Recent Papers on Mortality that use Spatial Methods

Ashley, P. and A. Tim. 2011. Neighborhood disparities in stroke and myocardial infarction mortality: A GIS and spatial scan statistics approach. BMC Public Health 11: 644656.

The purpose of this study was to investigate the spatial patterns and detect local neighborhood clusters of high risk of stroke and myocardial infarction mortality in 11 counties of the east Tennessee Appalachian region. In order to do this, a series of methods were utilized: (1) the authors use the Spatial Empirical Bayesian technique, which allows for the adjustment of spatial autocorrelation, population heterogeneity, and variance instability of the standardized risks of stroke and myocardial infarction. (2) spatial scan statistics were used to detect the presence of high risk stroke and myocardial infarction clusters among neighborhoods. (3) logistic regression models were used to investigate potential associations between the neighborhood being in a high risk stroke or myocardial infarction cluster and neighborhood level socioeconomic and demographic characteristics. The results showed spatial clusters of high mortality risk at the neighborhood level, indicating disparities in risk of death from stroke and myocardial infarction within counties in the east Tennessee Appalachian region.

Barufi, A.M., E. Haddad, and A. Paez. 2012. Infant mortality in Brazil, 19802000: A spatial panel data analysis. BMC Public Health 12(1): 181195.

In this paper, the authors investigate the factors that are associated with the infant mortality rate in Brazil from 1980 to 2000. Socioeconomic, infrastructure, and demographic data at the municipal-level is analyzed using four different models: a pooled model, a panel model with fixed spatial effects, a spatial error model, and a spatial lag model. The authors find that the panel model with spatial effects provides the best fit to the data. The results of this study show that the health care infrastructure and social policy measures are associated with reduced rates of infant mortality. In addition, there are regional benefits beyond the municipal-level as indicated by the spillover effects associated with health infrastructure and water and sanitation facilities.

Chen, V.Y.J., W.S. Deng, T.C. Yang, and S.A. Matthews. 2012. Geographically weighted quantile regression (GWQR): An application to U.S. mortality data. Geographical Analysis 44(2): 134150.

This paper uses geographically weighted quantile regression (GWQR), a method developed by the authors, which is an approach that allows for the analysis of spatially nonstationary relationships between the explanatory variables and the mortality rate within the framework of quantile regression. Specifically, the authors explore whether spatial nonstationarity is stable across the distribution of the mortality rate in US counties. The results show that the relationship between the mortality rate and its predictors vary spatially and that they simultaneously vary across the distribution of the mortality rate.

Chen, V.Y.J., P.C. Wu, T.C. Yang, and H.J. Su. 2010. Examining the non-stationary effects of social determinants on cardiovascular mortality after cold surges in Taiwan. Science of the Total Environment 408(9): 20422049.

This study analyzes the non-stationary effects of various social determinants on cardiovascular mortality after the cold surges in Taiwan using geographically weighted Poisson regression. This is a local spatial statistical modeling approach that allows for geographically smooth varying parameters. The results showed that right after the cold surges, on average, an immediate increase in cardiovascular mortality occurred, and the township-level social determinants (e.g., social disadvantage, stability, sensitive group, and rurality) vary across space for the post-cold-surge cardiovascular mortality in Taiwan.

Chin, B., L. Montana, and X. Basagaña. 2011. Spatial modeling of geographic inequalities in infant and child mortality across Nepal. Health & Place 17(4): 929936.

This paper uses data from the Nepal Demographic and Health Survey to explain the spatial pattern of child mortality in Nepal. The authors apply a flexible hierarchical survival model that allows for spatially correlated random effects so that they can account for potential spatial autocorrelation in the covariates at the community level. The results show that both mother and household level covariates are significant predictors of both infant and child mortality. However, after holding the mother and household characteristics constant, a significant spatial trend remains for infant mortality with increased mortality in Nepal’s Far-Western and Mid-Western development regions.

Gebreab, S.Y.and A.V. Diez Roux. 2012. Exploring racial disparities in CHD mortality between blacks and whites across the United States: A geographically weighted regression approach. Health & Place 18: 10061014.

This paper explores spatial heterogeneity in black-white differences in coronary heart disease (CHD) mortality across the US using geographically weighted regression (GWR). The GWR results showed significant spatial heterogeneity in black-white differences in CHD mortality. However, after controlling for county and race-specific poverty and segregation, significant race differences in CHD mortality were no longer present.

Hendryx, M., J. Conley, E. Fedorko, J. Luo, and M. Armistead. 2012. Permitted water pollution discharges and population cancer and non-cancer mortality: Toxicity weights and upstream discharge effects in us rural-urban areas. International Journal of Health Geographics 11(1): 923.

This is an exploratory paper that uses geographically weighted regression to examine whether: (1) greater amounts of permitted toxic chemical pollutants in surface water are associated with poorer population health, (2) there is evidence for pollution discharges affecting population health downstream from its source, and (3) these associations work differently across rural and urban environments. The results show that the effects for kidney mortality and total non-cancer mortality were stronger in rural areas than they were in urban areas. The geographically weighted regression results showed that the effects of both the chemical discharges and the covariates are not constant across space, and that the relative influence of chemical surface water discharges is small compared to the effects of covariates such as poverty or smoking rates. This shows how some rural areas may be impacted by upstream urban discharge.

James, W.L. and J.R. Porter. 2012. Inequality, health infrastructure, and spatial context: Understanding pathways to variations in the causal determinants of race-specific mortality rates. Sociological Spectrum 32(4): 322345.

This paper uses exploratory spatial data analysis (e.g., Moran’s I statistic and local indicator of spatial association (LISA) statistic) and spatially weighted path analysis to understand the mediating relationships of economic and social inequality, health infrastructure, and mortality. The spatially weighted path analysis approach allows for the identification of direct and mediating factors of inequality both through one another and through the investment in the local health infrastructure. The results show that while both economic and social inequality predicts mortality directly and indirectly; income inequality is a stronger determinant of mortality compared to segregation for both blacks and whites, regardless of the health infrastructure that is present.

Pedigo, A., T. Aldrich, and A. Odoi. 2011. Neighborhood disparities in stroke and myocardial infarction mortality: A gis and spatial scan statistics approach. BMC Public Health 11: 644656.

The purpose of this study was to investigate spatial patterns and detect local neighborhood clusters of high risk of stroke and myocardial infarction mortality in the East Tennessee Appalachian Region. For this investigation, a number of spatial techniques were used. In order to adjust for high variances due to the small number problem, spatial autocorrelation, and population heterogeneity, the raw age-adjusted risks were smoothed using Spatial Empirical Bayes in GeoDa. The authors went on to calculate the spatial scan statistic in SaTScan in order to detect the presence of high risk stroke and myocardial infarction clusters and identify their locations among neighborhoods. The results of this study showed that the risk of stroke and myocardial infarction mortality can be very high in some neighborhoods and that these risks can be missed in county-level analyses.

Sparks, J.P. and C.S. Sparks. 2010. An application of spatially autoregressive models to the study of us county mortality rates. Population, Space and Place 16(6):465481.

In order to control for the social and economic conditions that often affect mortality rates as well as the effects of the spatial structure of US counties, the authors apply spatially autoregressive models. The spatial lag model takes into account the spatial association between the mortality rate and the independent variables and the spatial error model takes into account the autocorrelation in the model residuals. The results show that after controlling for spatial structure in the data, several key variables become insignificant in the analysis. In addition, the authors conclude that the spatial pattern is largely the result of the existing autocorrelation among the omitted variables in the empirical model.

Sparks, J.P., C.S. Sparks, and J.J.A. Campbell. 2012. An application of Bayesian spatial statistical methods to the study of racial and poverty segregation and infant mortality rates in the U.S. GeoJournal: 117.

This paper uses two methods of spatial statistical analysis to compare the effects of both racial and poverty segregation on infant mortality risk in US counties. First, exploratory spatial data analysis was conducted by mapping the geographic distribution of the variables and calculating the Moran’s I statistics. Next, Bayesian Hierarchical regression models were estimated to examine the variation in associations between racial and poverty residential segregation and the county infant mortality risk. The results showed that when blacks live in close proximity to each other the infant mortality rate increases and when poor populations live in close proximity to each other the infant mortality rate is higher. However, when blacks interact with whites and poor individuals interact with non-poor individuals this decreases the infant mortality rate.

Yang, T.C., L. Jensen, and M. Haran. 2011. Social capital and human mortality: Explaining the rural paradox with county-level mortality data. Rural Sociology 76(3): 347374.

This study explores the rural paradox (i.e., the finding that standardized mortality rates are unexpectedly low in rural areas despite the economic and infrastructural disadvantages that are present there) by considering an underexplored predictor (social capital) and incorporating a spatial approach into the analysis. The authors examine whether social capital can be used as an explanation for the rural paradox by implementing a series of regression models (e.g., first-order spatial autoregressive, ordinary least squares, spatial autoregressive, spatial-error, and general spatial models). The spatial analysis improves the analytic results and the general spatial model had the best model fit. The authors find that the rural paradox can be partially explained by social capital and that a spatial perspective can further minimize the residential mortality differential.

by Carla Shoff, PhD
Research Associate
Population Research Institute
The Pennsylvania State University