Small-area racial disparity in stroke mortality: an application of bayesian spatial hierarchical modeling

Residential exposure to petroleum refining and stroke in the southern United States

BACKGROUND

The southern United States (U.S.) sustains a disproportionate burden of incident stroke and associated mortality, compared to other parts of the U.S. A large proportion of this risk remains unexplained. Petroleum production and refining (PPR) is concentrated within this region and emits multiple pollutants implicated in stroke pathogenesis. The relationship between residential PPR exposure and stroke has not been studied.

OBJECTIVE

We aimed to investigate the census tract-level association between residential PPR exposure and stroke prevalence for adults (≥18 years) in seven southern U.S. states in 2018.

METHODS

We conducted spatial distance- and generalized propensity score-matched analysis that adjusts for sociodemographic factors, smoking, and unmeasured spatial confounding. PPR was measured as inverse-distance weighted averages of petroleum production within 2.5km or 5km from refineries, which was strongly correlated with measured levels of sulfur dioxide, a byproduct of PPR.

RESULTS

The prevalence of self-reported stroke ranged from 0.4% to 12.7% for all the census tracts of the seven states. People with low socioeconomic status and of Hispanic ethnicity resided closer to petroleum refineries. The non-Hispanic Black population was exposed to higher PPR, while the non-Hispanic White population was exposed to lower PPR. Residential PPR exposure was significantly associated with stroke prevalence. One standard deviation increase in PPR within 5km from refineries was associated with 0.22 (95% confidence interval: 0.09, 0.34) percentage point increase in stroke prevalence. PPR explained 5.6% (2.4, 8.9) of stroke prevalence in the exposed areas. These values differed by states: 1.1% (0.5, 1.7) in Alabama to 11.7% (4.9, 18.6) in Mississippi, and by census tract-level: 0.08% (0.03, 0.13) to 25.3% (10.6, 40.0).

CONCLUSIONS

PPR is associated with self-reported stroke prevalence, suggesting possible links between pollutants emitted from refineries and stroke. The increased prevalence due to PPR may differ by sociodemographic factors.

How Socio-Environmental Factors Are Associated with Japanese Encephalitis in Shaanxi, China-A Bayesian Spatial Analysis

Evidence indicated that socio-environmental factors were associated with occurrence of Japanese encephalitis (JE). This study explored the association of climate and socioeconomic factors with JE (2006–2014) in Shaanxi, China. JE data at the county level in Shaanxi were supplied by Shaanxi Center for Disease Control and Prevention. Population and socioeconomic data were obtained from the China Population Census in 2010 and statistical yearbooks. Meteorological data were acquired from the China Meteorological Administration. A Bayesian conditional autoregressive model was used to examine the association of meteorological and socioeconomic factors with JE. A total of 1197 JE cases were included in this study. Urbanization rate was inversely associated with JE incidence during the whole study period. Meteorological variables were significantly associated with JE incidence between 2012 and 2014. The excessive precipitation at lag of 1–2 months in the north of Shaanxi in June 2013 had an impact on the increase of local JE incidence. The spatial residual variations indicated that the whole study area had more stable risk (0.80–1.19 across all the counties) between 2012 and 2014 than earlier years. Public health interventions need to be implemented to reduce JE incidence, especially in rural areas and after extreme weather.

Reporting of Bayesian analysis in epidemiologic research should become more transparent

OBJECTIVES

The objective of this systematic review is to investigate the use of Bayesian data analysis in epidemiology in the past decade and particularly to evaluate the quality of research papers reporting the results of these analyses.

STUDY DESIGN AND SETTING

Complete volumes of five major epidemiological journals in the period 2005-2015 were searched via PubMed. In addition, we performed an extensive within-manuscript search using a specialized Java application. Details of reporting on Bayesian statistics were examined in the original research papers with primary Bayesian data analyses.

RESULTS

The number of studies in which Bayesian techniques were used for primary data analysis remains constant over the years. Though many authors presented thorough descriptions of the analyses they performed and the results they obtained, several reports presented incomplete method sections and even some incomplete result sections. Especially, information on the process of prior elicitation, specification, and evaluation was often lacking.

CONCLUSION

Though available guidance papers concerned with reporting of Bayesian analyses emphasize the importance of transparent prior specification, the results obtained in this systematic review show that these guidance papers are often not used. Additional efforts should be made to increase the awareness of the existence and importance of these checklists to overcome the controversy with respect to the use of Bayesian techniques. The reporting quality in epidemiological literature could be improved by updating existing guidelines on the reporting of frequentist analyses to address issues that are important for Bayesian data analyses.

Does Place of Residence or Time of Year Affect the Risk of Stroke Hospitalization and Death? A Descriptive Spatial and Temporal Epidemiologic Study

Background

Identifying geographic areas with significantly high risks of stroke is important for informing public health prevention and control efforts. The objective of this study was to investigate geographic and temporal patterns of stroke hospitalization and mortality risks so as to identify areas and seasons with significantly high burden of the disease in Florida. The information obtained will be useful for resource allocation for disease prevention and control.

Methods

Stroke hospitalization and mortality data from 1992 to 2012 were obtained from the Florida Agency for Health Care Administration. Age-adjusted stroke hospitalization and mortality risks for time periods 1992–94, 1995–97, 1998–2000, 2001–03, 2004–06, 2007–09 and 2010–12 were computed at the county spatial scale. Global Moran’s I statistics were computed for each of the time periods to test for evidence of global spatial clustering. Local Moran indicators of spatial association (LISA) were also computed to identify local areas with significantly high risks.

Results

There were approximately 1.5 million stroke hospitalizations and over 196,000 stroke deaths during the study period. Based on global Moran’s I tests, there was evidence of significant (p<0.05) global spatial clustering of stroke mortality risks but no evidence (p>0.05) of significant global clustering of stroke hospitalization risks. However, LISA showed evidence of local spatial clusters of both hospitalization and mortality risks with significantly high risks being observed in the north while the south had significantly low risks of stroke deaths. There were decreasing temporal trends and seasonal patterns of both hospitalization and mortality risks with peaks in the winter.

Conclusions

Although stroke hospitalization and mortality risks have declined in the past two decades, disparities continue to exist across Florida and it is evident from the results of this study that north Florida may, in fact, be part of the stroke belt despite not being in any of the traditional stroke belt states. These findings are useful for guiding public health efforts to reduce/eliminate inequities in stroke outcomes and inform policy decisions. There is need to continually identify populations with significantly high risks of stroke to better guide the targeting of limited resources to the highest risk populations.

An examination of disparities in cancer incidence in Texas using Bayesian random coefficient models

Disparities in cancer risk exist between ethnic groups in the United States. These disparities often result from differential access to healthcare, differences in socioeconomic status and differential exposure to carcinogens. This study uses cancer incidence data from the population based Texas Cancer Registry to investigate the disparities in digestive and respiratory cancers from 2000 to 2008. A Bayesian hierarchical regression approach is used. All models are fit using the INLA method of Bayesian model estimation. Specifically, a spatially varying coefficient model of the disparity between Hispanic and Non-Hispanic incidence is used. Results suggest that a spatio-temporal heterogeneity model best accounts for the observed Hispanic disparity in cancer risk. Overall, there is a significant disadvantage for the Hispanic population of Texas with respect to both of these cancers, and this disparity varies significantly over space. The greatest disparities between Hispanics and Non-Hispanics in digestive and respiratory cancers occur in eastern Texas, with patterns emerging as early as 2000 and continuing until 2008.

Modelling changes in small area disability free life expectancy: trends in London wards between 2001 and 2011

Existing analyses of trends in disability free life expectancy (DFLE) are mainly at aggregate level (national or broad regional). However, major differences in DFLE, and trends in these expectancies, exist between different neighbourhoods within regions, so supporting a small area perspective. However, this raises issues regarding the stability of conventional life table estimation methods at small area scales. This paper advocates a Bayesian borrowing strength technique to model trends in mortality and disability differences across 625 small areas in London, using illness data from the 2001 and 2011 population Censuses, and deaths data for two periods centred on the Census years. From this analysis, estimates of total life expectancy and DFLE are obtained. The spatio-temporal modelling perspective allows assessment of whether significant compression or expansion of morbidity has occurred in each small area. Appropriate models involve random effects that recognise correlation and interaction effects over relevant dimensions of the observed deaths and illness data (areas, ages), as well as major spatial trends (e.g. gradients in health and mortality according to area deprivation category). Whilst borrowing strength is a primary consideration (and demonstrated by raised precision for estimated life expectancies), so also is model parsimony. Therefore, pure borrowing strength models are compared with models allowing selection of random age-area interaction effects using a spike-slab prior, and in fact borrowing strength combined with random effects selection provides better fit.

Socio-environmental drivers and suicide in Australia: Bayesian spatial analysis

BACKGROUND

The impact of socio-environmental factors on suicide has been examined in many studies. Few of them, however, have explored these associations from a spatial perspective, especially in assessing the association between meteorological factors and suicide. This study examined the association of meteorological and socio-demographic factors with suicide across small areas over different time periods.

METHODS

Suicide, population and socio-demographic data (e.g., population of Aboriginal and Torres Strait Islanders (ATSI), and unemployment rate (UNE) at the Local Government Area (LGA) level were obtained from the Australian Bureau of Statistics for the period of 1986 to 2005. Information on meteorological factors (rainfall, temperature and humidity) was supplied by Australian Bureau of Meteorology. A Bayesian Conditional Autoregressive (CAR) Model was applied to explore the association of socio-demographic and meteorological factors with suicide across LGAs.

RESULTS

In Model I (socio-demographic factors), proportion of ATSI and UNE were positively associated with suicide from 1996 to 2000 (Relative Risk (RR)ATSI = 1.0107, 95% Credible Interval (CI): 1.0062-1.0151; RRUNE = 1.0187, 95% CI: 1.0060-1.0315), and from 2001 to 2005 (RRATSI = 1.0126, 95% CI: 1.0076-1.0176; RRUNE = 1.0198, 95% CI: 1.0041-1.0354). Socio-Economic Index for Area (SEIFA) and IND, however, had negative associations with suicide between 1986 and 1990 (RRSEIFA = 0.9983, 95% CI: 0.9971-0.9995; RRATSI = 0.9914, 95% CI: 0.9848-0.9980). Model II (meteorological factors): a 1°C higher yearly mean temperature across LGAs increased the suicide rate by an average by 2.27% (95% CI: 0.73%, 3.82%) in 1996-2000, and 3.24% (95% CI: 1.26%, 5.21%) in 2001-2005. The associations between socio-demographic factors and suicide in Model III (socio-demographic and meteorological factors) were similar to those in Model I; but, there is no substantive association between climate and suicide in Model III.

CONCLUSIONS

Proportion of Aboriginal and Torres Strait Islanders, unemployment and temperature appeared to be statistically associated with of suicide incidence across LGAs among all selected variables, especially in recent years. The results indicated that socio-demographic factors played more important roles than meteorological factors in the spatial pattern of suicide incidence.

Metropolitan racial residential segregation and cardiovascular mortality: exploring pathways

Racial residential segregation has been associated with an increased risk for heart disease and stroke deaths. However, there has been little research into the role that candidate mediating pathways may play in the relationship between segregation and heart disease or stroke deaths. In this study, we examined the relationship between metropolitan statistical area (MSA)-level segregation and heart disease and stroke mortality rates, by age and race, and also estimated the effects of various educational, economic, social, and health-care indicators (which we refer to as pathways) on this relationship. We used Poisson mixed models to assess the relationship between the isolation index in 265 U.S. MSAs and county-level (heart disease, stroke) mortality rates. All models were stratified by race (non-Hispanic black, non-Hispanic white), age group (35-64 years, ≥ 65 years), and cause of death (heart disease, stroke). We included each potential pathway in the model separately to evaluate its effect on the segregation-mortality association. Among blacks, segregation was positively associated with heart disease mortality rates in both age groups but only with stroke mortality rates in the older age group. Among whites, segregation was marginally associated with heart disease mortality rates in the younger age group and was positively associated with heart disease mortality rates in the older age group. Three of the potential pathways we explored attenuated relationships between segregation and mortality rates among both blacks and whites: percentage of female-headed households, percentage of residents living in poverty, and median household income. Because the percentage of female-headed households can be seen as a proxy for the extent of social disorganization, our finding that it has the greatest attenuating effect on the relationship between racial segregation and heart disease and stroke mortality rates suggests that social disorganization may play a strong role in the elevated rates of heart disease and stroke found in racially segregated metropolitan areas.

Multivariate Bayesian spatial model of preterm birth and cardiovascular disease among Georgia women: Evidence for life course social determinants of health

BACKGROUND

There is epidemiologic evidence that women who experience preterm birth (PTB) are at elevated risk for cardiovascular disease (CVD) later in life. Each outcome independently has noted spatial and socioeconomic gradients; we test for spatial structure in the population correlation of the two.

METHODS

Exploratory spatial data analysis and multivariate Bayesian spatial models were fit to describe the spatial correlation of PTB with CVD among women in Georgia counties from 2002 to 2006.

RESULTS

Global Moran's I and local-indicators of spatial association statistics suggest significant co-occurrence of CVD and PTB. Bayesian posterior estimates for multivariate correlation of these outcomes range from r=0.11-0.34 for CVD and PTB. Significant spatial correlation persists with control for county covariates among whites but not blacks.

CONCLUSION

Modest evidence for spatial structure of the ecologic correlation of PTB and women's CVD is consistent with a lifecourse perspective on socially clustered determinants of health.

A statistical procedure to create a neighborhood socioeconomic index for health inequalities analysis

Introduction

In order to study social health inequalities, contextual (or ecologic) data may constitute an appropriate alternative to individual socioeconomic characteristics. Indices can be used to summarize the multiple dimensions of the neighborhood socioeconomic status. This work proposes a statistical procedure to create a neighborhood socioeconomic index.

Methods

The study setting is composed of three French urban areas. Socioeconomic data at the census block scale come from the 1999 census. Successive principal components analyses are used to select variables and create the index. Both metropolitan area-specific and global indices are tested and compared. Socioeconomic categories are drawn with hierarchical clustering as a reference to determine “optimal” thresholds able to create categories along a one-dimensional index.

Results

Among the twenty variables finally selected in the index, 15 are common to the three metropolitan areas. The index explains at least 57% of the variance of these variables in each metropolitan area, with a contribution of more than 80% of the 15 common variables.

Conclusions

The proposed procedure is statistically justified and robust. It can be applied to multiple geographical areas or socioeconomic variables and provides meaningful information to public health bodies. We highlight the importance of the classification method. We propose an R package in order to use this procedure.

Health service deficits and school-aged children with asthma: a population-based study using data from the 2007-2008 National Survey of Child Health

BACKGROUND

Asthma is one of the most common and costly illnesses of childhood. This study addresses health services deficits experienced by school-aged children with asthma.

METHODS

Analyzing data from the 2007-2008 National Survey of Child Health, this cross-sectional study used household income, race/ethnicity, and geographic residency as the primary independent variables and health service deficits as the dependent variable.

RESULTS

Multivariate analysis yielded that other/multiracial (odds ratio [OR], 1.234; 95% confidence interval [CI], 1.226-1.242) and Hispanic (OR, 2.207; 95% CI, 1.226-1.242) school-aged children with asthma had greater odds of having health services deficits as did both urban (OR, 1.106; 95% CI, 1.099-1.113) and rural (OR, 1.133; 95% CI, 1.124-1.142) school-aged children with asthma. Children with either moderate (OR, 1.195; 95% CI, 1.184-1.207) or mild (OR, 1.445; 95% CI, 1.431-1.459) asthma had greater odds of having a health services deficit than those with severe asthma. Low-income school-aged children with asthma had greater odds of having a health services deficit than high-income children (OR, 1.031; 95% CI, 1.026-1.036). At lesser odds of having a health service deficit were those who were African American, of middle-range income, male, or who were school-aged children with asthma in good to excellent health.

CONCLUSION

Both African American and other/multiracial school-aged children were at greater risk of having asthma than either Caucasian or Hispanic children. Three vulnerable subgroups of school-aged children with asthma-rural, Hispanic, and those of low income were the most likely to have health service deficits.

Exploring racial disparities in CHD mortality between blacks and whites across the United States: a geographically weighted regression approach

Coronary heart disease (CHD) mortality is one of the major contributors to racial disparities in health in the United States (US). We examined spatial heterogeneity in black-white differences in CHD mortality across the US and assessed the contributions of poverty and segregation. We used county-level, age-adjusted CHD mortality rates for blacks and whites in the continental US between 1996 and 2006. Geographically weighted regression was employed to assess spatial heterogeneity. There was significant spatial heterogeneity in black-white differences in CHD mortality (median black-white difference 17.7 per 100,000, 25th-75th percentile (IQR): 4.0, 34.0, P value for spatial non-stationarity <0.0001) before controlling for poverty and segregation. This heterogeneity was no longer present after accounting for county differences in race-specific poverty and segregation and interactions of these variables with race (median black-white difference -13.5 per 100,000, IQR: -41.3, 15.7,P value for spatial non-stationarity=0.4346). The results demonstrate the importance of spatial heterogeneity in understanding and eliminating racial disparities in CHD mortality. Additional research to identify the individual and contextual factors that explain the local variations in racial disparities is warranted.

A review of spatial methods in epidemiology, 2000-2010

Understanding the impact of place on health is a key element of epidemiologic investigation, and numerous tools are being employed for analysis of spatial health-related data. This review documents the huge growth in spatial epidemiology, summarizes the tools that have been employed, and provides in-depth discussion of several methods. Relevant research articles for 2000-2010 from seven epidemiology journals were included if the study utilized a spatial analysis method in primary analysis (n = 207). Results summarized frequency of spatial methods and substantive focus; graphs explored trends over time. The most common spatial methods were distance calculations, spatial aggregation, clustering, spatial smoothing and interpolation, and spatial regression. Proximity measures were predominant and were applied primarily to air quality and climate science and resource access studies. The review concludes by noting emerging areas that are likely to be important to future spatial analysis in public health.