Bivariate Local Indicators Of Spatial Association Research Articles

A district-level geospatial analysis of anaemia prevalence among rural men in India, 2019-21

BackgroundDespite its considerable impact on health and productivity, anemia among men has received limited attention. In a country as diverse as India, characterized by extensive geographic variations, there is a pressing need to investigate the nuanced spatial patterns of anemia prevalence among men. The identification of specific hotspots holds critical implications for policymaking, especially in rural areas, where a substantial portion of India’s population resides.MethodsThe study conducted an analysis on a sample of 61,481 rural men from 707 districts of India, utilizing data from the National Family Health Survey-5 (2019-21). Various analytical techniques, including Moran’s I, univariate LISA (Local Indicators of Spatial Association), bivariate LISA, and spatial regression models such as SLM (Spatial Lag Model), and SEM (Spatial Error Model) were employed to examine the geographic patterns and spatial correlates of anaemia prevalence in the study population.ResultsIn rural India, three out of every ten men were found to be anemic. The univariate Moran’s I value for anaemia was 0.66, indicating a substantial degree of spatial autocorrelation in anaemia prevalence across the districts in India. Cluster and outlier analysis identified five prominent ‘hotspots’ of anaemia prevalence across 97 districts, primarily concentrated in the eastern region (encompassing West Bengal, Jharkhand, and Odisha), the Dandakaranya region, the Madhya Pradesh-Maharashtra border, lower Assam, and select districts in Jammu and Kashmir. The results of SLM revealed significant positive association between anaemia prevalence at the district-level and several key factors including a higher proportion of Scheduled Tribes, men in the 49–54 years age group, men with limited or no formal education, individuals of the Muslim faith, economically disadvantaged men, and those who reported alcohol consumption.ConclusionsSubstantial spatial heterogeneity in anaemia prevalence among men in rural India suggests the need for region-specific targeted interventions to reduce the burden of anaemia among men in rural India and enhance the overall health of this population.

Contamination and source-specific health risk assessment of polycyclic aromatic hydrocarbons in soil from a mega iron and steel site in China

The iron and steel industry has always been a key and difficult point of environmental pollution control. In the present study, 493, 175, 153, 72, and 42 soil samples were collected from the soil depths of 0–0.5, 0.5–2, 2–3, 3–4, and 4–5 m (herein called the layers) of the Shougang Steel site, respectively. Compared with the evaluation criteria, the Shougang Steel surface soil was severely polluted by polycyclic aromatic hydrocarbons (PAHs). Inverse distance-weighted interpolation and the Kruskal-Wallis H test revealed that the soil PAH pollution in the iron-making area, especially the coking area, was severer than those in other areas. The PAH concentrations first decreased, and then, increased with the increase of depth. With the increase in depth, the contributions of 2- and 3-ring PAHs increased, while those of 4-, 5-, and 6-ring PAHs decreased. The bivariate local indicators of spatial association (LISA) analysis was used to identify the areas prone to soil PAH pollution due to atmospheric deposition of industrial waste gas and traffic emissions. The method could be used to analyze the impact of anthropogenic activities on soil's PAH pollution for other contaminated sites. Three main pollution sources of soil PAHs, the backfill source, the combustion of coal, and the traffic emissions, were identified based upon three diagnostic ratios, positive matrix factorization and the bivariate LISA analysis, and accounted for 53.8%, 23.5%, and 22.7%, respectively. The combination of bivariate LISA analysis and other source analysis methods could improve the accuracy of source analysis. Benzo[a]pyrene contributed the most to the total health risk among sixteen PAHs. The health risks related to the three pollution sources decreased in the order of backfill sources > coal combustion > traffic emissions. The incremental life-time carcinogenic risks were all below 10−4, indicating negligible or acceptable risks.

Assessing social vulnerability and identifying spatial hotspots of flood risk to inform socially just flood management policy.

This study presents the first nationwide spatial assessment of flood risk to identify social vulnerability and flood exposure hotspots that support policies aimed at protecting high-risk populations and geographical regions of Canada. The study used a national-scale flood hazard dataset (pluvial, fluvial, and coastal) to estimate a 1-in-100-year flood exposure of all residential properties across 5721 census tracts. Residential flood exposure data were spatially integrated with a census-based multidimensional social vulnerability index (SoVI) that included demographic, racial/ethnic, and socioeconomic indicators influencing vulnerability. Using Bivariate Local Indicators of Spatial Association (BiLISA) cluster maps, the study identified geographic concentration of flood risk hotspots where high vulnerability coincided with high flood exposure. The results revealed considerable spatial variations in tract-level social vulnerability and flood exposure. Flood risk hotspots belonged to 410 census tracts, 21 census metropolitan areas, and eight provinces comprising about 1.7 million of the total population and 51% of half-a-million residential properties in Canada. Results identify populations and the geographic regions near the core and dense urban areas predominantly occupying those hotspots. Recognizing priority locations is critically important for government interventions and risk mitigation initiatives considering socio-physical aspects of vulnerability to flooding. Findings reinforce a better understanding of geographic flood-disadvantaged neighborhoods across Canada, where interventions are required to target preparedness, response, and recovery resources that foster socially just flood management strategies.

Disparities in COPD Hospitalizations: A Spatial Analysis of Proximity to Toxics Release Inventory Facilities in Illinois.

Disproportionate distribution of air pollution is a major burden on the health of people living in proximity to toxic facilities. There are over 1000 Toxics Release Inventory (TRI) facilities distributed across the state of Illinois. This study investigates and spatially analyzes the relationship between chronic obstructive pulmonary disease (COPD) hospitalizations and toxic emissions from TRI facilities. In addition, this study investigates the connection between COPD hospitalizations and socioeconomic variables. Accounting for dispersion of air pollution beyond the TRI facilities source was attained using the inverse distance weighting interpolation approach. Multiple statistical methods were used including principal components analysis, linear regression, and bivariate local indicators of spatial association (BiLISA). The results from the linear regression model and BiLISA clustering maps show there is a strong connection between COPD hospitalizations and socioeconomic status along with race. TRI emissions were not statistically significant, but there are three major clusters of high COPD hospitalizations with high TRI emissions. Rural areas also seem to carry a higher burden of pollution-emitting facilities and respiratory hospitalizations.

The Influence of Aggregate Sociodemographic Characteristics on Fishing License Sales in Cook County, Illinois

AbstractAnglers have been among the major supporters of aquatic conservation activities in the USA through the Dingell–Johnson Act and other programs. Declining rates of participation in fishing have spurred significant efforts to recruit, retain, and reengage anglers, also known as “R3.” To make R3 strategies more effective, it is important to explore the broader sociodemographic patterns that underpin angling participation. Urban areas are variable with respect to residents’ socioeconomic status (SES), and scholars have observed lower rates of angling participation among urban residents. This heterogeneity highlights the need to understand angling participation among urban residents. Therefore, the objective of this study was to explore the influence of aggregate sociodemographic characteristics on fishing license sales in Cook County, Illinois, the main urban center of the state. We assessed the spatial association of sociodemographic characteristics and fishing license purchase and determined the relative influence of three components (SES, household mobility, and age) from principal components analysis on fishing license purchases. We used the bivariate LISA (local indicators of spatial association) method and regression models to test for differences in demographic effect. Results revealed statistically significant spatial dependence in patterns of license sales density and aggregate SES, household mobility, and age. Aggregate SES had a negative relationship with license sales density. Conversely, aggregate household mobility and age positively influenced the density of license sales. Interactions of the three components were also investigated. Census tracts with lower mobility tended to fish more with higher SES. Census tracts with a younger population tended to fish more with higher SES. Census tracts with higher SES tended to fish less with increasing age. Our findings provide assistance in developing R3 strategies aimed at increasing participation in recreational angling by showing how demographic factors influencing license sales can be examined using census data.

Modelling and analyzing spatial clusters of leptospirosis based on satellite-generated measurements of environmental factors in Thailand during 2013-2015

This study statistically identified the association of remotely sensed environmental factors, such as Land Surface Temperature (LST), Night Time Light (NTL), rainfall, the Normalised Difference Vegetation Index (NDVI) and elevation with the incidence of leptospirosis in Thailand based on the nationwide 7,495 confirmed cases reported during 2013-2015. This work also established prediction models based on empirical findings. Panel regression models with random-effect and fixed-effect specifications were used to investigate the association between the remotely sensed environmental factors and the leptospirosis incidence. The Local Indicators of Spatial Association (LISA) statistics were also applied to detect the spatial patterns of leptospirosis and similar results were found (the R2 values of the random-effect and fixed-effect models were 0.3686 and 0.3684, respectively). The outcome thus indicates that remotely sensed environmental factors possess statistically significant contribution in predicting this disease. The highest association in 3 years was observed in LST (random- effect coefficient = -9.787, P<0.001; fixed-effect coefficient = -10.340, P=0.005) followed by rainfall (random-effect coefficient = 1.353, P<0.001; fixed-effect coefficient = 1.347, P<0.001) and NTL density (random-effect coefficient = -0.569, P=0.004; fixed-effect coefficient = -0.564, P=0.001). All results obtained from the bivariate LISA statistics indicated the localised associations between remotely sensed environmental factors and the incidence of leptospirosis. Particularly, LISA's results showed that the border provinces in the northeast, the northern and the southern regions displayed clusters of high leptospirosis incidence. All obtained outcomes thus show that remotely sensed environmental factors can be applied to panel regression models for incidence prediction, and these indicators can also identify the spatial concentration of leptospirosis in Thailand.

The Spread of COVID-19 in China at Its Initial Stage: A Township-Level Analysis in Association with the Built Environment

Background: The built environment can contribute to the spread of the novel coronavirus disease (COVID-19) as well as other infectious diseases by facilitating human mobility and social contacts between infected and susceptible individuals. It can also provide a space that enables the direct transmission of pathogens to susceptible individuals. On the other hand, mobility data capturing interpersonal transmission at finegrain in large scale are not available. With the above background, we aim to assess the associations of key built environment factors, which create spaces for activities — “inferred activity” rather than “actually observed activity” — with the spread of COVID-19 in China at its initial stage. Method: In this nationwide study, we used a random forest (RF) approach to objectively assess the associations between the built environment factors and the spread of COVID-19. The spread is measured for 2 994 township-level administrative units in China by two indicators, the first of which is the ratio of cumulative infection cases (RCIC) that were confirmed until February 13, 2020 (14 days after stopping public transportation operations in all Chinese cities). The other indicator is the coefficient of variation of infection cases (CVIC) between February 6 and February 13, 2020, which reflects the policy effect in the initial stage of the spread. Accordingly, we selected 19 independent variables covering built environment attributes (e.g., urban facilities, transportation infrastructure, and land use), socioeconomic characteristics, and inter-city (from Hubei Province) population flow (PF). We also investigated the spatial agglomerations using a bivariate local indicators of spatial association (BiLISA) between the above two indicators of the spread. The data is from Amap, Resource and Environment Data Cloud Platform, Google Earth Engineer (GEE), Baidu direction lite API, Baidu Migration Production, Tencent, People's Daily, and other references.Findings: The spread of COVID-19 had obvious spatial agglomerations all across China. Clusters with both low RCIC and low CVIC at the same locations (Low-Low type) show the largest percentage, and are followed by the Low-High type. The Low-High type is more dangerous because clusters of low RCIC are surrounded by clusters with high CVIC which means that policy measures taken in the initial stage may be less effective to prevent/mitigate a rapid disease spread at these locations. The density of convenience shops, supermarkets, and shopping malls (DoCSS) along with PF were the two most important factors to RCIC, whereas PF was the most important factor to measure the policy effects (i.e., the CVIC). When the DoCSS as well as the density of road intersections (DoI), the density of gyms and sport centers (DoGSC) reach 21/km2, 72/km2, and 2/km2, respectively, positive associations of these density indicators with the RCIC become maximal and kept constant as the indicators further increase. Only after PF exceeds 75% its relationship with the RCIC starts showing a positive relationship with RCIC and the relationship keeps stable when the PF gets to 80%. For policy effects, the density of colleges and universities (DoCU) and the density of comprehensive hospitals (DoCH) had negative impacts on policy effects in the initial stage of the COVID-19 spread. Notably, when the DoCH was at less than 0.1/km2, it had a small positive trend. Accordingly, increasing betweenness centrality (BC) associated with increasing policy effect in the initial stage of COVID-19 spread before 5 000, and a reversed trend was observed after 5 000.Interpretation: First, the findings from this study do not support nationwide uniform measures against COVID-19 (e.g., lockdown and other restriction measures, or reopening of economic activities). Second, the built environment factors may be associated with mitigating or accelerating the spread of COVID-19. Third, the revealed thresholds of different built environment factors in association with the spread of COVID19 present insightful evidence on how to integrate urban and public health planning for future pandemics. Funding: This study is financially supported by two funds from Japan Science and Technology Agency. One of these funds is the J-RAPID Collaborative Research/Survey Program for Urgent Research framework which is titled, “Impacts of COVID-19 on the transport and logistics sector and countermeasures.” The other fund is the Ethical, Legal, and Social Implications/Issues (ELSI) research framework, “Responsible Innovation with Conscience and Agility,” which is titled, “Overcoming Vulnerability and Restoring Social Justice in Community and Re-designing Cities by Introducing Social Distancing."Declaration of Interests: The authors declare no competing interests.

The Spatio-Temporal Heterogeneity of Financial Agglomeration on Green Development in China Cities Using GTWR Model

By promoting financial agglomerations to support green development in a region is a keyway for China to resolve the sharp contradiction between economic growth and environmental protection. However, existing research only considered the promotion effect of financial agglomerations on green development, but the spatio-temporal non-stationarity of that effect has been overlooked. Using a panel data of 285 prefecture-level cities in China and based on the evaluation of green development by a Driving-Pressure-State-Impact-Response (DPSIR) model, this paper analyzes the spatial correlation of financial agglomeration on green development. The paper also investigates the differences in the spatio-temporal influence of financial agglomeration on green development from both global and local perspectives by employing a Bivariate Local Indicators of Spatial Association (BLISA) model and a Geographically and Temporally Weighted Regression (GTWR) model. The results indicate that: (1) There exists significant spatial dependency between financial agglomeration and green development from 2003 to 2015, with Low-Low (L-L) and Low-High (L-H) spatial clusters as the main cluster types. (2) From the local perspective, the promoting effect of financial agglomerations on green development has showed significant spatial heterogeneity with a gradually decreasing trend from the southeast coast to the northwest inland of China. This work can help to develop policies for supporting green development by formulating differential strategies for financial agglomerations.

An integrated methodology for improving heavy metal risk management in soil-rice system

The lack of linear correlation between regional soil and rice heavy metal (HM) content aggravates the difficulty of risk management in farmland. Hence, an effective HM contamination risk control strategy is urgently required for rice paddy. Here, a novel integrated spatial interaction and risk identification methodology was proposed. Bivariate local indicators of spatial association (BL-LISA) was used to analyze the spatial interaction between soil and rice HM. The mechanism of the spatial interaction pattern was elucidated with lead isotope ratios and redundancy analysis. The rice HM risk was predicted via a Bayesian decision tree (BDT). The spatial interaction patterns were mainly High-Low and Low-High. Thus, there was antagonism between soil and rice HM. Emission sources and sinks accounted for the observed spatial interaction patterns. The parent material contributed 69% to the soil HM content but only 10.2% to that of rice. Seven risk rules and 13 security rules were identified by BDT. The risk identification accuracy of these rules was 96.8% for the overall sample. BL-LISA mapping was combined with BDT to demarcate and classify the risk zones and project differentiated and refined management modes. The risk, potential risk and clean zones comprised 7.8%, 14.1% and 46.6% of the farmland, respectively. The integrated method was superior to other traditional techniques in terms of farm HM risk management and may enhance decision-making in HM risk management for soil-rice system.

Exploring the geography of serious mental illness and type 2 diabetes comorbidity in Illawarra-Shoalhaven, Australia (2010 -2017).

ObjectivesThe primary aim of this study was to describe the geography of serious mental illness (SMI)–type 2 diabetes comorbidity (T2D) in the Illawarra-Shoalhaven region of NSW, Australia. The Secondary objective was to determine the geographic concordance if any, between the comorbidity and the single diagnosis of SMI and diabetes.MethodsSpatial analytical techniques were applied to clinical data to explore the above objectives. The geographic variation in comorbidity was determined by Moran’s I at the global level and the local clusters of significance were determined by Local Moran’s I and spatial scan statistic. Choropleth hotspot maps and spatial scan statistics were generated to assess the geographic convergence of SMI, diabetes and their comorbidity. Additionally, we used bivariate LISA (Local Indicators of Spatial Association) and multivariate spatial scan to identify coincident areas with higher rates of both SMI and T2D.ResultsThe study identified significant geographic variation in the distribution of SMI–T2D comorbidity in Illawarra Shoalhaven. Consistently higher burden of comorbidity was observed in some urban suburbs surrounding the major metropolitan city. Comparison of comorbidity hotspots with the hotspots of single diagnosis SMI and T2D further revealed a geographic concordance of high-risk areas again in the urban areas outside the major metropolitan city.ConclusionThe identified comorbidity hotspots in our study may serve as a basis for future prioritisation and targeted interventions. Further investigation is required to determine whether contextual environmental factors, such as neighbourhood socioeconomic disadvantage, may be explanatory.Implications for public healthOurs is the first study to explore the geographic variations in the distribution of SMI and T2D comorbidity. Findings highlight the importance of considering the role of neighbourhood environments in influencing the T2D risk in people with SMI.

The Advantages of Comparative LISA Techniques in Spatial Inequality Research: Evidence from Poverty Change in the United States

Although scholarship regarding spatial inequality has grown in recent years, past research has seen limited use of spatial statistics—let alone comparison between spatial statistical techniques. Comparing and contrasting the application and use of spatial statistics is valuable in research because it allows for more precise identification of spatial patterns, and highlights results that may be hidden when only using a single method. This study serves as a demonstration on how the use of multiple LISA statistics can benefit inequality related research. Analyzing changes in county level poverty in the rural United States from 1990 to 2015 serves as a tool to demonstrate these techniques and this study examined how the geographic distribution of poverty has changed, and well as if there is evidence of diffusion effects. The three featured techniques utilized Local Indicators of Spatial Association (LISA) statistics. The techniques are Bivariate LISA, LISA Cluster Transitions, and LISA Diffusion Transitions, with the last technique specifically designed for this study. Each technique varies in how it reports the changes in the spatial structure of poverty. Bivariate LISA and LISA Cluster Transitions are complementary to each other—with the former technique providing a single global statistic while the latter is more easily interpretable. Diffusion Transitions show how the highest and lowest values of a variable may be spreading over time. The study also produces new findings regarding rural poverty, with poverty in Mountain-West and rural Sun Belt counties on the rise. Analysis shows a diffusion effect for poverty in Southeastern metropolitan fringe counties.

Space-time analysis of the incidence of human visceral leishmaniasis (VL) and prevalence of canine VL in a municipality of southeastern Brazil: Identification of priority areas for surveillance and control

Human and canine visceral leishmaniasis (HVL and CVL, respectively) represent serious public health issues in Brazil. The surveillance and control measures currently employed have had limited effect in impeding the territorial expansion of the disease and in reducing the number of cases. We have investigated the space-time distribution of HVL incidence rates and CVL prevalence in the coverage areas of the 148 primary healthcare units in Belo Horizonte (MG, Brazil) during a 6-year period in order to identify those that should be prioritized for disease control actions. Data were smoothed using the empirical Bayes method and analyzed by space-time scanning and application of univariate global Moran’s I index and local indicators of spatial association (LISA) statistics to identify spatial autocorrelations. Point data of CVL were analyzed using the Kernel method. Bivariate global Moran’s I and LISA techniques were employed to identify spatial correlations between HVL and CVL. Based on our results, we were able to formulate two proposals for establishing the prioritization of coverage areas, namely: (i) classification of maximum priority areas as identified by bivariate LISA for HVL and CVL, and (ii) combination of maximum priority areas with high priority areas as identified by univariate LISA for HVL. According to our proposals, 27 coverage areas in Belo Horizonte were categorized as maximum priority and a further 13 were classified as high priority. Our proposals, which are based on practical, feasible and inexpensive statistical tools, will contribute to a better understanding of VL distribution in urban settings and improving the efficiency of governmental control programs.

Dental workforce availability and dental services utilization in Appalachia: a geospatial analysis.

There is considerable variation in dental services utilization across Appalachian counties, and a plausible explanation is that individuals in some geographical areas do not utilize dental care due to dental workforce shortage. We conducted an ecological study on dental workforce availability and dental services utilization in Appalachia. We derived county-level (n = 364) data on demographic and socioeconomic characteristics and dental services utilization in Appalachia from the 2010 Behavioral Risk Factor Surveillance System (BRFSS) using person-level data. We obtained county-level dental workforce availability and physician-to-population ratio estimates from Area Health Resources File and linked them to the county-level BRFSS data. The dependent variable was the proportion using dental services within the last year in each county (ranging from 16.6% to 91.0%). We described the association between dental workforce availability and dental services utilization using ordinary least squares regression and spatial regression techniques. Spatial analyses consisted of bivariate local indicators of spatial association (LISA) and geographically weighted regression (GWR). Bivariate LISA showed that counties in the central and southern Appalachian regions had significant (P < 0.05) low-low spatial clusters (low dental workforce availability, low percent dental services utilization). GWR revealed considerable local variations in the association between dental utilization and dental workforce availability. In the multivariate GWR models, 8.5% (t-statistics > 1.96) and 13.45% (t-statistics > 1.96) of counties showed positive and statistically significant relationships between the dental services utilization and workforce availability of dentists and dental hygienists, respectively. Dental workforce availability was associated with dental services utilization in the Appalachian region; however, this association was not statistically significant in all counties. The findings suggest that program and policy efforts to improve dental services utilization need to focus on factors other than increasing the dental workforce availability for many counties in Appalachia.