P55 Creating a web-based interactive map visualising the geographic variations of the burden of Diabetes to inform policymaking: an example from Tasmania, Australia

Background: As diabetes imposes an increasing burden on the healthcare system, identifying regions with higher burden is crucial to tailor effective interventions to reduce the burden of diabetes.

Aim: With Tasmania, a state of Australia as an example, our study aimed to use mapping in combination with statistical analyses to visualise the geographic variations of diabetes burden and identify areas where targeted interventions are needed. Method: This study used a population-based dataset from 2004-2017 in Tasmania. Using diagnostic criteria supported by hospital codes, 51,324 diabetic people were identified. An interactive map visualising geographic distribution of diabetes prevalence, annual costs per diabetes person, and mortality rates in diabetic people were produced. The Getis-Ord Gi* method was performed based on statistical areas level 2 to identify areas with high (hot spot) and low (cold spot) diabetes burden. Fisher’s exact test was conducted to investigate the association of hot/low spots and Index of Relative Socio-economic Disadvantage (IRSD) quintiles.

Results: There were geographic variations in diabetes burden across Tasmania, with highest age-adjusted prevalence (6.1%-based on the Australian population in 2017), annual costs per person (AUD 5982) and age-adjusted mortality rates (20.7/10,000 people) in the West and Northwest. Among 98 areas, the Getis-Ord Gi* method identified 23 hot spots and 41 cold spots for annual costs per diabetes person, 14 hot spots and 11 cold spots for diabetes prevalence (p<0.1). 21/23 (91%) and 9/14 (64%) hot spots identified are in the West and Northwest. The map indicated similar distribution of hot clusters and IRSD level 1 areas (most disadvantaged) as well as cold clusters and IRSD level 5 areas (least disadvantaged). Fisher’s exact test demonstrated an association between hot/low spots and IRSD quintiles (p=0.01 and p=0.002, respectively).

Conclusion: The method presented in our study can be applied to any other regions and countries to identify areas where interventions are urgently needed to support evidence-based policymaking as well as enhance community’s awareness.