County-level characteristics associated with variation in end-stage kidney disease mortality in the United States, 2010-2018 Open Access
Snow, Kylie (Spring 2021)
Background: Once diagnosed with end-stage kidney disease (ESKD), mortality risk is high. Individual-level factors known to be associated with increased mortality in this patient population include age, race, gender, and multiple comorbidities. However, few studies have examined county-level characteristics associated with ESKD mortality. Therefore, using a national registry of ESKD patients receiving kidney replacement therapy (KRT) in the United States, we examined county-level variation in ESKD mortality and identified county-level characteristics associated with this variation.
Methods: Using the United States Renal Data System, we estimated county-level age-standardized mortality rates (ASMR) among all adults (N=1,516,742, aged 18-84) across 2,807 counties with ESKD receiving KRT between 2010 and 2018. County-level ASMRs were linked to county-level demographic (e.g., % female), socioeconomic (e.g., % unemployed), health care (e.g., % without health insurance), and health behavior (e.g., % current smokers) characteristics from publicly available census and survey data. Hierarchical linear mixed models, with a random intercept for state, identified county-level characteristics associated with ESKD-related ASMRs and quantified the percentage of variation explained by county-level characteristics.
Results: County-level ESKD-related ASMRs ranged from 25 to 509 per 1,000 person-years (PY). ASMRs were highest in counties located in the Tennessee Valley and Appalachia regions, and lowest in counties located in New England, the upper Midwest, and in some regions along the West coast. Overall, county-level characteristics explained 20% of variation in ESKD-related ASMRs. In fully adjusted models, a lower percentage of Black people (-5.33 deaths/1,000PY), higher health care expenditures (3.76 deaths/1,000PY), and lower percentage of people who drink excessively (-2.99 deaths/1,000PY) were significantly associated with ESKD mortality.
Conclusions: In this study, we show substantial variation in ESKD mortality by county, and find that approximately 20% of this variation is explained by county-level characteristics. Given that a large proportion of county-level variation (~80%) is unexplained, targeted vs. population-wide interventions may play a more important role in reducing ESKD mortality.
Table of Contents
Epidemiology of end-stage kidney disease in the United States
Geographic variation in end-stage kidney disease in the United States
Factors associated with geographic variation in end-stage kidney disease
Comparison to previous literature
Public health implications
Strengths and limitations
Tables and Figures
Table 1. Definitions of county-level characteristics and data sources
Table 2. Summary of U.S. county-level characteristics among 2,807 counties included in this analysis
Table 3. Associations of U.S. county-level characteristics with end-stage kidney disease (ESKD) mortality, 2010-2018
Figure 1. Trends in age, race, and sex adjusted end-stage kidney disease (ESKD) incidence, 1990-2018.
Figure 2. Flowchart of study cohort
Figure 3. Age-standardized end-stage kidney disease (ESKD) mortality rates across 2,807 counties in the United States, 2010-2018
Figure 4. State-level intercepts for age-standardized end-stage kidney disease (ESKD) mortality rates before and after adjustment for county-level characteristics in 2,807 counties in the United States, 2010-2018
Table S1. Variance inflation factors for county characteristics
Table S2. Correlation matrix for county characteristics
About this Master's Thesis
|Subfield / Discipline|
|Committee Chair / Thesis Advisor|
|County-level characteristics associated with variation in end-stage kidney disease mortality in the United States, 2010-2018 ()||2021-04-28 20:04:22 -0400||