Modeling Age-Adjusted Mortality in the United States
Brandon Dunbar, Paramahansa Pramanik, Haley Kate Robinson
TL;DR
This study assesses how total mortality relates to age-adjusted death rates ($AADR$) in the United States by analyzing NCHS data for the ten leading causes of death during 2012–2017. It combines descriptive visualizations with a log-linear regression framework, $\log(\text{Total Deaths}) = \hat{\beta}_0 + \hat{\beta}_1 \cdot \text{AADR} + \epsilon$, to quantify how standardized mortality reflects underlying risk beyond demographic aging. The analysis reveals a strong, positive association between crude deaths and $AADR$, while age adjustment exposes nuanced, cause-specific patterns and regional structure via empirical copulas and hierarchical clustering. The work demonstrates the methodological importance of demographic standardization in public health surveillance and points to future work incorporating morbidity, longer horizons, and multilevel models to better inform policy.
Abstract
This research explores how total mortality figures relate to age-standardized death rates within the United States, using the complete historical record of national mortality statistics. Through a detailed investigation of both all-cause and cause-specific mortality trends, the study evaluates the impact of demographic standardization on interpreting mortality data across different time periods and geographic regions. Results indicate a robust and persistent association between crude death totals and age-adjusted rates. However, the findings also demonstrate that without adjusting for age, comparisons over time or across locations may misrepresent underlying epidemiological shifts, largely due to evolving population age structures. The study underscores the critical role of age adjustment as a methodological tool for generating accurate, interpretable, and comparable measures of public health outcomes.