Degree Name

Doctor of Philosophy


Graduate School of Business


The pursuit of a more beneficial service mix in primary medical care is a worthwhile public goal. Public expenditure on diagnostic testing referred from general practice is a matter of public interest because of its potential benefit to the social welfare function.

To realise this potential, interventions must first reflect the evidence-base for enhancing clinical quality and promote discretionary increases in certain interventions (Eddy 1994[b] p.817; Rodwin 2004 p. 1328; Starfield 1998 p.406; Van Weel & Del Mar 2004 p.99). The effectiveness of primary care however is stratified by social class (Macinko, Starfield & Shi 2007 p.121; Starfield 1998 p.411). Therefore, services must also take into consideration any access barriers for vulnerable social groups and demonstrate a positive commitment to addressing the imbalance (Starfield 1998 p.406).

In practice, good clinical or scientific evidence alone is insufficient to achieve the optimum distribution of health services. The evidence must be matched by economic viability and sensitivity to the prevailing socio-political imperatives (Haas 2001 p.228; Van Der Weyden & Armstrong 2004 pp.607-608). Planning should explicitly consider marginal opportunities for changes in the balance of costs and benefits (Haas 1997 p.81).

The purpose of this study is to derive a model that levers redistribution of general practitioner-referred diagnostic services in favour of vulnerable social groups within Australia. The study operates within the boundaries of the dominant disease-state paradigm, because it focuses on systematically addressing nationally-prioritised epidemiological indicators for targeted populations.

The derived model relies on intermediaries representing groupings of general practices to drive the redistribution. It establishes an environment of nominal risk for the Divisions of General Practice network, acting as intermediaries. In turn, the actual risk to the Australian Government as the purchaser is limited to public funding through the Medicare Benefits Schedule of general practitioner-referred medical imaging (Category 5 [excluding Group I5]) and pathology tests (Category 6).

This is achieved by introducing a credit reserve ledger as a novel mechanism to track and reward Division performance. The ledger is a tool for the Australian Government to map the balance of benefits claimed on diagnostic services referred by general practitioners enrolled with each Division. Ledger balances depend on a separation of medical imaging and pathology items into three streams. The systematic streaming of items is according to whether they are over-, appropriately- or under-referred, according to the available evidence.

The key for Divisions to draw on their credit reserve ledger is the proportionate uptake of the evidence-based target items by identified vulnerable social groups within their catchment. This is compared with a target level of activity set for these groups to establish a specific performance ratio for each financial period.

The research design of this study tests the model’s effectiveness in the current health care environment, rather than its theoretical efficacy. The model acknowledges Australia’s current legislative and policy framework and its communities’ over-arching socio-political imperatives. No presumptions are made about changing the Medicare Benefits Schedule or its predominant fee-for-service mode of delivery.

The redistribution model is tested using a series of scenarios, and analysed in three parts. In the first part a macro-level analysis examines the net implications of the redistribution for the Australian Government, Divisions of General Practice and diagnostic providers as a whole across four different scenarios. In the second part, a meso-level analysis uses the existing Divisions’ network in a further three scenarios. Normative projections are developed across categories of geographic dispersion for each of the given scenarios. Thirdly, a micro-level analysis examines the absolute values of projected credit reserves within the same scenarios as the meso-level analysis for each of the Divisions.

The model results in a 0.02% increase in total tests with a 2.2% reduction in the total of benefits claimed. Within this ideal redistribution, there is an 18.4% reduction in uptake of over-referred items, a modelled 0.8% growth in uptake of appropriately referred items, and a substantial growth in uptake of the targeted, under-referred items (activity by 84.9% and benefits claimed by 94.2%).

The meso-level analysis demonstrates that the model has a defining normative bias in favour of increasing rurality and remoteness. This is consistent with the model’s aim of delivering supply-side incentives to service vulnerable social groups.

The meso-level results also indicate that a staged implementation of the model is required. This is because the overwhelming majority of Australia’s population live within the more metropolitan and regional Divisions that require the greatest effort to glean benefit from the model. Initially, they may be the most difficult to engage.

At the micro-level, the model is tested on estimated parameters matched to one-hundred and nineteen Divisions of General Practice. The result is that all the Divisions of General Practice within the existing network have sufficient critical mass to accumulate a material benefit from participation.

The study acknowledges that limitations in the model design may risk perverse incentives and unintended aberrations. For this reason, the model requires the protection of a regulatory framework to ensure its proper application in the field. Based on implementation experience, refinements may be required over time to reduce any unintended consequences.

There are limitations to the study design which give cause for further investigation and testing in the field. The analyses rely on secondary data, which risks artefacts within the results. Further, this is a study of marginal costs and benefits, rather than a true cost-effectiveness analysis because the utilisation targets used are interim measures of process, and not definitive measures of change in health status. Finally, this study is also limited by its inability to test the model against the actual parameters from identified Divisions of General Practice.

The study concludes that the model undergo further investigation and field testing in order to derive empirical results. It is also recommended that future studies test the generalisability of the model, with research into the redistribution modelling of general practice prescriptions and referrals to specialists for elective procedures. The consistent aim is to achieve marginal redistribution in the pursuit of an enhanced social welfare function.

02Whole.pdf (1906 kB)