Modeling County-Level Spatio-Temporal Mortality Rates Using Dynamic Linear Models

The lifestyles and backgrounds of individuals across the United States differ widely. Some of these differences are easily measurable (ethnicity, age, income, etc.) while others are not (stress levels, empathy, diet, exercise, etc.). Though every person is unique, individuals living closer together likely have more similar lifestyles than individuals living hundreds of miles apart. Because lifestyle and environmental factors contribute to mortality, spatial correlation may be an important feature in mortality modeling. However, many of the current mortality models fail to account for spatial relationships. This paper introduces spatio-temporal trends into traditional mortality modeling using Bayesian hierarchical models with conditional auto-regressive (CAR) priors. We show that these priors, commonly used for areal data, are appropriate for modeling county-level spatial trends in mortality data covering the contiguous United States. We find that mortality rates of neighboring counties are highly correlated. Additionally, we find that mortality improvement or deterioration trends between neighboring counties are also highly correlated.

Download Full-text

Association between county level cannabis dispensary counts and opioid related mortality rates in the United States: panel data study

BMJ ◽

10.1136/bmj.m4957 ◽

2021 ◽

pp. m4957 ◽

Cited By ~ 1

Author(s):

Greta Hsu ◽

Balázs Kovács

Keyword(s):

United States ◽

Confidence Interval ◽

The United States ◽

Mortality Rates ◽

Mortality Data ◽

Panel Regression ◽

County Level ◽

Regression Methods ◽

Synthetic Opioids ◽

Related Mortality

Abstract Objective To examine county level associations between the prevalence of medical and recreational cannabis stores (referred to as dispensaries) and opioid related mortality rates. Design Panel regression methods. Setting 812 counties in the United States in the 23 states that allowed legal forms of cannabis dispensaries to operate by the end of 2017. Participants The study used US mortality data from the Centers for Disease Control and Prevention combined with US census data and data from Weedmaps.com on storefront dispensary operations. Data were analyzed at the county level by using panel regression methods. Main outcome measure The main outcome measures were the log transformed, age adjusted mortality rates associated with all opioid types combined, and with subcategories of prescription opioids, heroin, and synthetic opioids other than methadone. The associations of medical dispensary and recreational dispensary counts with age adjusted mortality rates were also analyzed. Results County level dispensary count (natural logarithm) is negatively related to the log transformed, age adjusted mortality rate associated with all opioid types (β=−0.17, 95% confidence interval −0.23 to −0.11). According to this estimate, an increase from one to two storefront dispensaries in a county is associated with an estimated 17% reduction in all opioid related mortality rates. Dispensary count has a particularly strong negative association with deaths caused by synthetic opioids other than methadone (β=−0.21, 95% confidence interval −0.27 to −0.14), with an estimated 21% reduction in mortality rates associated with an increase from one to two dispensaries. Similar associations were found for medical versus recreational storefront dispensary counts on synthetic (non-methadone) opioid related mortality rates. Conclusions Higher medical and recreational storefront dispensary counts are associated with reduced opioid related death rates, particularly deaths associated with synthetic opioids such as fentanyl. While the associations documented cannot be assumed to be causal, they suggest a potential association between increased prevalence of medical and recreational cannabis dispensaries and reduced opioid related mortality rates. This study highlights the importance of considering the complex supply side of related drug markets and how this shapes opioid use and misuse.

Download Full-text

Outcomes for Children and Adolescents With Cancer: Challenges for the Twenty-First Century

Journal of Clinical Oncology ◽

10.1200/jco.2009.27.0421 ◽

2010 ◽

Vol 28 (15) ◽

pp. 2625-2634 ◽

Cited By ~ 593

Author(s):

Malcolm A. Smith ◽

Nita L. Seibel ◽

Sean F. Altekruse ◽

Lynn A.G. Ries ◽

Danielle L. Melbert ◽

...

Keyword(s):

United States ◽

Childhood Cancer ◽

Cancer Mortality ◽

Lymphoblastic Leukemia ◽

The United States ◽

Mortality Rates ◽

Mortality Data ◽

Childhood Cancers ◽

Essential Information ◽

The Impact

Purpose This report provides an overview of current childhood cancer statistics to facilitate analysis of the impact of past research discoveries on outcome and provide essential information for prioritizing future research directions. Methods Incidence and survival data for childhood cancers came from the Surveillance, Epidemiology, and End Results 9 (SEER 9) registries, and mortality data were based on deaths in the United States that were reported by states to the Centers for Disease Control and Prevention by underlying cause. Results Childhood cancer incidence rates increased significantly from 1975 through 2006, with increasing rates for acute lymphoblastic leukemia being most notable. Childhood cancer mortality rates declined by more than 50% between 1975 and 2006. For leukemias and lymphomas, significantly decreasing mortality rates were observed throughout the 32-year period, though the rate of decline slowed somewhat after 1998. For remaining childhood cancers, significantly decreasing mortality rates were observed from 1975 to 1996, with stable rates from 1996 through 2006. Increased survival rates were observed for all categories of childhood cancers studied, with the extent and temporal pace of the increases varying by diagnosis. Conclusion When 1975 age-specific death rates for children are used as a baseline, approximately 38,000 childhood malignant cancer deaths were averted in the United States from 1975 through 2006 as a result of more effective treatments identified and applied during this period. Continued success in reducing childhood cancer mortality will require new treatment paradigms building on an increased understanding of the molecular processes that promote growth and survival of specific childhood cancers.

Download Full-text

Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study

Frontiers in Public Health ◽

10.3389/fpubh.2021.766707 ◽

2021 ◽

Vol 9 ◽

Author(s):

Joshua J. Levy ◽

Rebecca M. Lebeaux ◽

Anne G. Hoen ◽

Brock C. Christensen ◽

Louis J. Vaickus ◽

...

Keyword(s):

United States ◽

Deep Learning ◽

Mortality Rate ◽

Satellite Images ◽

The United States ◽

Mortality Rates ◽

Image Features ◽

County Level ◽

Crude Mortality Rate ◽

Crude Mortality

What is the relationship between mortality and satellite images as elucidated through the use of Convolutional Neural Networks?Background: Following a century of increase, life expectancy in the United States has stagnated and begun to decline in recent decades. Using satellite images and street view images, prior work has demonstrated associations of the built environment with income, education, access to care, and health factors such as obesity. However, assessment of learned image feature relationships with variation in crude mortality rate across the United States has been lacking.Objective: We sought to investigate if county-level mortality rates in the U.S. could be predicted from satellite images.Methods: Satellite images of neighborhoods surrounding schools were extracted with the Google Static Maps application programming interface for 430 counties representing ~68.9% of the US population. A convolutional neural network was trained using crude mortality rates for each county in 2015 to predict mortality. Learned image features were interpreted using Shapley Additive Feature Explanations, clustered, and compared to mortality and its associated covariate predictors.Results: Predicted mortality from satellite images in a held-out test set of counties was strongly correlated to the true crude mortality rate (Pearson r = 0.72). Direct prediction of mortality using a deep learning model across a cross-section of 430 U.S. counties identified key features in the environment (e.g., sidewalks, driveways, and hiking trails) associated with lower mortality. Learned image features were clustered, and we identified 10 clusters that were associated with education, income, geographical region, race, and age.Conclusions: The application of deep learning techniques to remotely-sensed features of the built environment can serve as a useful predictor of mortality in the United States. Although we identified features that were largely associated with demographic information, future modeling approaches that directly identify image features associated with health-related outcomes have the potential to inform targeted public health interventions.

Download Full-text

The evolution of infant mortality inequality in the United States, 1960–2016

Science Advances ◽

10.1126/sciadv.aba5908 ◽

2020 ◽

Vol 6 (29) ◽

pp. eaba5908

Author(s):

Nick Turner ◽

Kaveh Danesh ◽

Kelsey Moran

Keyword(s):

United States ◽

Infant Mortality ◽

The United States ◽

Mortality Rates ◽

County Level ◽

Level Data ◽

Antipoverty Programs ◽

The Relationship ◽

Over Time ◽

Mortality Inequality

What is the relationship between infant mortality and poverty in the United States and how has it changed over time? We address this question by analyzing county-level data between 1960 and 2016. Our estimates suggest that level differences in mortality rates between the poorest and least poor counties decreased meaningfully between 1960 and 2000. Nearly three-quarters of the decrease occurred between 1960 and 1980, coincident with the introduction of antipoverty programs and improvements in medical care for infants. We estimate that declining inequality accounts for 18% of the national reduction in infant mortality between 1960 and 2000. However, we also find that level differences between the poorest and least poor counties remained constant between 2000 and 2016, suggesting an important role for policies that improve the health of infants in poor areas.

Download Full-text

Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Neighborhood Features

10.1101/2020.10.12.20211755 ◽

2020 ◽

Author(s):

Joshua J. Levy ◽

Rebecca M. Lebeaux ◽

Anne G. Hoen ◽

Brock C. Christensen ◽

Louis J. Vaickus ◽

...

Keyword(s):

United States ◽

Deep Learning ◽

Mortality Rate ◽

Satellite Images ◽

The United States ◽

Mortality Rates ◽

Image Features ◽

County Level ◽

Crude Mortality Rate ◽

Crude Mortality

AbstractWhat is the relationship between mortality and satellite images as elucidated through the use of Convolutional Neural Networks?BackgroundFollowing a century of increase, life expectancy in the United States has stagnated and begun to decline in recent decades. Using satellite images and street view images, prior work has demonstrated associations of the built environment with income, education, access to care and health factors such as obesity. However, assessment of learned image feature relationships with variation in crude mortality rate across the United States has been lacking. We sought to investigate if county-level mortality rates in the U.S. could be predicted from satellite images.MethodsSatellite images were extracted with the Google Static Maps application programming interface for 430 counties representing approximately 68.9% of the US population. A convolutional neural network was trained using crude mortality rates for each county in 2015 to predict mortality. Learned image features were interpreted using Shapley Additive Feature Explanations, clustered, and compared to mortality and its associated covariate predictors.ResultsPredicted mortality from satellite images in a held-out test set of counties was strongly correlated to the true crude mortality rate (Pearson r=0.72). Learned image features were clustered, and we identified 10 clusters that were associated with education, income, geographical region, race and age. Direct prediction of mortality using a deep learning model across a cross-section of 430 U.S. counties identified key features in the environment (e.g. sidewalks, driveways and hiking trails) associated with lower mortality.ConclusionsThe application of deep learning techniques to remotely-sensed features of the built environment can serve as a useful predictor of mortality in the United States. Although we identified features that were largely associated with demographic information, future modeling approaches that directly identify image features associated with health-related outcomes have the potential to inform targeted public health interventions.

Download Full-text

Trends in heart failure-related cardiovascular mortality in rural versus urban United States counties, 2011–2018: A cross-sectional study

PLoS ONE ◽

10.1371/journal.pone.0246813 ◽

2021 ◽

Vol 16 (3) ◽

pp. e0246813

Author(s):

Jacob B. Pierce ◽

Nilay S. Shah ◽

Lucia C. Petito ◽

Lindsay Pool ◽

Donald M. Lloyd-Jones ◽

...

Keyword(s):

United States ◽

Heart Failure ◽

Cause Of Death ◽

Negative Binomial ◽

The United States ◽

Mortality Rates ◽

County Level ◽

Physician Density ◽

Rural And Urban ◽

Related Mortality

Background Adults in rural counties in the United States (US) experience higher rates broadly of cardiovascular disease (CVD) compared with adults in urban counties. Mortality rates specifically due to heart failure (HF) have increased since 2011, but estimates of heterogeneity at the county-level in HF-related mortality have not been produced. The objectives of this study were 1) to quantify nationwide trends by rural-urban designation and 2) examine county-level factors associated with rural-urban differences in HF-related mortality rates. Methods and findings We queried CDC WONDER to identify HF deaths between 2011–2018 defined as CVD (I00-78) as the underlying cause of death and HF (I50) as a contributing cause of death. First, we calculated national age-adjusted mortality rates (AAMR) and examined trends stratified by rural-urban status (defined using 2013 NCHS Urban-Rural Classification Scheme), age (35–64 and 65–84 years), and race-sex subgroups per year. Second, we combined all deaths from 2011–2018 and estimated incidence rate ratios (IRR) in HF-related mortality for rural versus urban counties using multivariable negative binomial regression models with adjustment for demographic and socioeconomic characteristics, risk factor prevalence, and physician density. Between 2011–2018, 162,314 and 580,305 HF-related deaths occurred in rural and urban counties, respectively. AAMRs were consistently higher for residents in rural compared with urban counties (73.2 [95% CI: 72.2–74.2] vs. 57.2 [56.8–57.6] in 2018, respectively). The highest AAMR was observed in rural Black men (131.1 [123.3–138.9] in 2018) with greatest increases in HF-related mortality in those 35–64 years (+6.1%/year). The rural-urban IRR persisted among both younger (1.10 [1.04–1.16]) and older adults (1.04 [1.02–1.07]) after adjustment for county-level factors. Main limitations included lack of individual-level data and county dropout due to low event rates (<20). Conclusions Differences in county-level factors may account for a significant amount of the observed variation in HF-related mortality between rural and urban counties. Efforts to reduce the rural-urban disparity in HF-related mortality rates will likely require diverse public health and clinical interventions targeting the underlying causes of this disparity.

Download Full-text

Trends in Mortality Due to Cancer in the United States by Age and County-Level Income, 1999–2015

JNCI Journal of the National Cancer Institute ◽

10.1093/jnci/djz123 ◽

2019 ◽

Vol 111 (8) ◽

pp. 863-866 ◽

Cited By ~ 1

Author(s):

Diana R Withrow ◽

Amy Berrington de González ◽

Susan Spillane ◽

Neal D Freedman ◽

Ana F Best ◽

...

Keyword(s):

United States ◽

Cancer Mortality ◽

Death Certificate ◽

The United States ◽

Mortality Rates ◽

Income Quintile ◽

County Level ◽

Younger Adults ◽

Rate Ratios ◽

Over Time

Abstract Disparities in cancer mortality by county-level income have increased. It is unclear whether these widening disparities have affected older and younger adults equally. National death certificate data were utilized to ascertain cancer deaths during 1999–2015. Average annual percent changes in mortality rates and mortality rate ratios (RRs) were estimated by county-level income quintile and age (25–64 vs ≥65 years). Among 25- to 64-year-olds, cancer mortality rates were 30% higher (RR = 1.30, 95% confidence interval [CI] = 1.29 to 1.31) in the lowest-vs the highest-income counties in 1999–2001 and 56% higher (RR = 1.56, 95% CI = 1.55 to 1.57) in 2013–2015; the disparities among those 65 years and older were smaller but also widened over time (RR1999–2001 = 1.04, 95% CI = 1.03 to 1.05; RR2013–2015 = 1.14, 95% CI = 1.13 to 1.14). Widening disparities occurred across cancer sites. If all counties had the mortality rates of the highest-income counties, 21.5% of cancer deaths among 25- to 64-year-olds and 7.3% of cancer deaths in those 65 years and older would have been avoided in 2015. These results highlight an ongoing need for equity-focused interventions, particularly among younger adults.

Download Full-text

Abstract 223: National and Regional Trends in Hospitalizations for Hemorrhagic Stroke after Acute Myocardial Infarction in the United States: 1999-2010

Circulation Cardiovascular Quality and Outcomes ◽

10.1161/circoutcomes.6.suppl_1.a223 ◽

2013 ◽

Vol 6 (suppl_1) ◽

Author(s):

Ruizhi Shi ◽

Yun Wang ◽

Judith H Lichtman ◽

Kumar Dharmarajan ◽

Frederick A Masoudi ◽

...

Keyword(s):

United States ◽

Myocardial Infarction ◽

Acute Myocardial Infarction ◽

Hemorrhagic Stroke ◽

Temporal Trends ◽

The United States ◽

Mortality Rates ◽

Hospitalization Rates ◽

One Year ◽

Stroke Belt

Background: Elderly survivors of acute myocardial infarction (AMI) are at elevated risk for hemorrhagic stroke, which has a mortality rate of approximately 50%. Increasing use of warfarin for arterial fibrillation and anti-platelet agents for AMI combined with an increasing aging population may have influenced the risk of post-AMI strokes. We sought to characterize temporal trends in the risk for and mortality from hemorrhagic stroke over 12 years among older AMI survivors of different age, sex, race, revascularization status, and region within the US. Methods: We used 100% of Medicare inpatient claims data to identify all fee-for-service (FFS) patients aged> 64 years who were hospitalized for AMI in 1999-2010. We excluded patients who died during the hospitalization or were transferred. Revascularization procedures were identified during the index admission. We used a Cox proportional-hazards regression model to estimate the risk-adjusted annual changes in one-year hemorrhagic stroke hospitalization after AMI, overall and by subgroups. Changes were adjusted by age, gender, race, medical history and comorbidities. We calculated the 30-day mortality among patients readmitted for hemorrhagic stroke. Stroke belt regions were defined as the states with high stroke hospitalization rates in the southeast United States. Results: Among 2,433,036 AMI hospitalizations and 4,852 hemorrhagic stroke readmissions, the risk-adjusted one-year post-AMI hemorrhagic stroke rate remained stable from 1999 to 2010 (range, 0.2% to 0.3%). No significant trends were found for post-AMI stroke rates across all age-sex-race groups and all treatment groups (Figure). Thirty-day mortality rates for stroke after AMI did not show significant changes (1999, 46.7%, 95% CI 39.9%-53.7%; 2010, 50.7%, 95% CI 45.3%-56.1%; range: 46.5% to 54.6%). No difference was found in post-AMI hemorrhagic stroke rates between the stroke belt and non-stroke belt regions. Conclusions: From 1999 to 2010, the overall hospitalization rates of hemorrhagic stroke after AMI were relatively stable without significant changes across all subgroups. Thirty-day mortality rates remained largely unchanged over time. Stroke risk in the stroke belt was not found significantly higher comparing with non-stroke belt states.

Download Full-text

Infectious Disease-Related Deaths of Low Birth Weight Infants, United States, 1968 to 1982

PEDIATRICS ◽

10.1542/peds.84.2.296 ◽

1989 ◽

Vol 84 (2) ◽

pp. 296-303

Author(s):

Janine M. Jason

Keyword(s):

Infectious Disease ◽

United States ◽

Birth Weight ◽

Infant Mortality ◽

Low Birth Weight ◽

Infectious Diseases ◽

The United States ◽

Mortality Rates ◽

Mortality Data ◽

Tract Infections

Infant mortality rates in the United States are higher than in any other developed country. Low birth weight (LBW) is the primary determinant of infant mortality. Despite city, state, and federal programs to prevent LBW, decreases in infant mortality in the 1980s appear to be largely secondary to improved survival of LBW infants rather than to a decline in the rate of LBW births. Because prevention of mortality due to infectious disease is feasible, it was of interest to examine the role of infectious diseases in LBW infant mortality. US vital statistics mortality data for 1968 through 1982 were analyzed in terms of LBW infant mortality associated with infectious and noninfectious diseases. These analyses indicated that the rates of infectious disease-associated early neonatal and postneonatal LBW mortality increased during this time; late neonatal rates did not decline appreciably. Infectious diseases were associated with 4% of all LBW infant deaths in 1968; this had increased to 10% by 1982. Although LBW infant mortality rates associated with noninfectious diseases did not differ for white and black populations, infectious disease-associated mortality rates were consistently higher for blacks than whites in both metropolitan and nonmetropolitan areas. Chorioamnionitis was involved in 28% of infectious disease-associated early neonatal LBW deaths. Sepsis was an increasingly listed cause of death in all infant age periods, whereas respiratory tract infections were decreasingly listed. Necrotizing enterocolitis increased as a cause of late neonatal mortality. These data suggest that infectious diseases are an increasing cause of LBW infant mortality and these deaths occur more frequently in the black population targeted by prevention programs. More research concerning specific causes and prevention of infections in the LBW infant may help reduce US infant mortality.

Download Full-text

Lifestyle Effects on the Risk of Transmission of COVID-19 in the United States: Evaluation of Market Segmentation Systems

International Journal of Environmental Research and Public Health ◽

10.3390/ijerph18094826 ◽

2021 ◽

Vol 18 (9) ◽

pp. 4826

Author(s):

Esra Ozdenerol ◽

Jacob Seboly

Keyword(s):

United States ◽

The United States ◽

Mortality Rates ◽

Incidence Rates ◽

Statistical Testing ◽

Average Risk ◽

County Level ◽

Household Level ◽

The Impact ◽

The U.S

The aim of this study was to associate lifestyle characteristics with COVID-19 infection and mortality rates at the U.S. county level and sequentially map the impact of COVID-19 on different lifestyle segments. We used analysis of variance (ANOVA) statistical testing to determine whether there is any correlation between COVID-19 infection and mortality rates and lifestyles. We used ESRI Tapestry LifeModes data that are collected at the U.S. household level through geodemographic segmentation typically used for marketing purposes to identify consumers’ lifestyles and preferences. According to the ANOVA analysis, a significant association between COVID-19 deaths and LifeModes emerged on 1 April 2020 and was sustained until 30 June 2020. Analysis of means (ANOM) was also performed to determine which LifeModes have incidence rates that are significantly above/below the overall mean incidence rate. We sequentially mapped and graphically illustrated when and where each LifeMode had above/below average risk for COVID-19 infection/death on specific dates. A strong northwest-to-south and northeast-to-south gradient of COVID-19 incidence was identified, facilitating an empirical classification of the United States into several epidemic subregions based on household lifestyle characteristics. Our approach correlating lifestyle characteristics to COVID-19 infection and mortality rate at the U.S. county level provided unique insights into where and when COVID-19 impacted different households. The results suggest that prevention and control policies can be implemented to those specific households exhibiting spatial and temporal pattern of high risk.

Download Full-text