Analysing geographic variation in lower urgency emergency department presentations using readily – available administrative boundaries and a novel spatial smoothing technique

ABSTRACT

Misallocation of finite healthcare resources can occur when guided by maps that are produced using aggregate-level administrative units. Such maps are affected by the modifiable areal unit problem (MAUP), which describes how patterns may change depending on the particular choice of mapping unit. Smoothing can help avoid this unintended yet detrimental phenomenon. This paper compares the utility of aggregate-level administrative units and smoothing for exploring variation in lower urgency emergency department (ED) presentations across metropolitan Perth, Western Australia. Rates of such presentations were mapped using Australian Bureau of Statistics Statistical Areas Levels 1, 2, and 3 (SA1-3s) and the recently proposed Overlay Aggregation Method (OAM). Resulting maps were compared based on their ability to represent local variation in rates and optimise the targeting and logistical efficiency of geographically targeted resource allocation. SA1-level variation in rates was increasingly obscured by SA2s and SA3s. OAM helped avoid this pitfall, facilitating stable identification of SA1-resolution, high-rate regions while preserving privacy, mitigating the MAUP, and balancing the targeting and logistical efficiency of planned resource allocation to those regions. Routine application of smoothing can help avoid issues undermining maps of lower urgency ED presentations and other health outcomes that are based on aggregate-level administrative units.

KEYWORDS:

• Lower urgency ED presentations
• geographic variation
• spatial aggregation
• modifiable areal unit problem
• smoothing
• Overlay Aggregation Method

Acknowledgements

The authors would like to acknowledge the WA Data Linkage Branch and the WA Data Collections. We also gratefully acknowledge the use of the AZTool software (v1.0.3; 25/8/11; https://www.geodata.soton.ac.uk/software/AZTool), which is copyright David Martin, Samantha Cockings and University of Southampton.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Ethics Approval

The ED data used in this study are not publicly available but are available to investigators who obtain approval from the WA Data Collections.

Additional information

Funding

This study was funded by the Department of Health, Western Australia.

Notes on contributors

Mei Ruu Kok

Mei Ruu Kok is a PhD candidate in Medical School at The University of Western Australia, Australia and now works in the field of GIS.

Matthew Tuson

Matthew Tuson is a PhD candidate in Medical School at The University of Western Australia, Australia and now works in the field of mathematics and statistics.

Berwin Turlach

Berwin Turlach is an Associate Professor in the School of Physics, Maths and Computing at The University of Western Australia, Australia.

Bryan Boruff

Bryan Boruff is a Geographer and Senior Lecturer in the School of Agriculture and Environment at The University of Western Australia.

Alistair Vickery

Alistair Vickery is a Clinical Associate Professor in the Medical School at The University of Western Australia, Australia.

David Whyatt

David Whyatt is an Adjunct Senior Lecturer in the Medical School at The University of Western Australia, Australia.