Learning About the Incidence and Lethality of COVID-19 in Vulnerable Neighborhoods: The Case of Malaga (Spain)

Abstract

This research delves into the need to use granular analyses at the neighborhood level to study the preexisting conditions of vulnerability that best explain the waves of COVID-19 incidence and mortality. It seems most appropriate to use the comprehensive approach of the sustainable development with variables that analyze the economic, social, environmental, and governance dimensions, given the extensive literature that identifies each as a determining factor for the impact of disease. The work utilizes a composite vulnerability index that allows the city of Malaga to be divided into 434 census sections; waves of incidence and mortality for each section are constructed for the period of March 2020 to March 2021. Cluster analysis reveals that there are five different cluster incidence patterns, whereas the lethality waves are found to behave as a hot-spot phenomenon. The results reveal that neighborhoods that are the most vulnerable in terms of their demographic conditions (large proportion over 65 years of age and dependent) and socioeconomic conditions (severe material deprivation), have been the most affected by COVID-19 infection and mortality.

Keywords

clusters COVID-19 vulnerability robust statistical lethality

Introduction

The rapid expansion of the COVID-19 pandemic in cities has ratified the importance of cities in the context of present and future world development. As stated by UN-Habitat in its report on the response of cities to COVID (UN-Habitat 2020), more than 1430 cities in 210 countries have been affected by the pandemic and well above 95% of total cases have been located in urban areas.

The UN 2030 Agenda (2015) and the subsequent version developed by the Habitat Agency (New Urban Agenda, 2016) insisted that we need to build more inclusive, safe, resilient, and sustainable communities with the aim that we will “leave no one behind”. The localization of sustainable development goals in cities requires that neighborhood-level data be collected on the poverty and vulnerability conditions of its population, with a holistic vision of sustainability (UN System Staff College 2017). Here, we examine the importance of analyzing COVID-19 in light of the spatial inequalities in cities at the highest possible level of spatial and temporal data disaggregation (Mishra, Gayen, and Haque 2020).

Historically, the poorest populations are those that have been the most highly impacted by diseases and crises, exhibiting a higher prevalence of chronic diseases and high rates of morbidity and mortality (Stiglitz 2020; Wade 2020). Locating these people and groups in the urban fabric and understanding their living conditions will improve our understanding of the behavior of the SARS-CoV2 virus and the policies and measures needed to improve the resilience of these neighborhoods and populations (Matthew and McDonald 2006; Viezzer and Biondi 2021).

The differences between countries and regions in the evolution of COVID-19 have been studied by numerous disciplines, including regional science, and multiple articles have been written on its varied impacts (Karaye and Horney 2020; Macharia et al. 2020; Miramontes et al. 2021; Rodríguez-Pose and Burlina 2021; Shams, Haleem, and Javaid 2020). The example, Shams, Haleem and Javaid (2020) studied the 18 countries most highly affected by COVID-19. The disease has not spread homogeneously across or within countries, especially where the health system is decentralized. Some articles quantified the unevenness in the distribution of epidemiological variables across the regions, or within the provinces in the case of Spain, as Gutierrez, Inguanzo, and Orbe (2021) studied. Other authors found that countries with significant percentages of the older population have been vulnerable to a high number of deaths due to COVID-19 and proposed to provide specialized support for the self-isolation of people aged 75 and above.

The factors that have been identified as causes or vectors of the waves of contagion and lethality are multiple and, in some cases, even contradictory (air and water pollution, temperature, population mobility, lack of green spaces, previous morbidity and pathologies, ethnic characteristics, poverty and marginality, aging, etc.) (Auler et al. 2020; Bashir et al. 2020; Kodera, Rashed and Hirata 2020; Li et al. 2021; Liang et al. 2020; Prata, Rodrigues and Bermejo 2020).

Sharifi and Khavarian-Garmsir (2020) analyzed the literature published in scientific journals during the first 8 months after the contagion was first recognized in Wuhan, in reference to the causes and effects of COVID-19 on an urban scale. The authors found four main themes of research focus: environmental quality, socioeconomic impacts, management and governance, and transportation and urban design. In their conclusions, they emphasized the relative lack of research using an integrated approach in which the four dimensions of sustainable development (economic, social, environmental, and governance) were considered jointly in the analysis. This reflection encouraged us to examine the spatial and temporal disaggregation of COVID in cities from a comprehensive perspective such as the sustainable development.

As an example, we took the city of Malaga (570,000 inhabitants), which is located in southern Spain and is the capital of the widely populated Costa del Sol (1.5 million inhabitants). Recent studies have highlighted the clear spatial segregation of the neighborhoods of Malaga. For example, Bárcena-Martín et al. (2021) created a global vulnerability index and four sub-indices (socioeconomic, environmental, demographic, and assistance dimensions), and used them to establish a classification for the 434 census sections of the city. Based on individualized data provided by the Junta de Andalucía on the cases of infections and deaths during the first year of the pandemic, we visualized the waves of COVID-19 incidence and lethality for each census section. We then used the clustering method to identify five incidence patterns that are directly related to the previously classified vulnerability of the affected areas. In contrast, lethality did not exhibit a normal distribution; rather, this parameter was distributed as atypical cases or very small foci (hot spots). Analyzing the previous vulnerability conditions of the hot spots in comparison with the sections enabled us to determine that the neighborhoods with the highest mortality rates had a specific combination of demographic, socioeconomic, and healthcare factors.

The cluster analysis has been used by other authors to consider the incidence and lethality of COVID, as Bilal et al. (2021). They studied confirmed cases and mortality in New York, Philadelphia, and Chicago during the first 6 months of the pandemic and they found spatial clusters of high and low positivity, confirmed cases, and mortality co-located with clusters of low and high social vulnerability in the three cities.

Our research confirms that the preexisting conditions of vulnerability and poverty in neighborhoods have conditioned the results of the contagious in the city (Ahmed et al. 2020; Martins-Filho et al. 2020) but shows that COVID-19 lethality has not followed the same pattern. As a consequence, the resilience of neighborhoods in pandemic scenarios will not only imply betting on sanitary or preventive health measures, but also addressing measures to fight against inequality and spatial vulnerability in all its dimensions (Qian and Fan 2020; Jon 2020).

Theoretical Framework

Since the bibliographic review carried out by Sharifi and Khavarian-Garmsir (2020) on the impact of COVID in cities, accumulating studies have indicated that analyses should involve the granular disaggregation of data to the neighborhood level (Fu and Zhai 2021; Kayanan, Moore-Cherry, and Clavin 2021; Krellenberg and Koch 2021; Liu et al. 2021; Viezzer and Biondi 2021). Interesting analyses have looked at cities from both developed and developing countries, especially large metropolises such as New York, Chicago, Ontario, London, Rio de Janeiro, Barcelona, Madrid, etc. (Baldasano 2020; Credit 2020; Harris 2020). Although environmental and health dimensions have consistently prevailed in these analyses, newer studies have linked the social and economic vulnerability of neighborhoods with the evolution of COVID-19 contagion and lethality (Bambra et al. 2020; Patel et al. 2020; Qian and Fan 2020; Viezzer and Biondi 2021). Some authors examined the role of place-based versus diffusion factors in the geographic incidence of COVID-19, such as Florida and Mellander (2022) in the counties and neighborhood in Sweden. They concluded that in countries without lockdown policies the cases and lethality of the COVID- 19 is more related with diffusion factors, such as the location of nursing homes or the behavior of people.

Viezzer and Biondi (2021) looked at several Brazilian cities in the Atlantic Forest and confirmed the influence of urban, socioeconomic, and eco-environmental aspects on the lethality of COVID. Huang et al. (2021) studied cities in South Carolina and confirmed that preexisting conditions of social vulnerability, resilience, and urban or rural geography have influenced the spread of the pandemic and the severity of mortality rates. Credit (2020) focused on the neighborhoods of Chicago and New York and concluded that hospitalizations and deaths from COVID-19 at the individual case level was—to that point—strongly associated with minority groups and those with underlying conditions. The author’s findings suggested that higher socioeconomic status and the provision of a healthy and active built environment showed a significant negative association with COVID-19 infection rates, while several aspects of social vulnerability were significant positive predictors of COVID-19 infection rates. Although the impacts were context-specific, some social groups were found to be disproportionately affected, as previously suggested by Qian and Fan (2020) in China. The latter authors concluded that factors such as education, family income, Communist Party membership, and state-sector employment were important factors in determining people’s vulnerability to COVID-related financial troubles.

In an analysis of demographic and geographic factors in 182 countries, Nguimkeu and Tadadjeu (2021) found that the population density, urbanization rate, and proportion of people aged 65 and over were decisive factors that could explain why the number of COVID-19 cases differed across countries. Similarly, Khavarian-Garmsir, Sharifi and Moradpour (2021) found in Tehran that the mortality rate was higher in districts with a greater proportion of older people.

The results obtained from recent neighborhood-level studies have affirmed that this approach will be useful for performing space-time-based investigations and delineating vulnerable areas for COVID-19. For example, Rahman et al. (2021) proposed that a neighborhood-level focus could facilitate the formulation of effective management action plans to reduce and control the disease propagation and impacts of COVID-19.

Whitehead et al. (2020) recognized that there were clear differences between population groups due to preexisting vulnerability conditions, leading to the idea that we are not “all together” in this boat. The problems of marginality and vulnerability underlie some of the behavior exhibited by the virus, and only the organized efforts of society can help prevent marginalized and vulnerable groups from once again being disproportionately affected in the short and medium terms.

In Spain, Amate-Fortes and Guarnido-Rueda (2022) conducted a study in 819 Spanish municipalities with a linear cross-sectional model that allowed to conclude that a higher level of income inequality generates a higher rate of infections but not deaths, highlighting the importance of the Spanish National Health Service. Also, Aguilar- Palacio (2021) confirmed the effect of socioeconomic inequalities on the probability of COVID-19 infection, and its variations across the three pandemic waves in Aragon (a Spanish region) in individual and area levels. They insisted in the necessity to develop individual and area coordinated measures in the control, diagnosis and treatment of the epidemic to avoid an increase in the existing inequalities.

The present study applies an integrated approach that takes advantage of the battery of indicators and indices available for the spatial vulnerability conditions of the neighborhoods in the city of Malaga, Spain.

The existing studies on COVID-19 have varied widely in their statistical approach and analytic methods across authors and territorial spaces. In general, parametric measures have been used for regions with relatively reliable and continuous information sources, whereas non-parametric measures have been used in regions with lower-quality information (Nguimkeu and Tadadjeu 2021; Sannigrahi et al. 2020; Wu et al. 2020). However, efforts to study the spatial incidence of COVID-19 have required the use of non-parametric statistics, as this has enabled researchers to generate more robust and explanatory results without losing basic information (Khavarian-Garmsir, Sharifi and Moradpour 2021; Rahman et al. 2021).

Here, composite indices and the cluster technique were chosen to guarantee the robustness of the analysis while separately considering incidence and lethality and maintaining spatial vulnerability as a reference variable to explain the evolution of the other variables.

The Case of Malaga

This work studied the city of Malaga in light of its 434 census sections. Malaga is located on the shores of the Mediterranean; over its more than 3000 years of history, different cultures have succeeded one another, although the model of a compact and complex city has generally been maintained. This city is an enclave of high-quality natural wealth located between the sea and the Montes de Malaga Natural Park. The city economic is based on cultural tourism and the construction sector, although technological and logistical services are becoming increasingly important. The metropolitan environment of Malaga is marked by the tourist attraction of the Costa del Sol and the agricultural and gastronomic offerings of the provincial interior.

The city grew on the banks of the Guadalhorce River and later moved to the banks of the Guadalmedina River, both of which are located within the current urban area. Over the years, the city has advanced in different directions, generating a unique sociodemography that is linked to the characteristics of the territory. The historic center underwent a process of abandonment from the end of the 19th century to the middle of the 20th century, as industry and foreign trade moved from these areas to the outskirts. The valley of the Guadalhorce River to the northwest, which has more extreme temperatures and greater rural links, became the home of technology (Andalusia Technological Park) and knowledge (Campus of the University of Malaga). Along the eastern coastline, the original beachfront fishing neighborhoods were displaced by bourgeois houses and upper-middle class neighborhoods with sea views and a lower urban density. Along the western coastline, the remains of the 19th century industrial center gave way to middle and lower-middle working-class neighborhoods that were highly populated and relatively lacking in services.

During the second half of the 20th century, the degradation of the central areas was redirected, and a strategic urban revitalization movement transformed the configuration of the neighborhoods closest to the core of the city Centre. International tourism generated speculative pressures in many parts of the city and once again displaced vulnerable populations. However, one can still see the social, economic, environmental, and urban differences that have been consolidated in recent centuries.

Together, these waves of evolution have generated spatial and social segregation wherein some individuals and families live in areas with high standards of living but a large majority lives in neighborhoods with different types of inequalities and deficiencies. The arrival of the pandemic saw five-fold increases in the proportion families served by municipal social services and the investments necessary to prevent the marginalization of many others, providing further evidence that living conditions affect the resilience of individuals and neighborhoods against COVID-19.

Materials and Methods

During 2021, the CIEDES Foundation collaborated with the Malaga City Council and the University of Malaga to conduct an investigation examining the vulnerability of the different neighborhoods of Malaga in multiple dimensions (Bárcena-Martín et al. 2021). This study generated a global vulnerability index that consisted of four sub-indices: demographic, socioeconomic, social care, and territorial (spatial) vulnerability. This index allowed to examine the 362 neighborhoods and the 434 census sections (CSs) of Malaga using an in-house designed algorithm that grouped together the different spatial units. Analysis of more than 200 variables from official sources led to the selection of 19 variables from official sources and a short-term survey on living conditions, mostly coincides with those used by other studies and authors for vulnerability analysis in other cities in Spain (Ayuntamiento de Madrid 2018; Egea Jiménez and Soledad Suescún 2008). A normalization process was applied to these variables so all of them were ranged in [0, 1], being in all the cases 0 the best situation and 1 the worst. The variables were normalized and, despite being a maximizing or minimizing one, a higher value means a worse situation in that characteristic. Finally, the normalized variables were grouped in the previously mentioned four categories (demographic, socioeconomic, social care, and territorial) and a linear weighted aggregation was conducted to obtain a final vulnerability index. The authors chose to give the same weight to the dimensions and distribute the weights proportionally within each dimension. Note that this dimension’s aggregation method allows compensation between them. Therefore, authors assumed that, for example, a low score in the level of income can be compensated (according to the assigned weights) by an increase in the level of education and this compensation is constant.

This general vulnerability index was applied in another study analyzing the impact of the COVID pandemic in Malaga (paper under evaluation), wherein geostatistical analysis was used to delve as deeply as possible into the micro-scale of disease incidence and its correlation with vulnerability. The study found that the most vulnerable neighborhoods were the most highly affected by the COVID-19 pandemic.

We conducted a clustering analysis using the K-means, where we chose the number of clusters using the Elbow Method (that aims to minimize the within-cluster sum of squared errors-WSS) and trying to avoid having too small clusters. Using this approach, we selected five clusters (k = 5) as the best option when considering both the reduction of the WSS value and avoiding too small clusters. This methodological approach allowed us to obtain general patterns and to infer enough general conclusions, as well as analyzing the dissimilarities among clusters.

The results revealed that there were five different socio-spatial patterns for the propagation of the pandemic in Malaga, as the Table 1 shows. In this table, for each cluster we include the deviation of the cluster from the average of the whole population on the general vulnerability index, as well as from the vulnerability indexes corresponding to each of the four categories independently.

Table 1.

Deviation of the Clusters from the Average of the Whole Population on the Vulnerability Index.

Cluster	Global Vulnerability Index	Demographic Vulnerability Index	Socioeconomic Vulnerability Index	Social Care Vulnerability Index	Territorial Vulnerability Index
Cluster1	0.0069	−0.0073	0.0207	0.0152	−0.0010
Cluster2	−0.0083	0.0303	−0.0532	−0.0122	0.0018
Cluster3	0.0139	−0.0126	0.0566	0.0042	0.0072
Cluster4	0.0356	0.0056	0.0732	0.0450	0.0187
Cluster5	−0.0144	−0.0162	−0.0180	−0.0146	−0.0088

Source: Own elaboration based on Bárcena- Martín et al. (2021).

Cluster 1 (lower-middle class) included CSs with moderate socioeconomic status, welfare vulnerability, and low-income levels and work intensity. The incidence of COVID-19 was very high in cluster 1 CSs during the fourth wave, and the incidence showed a moderate rate of change on either side of the peak.

Cluster 2 (elderly and upper class) included CSs with the best socioeconomic conditions and quality of care (high income, high work intensity, advanced education, and little need for social interventions or material goods). These areas showed a low incidence of COVID-19 with relatively low numbers of cases during almost all waves.

Cluster 3 (young lower-middle class) included CSs with greater socioeconomic vulnerability than cluster 1, including low income, very low labor intensity, and a high poverty index. The pandemic manifested slowly in the CSs of this cluster, but there was a much greater increase in incidence during the fourth wave compared to those seen in clusters 1, 2, and 5.

Cluster 4 (elderly lower class) included a relatively small number of CSs, but these CSs were highly vulnerable. The neighborhoods in cluster 4 experienced the largest peaks during all four waves.

Cluster 5 (young upper-middle class) had positive conditions on all of the vulnerability sub-indices and almost all of the studied variables. The incidence of COVID-19 in CSs of cluster 5 increased during each of the waves, showing considerable intensity and slow recovery after the wave passed.

As this clustering analysis showed that COVID incidence was related to the vulnerability parameters of different areas, we next examined whether COVID-19 lethality was also related to vulnerability. To answer this question, we implemented the above-described methodology for lethality, and compared the results with those obtained for incidence.

The Andalusian Health Service provided daily data on the number of COVID-19 infections in Malaga by CS, and for each of these cases reported if the infected patient survived or not. The obtained information spanned from 5 March 2020 to 31 March 2021 for each CS, and thus covered the first four waves of the pandemic. This information was used for statistical, graphical, and cartographic analyses. The unit of spatial reference was the CS, which is smaller than the municipality and has easily identifiable limits, including natural terrain features, permanent buildings, and roads. Each CS had a population of about 1000–2500.

To facilitate the analysis, dates were relabeled to a modified Julian calendar. The study period started on 5 March 2020 (relabeled to the Julian calendar equivalent, day 64) and ended on 31 March 2021 (relabeled to day 455, to maintain the continuity of dates through the end of the year). The starting dates of the three waves were 5 March 2020, 21 June 2020, and 12 October 2020 (Table 2).

Table 2.

End Dates of Each Wave.

Wave 1	Wave 2	Wave 3
Till day 172	Till day 344	Till day 455
21/6/20	10/12/20	31/3/21

Source: Own elaboration based on data from Junta de Andalucía.

For each CS (CSi) on each day of a wave (dayj), the number of deaths per cases during the prior 14 days (C14dij) was calculated as the number of patients who finally died (maybe weeks later) among the total number of infections in the 14 days prior to dayj in the CSi, where i ranged from 1 to 434 and j ranged from 64 to 455. In other words, we have a record for each patient who got the COVID in a given date, and we know if that patient finally died, sometime in the future, due to COVID. So, for each CS, the number of diseases among the infected during the previous 14 days was obtained for each day from 5 March 2020 (day 64) to 31 March 2021 (day 455), as a measure of lethality on those 14 days. Each CS therefore had its own unique lethality profile, as indicated by the four representative CSs in Figure 1.

Figure 1.

Examples of lethality curves for four representative CSs. Source: Own elaboration based on data from Junta de Andalucía.

A K-means method-based clustering algorithm was applied to analyze the 434 curves with the Python package Kmeans. The K-means method (Hartigan and Wong, 1979) is one of the most widely used clustering methods (Chen et al. 2020). It aims to partition a set of n observations into k groups, in which each observation belongs to the group whose mean value is the closest to the observation. This method has been used in diverse fields, as market segmentation (Lichtenstein, Burton, and Netemeyer 1997), computer vision (Frigui and Krishnapuram 1999), geostatistics (Fouedjio 2016), and astronomy (Jang and Hendry 2007), and in many problems related to data mining (Berkhin 2006).

Here, the K-means algorithm was applied to the matrix of Euclidean distances between pairs of lethality curves, instead of incidence curves. Because each CS curve can be considered a 392-dimensional vector, the Euclidean distance in that 392-dimensional space was used to determine the similarity between each pair of curves. We selected k = 5 so the results would be comparable with those of vulnerability and contagion clusters, where analyses of the Elbow curve and the cardinality of clusters were conducted (paper under evaluation). However, the present work obtained very different results compared to the prior report. Firstly, according to the cardinality of the clusters, the incidence revealed well-balanced groups, whereas lethality revealed a single large cluster containing 372 CSs and four minor clusters (see Table 3).

Table 3.

Cardinals of Lethality and Incidence Clusters.

Lethality Cluster	Cardinal	Incidence Cluster	Cardinal
1	372	1	123
2	19	2	114
3	17	3	74
4	15	4	17
5	11	5	106

Source: Own elaboration.

If we look at the average curves of each cluster (Figure 2), the reason for the latter pattern becomes evident.

Figure 2.

Average curves of the lethality clusters. Source: Own elaboration based on data from Junta de Andalucía.

Note that there is one cluster of 372 CSs having quite low lethality throughout the study period, and four minor clusters including CSs that exhibited anomalously high lethality’s hot spots at certain time points (see below Figure 3). In this map, Cluster one (grayed) represents the low lethality normal areas, and the rest, clusters 2 to 5, represents the anomalously high lethality’s hot spots.

Figure 3.

Lethality hot-spots. Source: Own elaboration based on data from Málaga Municipality (OMAU) and CIEDES Foundation.

We next questioned whether these lethality hot spots were related to socioeconomic conditions and vulnerability, as seen for the analysis of incidence. To assess this, we followed two different paths. First, we analyzed each cluster individually according to the same set of variables used in our previous study (paper under evaluation), to establish the individual features of each cluster in the same framework used to study incidence. Second, we established two groups: the normal lethality group, which included all CSs from cluster 1; and the abnormal lethality group, which included those from clusters 2 through 5 (Figure 4).

Figure 4.

Normal and abnormal lethality curves (average). Source: Own elaboration based on data from Junta de Andalucía.

We then compared these groups according to the following set of variables (see Table 4) that were found to be closely related with vulnerability in our previous study (paper under evaluation) and in Bárcena-Martín et al. (2021). To compare the variables, we normalized data and use the criteria of “The more, the worse”. That means in Figures 1, 2, and 5 to 10, a bar to the right side means a worse value than the average of the whole population. The comparison of deviations from the mean of both groups (normal and abnormal lethality) concluded that the previous vulnerability conditions of these census sections are related with the evolution of the waves of contagion.

Table 4.

Examples of Lethality Profile for Four Representative CSs.

Variable	Description	Normalization
Vuln	Vulnerability index	The more the worse
COVIDInc	Total number of cases by 1000 inhab.	None
Age	Average of the population	None
Pop	Inhabitants of the CS	None
DepRate	Dependency rate	The more the worse
More75Alone	Older than 75 living alone	The more the worse
AgingRate	Aging rate	The more the worse
LifeExp	Life expectancy	The more the worse
HouseInc	House income	The more the worse
Illiterate	Illiterate	The more the worse
Unemployed	Unemployed	The more the worse
LaborInt	Labor intensity	The more the worse
AROPE	AROPE index	The more the worse
PaxAtt	People attended by social services	The more the worse
SocIntgr	Social integration needs	The more the worse
Homesize	Home size	The more the worse

Source: Own elaboration based on data from the Malaga Municipality (OMAU) and CIEDES Foundation.

Figure 5.

Characteristic of lethality Cluster 1. Source: Own elaboration based on data from Málaga Municipality (OMAU) and CIEDES Foundation.

Figure 6.

Characteristic of lethality Cluster 2. Source: Own elaboration based on data from Málaga Municipality (OMAU) and CIEDES Foundation.

Figure 7.

Characteristic of lethality Cluster 3. Source: Own elaboration based on data from Málaga Municipality (OMAU) and CIEDES Foundation.

Figure 8.

Characteristic of lethality Cluster 4. Source: Own elaboration based on data from Málaga Municipality (OMAU) and CIEDES Foundation.

Figure 9.

Characteristic of lethality Cluster 5. Source: Own elaboration based on data from Málaga Municipality (OMAU) and CIEDES Foundation.

Figure 10.

Deviation of normal and abnormal lethality clusters from the overall mean. Source: Own elaboration.

Results

As noted above, we first assessed each lethality cluster individually according to the same set of variables used in our previous study (paper under evaluation) to explain the incidence of the pandemic. For each cluster, we considered its deviation from the average of the whole population in Malaga for each variable and vulnerability index (general, demographic, socioeconomic, social care, and territorial). All variables and indexes were normalized: A negative deviation (left) indicates a better situation with respect to the average of the whole population, and a positive deviation (right) means a worse situation than the average of the whole population.

Cluster 1 summarizes 372 CSs and it was characterized by its low level of lethality at all time points, and by its normality. We observed only small deviations from the average values in all variables and vulnerability indices (Figure 5).

We have analyzed better the patterns in the other four abnormal clusters.

Cluster 2 presented overall negative results, with only a few variables of these components remaining positive, including the proportion of children under 16 years of age or good access to public services and green areas. The neighborhoods of this cluster were those experiencing very severe material poverty, with very low labor intensity and high unemployment. These neighborhoods were also characterized by low educational levels and quite old populations; this, along with the high proportion of minors, implies high dependency rates. The greatest COVID lethality in cluster 2 occurred during the first wave, when pronounced up-and-down curves were seen in almost all of the included neighborhoods. Lethality remained fairly controlled during the second wave. During the third wave, lethality rose again; the degree of change was relatively small but prolonged. These neighborhoods experienced a relatively late entry into each wave and showed the lowest level of lethality among the atypical clusters (Figure 6).

Cluster 2 neighborhoods had high proportions of those over 75 years of age and low proportions of those under 16 years of age; their dependency rates would thus be expected to be, although they showed relatively little utilization of public welfare and social services. These neighborhoods were characterized by high labor intensity, good employment conditions, and high levels of education and family income, but poor conditions of orientation, accessibility to services, and the most are located at an elevation that challenges accessibility, but the houses tend to be relatively large (Figure 7).

Cluster 4 neighborhoods were characterized by a low average income that would lead to severe material poverty, given that labor intensity is low, and unemployment is high. Although the population is not too old, the education level is low, and the use of social and welfare services is high. The houses are small in size and located in areas with high rainfall, although the housing orientation and temperature are good. The highest lethality occurred in the first wave, as in all clusters, but cluster 4 also showed lower-intensity peaks in the second and third waves. This cluster ranked third in lethality during the first wave but was the last to crest and resolve the situation. In the second wave, lethality remained fairly controlled; however, the slight increase was prolonged until almost the third wave, when this cluster experienced two significant peaks (Figure 8).

Cluster 5 included neighborhoods harboring large proportions of people over 75 years of age living alone and over 64 years of age, as well as some young people under 16 years of age. Family income was low. Although the educational level was not very low, there was high unemployment and low labor intensity. These were working-class neighborhoods without intensive use of social and assistance services, and very good access to public services. This cluster exhibited the highest lethality in the first wave of the pandemic and showed lethality during all subsequent waves, exhibiting curves with more pronounced crests than seen for the other clusters. It experienced the first wave ahead of the other clusters. During the second wave, it experienced a delay in the growth of lethality. In the third wave, there were multiple plateaus indicating prolonged periods of increased lethality (Figure 9).

We then unified these four abnormal clusters of lethality into a single atypical cluster and compared the vulnerability variables with respect to the average from the group of normal results.

As Figure 10 shows, this overall analysis revealed that outbreaks occurred in those census sections that, above all, have a relatively large proportion of individuals over 75 years living alone, along with high rates of aging and dependency. These neighborhoods are also typically characterized by especially small houses, a high rate of severe material deprivation, a low rate of education and an accordingly high need for social integration. Our work therefore shows that the CSs experiencing mortality peaks were generally characterized by high vulnerability, not only in general, but especially in aspects that are highly sensitive to COVID, such as high proportions of elderly individuals living alone or people living in particularly small houses.

Discussion

The territorial affectation processes of a phenomenon such as the COVID-19 pandemic can be complex, variable, and heterogeneous, often leading to the use of statistical tools that allow users to quickly understand what is happening. In this sense, the present work arose from previous studies that used synthetic indices to explain the vulnerability of neighborhoods to COVID 19 in terms of their economic, social, environmental, and governance dimensions. As note by Hyatt (2001) and Ebert and Welsch (2004), it is difficult to define indicators and there are important differences between simple indicators (combining two or more data sets) and synthetic indicators (treated with a mathematical function that synthesizes them) (EEA/AMAE, 2002). Pena Trapero (1977) and Zarzosa (1996) argued that although a constructed index generally fails to explain all the factors that can describe a latent variable, it will always approximate them. To measure spatial vulnerability in cities and its condition in a pandemic process, one must further introduce the temporal variable. Establishing a system of simple or synthetic indicators will often require information that is unavailable, making it necessary to introduce other, more robust statistical techniques such as the cluster method and tools, such as geographical information systems (Casas, Delmelle, and Varela 2010) or people movement mobile apps (Chen et al. 2022).

Despite the variability of data throughout the pandemic and the impossibility of strictly comparing data over time, the combination of synthetic indices and clusters utilized herein offers a quantitative tool that simplifies the attributes and weights of multiple variables to provide a broader explanation of the waves of incidence and case-fatality (lethality) that we wanted to assess. Nevertheless, in future research lines, we will improve our methodology considering the different techniques for choosing optimal partitions in the cluster analysis (Caballero et al. 2011; Scitovski et al. 2021).

A scan of the literature shows that the high transmissibility of the virus in cities has been influenced not only by specific characteristics of the virus, but also by factors related to social, economic, environmental, and governance determinants that previously existed, having evolved unevenly in the city’s urban geography (Bárcena-Martín et al. 2021; Credit 2020; Ríos Quituizaca et al. 2021).

Undoubtedly, the socioeconomic conditions, health resources, mechanisms for mobilizing the population, education, work, housing, basic services—in short, what we could call the way of life in the different urban areas, have modulated the territorial dynamics of the pandemic itself. This line of thought coincides with the work of Ríos Quituizaca et al. (2021), who proposed that the social characteristics of each territory should be add to individual factors to analyze the risks and vulnerability of neighborhoods in the COVID-19 pandemic and others respirational diseases. The study of Malaga has shown that social vulnerability is more than just exposure to risk; rather, it constitutes a combination of traits that combine potential adversities with the inability to respond and/or adapt to existing or potential risks (Escobar 2006). Indeed, vulnerability is affected by risk situations related to personal, family, professional, socioeconomic, and/or political environments, and these must be considered in neighborhood-level analyses (Salado Garcia, Rojas, and Cantergi 2008). An integrated sustainable urban development approach, especially one that also considers the importance of the governance dimension, is more consistent with the real dynamics of the pandemic in space and time (García-Peña, González-Medina, and Diaz-Sarachaga 2021). Future research lines could be related to the public policy and management involved in the resolution of this comprehensive problems, and the creation of tools for a new local governance (Huete García, Merinero Rodríguez, and Munoz Moreno 2016).

As indicated by Sharifi (2021), the resilience of cities will be linked to the capacities of cities and neighborhoods for planning, absorption, recovery, and adaptation. Characteristics such as flexibility, collaboration, diversity, redundancy, resourcefulness, and self-organization should be considered and enhanced at a granular level. In the future, resilience against pandemics will require pre-event planning, long-term visioning, early responses, integrated governance, community empowerment, and appropriate use of smart city solutions. Thus, micro-level analyses are an absolute necessity (Christopherson, Michie and Tyler 2010).

Conclusions

This research shows the importance of microscale spatial analysis, especially in historical moments such as the current one in which the COVID-19 pandemic has highlighted spatial inequalities, with the poorest suffering the worst consequences (Wade 2020). The comprehensive analysis of the sustainable development has been taken as a central approach for the statistical method and highlights the value of using economic, social, environmental, and governance indicators to understand the complexity of urban phenomena and their interrelationships (Del Castillo and Haarich 2013).

The interdisciplinary composition of the research team has enriched the conceptual field and has facilitated the elaboration of the method. Here, the spatial dimension of the sustainable development was addressed through the use of synthetic indices, georeferenced indicators at census-section scale, and parametric and non-parametric methods (such as cluster analysis). This multi-criterion approach to data analysis and decision-making enabled us to draw conclusions on the relationships between the waves of COVID-19 incidence and lethality and the preexisting conditions of poverty and vulnerability in cities.

During the waves of contagion, the incidence of disease was higher in neighborhoods that could be considered the most vulnerable from the economic and demographic points of view, whereas lethality exhibited hot-spot behavior that seemed to lack a direct correlation with the pattern of incidence. These findings are consistent with those previously reported by other authors (Credit 2020; Huang et al. 2021; Viezzer and Biondi 2021). This supports the accuracy of our approach and the effectiveness of the statistical methodology chosen to understand the urban dynamics of this complex phenomenon in space and time.

The use and study of all existing information allows us to learn from the exceptions or atypical values, which can help us identify patterns that can greatly facilitate the design of action plans. The results of our research make it possible to identify the main intervention needs in the city of Malaga with almost surgical precision and establish priority criteria that can be used in allocating resources to reduce poverty in all of its dimensions. The 2030 Agenda named the fight against all types of vulnerability as its main objective and a key to achieving sustainable development. The activation of territorial governance strategies based on statistics and data systems for monitoring, controlling, and evaluating public policies was identified by the UN as being key to localizing the 2030 Agenda in cities (UCLG 2019).

As noted by Abrams and Szefler (2020), the effects of COVID-19 have shed light on the broad disparities within our society, providing us with an opportunity to address those disparities moving forward. In the future, we must ensure that social and environmental dimensions of sustainability are considered in addition to economic development. Planners could capitalize on the pandemic to improve future urban development (Sharifi and Khavarian-Garmsir 2020), and this can have implications for public and private sector groups (Mathew and McDonald 2006).

As reflected by the UN-Habitat studies, one of the major action areas in response to the pandemic is the provision of “urban-evidence-based mapping and knowledge for informed decision-making (UN-Habitat 2020). Thus, it will be very important that we continue to publish studies on the dynamics of the pandemic in cities and the effects of government policies on these dynamics, especially given that deprivation and vulnerability are known to be key drivers of COVID-19 (Morrissey et al. 2021).

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by Junta de Andalucía under Grant ERDF (CV20-27760) within the framework of the “Post-Covid Vulnerability and Resilience in the Metropolitan Area of Malaga” project.

ORCID iD

Carmen García-Peña

References

Abrams

E. M.

Szefler

S. J.

2020. “COVID-19 and the Impact of Social Determinants of Health.” The Lancet Respiratory Medicine 8 (7): 659–661. doi: 10.1016/S2213-2600(20)30234-4

Aguilar-Palacio

Maldonado

Malo

Sanchez-Recio

Marcos-Campos

Magallon-Botaya

Rabanaque

M. J.

. 2022. “Inequality, Public Health, and COVID-19: An Analysis of the Spanish Case by Municipalities.” European Journal of Health Economics. https://www-mdpi-com-s.web.bisu.edu.cn/1660-4601/18/12/6607

Ahmed

Williamson

Hamid

M.A.

Ashraf

. 2020. “United States County-level COVID-19 Death Rates and Case Fatality Rates Vary by Region and Urban Status.” Healthcare 8: 330. doi: 10.3390/healthcare8030330

Amate-Fortes

Guarnido-Rueda

, 2022. “ICOVID-19 Inequalities: Individual and Area Socioeconomic Factors (Arag on, Spain).” International Journal of Environmental Research and Public Health 18 (12): 6607. doi: 10.1007/s10198-022-01455-9

Auler

A.C.

Cássaro

F. A. M.

da Silva

V. O.

Pires

L. F.

2020. “Evidence that High Temperatures and Intermediate Relative Humidity Might Favor the Spread Of COVID-19 in Tropical Climate: A Case Study for the Most Affected Brazilian Cities.” The Science of the Total Environment 729 (2020): 139090.

Ayuntamiento, de Madrid

2018. Metodología para la elaboración del índice de vulnerabilidad territorial de barrios y distritos de Madrid y ranking de vulnerabilidad. Madrid: Coordinación territorial y cooperación político-social.

Baldasano

J. M.

2020. “COVID-19 Lockdown Effects on Air Quality by NO2 in the Cities of Barcelona and Madrid (Spain).” Science of the Total Environment 741: 140353. doi: 10.1016/j.scitotenv.2020.140353

Bambra

Riordan

Ford

Matthews

. 2020. “The COVID-19 Pandemic and Health Inequalities.” J Epidemiol Community Health 74 (11): 964–968. doi: 10.1136/jech-2020-214401

Bárcena-Martín

García-Peña

Marín Cots

Molina Luque

Palomares Pastor

Ruiz Sinoga

J. D.

. 2021. Vulnerabilidad en los barrios de Málaga. Edición Fundación CIEDES.

10.

Bashir

M. F.

Bilal

Komal

Bashir

M. A.

Farooq

T. A.

Bashir

. 2020. “Correlation between Environmental Pollution Indicators and COVID-19 Pandemic: A Brief Study in Californian Context.” Environmental Research 187 (2020): 109652. doi: 10.1016/j.envres.2020.109652

11.

Berkhin

2006. “A Survey of Clustering Data Mining Techniques.” In Grouping Multidimensional Data, 25–71. Berlin, Heidelberg: Springer.

12.

Bilal

Tabb

L. P.

Barber

Diez Roux

A. V.

. 2021. “Spatial Inequities in COVID-19 Testing, Positivity, Confirmed Cases, and Mortality in 3 U.S. Cities.” Applied Economics 174 (7): 936–944. doi: 10.7326/m20-3936

13.

Caballero

Laguna

Martí

Molina

. 2011. “Scatter Tabu Search for Multiobjective Clustering Problems.” Journal of the Operational Research Society 62 (11): 2034–2046.

14.

Casas

Delmelle

Varela

. 2010. “A Space-Time Approach to Diffusion of Health Service Provision Information.” International Regional Science Review 33 (2): 134–156.

15.

Chen

Clark

W. A.

Shi

. 2022. “What Affects Perceived Health Risk Attitude During the Pandemic: Evidence From Migration and Dining Behavior in China.” International Regional Science Review.

16.

Chen

Igan

D. O.

Pierri

Presbitero

A. F.

Soledad

Peria

. 2020. “Tracking the economic impact of COVID-19 and mitigation policies in Europe and the United States.” IMF Working Papers 2020 (125).

17.

Christopherson

Michie

Tyler

. 2010. “Regional Resilience: Theoretical and Empirical Perspectives.” Cambridge Journal of Regions, Economy and Society 3 (1): 3–10.

18.

Credit

2020. “Neighbourhood Inequity: Exploring the Factors Underlying Racial and Ethnic Disparities in COVID-19 Testing and Infection Rates Using ZIP Code Data in Chicago and New York.” Regional Science Policy & Practice 12 (6): 1249–1271. doi: 10.1111/rsp3.12321

19.

Del Castillo

Haarich

. 2013. Desarrollo Urbano Sostenible cofinanciado por el FEDER en España 2014-2020: Directrices Estratégicas y Prioridades de Inversión. Final Report for DG Regio, European Commission. doi: 10.2776/40263

20.

Ebert

Welsch

. 2004. Meaningful environmental indices: a social choice approach. Journal of Environmental Economics and Management 47 (2): 270–283.

21.

Egea Jiménez

Soledad Suescún

J. I.

. 2008. “Migraciones y Conflictos: El Desplazamiento Interno En Colombia.” Convergencia 15 (47): 207–235.

22.

Escobar

Luis

. 2006. “Indicadores Sintéticos de Calidad Ambiental: Un Modelo General Para Grandes Zonas Urbanas.” Eure (Santiago) 32 (96): 73–98.

23.

Florida

Mellander

. 2022. “The Geography of COVID-19 in Sweden.” The Annals of Regional Science 68 (1): 125–150.

24.

Frigui

Krishnapuram

. 1999. “A Robust Competitive Clustering Algorithm with Applications in Computer Vision.” IEEE Transactions on Pattern Analysis and Machine Intelligence 21 (5): 450–465.

25.

Fouedjio

. 2016. “A hierarchical Clustering Method for Multivariate Geostatistical Data.” Spatial Statistics 18: 333–351.

26.

Zhai

. 2021. “Examining the Spatial and Temporal Relationship Between Social Vulnerability and Stay-at-Home Behaviors in New York City during the COVID-19 Pandemic.” Sustainable Cities and Society 67: 102757.

27.

García-Peña

González-Medina

Diaz-Sarachaga

J. M.

. 2021. “Assessment of the Governance Dimension in the Frame of the 2030 Agenda: Evidence from 100 Spanish Cities.” Sustainability 13: 5519. DOI: 10.3390/su13105519

28.

Gutierrez

M. J.

Inguanzo

Orbe

. 2021. “Distributional Impact of COVID-19: Regional Inequalities in Cases and Deaths in Spain during the First Wave.” Applied Economics 52 (31): 3626–3657. DOI: 10.1080/00036846.2021.1884838

29.

Harris

2020. “Exploring the Neighbourhood-Level Correlates of Covid-19 Deaths in London Using a Difference Across Spatial Boundaries Method.” Health & Place 66: 102446. doi: 10.1016/j.healthplace.2020.102446

30.

Hartigan

J. A.

Wong

M. A.

1979. Algorithm AS 136: A k-medias Clustering Algorithm. Journal of the Royal Statistical Society, Series C (Applied Statistics) 28(1): 100–108.

31.

Huang

Jackson

Derakhshan

Lee

Pham

Jackson

Cutter

S. L.

2021. “Urban-Rural Differences In COVID-19 Exposures and Outcomes in the South: A Preliminary Analysis of South Carolina.” PLoS ONE 16 (2): e0246548. doi: 10.1371/journal.pone.0246548

32.

Huete García

M. A.

Merinero Rodríguez

Munoz Moreno

. 2016. “Urban Regeneration Policy from the Model of Integrated Urban Development in the European Union: An Analytical Approach Based on the Study of Spanish Cities.” Local Government Studies 42 (2): 267–286. doi: 10.1080/03003930.2015.1110520

33.

Hyatt

S. B

. 2001. From citizen to volunteer. The new poverty studies: The ethnography of power, politics, and impoverished people in the United States. 201–235.

34.

Jang

Hendry

. 2007. “Cluster Analysis of Massive Datasets in Astronomy.” Statistics and Computing 17 (3): 253–262.

35.

Jon

. 2020. “A Manifesto for Planning After the Coronavirus: Towards Planning of Care.” Planning Theory 19 (3): 329–345. doi: 10.1177/1473095220931272

36.

Karaye

Horney

. 2020. “The Impact of Social Vulnerability on COVID-19 in the U.S.: An Analysis of Spatially Varying Relationships.” American Journal of Preventive Medicine 59 (3): 317–325. doi: 10.1016/j.amepre.2020.06.006

37.

Kayanan

Moore-Cherry

Clavin

. 2021. “Cities Under Lockdown: Public Health, Urban Vulnerabilities and Neighbourhood Planning in Dublin.” Global Reflections on Covid 19: 189–198.

38.

Khavarian-Garmsir

A. M.

Sharifi

Moradpour

. 2021. “Are High-Density Districts More Vulnerable to the COVID-19 Pandemic?” Sustainable Cities and Society 70: 102911. doi: 10.1016/j.scs.2021.102911

39.

Kodera

Rashed

E.A.

Hirata

. 2020. “Correlation Between COVID-19 Morbidity and Mortality Rates in Japan and Local Population Density, Temperature, and Absolute Humidity.” International Journal of Environmental Research and Public Health 17 (15): 5477. doi: 10.3390/ijerph17155477

40.

Krellenberg

Koch

. 2021. “Conceptualizing Interactions Between Sdgs and Urban Sustainability Transformations in Covid-19 Times.” Politics and Governance 9 (1): 200–210.

41.

Peng

Wang

Feng

. 2021. “Built Environment and Early Infection of COVID-19 in Urban Districts: A Case Study of Huangzhou.” Sustainable Cities and Society 66 (2021): 102685. doi: 10.1016/j.scs.2020.102685

42.

Liang

Shi

Zhao

Liu

Sarnat

J. A.

Gao

Schwartz

Ebelt

S. T.

Liu

Scovronick

Chang

H. H.

2020. “Urban Air Pollution May Enhance Covid-19 Case-Fatality and Mortality Rates in the United States.” The Innovation 1 (3): 100047. doi: 10.1016/j.xinn.2020.100047

43.

Lichtenstein

D. R.

Burton

Netemeyer

R. G.

. 1997. “An Examination of Deal Proneness Across Sales Promotion Types: A Consumer Segmentation Perspective.” J Retailing 73: 283–297.

44.

Liu

Zhou

Yao

Zhang

. 2021. “Impact of Meteorological Factors on the COVID-19 Transmission: A Multi-City Study in China.” The Science of the Total Environment 726: 138513. doi: 10.1016/j.scitotenv.2020.138513

45.

Macharia

P. M.

Joseph

N. K.

Okiro

E. A.

. 2020. A vulnerability index for COVID-19: spatial analysis at the subnational level in Kenya. BMJ Global Health 5 (8): e003014.

46.

Martins-Filho

P. R.

de Souza Araújo

A. A.

Quintans-Júnior

L. J.

Santos

V. S.

. 2020. “COVID-19 Fatality Rates Related to Social Inequality in Northeast Brazil: A Neighbourhood-Level Analysis.” Journal of Travel Medicine 27 (7): taaa128. doi: 10.1093/jtm/taaa128

47.

Matthew

R. A.

McDonald

. 2006. “Cities Under Siege: Urban Planning and the Threat of Infectious Disease.” Journal of the American Planning Association 72 (1): 109–117. doi: 10.1080/01944360608976728

48.

Miramontes Carballada

Á.

Lois González

Perles Roselló

M. J.

Miranda Páez

. 2021. Riesgo de contagio por COVID-19 en un entorno urbano (Malaga, noviembre del 2020). Boletín De La Asociación De Geógrafos Españoles (91)

49.

Mishra

S. V.

Gayen

Haque

S. M.

. 2020. “COVID-19 and Urban Vulnerability in India.” Habitat International 103: 102230. doi: 10.1016/j.habitatint.2020.102230

50.

Morrissey

Spooner

Salter

Shaddick

. 2021. “Area Level Deprivation and Monthly COVID-19 Cases: The Impact of Government Policy in England.” Social Science & Medicine 289: 114413. doi: 10.1016/j.socscimed.2021.114413

51.

Nguimkeu

Tadadjeu

. 2021. “Why is the Number of COVID-19 Cases Lower than Expected in Sub-Saharan Africa? A Cross-Sectional Analysis of the Role of Demographic and Geographic Factors.” World Development 138: 105251. doi: 10.1016/j.worlddev.2020.105251

52.

Patel

J. A

Nielsen

F. B. H.

Badiani

A. A.

Ass

Unadkat

Patel

Ravindrane

Wardle

. 2020. “Poverty, Inequality and COVID-19: The Forgotten Vulnerable.” Public Health 183: 110–111.

53.

Prata

D. N.

Rodrigues

Bermejo

P. H.

. 2020. “Temperature Significantly Changes COVID-19 Transmission in (Sub)Tropical Cities of Brazil.” The Science of the Total Environment 729 (2020): 138862. doi: 10.1016/j.scitotenv.2020.138862

54.

Qian

Fan

. 2020. “Who Loses Income During the COVID-19 Outbreak? Evidence from China.” Research in Social Stratification and Mobility 68: 100522. doi: 10.1016/j.rssm.2020.100522

55.

Rahman

M. R.

Islam

A. H. M. H.

Islam

M. N

. 2021. “Geospatial Modelling on the Spread and Dynamics of 154 Day Outbreak of the Novel Coronavirus (COVID-19) Pandemic in Bangladesh Towards Vulnerability Zoning and Management Approaches.” Model Earth Syst Environ 7: 2059–2087. doi: 10.1007/s40808-020-00962-z

56.

Ríos Quituizaca

Calderón

Piedra

Guerrero

Narváez

. 2021. “Propuesta De Análisis Territorial Para Enfrentar La Pandemia Por SARS-Cov-2 Basado En El Perfil De Neumonía e Influenza en Ecuador 2016-2018.” Atención Primaria 53 (5): 102021. doi: 10.1016/j.aprim.2021.102021

57.

Rodríguez‐Pose

Burlina

. 2021. Institutions and the uneven geography of the first wave of the COVID‐19 pandemic. Journal of Regional Science 61 (4): 728–752.

58.

Salado García

María Jesús

Rojas

Alejandra

Cantergiari

Carolina

. 2008. “Modelos Territoriales Y Sistemas De Indicadores De Sostenibilidad: Valoración Crítica De Algunas Experiencias Concretas. Serie Geográfica - Profesora María de los Ángeles Díaz Muñoz.” In Memoriam Número 14 - 131 – 145.

59.

Sannigrahi

Pilla

Basu

A. S.

Molter

. 2020. “Examining the Association Between Socio-Demographic Composition and COVID-19 Fatalities in the European Region Using Spatial Regression Approach.” Sustainable Cities and Society 62: 102418. doi: 10.1016/j.scs.2020.102418

60.

Scitovski

Sabo

Martínez-Álvarez

Ungar

Š.

. 2021. Cluster Analysis and Applications, 08–23. Cham: Springer. doi: 10.1007/978-3-030-74552-3

61.

Shams

S. A.

Haleem

Javaid

. 2020. “Analyzing COVID-19 Pandemic for Unequal Distribution of Tests, Identified Cases, Deaths, and Fatality Rates in the Top 18 Countries.” Diabetes & Metabolic Syndrome: Clinical Research & Reviews 14 (5): 953–961. doi: 10.1016/j.dsx.2020.06.051

62.

Sharifi

. 2021. “The COVID-19 Pandemic: Lessons for Urban Resilience.” In COVID-19: Systemic Risk and Resilience. Risk, Systems and Decisions, edited by Linkov

Keenan

J. M.

Trump

B. D.

, eds Cham: Springer. doi: 10.1007/978-3-030-71587-8_16

63.

Stiglitz

J. E

. 2020. Priorities for the COVID-19 Economy. Project Syndicate.

64.

Sharifi

Khavarian-Garmsir

A. R.

. 2020. “The COVID-19 Pandemic: Impacts on Cities and Major Lessons for Urban Planning, Design, and Management.” Science of the Total Environment 749: 142391. doi: 10.1016/j.scitotenv.2020.142391

65.

UCLG . 2019. United Cities and Local Governments. Towards the localization of the SDGs. Local and Regional Governments’ Report to the 2019 HLPF, 3rd Report.

66.

UN-Habitat . 2020. UN-Habitat COVID-19 Response Plan. https://unhabitat.org/sites/default/files/2020/04/final_un-habitat_covid-19_response_plan.pdf

67.

UN System Staff College . 2017. Understanding the Dimensions of Sustainable Development. https://www.unssc.org/news-and-insights/news/understanding-dimensions-sustainable-development-now-available-8-languages/. accessed on 9 May 2020.

68.

Viezzer

Biondi

. 2021. “The Influence of Urban, Socio-Economic, and Eco-Environmental Aspects on COVID-19 Cases, Deaths and Mortality: A Multi-City Case in the Atlantic Forest, Brazil.” Sustainable Cities and Society 69: 102859. doi: 10.1016/j.scs.2021.102859

69.

Wade

2020. An Unequal Blow. doi: 10.1126/science.368.6492.700

70.

Whitehead

Barr

Taylor-Robinson

. 2020. COVID-19: We are not “all in it together”—less privileged in society are suffering the brunt of the damage. Available online: https://blogs.bmj.com/bmj/2020/05/22/COVID-19-we-are-not-all-in-it-together-less-privileged-in-society-are-suffering-the-brunt-of-the-damage/

71.

Nethery

R. C.

Sabath

B. M.

Braun

Dominici

. 2020. “Air Pollution and COVID-19 Mortality in the United States: Strengths and Limitations of an Ecological Regression Analysis Science.” Advances 6 (45): eabd4049. doi: 10.1126/sciadv.abd4049