The timing of geographic power

Area of Science:

• Environmental epidemiology
• Spatial statistics
• Cancer research

Background:

• Spatial risk studies often use current residential data to find disease clusters.
• These methods frequently miss significant spatial patterns due to exposure timing and location mismatches.
• Residential mobility and long cancer latencies can obscure historical environmental exposure signals.

Purpose of the Study:

• To investigate how residential mobility affects the detection of historical environmental exposures using spatial models.
• To assess the impact of exposure timing and duration on identifying elevated risk areas.
• To evaluate the effectiveness of generalized additive models in detecting spatial disease risk signals in mobile populations.

Main Methods:

• Simulation studies were conducted using residential histories from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) non-Hodgkin lymphoma study.
• Data were generated by varying the duration and intensity of simulated environmental exposures.
• Generalized additive models were employed to analyze the simulated data and assess model performance.

Main Results:

• The probability of detecting elevated risk areas decreased as the time since exposure increased.
• Mean spatial sensitivity and specificity of the models declined with increasing time lag between exposure and diagnosis.
• These findings indicate that historical, high-intensity, short-duration exposures become harder to detect over time in mobile populations.

Conclusions:

• Detecting spatial disease risk linked to past environmental exposures is challenging in populations with residential mobility and long disease latencies.
• Current spatial epidemiological methods relying on diagnosis-time residences may fail to identify significant risk areas for diseases with long latency periods.
• Future research should consider advanced methods that account for residential history and exposure timing to improve spatial risk assessment.