Creating a Spatial Vulnerability Index for Environmental Health

Aiden Price, Kerrie Mengersen, Michael Rigby, Paula FiévezarXiv 2024 · DOI · arXiv

TL;DR

Researchers have developed a high-resolution, open-access "Spatial Vulnerability Index" for Australia that specifically weights environmental factors based on their actual correlation with mortality rates. By moving beyond standard "equal-weight" models to a health-outcome-weighted approach, this tool provides a more accurate, weekly view of how extreme heat, cold, and air pollution impact different communities.

What Problem Does This Paper Address?

While we know that environmental factors like heatwaves and poor air quality drive significant mortality, measuring a community's "vulnerability" is notoriously difficult. Most existing Vulnerability Indices (VIs) rely on a "one-size-fits-all" approach, either weighting all variables equally or using statistical techniques like Principal Component Analysis (PCA) that prioritize data variance over actual health impacts.

Furthermore, environmental health data in Australia is often siloed, making it time-consuming for policy-makers to compare regions or track changes over time. There is a critical gap between having raw environmental data and having an actionable, health-aligned metric that accounts for a population's specific sensitivities (like pre-existing conditions) and adaptive capacities (like access to hospitals).

Methods

The authors utilize the IPCC framework, which defines vulnerability as a combination of Exposure, Sensitivity, and Adaptive Capacity.

The paper’s core innovation is the Health Outcome Weighted (HOW) methodology. Instead of assuming every variable is equally important, the researchers calculated the Kendall’s Tau correlation between individual variables (e.g., percentage of elderly residents, canopy cover) and age-standardized mortality rates. These correlations were then used as weights to build the sub-indices.

Key technical highlights include:

  • Temporal Granularity: Indices are produced at weekly, monthly, and yearly resolutions, allowing for the detection of short-term "spikes" in risk.
  • Spatial Granularity: Data is mapped to various Australian Bureau of Statistics (ABS) levels (SA2, SA3, SA4) and Local Government Areas (LGAs).
  • Missing Data Handling: The authors implemented a modified sub-index theme approach to ensure that missing observations in one variable wouldn't "zero out" an entire region's score.

Datasets

The study draws from a massive array of sources to create the AusEnHealth repository:

  • Climate: 20 years of daily min/max temperatures from the Bureau of Meteorology (BOM).
  • Air Quality: 3-hourly pollutant data (NO, NO2, SO2, CO, O3, and PM1, 2.5, and 10) from the Copernicus Atmosphere Monitoring Service (CAMS).
  • Built Environment: Data on hospital counts, green space, and canopy cover from the ABS, Geoscience Australia, NASA (MODIS), and TERN.
  • Demographics & Health: 20+ variables from the ABS Census and the Public Health Information Development Unit (PHIDU), including rates of asthma, COPD, and cardiovascular disease.
  • Mortality: Publicly available "Mortality Over Regions and Time" (MORT) books from the Australian Institute of Health and Welfare (AIHW).

Key Insights and Ideas

  • Sensitivity can trump Exposure: In a case study of Bundaberg, the researchers found high vulnerability despite low environmental exposure. This was driven by "Sensitivity" factors: high percentages of elderly residents and people with chronic respiratory diseases.
  • Averages hide the danger: Annual vulnerability scores often mask lethal short-term events. The weekly resolution revealed that some regions hit the 100th percentile of vulnerability during specific weeks, even if their yearly average appeared moderate.
  • Health-Weighting works: By comparing their weighted index (wHVI) against a standard baseline (HVI), the authors showed that the weighted version aligns much more closely with actual mortality data (as seen in the tighter clusters in their scatter plots).
  • Acclimatization matters: The heat exposure index uses "historical temperature percentiles," meaning it measures heat relative to what a specific region is used to, rather than a national absolute temperature.

Key Results

  • Improved Correlation: The weighted heat vulnerability index (wHVI) showed a substantively higher correlation with age-standardized all-cause mortality compared to the baseline equal-weight index.
  • Spatial Insights: Choropleth maps demonstrated that the difference between actual mortality and predicted vulnerability was significantly smaller when using the weighted methodology, particularly in remote and regional areas.
  • The AusEnHealth Platform: The project resulted in a functional, interactive "Shiny" R application that allows users to explore nearly a decade of time-series environmental health data across Australia.

Limitations and Future Work

The authors acknowledge several areas for refinement:

  • Association vs. Causation: The weights are based on empirical associations (correlations) with mortality, which does not strictly prove a causal link.
  • Data Recency: The study primarily used 2011 and 2016 Census data; updating the model with 2021 Census data is a priority.
  • Variable Inter-dependency: Future iterations could explore more complex models that account for how different variables (like income and education) interact with one another.
  • Expanded Variables: There is potential to include more granular urban data, such as building density and specific road traffic volumes, to further refine the air quality and heat-island components.

Summary updated

Abstract

Extreme natural hazards are increasing in frequency and intensity. These natural changes in our environment, combined with man-made pollution, have substantial economic, social and health impacts globally. The impact of the environment on human health (environmental health) is becoming well understood in international research literature. However, there are significant barriers to understanding key characteristics of this impact, related to substantial data volumes, data access rights and the time required to compile and compare data over regions and time. This study aims to reduce these barriers in Australia by creating an open data repository of national environmental health data and presenting a methodology for the production of health outcome-weighted population vulnerability indices related to extreme heat, extreme cold and air pollution at various temporal and geographical resolutions. Current state-of-the-art methods for the calculation of vulnerability indices include equal weight percentile ranking and the use of principal component analysis (PCA). The weighted vulnerability index methodology proposed in this study offers an advantage over others in the literature by considering health outcomes in the calculation process. The resulting vulnerability percentiles more clearly align population sensitivity and adaptive capacity with health risks. The temporal and spatial resolutions of the indices enable national monitoring on a scale never before seen across Australia. Additionally, we show that a weekly temporal resolution can be used to identify spikes in vulnerability due to changes in relative national environmental exposure.
BibTeX
@misc{price2024creating,
  author = {Aiden Price and Kerrie Mengersen and Michael Rigby and Paula Fiévez},
  title = {{Creating a Spatial Vulnerability Index for Environmental Health}},
  year = {2024},
  eprint = {2403.14954},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2403.14954},
}
Figure