Background and Methods

Background | Methods

Background

To date, the analysis of cost-effective health care has primarily been conducted within the sphere of Health Technology Assessment (HTA). Such analyses generally estimate cost-effectiveness in isolation from the realities of clinical practice, availability of resources, and service delivery models. Reimbursement decisions informed by cost-effectiveness analyses incorporate subjective thresholds for cost-effectiveness. It is then implicitly assumed that if a set of cost-effective interventions are specified, the health service will contrive to deliver a cost-effective health care system.

In practice, the organisation and availability of resources will affect the cost-effectiveness of supplied interventions. In hospitals, service delivery factors include the allocation of resources to outpatient clinics and surgical procedures, booking processes, as well as the interventions offered to different patient sub-groups. Activities in primary care have significant effects on hospital service delivery via referral rates, but also in terms of the timeliness of referrals and the accompanying referral information. Service delivery also affects patient decisions to seek treatment in the private sector.

The assessment of health service efficiency should evaluate costs and health outcomes associated with the joint provision of health technologies within clinical specialities. A whole of system perspective, which incorporates processes in primary and secondary care that impact observed outcomes, facilitates analyses of the impact of improving individual elements of the system.

The framework will be developed in the Ophthalmology Department at the Royal Adelaide Hospital (RAH). Ophthalmology is chosen because four ophthalmic disorders (age-related macular degeneration (AMD), diabetic retinopathy, glaucoma, and cataracts) cover the majority of separations within the speciality (data from RAH). The inclusion of just 4 conditions provides a less complex area for developing the framework. It is also predicted that the number of Australians with low vision and blindness will double by 2024 and that the 4 conditions account for 89% of non-correctable blindness in Australia,[Taylor et al, 2005] and so demand for related services will also increase significantly over the next 2 decades.

The need and potential to improve the efficiency of ophthalmic services in Australia has been recognised,[Victorian Government, 2005] but accompanying recommendations have been restricted to general statements regarding the “development and expansion of models of care that promote effective and efficient delivery of eye care services” (pg51).

Methods

Separate disease models will describe incidence, natural history and treatment pathways and effects for the four ophthalmic conditions as functions of relevant patient characteristics (including indigenous status) and prognostic indicators. Literature reviews will identify existing models, and inform disease parameters using established methods.[NHS CRD, 2001]

The four disease models will be combined in a linked model that jointly represents contacts with the health service, disease progression, and health-related quality of life (QoL). Figure 1 illustrates the modelling framework for a generic eye disease, developed iteratively by the research team. The decision tree represents alternative pathways through the health care system. The timing between contacts is important and the notes summarise the data sources to populate the model. The bold line represents the pathway for a hypothetical individual who presents in primary care in disease state A (an early stage of disease), which progresses to disease state B whilst waiting for an initial hospital appointment, and subsequently to disease state C whilst waiting for surgery. QoL declines until surgery, after which it improves.

The full model will represent all individuals with the four conditions in South Australia, focussing on those referred to the RAH. Patient interactions will be modelled, i.e. the use of scarce resources by one patient delays use by another patient. The model will extend out to 30 years using predicted changes in disease incidence and the health workforce to explore current and future service delivery issues, costs and outcomes. 

The model will be developed as an individual sampling simulation model using commercially available software (Simul8, Simul8 Corporation), which has been successfully used to model a wide range of complex health care evaluations.[NICE, 2006; Campbell et al, 2001; Karnon & Brown, 2002]

The locally relevant data collected from patient records will inform natural history (at visits prior to treatment being provided), treatment options, treatment effectiveness (after treatment is provided), and help define local prioritisation approaches. The data will be analysed by discrete disease stage categories, defined in accordance with the disease models.

The QoL survey will inform the estimation of the main outcome measure, the quality adjusted life year (QALY).