Independent Relationship of Changes in Death Rates with Changes in US Presidential Voting

Abstract

Background: The outcome of the 2016 presidential election is commonly attributed to socioeconomic and ethnic/racial issues, but health issues, including "deaths of despair," may also have contributed.

Objective: To assess whether changes in age-adjusted death rates were independently associated with changes in presidential election voting in 2016 vs. 2008.

Design: We used publicly available data in each of 3112 US counties to correlate changes in a county's presidential voting in 2016 compared with 2008 with recent changes in its age-adjusted death rate, after controlling for population and rural-urban status, median age, race/ethnicity, income, education, unemployment rate, and health insurance rate.

Design setting: Cross-sectional analysis of county-specific data.

Setting/participants: All 3112 US counties.

Main measures: The independent correlation of a county's change in age-adjusted death rate between 2000 and 2015 with its net percentage Republican gain or loss in the presidential election of 2016 vs. 2008.

Key results: In 2016, President Trump increased the Republican presidential vote percentage in 83.8% of counties compared with Senator McCain in 2008. Counties with an increased Republican vote percentage in 2016 vs. 2008 had a 15% higher 2015 age-adjusted death rate than counties with an increased Democratic vote percentage. Since 2000, overall death rates declined by less than half as much, and death rates from drugs, alcohol, and suicide increased 2.5 times as much in counties with Republican gains compared with counties with Democratic gains. In multivariable analyses, Republican net presidential gain in 2016 vs. 2008 was independently correlated with slower reductions in a county's age-adjusted death rate. Although correlation cannot infer causality, modest reductions in death rates might theoretically have shifted Pennsylvania, Michigan, and Wisconsin to Secretary Clinton.

Conclusions: Less of a reduction in age-adjusted death rates was an independent correlate of an increased Republican percentage vote in 2016 vs. 2008. Death rates may be markers of dissatisfactions and fears that influenced the 2016 Presidential election outcomes.

Keywords: 2016 election; age-adjusted death rate; deaths of despair; presidential election; rural public health.

Figure 1

Net Republican gain vs. 2015 age-adjusted death rates. The net Republican gain was greater in rural counties with higher age-adjusted death rates.

Figure 2

Age-adjusted death rates in counties by urban–rural classification, 2000–2015. Overall age-adjusted death rates are progressively lower in more populous urban counties than in less populous rural counties, and this gap has widened since 2000. Large central metro and large fringe metro counties have a population > 1,000,000; medium metro counties have a population 250,000–999,999; small metro counties have a population < 250,000; micropolitan counties have a non-metro population 10,000–49,999; and noncore non-metro counties have a population < 50,000. Data sources: 2013 NCHS Urban–Rural Classification Scheme for Counties, available at https://www.cdc.gov/nchs/data/series/sr_02/sr02_166.pdf; and the CDC WONDER database, available at https://wonder.cdc.gov/ucd-icd10.html.