Diabetes is a key non-communicable disease (NCDs) targeted for global action (1, 2). The prevalence of diabetes has increased from 4.7% in 1980 to 8.5% in 2014 in the adult population globally (1). The World Health Organization (WHO) projected that the number of diabetes cases worldwide would reach 366 million by 2030 (3). Type 2 diabetes mellitus (T2DM), diagnosed when the pancreas could no longer produce sufficient insulin, accounts for 90-95% of all diabetic cases (4). The management of T2DM is complex and time-consuming, often involving regular health consultations, lifestyle modifications, frequent blood glucose and podiatry monitoring and complex medication regimes (5–8). The estimated annual spending on T2DM in Australia is around $6 billion. The average annual healthcare cost per person is as high as $4,025, even if there are no associated complications (9).

Previous studies (10–12) have revealed that the development of T2DM results from complex interactions between genetic, environmental, lifestyle and other risk factors. It is also revealed that obesity and physical inactivity are highly correlated with T2DM (13, 14). Most of these associated studies have been conducted on an individual level, providing essential information to identify key modifiable health behaviours for the general health of people living with T2DM. However, understanding the geographical trends of T2DM at a population level may provide important evidence to inform better health policies and population-based prevention programs.

Several existing studies have reported the geographical variance in T2DM disease burden (15–19). Angela et al. (16) evaluated the geographical variations of T2DM incidence among teenagers in the United States. Douglas et al. (17) identified regions with a high disease burden of T2DM in the city of London, UK. Similar studies were also conducted in Ukraine (19) and China (20). Geospatial correlation studies that focused on the association between the risk factors and population effects on T2DM were also conducted in Europe, the US and China as evidence to facilitate political commitment and implementation of community-based programs to curb the epidemic of T2DM (21–23). They reported the association of T2DM with obesity and physical activity, and some studies also highlighted that these geographical variances could result from social determinants, such as income and employment (24–26). In Australia, a typical developed country with the world’s 12^th largest economy, several government-issued reports unveiled the geographical variations of T2DM incidence nationwide but without in-depth spatial analysis of T2DM and its interaction with population risk factors (9, 27–29). In particular, more than 90% of T2DM in Australia were found in the middle-aged and elderly population of more than 45 years old. Therefore, understanding the geographical trends and the relevant population factors that contribute to T2DM in this group is key for the prevention of T2DM in the ageing Australian population.

We hypothesised that there are significant geographical variations in T2DM incidence across the Australian state of New South Wales, and these variations are potentially associated with the obesity prevalence and physical inactivity rate in the population. We used demographic, behavioural, medical, and pharmaceutical data from a ten-year follow-up study cohort with 266,896 participants to explore the geographical variation in the incidence of T2DM and associated population factors. Findings from this study will add new evidence to inform health policies to be modified based on geographical variations across Australia at a population level.

SAS version 9.4 (SAS Institute Incorporation) and R Studio was used for non-spatial data analyses. To understand the distributions in diabetes incidence across different regions in Australia, we calculated the range, mean and standard deviation for each potential association factor, including obesity prevalence, percentage of the population with insufficient physical activities and percentage of the population walking to walk. We further analysed the non-spatial correlation between diabetes incidence and each factor using Spearman correlation.

For spatial analysis, Choropleth maps of the outcome and associated factors were plotted using ArcGIS 10.4 (ESRI, Redlands, CA) and GeoDa 1.12. We applied univariate Moran’ I to identify if there are significant geographical variations in diabetes incidence, obesity prevalence, percentage of the population with insufficient physical activities and percentage of the population walking to walk, respectively. If a univariate Moran’ I value is close to 1, it indicates a significant geographical variation, whereas a value close to 0 indicates no clear geographical variations. To assess colocation between T2DM incidence and each of the association factors, we calculated the bivariate Moran’s I and intraclass correlation coefficient for the associated factors and diabetes incidence. Bivariate Moran’s I was a global measure of spatial correlation to measure the influence one variable has on the occurrence of another variable in close proximity (39). A value close to 1 represents a clear spatial correlation, whereas a value close to 0 means no clear spatial correlation.

In this large-cohort GIS study of Australians aged 45 and above, we identified significant geographical variations in T2DM incidence across NSW, Australia. We demonstrate that T2DM incidence is significantly correlated with obesity prevalence and physical inactivity rate at a population level.

The geographical variation and spatial clustering of T2DM incidence identified in this study are consistent with the results from previous studies that compare the diabetes prevalence among major cities and remote areas in Australia (40). Besides, the very high T2DM incidence in the outback region of NSW could result from the arid local climate and the lack of health resources in these regions (41).

Our finding that the onset of T2DM was closely associated with obesity and physical inactivity is consistent with the findings from existing studies (13, 42, 43). Although multiple epidemiological studies identified obesity as the most significant associated factor for T2DM due to its pro-inflammatory contribution to the development of insulin resistance and disease progression (44, 45), our study revealed that physical inactivity rate is more important contributing factor to T2DM than obesity in an Australian population. This is supported by Ansari (42) who found that physical activities, including stair climbing and cycling, may reduce the risk of T2DM at a population level.

Previous studies have identified that the spatial variance of T2DM, obesity and physical inactivity can result from a few socioeconomic factors, such as income, education and occupational status (46, 47). It is found that one’s education level is most likely to affect one’s understanding and knowledge of the health benefits of preventative behaviors (48). Low income and education level have also shown to be associated with a high risk of metabolic system, leading to over-weight and obesity (49, 50). In addition, employment status categorized by occupation, has shown an inverse relation with glucose intolerance (47, 51). Therefore, more educational resources and occupational guidance can be provided to population in the high-risk regions for T2DM control.

The major contribution of this study is that we identified a significant colocation of T2DM incidence with obesity prevalence and physical inactivity rate at a population level. A population with a high prevalence of obesity (BMI>30kg/m²) and a high percentage of insufficient activity (physical activities of <5 sessions/week) was strongly associated with a high incidence of T2DM. To our knowledge, our study was the first to investigate the spatial correlation of T2DM with its associated lifestyle factors, contributing to the sociocultural perspective of T2DM prevention. The advances in technology and transportation, long hours of sedentary office work, and increasing access to processed food increase the risk of obesity and physical inactivity (52), leading to an increased risk of T2DM onset in adults. For instance, in our study, the Mount Druitt region has the highest obesity prevalence, the second-highest percentage of population physical inactivity; therefore, substantial changes in public policies to create an environment that promotes the wellbeing of the whole community is a priority in the region. Further, the government may consider allocating resources for health promotion campaigns and increasing accessible exercise facilities at the workplace to reduce T2DM incidence (53). Another strength of this study is that the use of a sizable study cohort with 232,064 participants in Australia with a long follow-up period, and the comprehensiveness of the captured information for each participant, can lead to more reliable study findings. T2DM incidence was captured based on the Medicare record system, which effectively minimized the recall bias from the participants, especially the elderly.

However, the study has a number of limitations. First, our definition of T2DM might overlook cases of gestational diabetes. Second, our study was conducted based on the assumption that participants did not move to a new location, and their body weights were relatively stable during the study period. Third, the data of working to work used in this study only relates to the regions around the City of Sydney not the whole of NSW and is limited to the people who are employed. Fourth, data for obesity prevalence and physical inactivity rate were collected from individual-level questionnaires with the response rate of only 18%, raising issues regarding the overall representativeness of the recruited participants from each SA3 region. But we found that the characteristics of participants in our study are, in fact, very similar to the characteristics presented in the New South Wales Population Health Survey conducted by the NSW government (54).

In conclusion, this study is a large-scale GIS study that addressed the geographical disparities in T2DM incidence and its associations with the prevalence of obesity and physical inactivity across an Australian state. This study highlights that a high prevalence of obesity and physical inactivity in a population may contribute to a high incidence of T2DM in the population. Community-based intervention on healthy lifestyles and behaviors should be prioritized to help control T2DM incidence in a population.

The data analyzed in this study is subject to the following licenses/restrictions: This research was completed using data collected through the 45 and Up Study (www.saxinstitute.org.au) supplied by Services Australia and the Australian Bureau of Statistics (ABS). The 45 and Up Study is managed by the Sax Institute. Requests to access these datasets should be directed to www.saxinstitute.org.au.

JW was in charge of the project experiments and analysis, and writing the draft for the paper. YW was responsible for assuring the experiment the results are consistent and modifying the paper. XX was providing method advice and tool advice to make sure appropriate data analysis. XS is responsible for data preprocessing to extract proper data for our project. MH and LZ supervised the project, gave advice on the techniques and direction advice on the project, and helped in paper revision. All authors contributed to the article and approved the submitted version.

MH receives support from the University of Melbourne at Research Accelerator Program and the CERA Foundation. The Centre for Eye Research Australia (CERA) receives Operational Infrastructure Support from the Victorian State Government. The specific project is funded by the Australia China Research Accelerator Program at CERA. MH is also supported by the Fundamental Research Funds of the State Key Laboratory in Ophthalmology, National Natural Science Foundation of China (81420108008).LZ is supported by the National Natural Science Foundation of China (Grant number: 81950410639); Outstanding Young Scholars Support Program (Grant number: 3111500001); Xi’an Jiaotong University Basic Research and Profession Grant (Grant number: xtr022019003, xzy032020032); Epidemiology modeling and risk assessment (Grant number: 20200344) and Xi’an Jiaotong University Young Scholar Support Grant (Grant number: YX6J004).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.