Spatial and temporal variation of mortality and deprivation 2: statistical modelling

Area of Science:

• Public Health
• Biostatistics
• Sociology

Background:

• Tabular analyses are common in medical literature for examining mortality and deprivation.
• A formal statistical approach is needed to analyze the dynamic relationship between mortality and deprivation over time.

Purpose of the Study:

• To apply generalized linear models for a formal statistical analysis of mortality-deprivation relationships.
• To investigate the changes in mortality-deprivation relationships over time using longitudinal data.
• To explore nonlinear effects of deprivation on mortality and connect them to tabular methods.

Main Methods:

• Utilized generalized linear models, specifically Poisson and logit fixed-effects models.
• Employed ward-level data from Wales, consistent with prior tabular analyses.
• Incorporated dummy variables for deprivation categories to model nonlinear effects.

Main Results:

• Estimated cross-sectional and repeated-measures Poisson models to analyze mortality-deprivation at single time points and over time.
• Estimated logit models to specifically focus on temporal changes in these relationships.
• Established a connection between formal statistical models and tabular approaches for analyzing nonlinear deprivation effects.

Conclusions:

• Generalized linear models provide a robust statistical framework for analyzing mortality-deprivation dynamics.
• The study quantifies temporal changes in mortality-deprivation relationships, advancing understanding of health inequalities.
• The findings bridge formal statistical modeling with established tabular methods for analyzing deprivation impacts on health outcomes.