key: cord-0945889-moyugv7k
authors: Rahman, Md. Hamidur; Zafri, Niaz Mahmud; Ashik, Fajle Rabbi; Waliullah, Md; Khan, Asif
title: Identification of risk factors contributing to COVID-19 incidence rates in Bangladesh: A GIS-based spatial modeling approach
date: 2021-02-10
journal: Heliyon
DOI: 10.1016/j.heliyon.2021.e06260
sha: 9957bac6dbb9cf718e2a02d4228b76d33f70df42
doc_id: 945889
cord_uid: moyugv7k

BACKGROUND: COVID-19 pandemic outbreak is an unprecedented shock throughout the world, which has generated a massive social, human, and economic crisis. Identification of risk factors is crucial to prevent the COVID-19 spread by taking appropriate countermeasures effectively. Therefore, this study aimed to identify the potential risk factors contributing to the COVID-19 incidence rates at the district-level in Bangladesh. METHOD: ology: Spatial regression methods were applied in this study to fulfill the aim. Data related to 28 demographic, economic, built environment, health, and facilities related factors were collected from secondary sources and analyzed to explain the spatial variability of this disease incidence. Three global (ordinary least squares (OLS), spatial lag model (SLM), and spatial error model (SEM)) and one local (geographically weighted regression (GWR)) regression models were developed in this study. RESULTS: The results of the models identified four factors: percentage of the urban population, monthly consumption, number of health workers, and distance from the capital city, as significant risk factors affecting the COVID-19 incidence rates in Bangladesh. Among the four developed models, the GWR model performed the best in explaining the variation of COVID-19 incidence rates across Bangladesh with an R(2) value of 78.6%. CONCLUSION: Findings and discussions from this research offer a better insight into the COVID-19 situation, which helped discuss policy implications to negotiate the future epidemic crisis. The primary policy response would be to decentralize the urban population and economic activities from and around the capital city, Dhaka to create self-sufficient regions throughout the country, especially in the north-western region.

Coronavirus disease is an exceptionally infectious disease caused by the 4 SARS-CoV-2 virus. The first human cases of COVID-19 were reported in Wuhan City, China, in 5

December 2019 (1). Within a few weeks, the outbreak of COVID-19 spread globally; thus, the 6 World Health Organization (WHO) declared this a public health emergency of international 7 concern (PHEIC) on 30 January 2020 and a pandemic on 11 March 2020. As of 05 February 8 2021, there have been about 105 million reported cases resulting in approximately 2.29 million 9 deaths (2). According to the World Bank, COVID-19 has triggered a global crisis like no other. 10

The disease is leading to the deepest global recession since the Second World War. The baseline 11 forecast envisions a 5.2% contraction in the global GDP in 2020-the deepest global recession 12 in the last eight decades, despite unprecedented policy support (3) . About 1.6 billion informal 13 workers, having little to no savings and no social protection access, lost 60% of their income. 14 This pandemic will push 40-60 million people into extreme poverty (4) . 15

Identification of possible demographic, economic, built environment, health, and service 16 facilities related risk factors of infectious diseases is crucial at each phase of the epidemic to 17 effectively prevent further spread through appropriate interventions (5). Disadvantaged 18 demographic pattern, economic, environmental, and health conditions have been established as 19 potential determinants of infectious diseases (e.g., Tuberculosis, HIV, Pertussis, Pneumonia, 20 SARS, Hand-foot-mouth, and Influenza) in general (5-10). These findings suggest that similar 21 results might be obtainable for the newly emerged coronavirus disease. However, these 22 J o u r n a l P r e -p r o o f 1 The first three coronavirus cases were confirmed in Bangladesh on 8 March 2020 by the 2 Institute of Epidemiology Disease Control and Research (21) . COVID-19 infected persons 3 returned from Italy to join their family at their native places. Bangladesh reported its first 4 coronavirus death on 18 March 2020 (22). Following the initial detection of coronavirus cases in 5 Dhaka, the virus transmitted from the capital city to other major administrative areas of 6 Bangladesh rapidly. To intercept the disease's spread, the government declared a nationwide 7 'lockdown' from 26 March to 30 May 2020 (23) , but unlike the developed countries, the 8 enforcement of the lockdown has been weak. To prevent viral transmission from higher infected 9

to lower infected areas, the Bangladesh government took measures like the closure of 10 educational institutions, declaration of a general holiday, restriction on religious gathering, 11 suspension of commercial activities, closure of garments factories, the prohibition of inter-12 country, inter-district, and intra-district travels (24). To date, the education institutes are closed, 13 but the government and non-government offices, as well as commercial and industrial entities, 14 were opened after being shut down until 30 May 2020. Inter-city and intra-city travel restrictions 15 were also lifted. 16

As of 31 January 2021, 0.535 million confirmed cases with 8,127 deaths (Infection 17 

The use of geospatial and statistical tools is critical to explore the association between 2 COVID-19 incidence and its contributing factors as it is a process that occurs in geographical 3 space (29) . Traditional statistical approaches used in epidemiological studies, e.g., factor analysis 4 (30) , principal component analysis (31) , cluster analysis (32), regression analysis (33) fail to take 5 into account the spatial dependency and autocorrelation in parameters estimation. To address this 6 functional lag, the spatial regression model (SRM), i.e., spatial lag model (SLM), spatial error 7 model (SEM), and geographically weighted regression (GWR) have been widely used in 8 epidemiological studies (14) . Besides, geographic information system (GIS) is an essential tool 9 that can assist in the process of combating a pandemic as well as improve the quality of care 10 through examining the spatial distribution of infectious diseases (29, 34, 35) . 11

In this study, data of explanatory variables are inherently spatial. Hence, it is plausible 12 that data would be positively spatially dependent, which means areas located nearby tend to be 13 more similar than those separated by great distances (36). As OLS regression cannot take spatial 14 dependence into account due to the assumption of homogeneity and spatial non-variability (37, 15 38), we used SEM and SLM along with OLS. Nevertheless, global models still have limitations as they cannot account for a spatial non-stationarity issue. The spatial non-stationarity issue 1 explains how the dependent and independent variables might vary over space (14, 39) . Thus, as a 2 local regression model, GWR was used to address the spatial non-stationarity issue. A detailed 3 description of this model has been avoided to focus more on discussions than methodological 4 exercises. Information on these models is available in the studies of Mollalo, Vahedi (17) and 5 Sannigrahi, Pilla (14) . 6

At first, for developing multivariate models, a univariate analysis was performed to 8 identify the potential explanatory variables from the collected data of many factors. Here, OLS 9 models were developed in ArcGIS to explain each variable's impact on the COVID-19 incidence 10 rates individually. Factors found insignificant in this stage were excluded for further 11 consideration as explanatory variables. From these selected explanatory variables, the stepwise 12 forward procedure was applied to eliminate non-significant explanatory variables and develop an 13 overall global multivariate OLS model with the best fit statistics. Then, Pearson's correlation 14 analysis was conducted to examine the correlations between the significant factors in univariate 15 analysis. For detecting multicollinearity in the model, the Variance Inflation Factor (VIF) was 16 used, and therefore, the uncorrelated factors were selected as the input of the final OLS 17 regression model. After that, SLM, SEM, and GWR were developed using the final OLS model's 18 significant explanatory variables for comparison purposes. Two global models (SLM and SEM) 19

were developed in GeoDa 1.14 software. Based on the first-order Queen's contiguity, the weight 20 matrix was generated for developing the SLM and SEM. 21

On the other hand, the local model (GWR) was run in ArcGIS software. Adaptive kernel 22 type and AICc bandwidth were selected to run the GWR model. Finally, The R 2 and AICc values 23 of the four developed models were used to compare the models' performances in explaining 1 COVID-19 incidence rates across Bangladesh. 2

The results of the univariate analysis are presented in Table 2 . Out of 28 factors, 17 were 4 found statistically significant and considered explanatory variables for developing a multivariate 5 model later. All the demographic factors were found significant; whereas, most of the 6 community facilities related factors found insignificant. From Table 2 , it is also clear that most 7 of the demographic and health-related factors have comparatively higher R 2 values, which means 8 these factors could explain a good portion of the variation in the COVID-19 incidence rates 9 across Bangladesh. On the other hand, a relatively lower R 2 was found for other factors. 10 Therefore, variation in COVID-19 incidence rates could mostly be described by demographic 11 and health-related factors. 12 13 J o u r n a l P r e -p r o o f J o u r n a l P r e -p r o o f 1 four variables were included in the final global multivariate OLS model (Figure 2) . Several 2 factors, which had high R 2 and were strongly significant in univariate analysis, were not 3 considered for the model development to diminish multicollinearity in the model. Those factors 4 were: total population, population density, rented tenancy percentage, internal migrant 5 population, and the number of an economic unit. The final model included only four significant 6 factors-urban population percentage, monthly consumption, number of health workers, and 7 distance from the capital ( Table 3) . 8

The model has relatively low multicollinearity since the highest VIF value among the 9 four factors (VIF= 2.9) is far less than the threshold of 7.5. The association's direction suggested 10 that three factors positively associate with the COVID-19 incidence rates except for distance 11 from the capital factor. This model was found to be statistically significant and has an R 2 value 12 of 0.673. This value means that about 67.3% of the COVID-19 incidence rates across 13

Bangladesh are associated with the model's four factors. The rest of the 32.7% incidence rates 14 are caused by unknown factors to the model and probably for the local variations, which could 15 not be captured by the global OLS model. 16

1 Figure 2 . Results of the correlation analysis 2 GWR was used to model the COVID-19 incidence rates on a local scale. From Table 5 , 3 R 2 value was found to be the highest for the SEM model among the global models. This value 4 increased from 0.722 in the SEM to 0.786 in the GWR model. Therefore, it is evident that the 5 GWR model could explain 78.6% of the variations of COVID-19 incidence rates across 6

Bangladesh. In addition, the AICc value was also found the lowest in the GWR model (AICc= 7 340.49) compared to the global models, indicating that the GWR model is the most effective in 8 this context. 9

The spatial distributions of the coefficient values of the GWR model are presented in 10 percentage. For the eastern and southern regions of Bangladesh, monthly consumption and the 13 number of health workers were found as influential factors, and urban population percentage was 14 found as a weak factor in explaining the COVID-19 incidence rates. On the other hand, for the 15 north-western regions of the country, the opposite findings were observed. Furthermore, distance 16 from the capital was an influential factor in the country's central parts. It was less influential in 17 the south-eastern region of the country. The discussion section explains why and how these 18 factors better explain COVID-19 incidence in a developing country. 19 20 Figure 3 . Spatial distribution of the coefficient values of urban population percentage, monthly 2 consumption, number of health workers, and distance from the capital in describing COVID-19 3

incidence rates using GWR model 4

The results of mapping local R 2 values of the GWR model are demonstrated in Figure 4 . 1 Though decent local R 2 values were found for all districts, the districts located in the southern 2 regions of the country have comparatively lower local R 2 values than northern regions of the 3 country, indicating a decent prediction of the model across the country, especially in northern 4 districts. The reason behind the slight variation in the local R 2 value might be unique 5 topographical differences between the two regions. In the southern parts of Bangladesh, coastal 6 districts are located, which are mostly isolated by rivers from their surrounding districts. In 7 addition, southern-east parts of Bangladesh are hilly areas where communication from one 8 district to another district is very difficult. Whereas, central, northern, and north-western parts of 9 the country are relatively flat land areas. Therefore, communication among them is relatively 10 easier. As the GWR model estimated the local R 2 value based on the surrounding districts, 11 therefore, for the southern part of the country, surrounding districts cannot influence much on the 12 COVID-19 incidence rates of the concerned district due to poor communication. Therefore, it is 13 likely that for these reasons the GWR model performed slightly inferior in that region compared 14 to other regions. 15 

The percentage of the urban population of districts was positively related to COVID-19 4 incidence rates, which is similar to the findings of Hamidi, Sabouri (15), where researchers 5 found that a high urban population leads to increase movement and activities of people in a high-6 density urban area. The higher the population density, the more likely it is to contact an infector 7 and infectee (40). This study also found population density as a significant variable in univariate 1 analysis. However, it could not be considered for the final model due to the multicollinearity 2 issue. In addition to that, the impact of the urban population percentage on the COVID-19 3 incidence rates was found lower (lower coefficient value in GWR model) in the north-western 4 region (e.g., Rajshahi, Rangpur, and Mymensingh division) of Bangladesh compared to the 5 south, central, and south-eastern regions (Figure 3) . Here, the geographical context of the south-6 central-eastern regions of Bangladesh suggests that these regions cover three major urban areas. 7

The first one is Dhaka megacity, which comprises 38% (12 million people) of the country's total 8 urban population. Next is Chottogram, the second-largest city located south-eastern region of 9

Bangladesh, which has one-third population of Dhaka. Lastly, Khulna, the third-largest city of 10 the country, is also located in the southern region (41). All of these three cities are densely 11 populated and have greater economic and administrative agglomeration than other parts of the 12 country. On the other hand, the north-western region's major urban areas are not as vibrant as the 13 mentioned three cities as they are far behind in employment opportunities, commercial activities, 14 and industrial activities (41). As a result, COVID-19 incidence rates were less influenced by the 15 urban population in the north-western region than the country's other regions. 16

In this situation, planned and decentralized urbanization needs to be ensured and physical 17 and social infrastructure need to be better distributed across the country so that people do not 18 concentrate in a few districts, raising their densities beyond sustainable limits. For decentralized 19 urban development, existing cities of the north-western region of Bangladesh need to be 20 prioritized as alterative urban areas of Dhaka and Chottogram to reduce pressure from these two 21 cities. 22

The monthly consumption factor positively influenced COVID-19 incidence rates. 1

Households having higher monthly consumption tend to purchase goods from commercially 2 vibrant places, which increases the potential of being affected by . Furthermore, 3 higher consumption is associated with higher income levels, lower poverty, and higher 4 employment rates. Therefore, income and employment-related activities trigger frequent travel 5 and physical contact with people, increasing the risk of transmission of COVID-19 (42). Another 6 explanation behind this could be that low-income people might have less access to testing 7 facilities, resulting in the underreporting of the positive COVID-19 cases in low-income areas 8 (11, 43) . In addition to that, monthly consumption had a higher impact on COVID-19 incidence 9 rates in Bangladesh's north-western region than in other regions (Figure 3 ). This north-western 10 region of the country is lagging behind than other regions due to higher poverty and inequality 11 (44). This region suffers from severe "Monga", referred to as a seasonal phenomenon of poverty 12 and hunger, during the time period of September-November and March-April (45). So, after the 13 March-April "Monga", people are compelled to take part in economic activities to secure their 14 needs ahead of the upcoming "Monga" which might lead to the higher impact of monthly 15 consumption on COVID-19 incidence rate in this particular region. 16

As a recommendation, economic activities need to be diversified and dispersed around 17 the country so that high-income people and employment opportunities are not concentrated in a 18 few districts. A large portion of the north-western region's poor working group migrates to 19 Dhaka for work due to a lack of employment in their native place (46). They are mostly engaged 20 in informal jobs in Dhaka. In the period leading from the first detection of the COVID-19 case 21 in Bangladesh, quite a few people lost their jobs while many others experienced pay cuts (47). 22

As a result, many households left the capital, Dhaka, facing affordability issues, and they mostly 23 relocated to their village homes (48, 49) . This is the time to seize the opportunity to create 1 employment for these people by providing them monetary support to engage in agricultural 2 activities and establish industries, which can be based in remote areas, especially the north-3 western region. This will ensure that these people will be less likely to re-migrate to the capital; 4 instead, others living in Dhaka and other major cities might be attracted to return to their village 5 homes. 6

The number of health workers positively influenced the COVID-19 incidence rates as 7 hospitals became the hotspots for COVID-19 circulation. This result was found to be consistent 8 with the findings of Mollalo, Vahedi (17) . From correlation analysis, it can also be observed that 9 the number of health workers factor had a strong positive relationship with total population and 10 population density (Figure 2) . This direction of correlation suggested that areas with a larger 11 population with higher population density have comparatively more health facilities and, 12 consequently, have a higher number of health workers and a number of COVID-19 testing 13 centers. Therefore, number of positive COVID-19 cases was also more in those areas. In 14 addition, this factor was found to have a more substantial influence on COVID-19 incidence 15 rates in the north-western region of Bangladesh than other regions (Figure 3 ). There was a 16 severe shortage of personal protection equipment (PPE) in Bangladesh at the beginning of the 17 pandemic. Therefore, a significant percentage (25%) of front-line health workers were obliged to 18 tackle COVID-19 without any protection, and thereby, a large number of health workers became 19 infected by . The infected health professionals are likely to have passed on the 20 disease to others, including patients visiting the hospitals for non-COVID-19 related issues and 21 people accompanying them. Furthermore, PPEs were distributed in the hospitals of important 22 urban areas first. Therefore, it might be assumed that hospitals in the north-western region 23 received PPE later. Due to this, COVID-19 might have spread from hospitals through health 1 workers in this region. 2

In a situation like this, the supply-chain system must be uninterrupted so that proper 3 protective equipment can be delivered to the health workers on a priority basis. The demand for 4 health services escalates during pandemic times. Steps need to be taken to reduce the pressure on 5 existing hospitals and prevent them from becoming COVID-19 hotspots. A few places in the 6 major cities can be earmarked for erecting temporary hospitals to tackle a pandemic like the 7 COVID-19. These places should have good transport access, and they can be used as recreational 8 open spaces at other times. 9

Distance from the capital was found inversely related to COVID-19 incidence rates. 10 Dhaka, the capital and a megacity of Bangladesh, is the epicenter of the country's pandemic and 11 has the most infected population (Figure 1 ). In addition, Dhaka is the central commercial and 12 administrative hub of the country; hence, it attracts people from all over Bangladesh, resulting in 13 many trips to and from Dhaka. Since Dhaka is the economic hub of Bangladesh, a lot of people 14 live in surrounding districts to save on housing costs and commute to the city daily. During this 15 pandemic, public transport was under restriction across the country. People residing in districts 16 in closer proximity to Dhaka could somehow manage to travel to this district more frequently 17 than districts located further. These might be the reasons behind the result-the shorter the 18 distance of a district from the capital, the higher the likelihood of transmission of Moreover, correlation analysis suggested that distance from the capital was negatively correlated 20 with population density while positively correlated with poverty rate. The more the districts are 21 far away from the capital city, the greater is their potential to have low density and higher 22 poverty incidence which negatively affected COVID-19 incidence rates in Bangladesh ( Figure  23 2). Furthermore, this factor was less influential in the south-eastern region, southern region, and 1 the north-eastern region of Bangladesh compared to other regions. The reason behind that south-2 eastern region is a hilly area, and the south region comprises a large number of islands where 3 water transportation is the primary mode for communication. Therefore, accessibility between 4 the capital city and these locations is not as good as with other regions. 5

Additionally, travel between a lagged north-western region and Dhaka might be less 6 required during the pandemic. For these reasons, fewer viral transmissions occurred in those 7 regions from Dhaka, and fewer COVID-19 incidences were reported there. From literature, 8 connectivity was also found more important than density in 913 US metropolitan counties (15). 9

The policy response of diversifying and dispersing economic activities across Bangladesh should 10 be able to take care of this issue and reduce pressure on the capital city. This would help to create 11 self-sufficient regions and might reduce the demand for inter-region travel during critical 12

This study showed that the impacts of most of the community facilities, e.g., primary 14 school, secondary school, college, growth center, rural market, and religious establishment on 15 COVID-19 incidence rates were insignificant as these facilities were shut down in the initial 16 stage of COVID-19 pandemic. However, the univariate analysis results show that the number of 17 transit stations and the number of police station factors profoundly influenced increasing 18 COVID-19 incidence rates. Though transit stations were also controlled through lockdown 19 measures, there was a lax during the festive periods, and garment factories were hastily opened 20 and immediately closed, which resulted in the movement of a large number of people (51). In 21 addition, emergency activities, such as health care, food delivery, and necessary regular goods 22 marketing were out of the scope of lockdown measures and involved the use of transit stations. 23

As a result, transit stations might have become crowded, which paved the way to transmit 1 COVID-19. Police had the responsibility to implement lockdown and social distancing at the 2 field level during the pandemic. Enough protection equipment was not available for the police 3 personnel to carry out their duties safely. Moreover, police officers' overcrowding travel in a 4 single police van made them more vulnerable to the virus (52). The design of transit stations 5 needs to take the physical distance factor into account. The design should discourage the 6 clustering of people and ensure good natural air circulation. Like the health professionals, police 7 personnel and all other front-line workers should be given priority for the distribution of PPE 8 after ensuring an adequate supply of this equipment through an efficient supply-chain system. 9

6. Limitation of the study 10 One of the limitations of our study was the lack of appropriate data for some variables. 11

Due to the unavailability of individual or community level data, it is not logical to draw 12 inferences at the individual or community level. Another reservation is the COVID-19 testing 13 center's spatial availability. There are few opportunities to test COVID-19 for people living in 14 remote areas in Bangladesh. Besides, there is a tendency among people not to test COVID-19 15 even though they have symptoms. Therefore, there might be an underestimation of COVID-19 16 cases. Furthermore, the influence of lockdown and other containment measures on COVID-19 17 incidence rates were not considered in this study. There is likely to be some variations in 18 lockdown related policies and their implementation efficiency within a district and between 19 districts. It might play an essential role at the district-level to control the COVID-19 incidence 20 rates, but analyzing this influence was out of this research scope. 21

1 Identification of possible virus transmission and spread determinants is crucial, especially 2 for coronavirus disease , which brought unprecedented shock globally. This study 3 aimed to identify potential risk factors contributing to the district-level COVID-19 incidence 4 rates across Bangladesh. To fulfill the aim, three global (OLS, SLM, and SEM) and one local 5 (GWR) models were developed in this study to identify potential demographic, economic, built 6 environment, health, and facilities related factors affecting the COVID-19 incidence rates. 7

The models' results showed four factors-urban population percentage, monthly 8 consumption, number of health workers, and distance from the capital-were statistically 9 significant. The R 2 value was found to be 0.673 for the overall OLS model, which increased to 10 0.78 in the GWR model. The findings of the study showed that the COVID-19 incidence rates 11 increased with an increase in urban population percentage, monthly consumption, and the 12 number of health workers within a district. At the same time, the incidence rates decreased with 13 an increase in the distance between the capital city with districts. Discussions on the findings 14 reveal that higher level dependency on and concentration of economic as well as industrial 15

activities within a few districts, ineffective supply-chain system, lack of self-sufficient regions, 16 poverty and inequality, especially in the north-western region of the country, were some of the 17 main causes which led to higher COVID-19 incidence rates. 18

The recommended policy response suggests that increasing dependency on the capital 19 city should be lessened by diversifying and decentralizing economic activities across 20

Bangladesh, especially the north-western region of Bangladesh, which needs to be prioritized as 21 alterative urban centers Dhaka and Chottogram. This would help create self-sufficient regions 22

and might reduce the demand for inter-region travel during critical situations and, consequently, 23 help control any pandemic like the COVID-19 in the future. Theoretical investigations and 1 empirical observations from this research offer an alternate view of the joint importance of health 2 and non-health determinants, which will help planners and local governments for effective policy 3 development to tackle future epidemic crises. Future studies incorporating individual or 4 community level data having temporal consistency with COVID-19 transmission time could 5 produce better directives analyzing risk factors at lower geographic unit and finding solutions for 6 different socio-economic groups. 7

The research work is facilitated by the support of Bangladesh University of Engineering and 9

Technology (BUET) through their research lab facilities and infrastructures. Hence, the authors 10 are happy to acknowledge that support. In addition, the authors also like to thank Madiha 11 Chowdhury (URP'15, BUET) for proofreading the manuscript. 12

World Health Organization; 2020. 15 2. WHO. WHO Coronavirus Disease (COVID-19) Dashboard

Global Economic Prospects

BRIEF#2: PUTTING THE UN FRAMEWORK FOR SOCIO-ECONOMIC

Importance of 22 collecting data on socioeconomic determinants from the early stage of the COVID-19 outbreak 23 onwards

The Spread of COVID-19 and the BCG Vaccine: A Natural 25 Experiment in Reunified Germany

determinants of HIV incidence: evidence from a longitudinal, population-based study in rural 2

Assessing the social and environmental determinants of pertussis epidemics in Queensland

Australia: a Bayesian spatio-temporal analysis

Risk factors for community-acquired 8 pneumonia diagnosed upon hospital admission

Tropical medicine & international health

Slum health: 14 arresting COVID-19 and improving well-being in urban informal settlements

What will be the economic impact of covid-19 in the us? rough estimates of 17 disease scenarios

The impact of COPD and 19 smoking history on the severity of COVID-19: a systemic review and meta-analysis

Examining the association between 22 socio-demographic composition and COVID-19 fatalities in the European region using spatial 23 regression approach

Does Density Aggravate the COVID-19 Pandemic? 25

COVID-19 transmission: A multi-city study in China

GIS-based spatial modeling of COVID-19 incidence 30 rate in the continental United States

Mainland China is associated with temperature and humidity: A time-series analysis. Science of 2 the Total Environment

GIS-based spatial modelling of COVID-19 death incidence in

COVID-19 pandemic: environmental and social factors 6 influencing the spread of SARS-CoV-2 in São Paulo, Brazil. Environmental Science and 7 Pollution Research

Covid-19 Update

Bangladesh confirms first coronavirus death Dhaka, Dhaka

Facility and Purchase under the Rapid Financing Instrument-Press Release

Statement by the Executive Director for Bangladesh

COVID-19 and Bangladesh: Challenges and How to 16

Address Them. Frontiers in Public Health

COVID-19 Bangladesh situation reports. 2020. 18 26. Preetha SS, Islam Z. Is foul play the new normal? The Daily Star

Coronavirus and Impact on the Bangladesh Economy: Assessing the Damage of 20 the Black Swan of 2020

A GIS-based artificial neural network model for 23 spatial distribution of tuberculosis across the continental United States. International journal of 24 environmental research and public health

Invited Commentary: Insulin Resistance Syndrome? Syndrome X? Multiple 26

Metabolic Syndrome? A Syndrome At All? Factor Analysis Reveals Patterns in the Fabric of 27

Assessment of dietary patterns in nutritional epidemiology: principal component analysis

A brief conceptual 3 tutorial of multilevel analysis in social epidemiology: using measures of clustering in multilevel 4 logistic regression to investigate contextual phenomena

The changing epidemiology of acute 7 and subacute haematogenous osteomyelitis in children

Machine learning 10 approaches in GIS-based ecological modeling of the sand fly Phlebotomus papatasi, a vector of 11 zoonotic cutaneous leishmaniasis in Golestan province

Using 13 geographical information systems and cartograms as a health service quality improvement tool

Geographically weighted regression. Handbook of applied spatial 16 analysis

mgwr: A Python implementation 18 of multiscale geographically weighted regression for investigating process spatial heterogeneity 19 and scale

Geographically weighted regression: a 22 method for exploring spatial nonstationarity. Geographical analysis

Distribution of COVID-19 Morbidity Rate in Association with 24

Social and Economic Factors in Wuhan, China: Implications for Urban Development

Urbanization and sustainability in Asia: Case studies of good 27 practice: Asian Development Bank

The Geography of COVID-19 growth in the US: 29 Counties and Metropolitan Areas. Available at SSRN 3570540

the COVID-19 pandemic in informal urban settlements

Bangladesh Poverty Assessment: Assessing progress from

Understanding the Monga in Northwest Bangladesh: Household

Perceptions and Perceptual Connotations

Poverty Level of Bangladesh: Special Case

Covid-19 job cuts forcing people to leave Dhaka. Dhaka Tribune

Low-income people leaving Dhaka. The Business Standard

Families leaving Dhaka as coronavirus crisis hits livelihood

Hundreds of doctors in Bangladesh infected with coronavirus

Dhaka bracing for homebound rush for Eid amid pandemic

Police in Bangladesh at great risk from pandemic. Anadolu Agency