Impact of the modifiable areal unit problem in assessing determinants of emergency department demand

Abstract

Objective To examine the impact of the modifiable areal unit problem (MAUP) in an investigation of factors associated with ED demand in Perth, Western Australia, in 2016. Furthermore, to advocate a means of avoiding this impact. Methods ED presentations were classified as: urgent medical, non‐urgent medical, urgent trauma or non‐urgent trauma. In each group, sex‐stratified, age‐adjusted multivariate associations with socio‐economic status and distance to the nearest ED and general practitioner (GP) were estimated. Modelling was undertaken using different sets of spatial units: Australian Bureau of Statistics (ABS) Statistical Areas Level 1 (SA1s) and numerous aggregate‐level zonations of SA1s (ABS SA2s and others). Results Estimates obtained using the different units often varied widely: for seven (30%) of 24 strata defined by combinations of sex, ED type and covariate, the smallest and largest effect sizes differed in terms of direction; further, for 11 (65%) of the remaining 17 strata, the largest effect size was at least twice as high as the smallest. This demonstrates the MAUP's impact and that analyses based on a single set of spatial units are unreliable. To resolve the observed variation, we highlight the SA1‐level estimates. Conclusions When formulating interventions targeting reduced ED utilisation, policy planners should be guided by evidence based on analysis of appropriate spatial units. This ideal is undermined by the widespread lack of acknowledgement of the MAUP in studies examining drivers of ED demand using spatially aggregated data. To avoid the MAUP, only estimates obtained through examining a minimal geographic unit should be relied upon.

Assessing the influence of the modifiable areal unit problem on Bayesian disease mapping in Queensland, Australia

“…This aligns with the findings in Fontanet et al [14] where the smallest level failed to provide meaningful inference due to smaller populations and counts. Kok et al [9] also found this to be the case when analysing hospitalisation rates for foot-related issues among the Indigenous population of Australia. Our findings, in conjunction with theirs, suggest that while aiming for the highest level of resolution may be optimal to reduce the biases of the MAUP, using mesh blocks can introduce other problems such as small numbers, zero populations, and computational inefficiencies.…”

“…One proposed method to minimising the MAUP is through using high spatial resolution [9]. However, we found that meshblocks, which typically represent 30 to 60 dwellings [28], were too sparse for a condition as rare as cancer.…”

“…As a result, the MAUP may have different implications for health data analysis in Australia compared to other countries. To date, there are only a handful of published works [4,9,12] investigating the impact of the MAUP in the Australian context.…”

“…There were 3890 SA1s and 164 SA2s in Perth in 2011, considered in this study, with median populations of 400 (for statistical areas level 1 (SA1s)), and 10000 (for larger SA2s). These studies have proposed various solutions to the MAUP, including using an overlay aggregation method for disease mapping by incorporating information from two nested aggregate levels [12], or minimising the influence of the MAUP through only reporting results at very high resolution, such as SA1s [9]. However, there are far greater numbers of people who present to hospital emergency departments or are admitted to hospitals than experience less common diseases or conditions.…”

Artificial Intelligence in Emergency Medicine: Viewpoint of Current Applications and Foreseeable Opportunities and Challenges

“…This bias results in different trends learned when the data are aggregated at different spatial scales. For example, when designing an AI system for ambulance demand, only estimates based on minimal-resolution data should be relied upon, as ambulance demand using areal data is potentially misleading owing to the modifiable areal unit problem [ 117 ].…”