Spatial Bayesian distributed lag non-linear models (SB-DLNM) for small-area exposure-lag-response epidemiological...

Area of Science:

• Environmental Epidemiology
• Spatial Statistics
• Biostatistics

Background:

• Distributed Lag Non-linear Models (DLNMs) are standard for lagged non-linear associations, typically in large studies.
• Small-area analyses often lack lagged non-linear effects or geographically-varying risks.
• Previous methods downscaled risks or were infeasible due to low statistical power.

Purpose of the Study:

• To propose Spatial Bayesian DLNMs (SB-DLNMs) for reliable small-area lagged non-linear association estimation.
• To demonstrate SB-DLNMs using the temperature-mortality relationship in Barcelona neighbourhoods.
• To address limitations of existing models in small-area statistical analysis.

Main Methods:

• Generalized location-independent DLNMs to a Bayesian framework (B-DLNMs).
• Extended B-DLNMs to SB-DLNMs by incorporating spatial models in a single-stage approach.
• Accounted for spatial dependence between risks in the model.

Main Results:

• SB-DLNMs demonstrated benefits for small-area analysis by incorporating spatial components.
• Independent B-DLNMs yielded unstable estimates in areas with low death counts.
• SB-DLNMs produced more plausible and coherent estimates, revealing spatial patterns.

Conclusions:

• SB-DLNMs effectively model spatial structures in risk associations across small areas.
• Facilitate direct estimation of non-linear exposure-response lagged associations at the small-area level.
• Applicable even in areas with minimal data (e.g., 19 deaths), with reproducible code provided.