Overcoming inefficiencies arising due to the impact of the modifiable areal unit problem on single-aggregation disease maps

Abstract

Background: In disease mapping, fine-resolution spatial health data are routinely aggregated for various reasons, for example to protect privacy. Usually, such aggregation occurs only once, resulting in 'single-aggregation disease maps' whose representation of the underlying data depends on the chosen set of aggregation units. This dependence is described by the modifiable areal unit problem (MAUP). Despite an extensive literature, in practice, the MAUP is rarely acknowledged, including in disease mapping. Further, despite single-aggregation disease maps being widely relied upon to guide distribution of healthcare resources, potential inefficiencies arising due to the impact of the MAUP on such maps have not previously been investigated.

Results: We introduce the overlay aggregation method (OAM) for disease mapping. This method avoids dependence on any single set of aggregate-level mapping units through incorporating information from many different sets. We characterise OAM as a novel smoothing technique and show how its use results in potentially dramatic improvements in resource allocation efficiency over single-aggregation maps. We demonstrate these findings in a simulation context and through applying OAM to a real-world dataset: ischaemic stroke hospital admissions in Perth, Western Australia, in 2016.

Conclusions: The ongoing, widespread lack of acknowledgement of the MAUP in disease mapping suggests that unawareness of its impact is extensive or that impact is underestimated. Routine implementation of OAM can help avoid resource allocation inefficiencies associated with this phenomenon. Our findings have immediate worldwide implications wherever single-aggregation disease maps are used to guide health policy planning and service delivery.

Keywords: Disease mapping; Modifiable areal unit problem; Resource allocation efficiency; Single-aggregation disease maps; Zonation-dependence.

Fig. 1

Minimal-resolution efficiency results for the simulated dataset. a Simulated point location disease cases with the minimal units overlaid (grey squares). b Minimal-resolution disease map of crude rates. c Targeting efficiency map associated with b. d Logistical efficiency map based on a target case percentage of 50%

Fig. 2

Targeting and logistical efficiency curves for different mapping strategies applied to the simulated dataset. a Targeting efficiency curves. b Logistical efficiency curves. Curves are shown for the minimal-resolution (Min.); single-aggregation (Agg.); and OAM targeting strategies

Fig. 3

Single-aggregation efficiency results for the simulated dataset. a Simulated point location disease cases with the single-aggregation units overlaid (large grey squares). b Single-aggregation disease map of crude rates. c Targeting efficiency map associated with b. d Logistical efficiency map based on a target case percentage of 50%

Fig. 4

OAM efficiency results for the simulated dataset. a–c Three of OAM’s zonations. d–f Crude rate disease maps based on a–c. g Map of population-weighted mean crude rates produced using OAM. h Targeting efficiency map associated with g. i Logistical efficiency map based on a target case percentage of 50%

Fig. 5

Hotspot analysis results for the simulated dataset. a–c Hotspots classified based on Figs. 4d–f. d Minimal-resolution map of hotspot counts

Fig. 6

Administrative geography and population density of Perth in 2016. a SA1 and SA2 boundaries. b SA2-resolution population density

Fig. 7

Map of population-weighted mean RRs for stroke

Fig. 8

Targeting and logistical efficiency curves for stroke. a Targeting efficiency curves. b Logistical efficiency curves. Curves shown correspond to maps produced by SA1; SA2; or using OAM

Fig. 9

Logistical efficiency maps for stroke based on a target case percentage of 15%. a SA1 map. b SA2 map. c OAM map

Fig. 10

Hotspot analysis results for stroke. a SA2 hotspots. b, c Hotspots based on two of OAM’s zonations. d SA1-resolution hotspot counts