U.S. cancer mortality 1950-1978: a strategy for analyzing spatial and temporal patterns.

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Monitoring local area death rates presents statistical challenges in identifying true elevated risks.
• Existing methods often focus on outliers, potentially missing broader patterns.

Purpose of the Study:

• To propose and apply a two-stage analytic strategy for monitoring temporal and spatial variations in local area death rates.
• To address the statistical concern of identifying areas with genuinely elevated mortality risks amidst numerous comparisons.

Main Methods:

• A two-stage analytic strategy was developed.
• Stage one examines the entire distribution of local area death rates, not just outliers.
• Stage two relates temporal changes in death rate distributions to disease process models.

Main Results:

• The strategy was applied to U.S. county cancer mortality data from 1950-1978.
• Striking mortality patterns were identified across the U.S., regions, and states.
• A key finding was nonmetropolitan respiratory cancer mortality rates catching up to metropolitan rates.

Conclusions:

• The proposed analytic strategy effectively identifies significant mortality patterns.
• The "catching up" of nonmetropolitan respiratory cancer mortality suggests evolving population health dynamics.
• This approach enhances the ability to monitor and understand geographic variations in mortality risks.