Research to determine the health outcomes of a population based on the access they have to health services must incorporate a way of attributing outcomes to a particular location or area (often hospital). This area is usually described as the hospital catchment; however, the best definition of hospital catchment is debated in the health services literature. Methods involving relatively easy calculation such as a fixed radius around each hospital do not take into account geographic or political boundaries which may present barriers to obtaining care. Other methods involving geographic information systems (GIS) calculated surface travel time, although accurate, are computationally and time intensive.
Another method involves analysing hospital discharge data to determine where people from a given area (often defined in terms of postal code) choose to obtain their care. This technique requires the analysis of large amounts of data, but this data is readily available from existing databases. Additionally, because a patients choice of hospital will usually incorporate the limitations imposed by geopolitical boundaries, this method incorporates the advantages of the GIS method mentioned above, while still being relatively straightforward computationally. For the purposes of rural hospital services research there are two main limitations to this method: First, it does not allow identification of the few areas of the country that are very remote where patients must travel long distances to obtain care at the closest hospital. Secondly, because hospital discharges are analysed, if the service being examined is not offered at a particular hospital, then the hospital will not have a catchment area for that service.
We are interested in determining the health outcomes for patients stratified by the degree of access to services they have at their local hospital, including hospitals that dont have routine access to the service in question. Thus, we must find a time and resource efficient way to analyse extremely large datasets in order to categorize patients by the level of service available at their home hospital.
In this project, we will use previously gathered hospital administrative data to determine catchment areas based on the methods described above. We will then compare the degree to which these definitions agree with one another and determine the method with the best combination of efficiency and effectiveness for the purposes of larger studies we are proposing for the future.