An Introduction to Macro- Level Spatial Nonstationarity: a Geographically Weighted Regression Analysis of Diabetes and Poverty

Mortality from type 2 diabetes mellitus across municipalities in Mexico

BACKGROUND

One in six Mexican adults' lives with type 2 diabetes mellitus (T2DM), which is the third leading cause of death in the country. Analyzing the geographic distribution of T2DM mortality helps identify regions with higher mortality rates. This study aimed to examine the spatial patterns of mortality from type 2 diabetes mellitus (T2DM) across municipalities in Mexico and to analyze the main contextual factors linked to this cause of death in 2020.

METHODS

We employed a spatial Bayesian hierarchical regression model to estimate the risk and probability of death from type 2 diabetes mellitus (T2DM) across Mexico's municipalities.

RESULTS

The SMR results revealed geographic and age-specific patterns. Central Mexico and the Yucatán Peninsula exhibited the highest excess mortality rates. For the population under 50 years of age, municipalities in Oaxaca had the highest T2DM mortality rates, whereas those aged 50 years old and older had the highest rates in Tlaxcala and Puebla. Socioeconomic factors such as low levels of educational attainment, lack of health services, dietary deficiency, and marginalization were positively associated with increased T2DM mortality risk. By contrast, GDP per capita showed a negative association. High-risk areas for T2DM mortality were prominent along the south of the Pacific Coast, the Bajío, Central Mexico, and southern Yucatán for those under 50, and along a central strip extending to the Yucatán Peninsula for the older population. Significant uncertainties in mortality risk were identified, with Central Mexico, Oaxaca, Chiapas, and Tabasco showing high probabilities of excess risk for those under 50 years of age and extended risk areas along the Gulf of Mexico for those 50 years old and older.

CONCLUSIONS

The assessment and identification of spatial distribution patterns associated with T2DM mortality, and its main contextual factors, are crucial for informing effective public health policies aimed at reducing the impact of this chronic disease in Mexico.

Geospatial Modeling Methods in Epidemiological Kidney Research: An Overview and Practical Example

Geospatial modeling methods in population-level kidney research have not been used to full potential because few studies have completed associative spatial analyses between risk factors and exposures and kidney conditions and outcomes. Spatial modeling has several advantages over traditional modeling, including improved estimation of statistical variation and more accurate and unbiased estimation of coefficient effect direction or magnitudes by accounting for spatial data structure. Because most population-level kidney research data are geographically referenced, there is a need for better understanding of geospatial modeling for evaluating associations of individual geolocation with processes of care and clinical outcomes. In this review, we describe common spatial models, provide details to execute these analyses, and perform a case-study to display how results differ when integrating geographic structure. In our case-study, we used U.S. nationwide 2019 chronic kidney disease (CKD) data from Centers for Disease Control and Prevention's Kidney Disease Surveillance System and 2006 to 2010 U.S. Environmental Protection Agency environmental quality index (EQI) data and fit a nonspatial count model along with global spatial models (spatially lagged model [SLM]/pseudo-spatial error model [PSEM]) and a local spatial model (geographically weighted quasi-Poisson regression [GWQPR]). We found the SLM, PSEM, and GWQPR improved model fit in comparison to the nonspatial regression, and the PSEM model decreased the positive relationship between EQI and CKD prevalence. The GWQPR also revealed spatial heterogeneity in the EQI-CKD relationship. To summarize, spatial modeling has promise as a clinical and public health translational tool, and our case-study example is an exhibition of how these analyses may be performed to improve the accuracy and utility of findings.

Exploring regional and sociodemographic disparities associated with unenrollment for the disease management program for type 2 Diabetes Mellitus using Bayesian spatial modelling

BACKGROUND

The disease management program (DMP) for type 2 Diabetes Mellitus (T2DM) is the largest DMP in Germany. Our goal was to analyze regional differences in unenrollment rates, suggest areas for intervention and provide background information, which population groups in which locations are currently not enrolled in the DMP for T2DM.

METHODS

In this study, we used data of the 1.7 mil. insurants of the AOK Nordost health insurance. For the visualization of enrollment potential, we used the Besag-York-Mollie model (BYM). The spatial scan statistic (SaTScan) was used to detect areas of unusually high rates of unenrolled diabetics to prioritize areas for intervention. To explore sociodemographic associations, we used Bayesian spatial global regression models. A Spatially varying coefficient model (SVC) revealed in how far the detected associations vary over space.

RESULTS

The proportion of diabetics currently not enrolled in the DMP T2DM was 36.8% in 2019 and varied within northeastern Germany. Local clusters were detected mainly in Mecklenburg-West-Pomerania and Berlin. The main sociodemographic variables associated with unenrollment were female sex, younger age, being unemployed, foreign citizenship, small household size and the proportion of persons commuting to work outside their residential municipality. The SVC model revealed important spatially varying effects for some but not all associations.

CONCLUSION

Lower socioeconomic status and foreign citizenship had an ubiquitous effect on not being enrolled. The DMP T2DM therefore does currently not reach those population groups, which have a higher risk for secondary diseases and possible avoidable hospitalizations. Logically, future interventions should focus on these groups. Our methodology clearly suggests areas for intervention and points out, which population group in which locations should be specifically approached.

Spatial epidemiology of diabetes: Methods and insights

Diabetes mellitus (DM) is a growing epidemic with global proportions. It is estimated that in 2019, 463 million adults aged 20-79 years were living with DM. The latest evidence shows that DM continues to be a significant global health challenge and is likely to continue to grow substantially in the next decades, which would have major implications for healthcare expenditures, particularly in developing countries. Hence, new conceptual and methodological approaches to tackle the epidemic are long overdue. Spatial epidemiology has been a successful approach to control infectious disease epidemics like malaria and human immunodeficiency virus. The implementation of this approach has been expanded to include the study of non-communicable diseases like cancer and cardiovascular diseases. In this review, we discussed the implementation and use of spatial epidemiology and Geographic Information Systems to the study of DM. We reviewed several spatial methods used to understand the spatial structure of the disease and identify the potential geographical drivers of the spatial distribution of DM. Finally, we discussed the use of spatial epidemiology on the design and implementation of geographically targeted prevention and treatment interventions against DM.

Geographically weighted machine learning model for untangling spatial heterogeneity of type 2 diabetes mellitus (T2D) prevalence in the USA

Type 2 diabetes mellitus (T2D) prevalence in the United States varies substantially across spatial and temporal scales, attributable to variations of socioeconomic and lifestyle risk factors. Understanding these variations in risk factors contributions to T2D would be of great benefit to intervention and treatment approaches to reduce or prevent T2D. Geographically-weighted random forest (GW-RF), a tree-based non-parametric machine learning model, may help explore and visualize the relationships between T2D and risk factors at the county-level. GW-RF outputs are compared to global (RF and OLS) and local (GW-OLS) models between the years of 2013-2017 using low education, poverty, obesity, physical inactivity, access to exercise, and food environment as inputs. Our results indicate that a non-parametric GW-RF model shows a high potential for explaining spatial heterogeneity of, and predicting, T2D prevalence over traditional local and global models when inputting six major risk factors. Some of these predictions, however, are marginal. These findings of spatial heterogeneity using GW-RF demonstrate the need to consider local factors in prevention approaches. Spatial analysis of T2D and associated risk factor prevalence offers useful information for targeting the geographic area for prevention and disease interventions.

Influencing Indicators and Spatial Variation of Diabetes Mellitus Prevalence in Shandong, China: A Framework for Using Data-Driven and Spatial Methods

To control and prevent the risk of diabetes, diabetes studies have identified the need to better understand and evaluate the associations between influencing indicators and the prevalence of diabetes. One constraint has been that influencing indicators have been selected mainly based on subjective judgment and tested using traditional statistical modeling methods. We proposed a framework new to diabetes studies using data-driven and spatial methods to identify the most significant influential determinants of diabetes automatically and estimated their relationships. We used data from diabetes mellitus patients' health insurance records in Shandong province, China, and collected influencing indicators of diabetes prevalence at the county level in the sociodemographic, economic, education, and geographical environment domains. We specified a framework to identify automatically the most influential determinants of diabetes, and then established the relationship between these selected influencing indicators and diabetes prevalence. Our autocorrelation results showed that the diabetes prevalence in 12 Shandong cities was significantly clustered (Moran's I = 0.328, p < 0.01). In total, 17 significant influencing indicators were selected by executing binary linear regressions and lasso regressions. The spatial error regressions in different subgroups were subject to different diabetes indicators. Some positive indicators existed significantly like per capita fruit production and other indicators correlated with diabetes prevalence negatively like the proportion of green space. Diabetes prevalence was mainly subjected to the joint effects of influencing indicators. This framework can help public health officials to inform the implementation of improved treatment and policies to attenuate diabetes diseases.

Geographical clustering and socioeconomic factors associated with hypoglycemic events requiring emergency assistance in Andalusia (Spain)

INTRODUCTION

The geographical distribution of hypoglycemic events requiring emergency assistance was explored in Andalusia (Spain), and potentially associated societal factors were determined.

RESEARCH DESIGN AND METHODS

This was a database analysis of hypoglycemia requiring prehospital emergency assistance from the Public Company for Health Emergencies (Empresa Pública de Emergencias Sanitarias (EPES)) in Andalusia during 2012, which served 8 393 159 people. Databases of the National Statistics Institute, Basic Spatial Data of Andalusia and System of Multiterritorial Information of Andalusia were used to retrieve spatial data and population characteristics. Geographic Information System software (QGIS and GeoDA) was used for analysis and linkage across databases. Spatial analyses of geographical location influence in hypoglycemic events were assessed using Moran's I statistics, and linear regressions were used to determine their association with population characteristics.

RESULTS

The EPES attended 1 137 738 calls requesting medical assistance, with a mean hypoglycemia incidence of 95.0±61.6 cases per 100 000 inhabitants. There were significant differences in hypoglycemia incidence between basic healthcare zones attributable to their geographical location in the overall population (Moran's I index 0.122, z-score 7.870, p=0.001), women (Moran's I index 0.088, z-score 6.285, p=0.001), men (Moran's I index 0.076, z-score 4.914, p=0.001) and aged >64 years (Moran's I index 0.147, z-score 9.753, p=0.001). Hypoglycemia incidence was higher within unemployed individuals (β=0.003, p=0.001) and unemployed women (β=0.005, p=0.001), while lower within individuals aged <16 years (β=-0.004, p=0.040), higher academic level (secondary studies) (β=-0.003, p=0.004) and women with secondary studies (β=-0.005, p<0.001). In subjects aged >64 years, lower rate of hypoglycemia was associated with more single-person homes (β=-0.008, p=0.022) and sports facilities (β=-0.342, p=0.012).

CONCLUSIONS

This analysis supports the geographical distribution of hypoglycemia in the overall population, both genders and subjects aged >64 years, which was affected by societal factors such as unemployment, literacy/education, housing and sports facilities. These data can be useful to design specific prevention programs.

Global and local correlations of Hajnal's household formation markers in historical Europe: A cautionary tale

Previous scholarship has assumed global correlations between premarital service in husbandry, marriage age, the extent to which marriage coincided with the attainment of household headship, and the nuclear household structure. According to John Hajnal, these were the four core principles of historical household formation systems. However, whether such correlations applied universally across Europe remains uncertain. We test this possibility by applying both global and local (geographically weighted) measures of correlation to data for 256 rural populations from historical Europe. We demonstrate that local correlations diverge considerably from the global results. The mutual associations between household formation markers exhibit considerable spatial drifts and important spatial gradients, and the numbers of joint combinations of these associations far exceed those predicted by Hajnal. We argue that the global relationship patterns that Hajnal promoted may lead to incorrect interpretations of historical family systems, and may detract from our understanding of their actual mechanics.

Small-scale spatial analysis of intermediate and definitive hosts of Angiostrongylus cantonensis

BACKGROUND

Angiostrongyliasis is a food-borne parasitic zoonosis. Human infection is caused by infection with the third-stage larvae of Angiostrongylus cantonensis. The life cycle of A. cantonensis involves rodents as definitive hosts and molluscs as intermediate hosts. This study aims to investigate on the infection status and characteristics of spatial distribution of these hosts, which are key components in the strategy for the prevention and control of angiostrongyliasis.

METHODS

Three villages from Nanao Island, Guangdong Province, China, were chosen as study area by stratified random sampling. The density and natural infection of Pomacea canaliculata and various rat species were surveyed every three months from December 2015 to September 2016, with spatial correlations of the positive P. canaliculata and the infection rates analysed by ArcGIS, scan statistics, ordinary least squares (OLS) and geographically weighted regression (GWR) models.

RESULTS

A total of 2192 P. canaliculata specimens were collected from the field, of which 1190 were randomly chosen to be examined for third-stage larvae of A. cantonensis. Seventy-two Angiostrongylus-infected snails were found, which represents a larval infection rate of 6.1% (72/1190). In total, 110 rats including 85 Rattus norvegicus, 10 R. flavipectus, one R. losea and 14 Suncus murinus were captured, and 32 individuals were positive (for adult worms), representing an infection rate of 29.1% of the definitive hosts (32/110). Worms were only found in R. norvegicus and R. flavipectus, representing a prevalence of 36.5% (31/85) and 10% (1/10), respectively in these species, but none in R. losea and S. murinus, despite testing as many as 32 of the latter species. Statistically, spatial correlation and spatial clusters in the spatial distribution of positive P. canaliculata and positive rats existed. Most of the spatial variability of the host infection rates came from spatial autocorrelation. Nine spatial clusters with respect to positive P. canaliculata were identified, but only two correlated to infection rates. The results show that corrected Akaike information criterion, R², R² adjusted and σ² in the GWR model were superior to those in the OLS model.

CONCLUSIONS

P. canaliculata and rats were widely distributed in Nanao Island and positive infection has also been found in the hosts, demonstrating that there was a risk of angiostrongyliasis in this region of China. The distribution of positive P. canaliculata and rats exhibited spatial correlation, and the GWR model had advantage over the OLS model in the spatial analysis of hosts of A. cantonensis.

Racial differences in spatial patterns for poor glycemic control in the Southeastern United States

PURPOSE

Evidence consistently shows poor outcomes in racial minorities, but there is limited understanding of differences that are explained by spatial variation. The goal of this analysis was to examine contribution of spatial patterns on disparities in diabetes outcomes in the Southeastern United States.

METHODS

Data on 64,022 non-Hispanic black (NHB) and non-Hispanic white (NHW) veterans with diabetes living in Georgia, Alabama, and South Carolina were analyzed for 2014. Hemoglobin A1c (HbA1c) was categorized as controlled (less than 8%) and uncontrolled (greater than or equal to 8%). Logistic regression was used to understand the additional explanatory capability of spatial random effects over covariates such as demographics, service connectedness, and comorbidities. Data aggregated at the county level were used to identify hotspots in distribution of uncontrolled HbA1c and tested using local Moran's I test.

RESULTS

Overall percent uncontrolled HbA1c was 36.5% (40.8% in NHB and 33.4% in NHW). In unadjusted analyses, NHB had 37% higher odds of uncontrolled HbA1c (odds ratio [OR]: 1.37, 95% confidence interval, 1.32, 1.41). After adjusting for demographics and comorbidities, the OR decreased to 1.09 but remained significant (95% confidence interval, 1.05, 1.13). The OR further decreased after incorporating spatial effects (OR: 1.07, 95% confidence interval, 1.03, 1.11) but remained statistically significant. Hotspots of high HbA1c were detected, and spatial patterns differed across racial groups.

CONCLUSIONS

Differences in spatial patterns in glycemic control exists between NHB and NHW veterans with type 2 diabetes. Incorporating spatial effects helps explain more of the disparity in uncontrolled HbA1c than adjusting only for demographics and comorbidities, but significant differences in uncontrolled HbA1c remained.

Who is where at risk for Chronic Obstructive Pulmonary Disease? A spatial epidemiological analysis of health insurance claims for COPD in Northeastern Germany

Background

Chronic obstructive pulmonary disease (COPD) has a high prevalence rate in Germany and a further increase is expected within the next years. Although risk factors on an individual level are widely understood, only little is known about the spatial heterogeneity and population-based risk factors of COPD. Background knowledge about broader, population-based processes could help to plan the future provision of healthcare and prevention strategies more aligned to the expected demand. The aim of this study is to analyze how the prevalence of COPD varies across northeastern Germany on the smallest spatial-scale possible and to identify the location-specific population-based risk factors using health insurance claims of the AOK Nordost.

Methods

To visualize the spatial distribution of COPD prevalence at the level of municipalities and urban districts, we used the conditional autoregressive Besag–York–Mollié (BYM) model. Geographically weighted regression modelling (GWR) was applied to analyze the location-specific ecological risk factors for COPD.

Results

The sex- and age-adjusted prevalence of COPD was 6.5% in 2012 and varied widely across northeastern Germany. Population-based risk factors consist of the proportions of insurants aged 65 and older, insurants with migration background, household size and area deprivation. The results of the GWR model revealed that the population at risk for COPD varies considerably across northeastern Germany.

Conclusion

Area deprivation has a direct and an indirect influence on the prevalence of COPD. Persons ageing in socially disadvantaged areas have a higher chance of developing COPD, even when they are not necessarily directly affected by deprivation on an individual level. This underlines the importance of considering the impact of area deprivation on health for planning of healthcare. Additionally, our results reveal that in some parts of the study area, insurants with migration background and persons living in multi-persons households are at elevated risk of COPD.

Exploring the small-scale spatial distribution of hypertension and its association to area deprivation based on health insurance claims in Northeastern Germany

BACKGROUND

Hypertension is one of the most frequently diagnosed chronic conditions in Germany. Targeted prevention strategies and allocation of general practitioners where they are needed most are necessary to prevent severe complications arising from high blood pressure. However, data on chronic diseases in Germany are mostly available through survey data, which do not only underestimate the actual prevalence but are also only available on coarse spatial scales. The discussion of including area deprivation for planning of healthcare is still relatively young in Germany, although previous studies have shown that area deprivation is associated with adverse health outcomes, irrespective of individual characteristics. The aim of this study is therefore to analyze the spatial distribution of hypertension at very fine geographic scales and to assess location-specific associations between hypertension, socio-demographic population characteristics and area deprivation based on health insurance claims of the AOK Nordost.

METHODS

To visualize the spatial distribution of hypertension prevalence at very fine geographic scales, we used the conditional autoregressive Besag-York-Mollié (BYM) model. Geographically weighted regression modelling (GWR) was applied to analyze the location-specific association of hypertension to area deprivation and further socio-demographic population characteristics.

RESULTS

The sex- and age-adjusted prevalence of hypertension was 33.1% in 2012 and varied widely across northeastern Germany. The main risk factors for hypertension were proportions of insurants aged 45-64, 65 and older, area deprivation and proportion of persons commuting to work outside their residential municipality. The GWR model revealed important regional variations in the strength of the examined associations.

CONCLUSION

Area deprivation has only a significant and therefore direct influence in large parts of Mecklenburg-West Pomerania. However, the spatially varying strength of the association between demographic variables and hypertension indicates that there also exists an indirect effect of area deprivation on the prevalence of hypertension. It can therefore be expected that persons ageing in deprived areas will be at greater risk of hypertension, irrespective of their individual characteristics. The future planning and allocation of primary healthcare in northeastern Germany would therefore greatly benefit from considering the effect of area deprivation.

Do the risk factors for type 2 diabetes mellitus vary by location? A spatial analysis of health insurance claims in Northeastern Germany using kernel density estimation and geographically weighted regression

BACKGROUND

The provision of general practitioners (GPs) in Germany still relies mainly on the ratio of inhabitants to GPs at relatively large scales and barely accounts for an increased prevalence of chronic diseases among the elderly and socially underprivileged populations. Type 2 Diabetes Mellitus (T2DM) is one of the major cost-intensive diseases with high rates of potentially preventable complications. Provision of healthcare and access to preventive measures is necessary to reduce the burden of T2DM. However, current studies on the spatial variation of T2DM in Germany are mostly based on survey data, which do not only underestimate the true prevalence of T2DM, but are also only available on large spatial scales. The aim of this study is therefore to analyse the spatial distribution of T2DM at fine geographic scales and to assess location-specific risk factors based on data of the AOK health insurance.

METHODS

To display the spatial heterogeneity of T2DM, a bivariate, adaptive kernel density estimation (KDE) was applied. The spatial scan statistic (SaTScan) was used to detect areas of high risk. Global and local spatial regression models were then constructed to analyze socio-demographic risk factors of T2DM.

RESULTS

T2DM is especially concentrated in rural areas surrounding Berlin. The risk factors for T2DM consist of proportions of 65-79 year olds, 80 + year olds, unemployment rate among the 55-65 year olds, proportion of employees covered by mandatory social security insurance, mean income tax, and proportion of non-married couples. However, the strength of the association between T2DM and the examined socio-demographic variables displayed strong regional variations.

CONCLUSION

The prevalence of T2DM varies at the very local level. Analyzing point data on T2DM of northeastern Germany's largest health insurance provider thus allows very detailed, location-specific knowledge about increased medical needs. Risk factors associated with T2DM depend largely on the place of residence of the respective person. Future allocation of GPs and current prevention strategies should therefore reflect the location-specific higher healthcare demand among the elderly and socially underprivileged populations.

Spatial Variation in the Relationship between Hispanic Concentration and County Poverty: A Migration Perspective

Racial/ethnic minority concentration is generally positively related to county poverty. Yet, spatial variation in this relationship may call into question the meaning attached to racial/ethnic concentration. We argue that racial/ethnic concentration reflects more than just the concentration of individuals from a disadvantaged group. In addition, we extend previous work by taking a migration perspective to explain spatial non-stationarity in racial/ethnic concentration’s relationship with county poverty. Factors related to the migration process, including migrant selectivity and spatial differentiation in place characteristics, could alter the relationship between a minority group’s concentration and poverty. We employ spatially informed methods and 2006–2010 American Community Survey data to examine the relationship between Hispanic concentration and county poverty. The GWR results indicate significant spatial variation in the percent Hispanic-county poverty relationship. Hispanic migration regimes capture some of the observed relationship non-stationarity, suggesting migration-related processes partially drive Hispanic-county poverty relationship non-stationarity. However, we discuss other explanations that should be considered in future research. This work advances research on spatial inequality by examining the social implications of migration and by investigating the role of place in shaping the meaning of minority concentration.

Spatial analysis and correlates of county-level diabetes prevalence, 2009-2010

Introduction

Information on the relationship between diabetes prevalence and built environment attributes could allow public health programs to better target populations at risk for diabetes. This study sought to determine the spatial prevalence of diabetes in the United States and how this distribution is associated with the geography of common diabetes correlates.

Methods

Data from the Centers for Disease Control and Prevention and the US Census Bureau were integrated to perform geographically weighted regression at the county level on the following variables: percentage nonwhite population, percentage Hispanic population, education level, percentage unemployed, percentage living below the federal poverty level, population density, percentage obese, percentage physically inactive, percentage population that cycles or walks to work, and percentage neighborhood food deserts.

Results

We found significant spatial clustering of county-level diabetes prevalence in the United States; however, diabetes prevalence was inconsistently correlated with significant predictors. Percentage living below the federal poverty level and percentage nonwhite population were associated with diabetes in some regions. The percentage of population cycling or walking to work was the only significant built environment–related variable correlated with diabetes, and this association varied in magnitude across the nation.

Conclusion

Sociodemographic and built environment–related variables correlated with diabetes prevalence in some regions of the United States. The variation in magnitude and direction of these relationships highlights the need to understand local context in the prevention and maintenance of diabetes. Geographically weighted regression shows promise for public health research in detecting variations in associations between health behaviors, outcomes, and predictors across geographic space.