of the hospitals that provide data to the National Stroke Registry. Therefore, some of the data provided to the registry is the same as the data used in this project. This project was focused on only a single geographical region, as anything more was beyond the scope and funded capacity of the project. Future research could use this same geospatial mapping approach with a larger geographical region and use a greater volume of data from the registry (therefore providing information about a 90-day follow-up).
Conclusion
This project is the first step in determining whether current primary healthcare services are accessible by community-dwelling patients with stroke. This study suggests that primary healthcare services are of variable accessibility for patients with stroke, with a trend towards low levels of accessibility for many patients. This information may assist primary healthcare services with the planning of future services for community-dwelling patients with stroke to ensure that new services are located in areas that currently experience low access to primary healthcare services. The project also highlights that there is an urgent need for more research into the access of primary healthcare services for community-based patients with stroke.
Acknowledgements
The authors would like to thank the Brisbane South PHN and the Princess Alexandra Hospital for the data analysed in this study. They would also like to thank Adele Coleman and K. C. Kiran for assistance with data collection and processing.
Financial support
This research was supported by a University of Queensland Collaborative Industry and Engagement Fund FirstLink grant.
Conflict of interest
The authors have no conflict of interest to disclose.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
