This study has shown, that where people live (at the local area-level) matters for the GP and specialist services they receive, independent of their personal characteristics. This was the case across all remoteness categories - major cities, regional and more remote areas - in New South Wales. Further, having accounted for where people live, use of GP services and quality of care was equitable, in that disadvantaged people were more likely to use more services on average, and to have continuity of care and care planning. However, the finding that advantaged people were more likely to see a specialist or have a long consultation suggests a potential source of inequity.
This is the first study in Australia and one of few internationally to quantify area-level variation in GP and specialist use, independent of the characteristics of people who lived in these areas. The amount of variation between areas quantified in this study is comparable to that previously reported when examining other healthcare outcomes in Australia (e.g. hospitalisations ), and internationally . More use of GP services and care planning and greater continuity of care among people of lower SEP has been previously shown [7–9, 27] and this study confirms that this holds having accounted for where people live. People of lower SEP are more likely to have multiple and complex health and psychosocial care needs  than their advantaged counterparts; continuity and care planning are essential for enabling these needs to be met. International data from countries without gate-keeping mechanisms in place have found inequity of specialist use , independent of where people live. Our study demonstrates this was also found within a setting with gate-keeping policies in place.
We found that individual use of GP and specialist services varied across small-areas, for all remoteness categories, beyond what could be explained by the characteristics of people living in those areas. This suggests that there are aspects within peoples’ local context that systematically shape the care of all who live in that area. The specific reasons are unknown but may relate to how services are organised and delivered (including availability of providers) within an area or structural policies determining the geographical distribution of services and providers. International multilevel studies in countries with  and without  a gate-keeping mechanism have shown that availability of GPs and specialists within an area was associated with specialist use. This has not been investigated for GP service use or quality of care. Importantly, how services are organised can be changed (through policy and practice) and doing so may contribute to reducing the unwarranted variation across areas.
There are likely multiple reasons why socioeconomically disadvantaged people use less specialist services for a given level of need. Unlike GP care, in Australia there are no bulk-billing incentives for specialists. Out-of-pocket costs for specialist services doubled in the decade prior to the study period  and have continued to rise since. Further, private health insurance has been shown to contribute to pro-high income use of specialist services ; yet government funded rebates for private health insurance have remained in place. Other possible reasons include: differences in propensity to seek care due to differences in health literacy, attitudes and beliefs; or, due to negatively biased behaviours from providers, disadvantaged people are less likely to seek specialist care . However, if this was the case a similar finding would be expected with use of GP services. Further, studies examining propensity to seek care  or rates of completion of specialist referrals have not found differences between socioeconomic groups.
Alternatively, these differences may be due to provider preferences and bias. International evidence also suggests providers offer fewer services to those of low SEP  and are more likely to refer higher SEP individuals to a specialist . Irrespective of the reasons, differences in use does not reflect need for care and hence is inequitable and unjust.
The reason why people of high education were more likely to have a long consultation is unknown. Possibly, low educated people are more likely to be bulk-billed, and given current financing arrangements in Australia, the benefit per minute falls with longer consultations. These findings may also reflect differences in health literacy. More highly educated people may be more likely to anticipate and expect a range a health issues to be addressed in a single episode and request a consultation length to that effect, or actively seek out practitioners with characteristics associated with longer consultations.
A strength of our study is the multilevel analytical design, which allowed modelling of nested levels of data and quantification of area- and individual-level variation. Further, the large sample linked to MBS service use, allowed quantification of observed use (rather than self-report) after accounting for a range of factors. While MBS data will capture nearly all GP services, there are some settings where services provided do not attract an MBS claim. For example, publicly funded community health centres and some GP services provided in emergency departments in rural and remote areas. In addition, a substantial proportion of specialist services in Australia are provided in publicly funded hospital-based outpatient clinics, which generally do not attract an MBS rebate. Low-SEP people are more likely to use these community and hospital-based services  and exclusion of these services may bias estimates for SEP gradients to be pro-high SEP. However, previous studies found this did not alter estimates of socioeconomic variation in GP and ambulatory specialist care.
Implications for policy/public health.
An effective PHC system requires ready and reliable access to secondary level care. This has not been equitably achieved in Australia–despite the presence of universal health insurance–undermining the equity that exists in the PHC system. National structural policies, such as minimising out-of-pocket costs for example through bulk-billing incentives, would go some way to redressing inequitable use of specialist services. Given that private health insurance contributes to pro-high SEP use of specialist services, offsetting government rebates in favour of lower income or disadvantaged individuals could also contribute to reducing this inequity. It could be argued that the inequity in community-based specialist services is balanced by a pro-low SEP preference for specialist outpatient services through the public hospital sector. However, waiting times for less urgent and more discretionary health needs (and in some instances for more urgent health needs) in the public sector are understood to exceed that in the private sector , although actual wait times are not published. This increases the impact of illness on recovery and quality of life, affecting those who are disadvantaged to a greater extent. As such, addressing inequalities in access to specialist care is even more pressing.
The unwarranted variation in both GP and specialist use suggests that additional policy approaches are needed that are directed to local contexts, rather than at individuals. For example, it may be that availability of providers (both GPs and specialists) may need to be addressed, as international studies have shown this explains some of the area-level variation in care. Similarly, there may be other aspects of how services are organised and delivered at the local area-level that may determine their use of services. The specific drivers and hence policy solutions to addressing the unwarranted area-level variation requires further exploration.