A causal inference framework for cancer cluster investigations using publicly available data

Area of Science:

• Environmental Epidemiology
• Causal Inference
• Biostatistics

Background:

• Communities often suspect environmental hazards cause observed high cancer rates.
• Standardized incidence ratio (SIR) analysis is a common but limited tool for investigating cancer clusters.

Purpose of the Study:

• To propose a novel causal inference framework for cancer cluster investigations.
• To introduce the causal SIR (cSIR) as an improved metric for assessing hazard-related cancer incidence.

Main Methods:

• Utilized the potential outcomes framework and matching to identify comparable unexposed populations.
• Developed a Bayesian hierarchical model to impute cancer incidence at a fine spatial resolution.
• Proposed the causal SIR (cSIR) estimand for causal inference.

Main Results:

• The novel statistical approach demonstrated significantly reduced bias and improved coverage in simulations compared to standard SIR analyses.
• The causal SIR (cSIR) framework provides a more robust method for evaluating potential environmental causes of cancer.

Conclusions:

• The proposed causal inference framework offers a more accurate and reliable method for cancer cluster investigations.
• This approach can better inform public health decisions regarding environmental exposures and cancer risks.