Approche spatiale des inégalités d’exposition environnementale et socioéconomiques. Quelle influence sur les inégalités de cancer ? Projet cancer inégalités régionales, cantonales et environnement (CIRCE)

Danièle Valdés-Lao, Christophe Declercq, André Cicolella, Annie J. Sasco, Olivier Blanchard, Luc Mosqueron, Alain Trugeon, Isabelle Grémy, P. Pépin, Olivier Guye, Abdoul Sonko, Marc Colonna, Julien Caudeville, Jean‐Baptiste Masson, G. GovaertRevue 2008 · DOI

TL;DR

This paper explores the spatial relationship between environmental exposures, socioeconomic status, and cancer mortality in the Nord-Pas-de-Calais region of France. By using Geographic Information Systems (GIS) to map these factors at the cantonal level, the study aims to understand how environmental and socioeconomic inequalities might contribute to cancer disparities. The findings suggest that certain areas exhibit a clustering of social, environmental, and health inequalities, highlighting the complex interplay between these factors.

What Problem Does This Paper Address?

The paper tackles the growing concern over the increasing incidence of cancer, particularly in developed countries. While genetic and individual lifestyle factors are acknowledged, the authors posit that environmental exposures and socioeconomic status play a significant, yet often under-quantified, role. The challenge lies in understanding how these broad-scale factors translate into spatial inequalities in cancer risk. Traditional epidemiological studies often rely on experimental data or case-control studies, but this research proposes an ecological correlation approach using GIS to analyze geographical disparities and their potential links to cancer. Specifically, it seeks to answer: "What influence do environmental and socioeconomic inequalities have on cancer inequalities?"

Methods

The study employs a spatial ecological correlation approach using Geographic Information Systems (GIS). The core methodology involves:

  • Developing Proxies for Environmental Exposure: Instead of direct risk assessment, the researchers use "proxies" – indirect indicators – to estimate population exposure. These include:
  • Air Pollution: Modeled concentrations of PM10, PM2.5, O3, NO, and NO2 derived from the PREVAIR model, gridded at 10km x 15km and attributed to cantons.
  • Land Use: Percentage of urbanized areas (proxy for domestic pollution) and arable land (proxy for agricultural pollution) from the Corine Land Cover database.
  • Soil Contamination: Density of polluted industrial and commercial sites from the BASIAS database, serving as a proxy for historical exposure.
  • Industrial Pollution: An index of air concentration based on emissions data from the EPER database, considering distance to emission sources and emission flux, modeled using an exponential decay function.
  • Quantifying Socioeconomic Status: The Townsend Deprivation Index is used, calculated from INSEE census data. This index aggregates standardized values of unemployment rates, household size per room, lack of home ownership, and car ownership.
  • Assessing Cancer Risk: Relative risks of mortality for 20 cancer sites are mapped using data from the CépiDc (Inserm) and smoothed using a local risk estimation method to account for low statistical power in sparsely populated cantons.
  • Spatial Analysis: All data is mapped at the cantonal level using GIS (ArcGIS). The study then proposes to analyze the relationships between these environmental, socioeconomic, and health indicators using spatial statistical methods, including Bayesian hierarchical models.

Datasets

The study utilizes several publicly available datasets:

  • Environmental Data:
    • PREVAIR model outputs for air pollutants (PM10, PM2.5, O3, NO, NO2).
  • Corine Land Cover database for land use information.
  • BASIAS database for polluted sites and soils.
  • EPER (European Pollutant Emission Register) for industrial emissions.
  • Socioeconomic Data:
    • INSEE census data for the calculation of the Townsend Deprivation Index.
  • Health Data:
    • CépiDc (Inserm) data on cancer mortality.
  • Regional Health Observatory (ORS) data for Nord-Pas-de-Calais, Picardie, Île-de-France, and Rhône-Alpes.

The study focuses on the Nord-Pas-de-Calais region for illustrative results, mapping data at the cantonal level.

Key Insights and Ideas

  • Spatial Clustering of Inequalities: The study reveals that certain cantons, particularly the Lille metropolitan area and the mining basin, tend to concentrate social, environmental, and health inequalities. This suggests that these factors are not randomly distributed but are spatially linked.
  • GIS as a Tool for Understanding Complex Relationships: The use of GIS is presented as a novel approach to visualize and analyze the complex, multi-faceted relationships between environmental exposures, socioeconomic status, and cancer. It allows for a geographical perspective that might be missed in traditional analyses.
  • Proxy-Based Exposure Assessment: The paper highlights the practical challenges of direct exposure measurement and advocates for the use of well-chosen proxies (like land use or industrial site density) to estimate population exposure to environmental hazards.
  • Interplay of Factors: The research implicitly suggests that understanding cancer disparities requires considering the combined influence of multiple factors, rather than isolating single causes. The clustering of inequalities implies that these factors may interact and amplify each other.

Key Results

The paper presents extracted maps (Figure 2) illustrating the spatial distribution of socioeconomic indicators, environmental proxies, and relative cancer mortality risks at the cantonal level in Nord-Pas-de-Calais. These maps demonstrate significant spatial disparities within the region. For instance:

  • The Townsend Deprivation Index highlights the Lille area and the mining basin as particularly disadvantaged.
  • Environmental exposure indices (e.g., density of polluted sites, urbanized areas) also show concentrations in similar areas, suggesting a potential overlap between environmental burdens and socioeconomic disadvantage.
  • Relative cancer mortality risks vary by cancer type and location, with recurrent situations of excess mortality observed around major urban centers.
  • Crucially, the study notes that some cantons appear to cumulate social, environmental, and health inequalities, indicating a complex web of contributing factors.

Limitations and Future Work

The authors acknowledge several limitations:

  • Population Mobility: The study does not account for short-term (workplace vs. residence) or long-term (past residences) population mobility, which can affect actual exposure.
  • Cancer Latency: The current exposure data is contemporary, while cancers have long latency periods (years to decades). This temporal mismatch is a significant challenge.
  • Scale of Analysis: While cantonal data is used, the authors suggest that finer scales (e.g., municipal) might offer more relevant insights.
  • Data Type: The study relies on mortality data, whereas incidence data is considered more pertinent for studying the cancer-environment relationship.

Future work suggested by the project CIRCE includes mapping cancer incidence, which would provide a more direct measure of disease occurrence and potentially strengthen the analysis of environmental and socioeconomic influences. The authors emphasize that despite these limitations, ecological correlation studies using GIS offer a valuable new perspective on spatial disparities and help in understanding the complexity of the environment-social-health nexus.

Summary updated

BibTeX
@inproceedings{valdeslao2008approche,
  author = {Danièle Valdés-Lao and Christophe Declercq and André Cicolella and Annie J. Sasco and Olivier Blanchard and Luc Mosqueron and Alain Trugeon and Isabelle Grémy and P. Pépin and Olivier Guye and Abdoul Sonko and Marc Colonna and Julien Caudeville and Jean‐Baptiste Masson and G. Govaert},
  title = {{Approche spatiale des inégalités d’exposition environnementale et socioéconomiques. Quelle influence sur les inégalités de cancer ? Projet cancer inégalités régionales, cantonales et environnement (CIRCE)}},
  booktitle = {Revue},
  year = {2008},
  doi = {10.1016/j.respe.2008.07.036},
}
Figure