Risk factors and spatial distribution associated with deaths due to COVID-19: an integrative review

Background and objectives: understanding the clinical-epidemiological and environmental factors related to deaths due to COVID-19 and their distribution in space can serve as subsidies to direct and implement more effective health actions for vulnerable populations. Thus, the objective was to synthesize the scientific evidence related to risk factors and spatial distribution of deaths due to COVID-19 in the world. Content: this is an integrative literature review, and the following guiding question emerged: what is the scientific evidence related to risk factors and spatial distribution of deaths due to COVID-19 in the world? Searches were carried out in the Scientific Electronic Library Online (SciELO) and the Scopus, Web of Science and National Library of Medicine (PubMed) databases in June 2021. Original studies in Portuguese, English or Spanish, without time frame, excluding studies with a specific age group or with an audience with specific comorbidity, were used. A total of 25 studies were included, with findings in different scenarios around the world. Factors such as age, sex, pre-existing diseases were associated with


INTRODUCTION
COVID-19 is a disease that presents rapid spread, mainly affects the respiratory system 1 , but can affect other organs and systems of the human body. 2 Since it demonstrates high contagion, accelerated transmission between humans 3 and dissemination in many countries on several continents, the World Health Organization (WHO) declared, on March 11, 2020, a pandemic state. 4 It is a viral, infectious disease with respiratory symptoms, whose etiological agent is SARS-CoV-2, a type of beta coronavirus. The predominant symptoms are fever, malaise and cough, with mild course in most infected 5 . However, some patients develop the most severe form, and when the organism cannot recover, they evolve to death. 6 It is in this context and pandemic scenario that many cases and deaths related to the disease worldwide have been and continue to be recorded. Assessing the overview of COVID-19 deaths worldwide, until October 24, 2021, there were a total of 4,927,723 deaths. To date, the United States had the highest cumulative number of deaths, with 726,846, followed by Brazil, with 604,228, India, with 453,042, Mexico, with 285,347, and Russia, with 228, 453. 4 As for the global mortality rate, until October 23, 2021, a rate of 627.8 deaths/1 million inhabitants was recorded. Among countries with a population of more than 1 million, Peru had the highest rate, with 5,995.9 deaths/1 million inhabitants, followed by Bosnia, with 3,469.3/1 million inhabitants. 7 It is believed that individual and collective factors can interfere with these rates and influence the outcome of discharge or death, such as socioeconomic factors and vulnerabilities related to housing, population income conditions, aging and level of social exclusion. 8 Moreover, the approach, clinical guidance and previous identification of comorbidities are equally important, as these are significant points in the course of a disease.
The clinical management of SARS-CoV-2 infection still requires further clarification to obtain better control over the disease, and, as in other pathologies, it is inferred that early diagnosis and follow-up act positively and can prevent the most serious manifestations of COVID-19. The virus causes an acute respiratory syndrome that ranges from mild to very severe cases, with evolution to severe respiratory failure, and its lethality differs according to associated comorbidities and age. 9 With regard to comorbidities, the most severe cases have been recorded in older adults who have some associated disease, especially hypertension and diabetes mellitus, but this association was also verified in cardiac and respiratory diseases. 10 Regarding age, it is verified that over 50 years old is positively associated with cases of deaths. 11,12 It is also noteworthy that knowledge of how diseases are distributed and disseminated in time and space is a central point of spatial epidemiology and health geography. Spatial analysis is essential to understand the spatial spread of an infection and its association with the community and the environment 13 , and can thus be used in the field of health research, as it brings a significant contribution to the observation of diseases and injuries in an area or region, making it possible to verify how each territory's particularities influence the dissemination of a given disease.
Understanding the clinical-epidemiological and environmental factors and their distribution in space can serve as an instrument for directing more effective health actions to vulnerable populations. Considering that conducting literature review studies related to risk factors and their associations with COVID-19 mortality can collaborate with professionals, managers and health services in promoting a more efficient clinical and epidemiological management, the objective was to synthesize the scientific evidence related to risk factors and spatial distribution of COVID-19 deaths worldwide.

METHODS
This is an integrative literature review, which aims to synthesize the knowledge about a given subject, making use of a systematic process and with scientific rigor. 14 The construction of this review consisted of the following phases: research question formulation; database search and primary study selection; study screening; analysis and synthesis of selected studies; and presentation of results.
For data collection, the PICo strategy (Population or problem, Phenomenon of Interest and Context) was used. 15 Problem (P) covered the deaths due to COVID-19, Interest (I), risk factors and spatial analysis, and Context (Co), studies published worldwide. Thus, the following guiding question was formulated: what is the scientific evidence related to risk factors and spatial distribution of COVID-19 deaths in the world?
The searches for the studies took place in June 2021, in the Web of Science, Scopus, National Library of Medicine (PubMed) and in the Scientific Electronic Library Online (SciELO) databases, using the Descriptors in Health Sciences (DesC) in Portuguese: "COVID- 19", "Mortalidade", "Fatores de Risco", "Análise Espacial" e "Saúde Global". The corresponding terms in English of the Medical Subject Headings (MeSH) of the National Library were used in the databases: "COVID-19", "Mortality", "Risk Factors", "Spatial Analysis" e "Global Health". The descriptors were combined with Boolean operators AND and OR. The crosses were performed as follows: (COVID-19) AND (mortality) AND ((Risk Factors) OR (Spatial Analysis)) AND (Global Health).
In study selection, we included original studies, made available free of charge, in Portuguese, English or Spanish, and without time frame. Editorials, letters to the editor, expert opinions, reviews (literary, integrative and systematic), theses, dissertations, studies that did not answer the guiding research question and studies in which research was conducted with groups of patients with specific age and/or comorbidity were excluded.
For the initial analysis of pre-selected articles, the Rayyan Systems Inc instrument was used, a free technological application web version, which presents in its interface the studies' main information, collaborating with the initial screening of the articles through a semiautomated process. 16 At this stage, a thorough reading of titles and abstracts was carried out in order to verify which were related to the research question and the inclusion and exclusion criteria adopted.
The strategy adopted in article search and selection was based on the Preferred Reporting Items for Systematic reviews and Meta-Analyzes (PRISMA) model. 17 The studies selected for the application of eligibility criteria were read and analyzed in full, in order to identify which met the objective of this review and thus select those that composed the final sample.
Thus, study characterization was carried out through an instrument designed to present the data related to the objective of this review. The data for descriptive analysis of this instrument are authorship, year of publication, country where the study was conducted, population, data source, objective(s), main findings and study limitations.
Regarding selected studies' limitations, incomplete feeding of some variables in consulted databases, possibility of underreporting and ecological fallacy stood out.
Chart 1 shows the synthesis of studies included in this review. Brown ethnicity is more prone to death due to COVID-19 than other ethnicities, it appeared as the second variable more associated with mortality, being the first age. As for comorbidities, CVD was the most associated with death (77.4%). Pattern of increased deaths at a median age of 65.3 years and higher percentages of death in males (67.2%). Sorci, Faivre, Morand. World, 2020. 11 3,778 COVID-19 patients from 72 countries. Secondary Data by 11 June 2020 from the European Centre for Disease Prevention and Control (ECDC).
Analyze the factors that may cause variability in COVID-19 MR between countries.
Worldwide lethality rate due to COVID-19 was around 4%. Countries with the lowest CVD MR had the lowest fatality rate, and countries with the highest MR from smoking and people over the age of 70 years had the highest mortality. Analyze the progression of COVID-19 in the first 7 days in the ICU and the prognostic factors of mortality in patients.
It showed moderate mortality in the ICU of 24%. It suggests that creatinine, ddimer, lactate, potassium, and ischemic heart disease are predictors of mortality in these patients. Thus, 58 of mechanically ventilated patients died, with a predominance of males and ages ranging from 70 to 81 years. Cao Y., Hiyoshi A., Montgome ry S., World, 2020. 29 10, 445,656 confirmed COVID-19 cases and 511,130 deaths due to two hundred and nine countries and territories worldwide. Secondary data from the Our World in Data website.
Investigate the influence of demographic and socioeconomic factors on COVID-19 case fatality rate (CFR) globally.
The mean COVID-19 MR of countries and territories is about 2% to 3% worldwide. Risk factors associated with deaths were CVD and DM. SA: used spatial autocorrelation (Moran's Index), Pearson's correlation and multivariate linear regression. There were spatial clusters of deaths in Yemen (Asia), Western and Northern Europe and North America. The lethality rate of COVID-19 was higher in European and North American countries than that observed in Asia. The USA and the United Kingdom had high mortality (7.84%) compared to Japan and Singapore (1.54%). Obesity, DM and age are positively associated with MR. Ambient air pollution was a risk factor for higher lethality rates due to COVID-19. The highest number of deaths due to COVID-19 was mainly in the western part of India. Associated factors were socioeconomic level and age greater than 50 years. SA: used the OLS (ordinary least square) and GWR (geographicallyweighted regression) models and spatial correlation (Moran's Index) to verify associations of deaths. Significant clusters of deaths due to the disease were identified in the state of West Bengal. Castro et al, Brazil, 2021. 34 5,082,637 confirmed cases and 150,198 deaths in the period from February 23, 2020 to October 10, 2020. Secondary data from state health departments.

Middya
Understand, measure and compare the pattern of spread of COVID-19 cases and deaths in Brazil at thin spatial and temporal scales.
The first cluster of COVID-19 deaths was in May 2020 in Recife. Afterwards, five others emerged until June 2020. The clusters were around Fortaleza, Rio de Janeiro, Amazonas, Pará and Amapá. In January 2021, Manaus (the capital of Amazonas) had a cluster with significant cases and deaths, with collapse of the hospital network and death rates 39.8% higher than in 2020. SA: used spatiotemporal scanning and spatial correlation techniques (Moran's Index), presenting thematic maps with clusters of cases and deaths in Brazilian states by epidemiological weeks. Mean age 62 years, most were male (57%) with a history of hypertension (60%), obesity (41%), and diabetes (42%). SA: LASSO. The geospatial distribution of patients' homes in relation to population density was mapped by USA regions, and the analysis showed that geographic residences of patients hospitalized with COVID-19 were associated with greater population housing or were located in regions with clusters. Characterize COVID-19 patients in US hospitals and examine risk factors associated with in-hospital mortality.
The in-hospital MR was 20.3%. The use of hydroxychloroquine and azithromycin was associated with a higher chance of mortality. Identify the risk factors associated with mortality and survival due to COVID-19 in a state in northeastern Brazil.
6.3% died. Variables that potentiated MR were elderly, CVD, DM, neurological diseases and lung diseases. The difference between the sexes was not significant, but the male had a higher death (relative risk of 1.1, while for females, 1.0). Analyze the factors at the municipal level associated with a high MR due to COVID-19 in Mexico.

Contreras-
At the municipal level, the prevalence of DM, obesity, population density and others influenced mortality. In 82 municipalities, MR was higher than expected, with 13.8 deaths/100,000 inhabitants. SA: the Poisson regression model was used to assess the binary outcome of death by COVID-19 and in the independent variables analyzed other characteristics and comorbidities. Alamdari NM et al, Iran, 2020. 18 It included 459 hospitalizations due to COVID-19 with 63 deaths. Data extracted from forms validated from electronic health records between January 30 and April 5, 2020.
Determine prognostic factors associated with COVID-19 mortality in patients admitted to a hospital in Tehran, Iran.
Older age, higher BMI, presence of lymphopenia, hypomagnesemia and elevated creatinine on admission have a higher risk of mortality. Predominance of male patients in hospitalizations 69.7% and deaths 77.8%. The MR was higher in patients with comorbidities of DM, malignancy, CKD and obesity. Males, indigenous (speakers of the indigenous language) and older were more likely to die. Those with DM, obesity, CKD, hypertension and use of immunosuppressants had a higher risk of death. CVD were not associated with higher mortality. The interaction of DM with other comorbidities such as CKD and hypertension doubles the risk of COVID-19 mortality.

Alshogran
OY et al, World, 2021. 19 Confirmed cases and deaths due to COVID-19 in 113 countries that had more than 100 positive cases by April 13, 2020. Secondary data collected from the Johns Hopkins University coronavirus resource center.
Check the association of social, demographic, non-communicable diseases and health care factors with deaths due to COVID-19 worldwide.
The MR of COVID-19 varies between countries by 4.2 ± 3.8%, about half of the countries had a MR > 3.2%. There was no significant positive association at the country level between comorbidities hypertension and DM and deaths. SA: multivariate logistic regression and Spearman correlation. The analyses showed that the proportion of people over 60 years of age was positively correlated with fatal cases.
It was identified in the global context that several factors appeared as potentiators or associated with deaths due to COVID-19. Some of these factors were distinct or had significant differences depending on the region, country, or continent. Variables such as advanced age and male gender were associated with the risk of death due to the disease in almost all studies.
COVID-19 mortality rates in the world were heterogeneous and varied between 2% and 4.2% of those affected by the disease. 11,19,29 Among patients with worsening conditions and hospitalized in the Intensive Care Unit (ICU), these rates were much more expressive and changed from one health institution to another. Regarding mortality rates among hospitalized patients, there was a variation between 20.3% and 21.7% in the USA, 12,38 24% in Europe and 25% in Africa. 22,23 In Spain and Italy, these rates were higher, respectively 42.4% and 44.3%. 21,26 It was noted that, in outpatients, the mortality rate was 13% in the USA, 12,38 significantly lower than in patients in hospital. This discrepancy between the rates can be justified due to age. Outpatients were mostly younger than hospitalized patients. Furthermore, in the hospitalized group, there were people more likely to have pre-existing diseases (asthma, chronic obstructive pulmonary disease, hypertension, obesity, DM), changes in vital signs 12 and more severe conditions, such as severe acute respiratory syndrome, requiring the use of mechanical ventilation. 23,38 Specifically, regarding age, all studies that evaluated this variable demonstrated the association of higher death rates among older patients. A study that analyzed several countries found that 80% of deaths caused by COVID-19 occurred among adults aged ≥ 65 years. 10 Other studies showed similar results, thus showing that older adults aged 60 years and over had a higher risk of death. [18][19][20][22][23][24]26,28,31 The greater vulnerability to COVID-19 in this age group may be associated with the weakening of the immune system in the fight against infections, 39,40 when the functions of T and B cells become potentially more defective with high production of type 2 cytokines, thus causing a deficiency in the control of viral replication, which is possibly related to the worse clinical prognosis in this age group. 28,41 Gender was another factor that was associated with higher mortality due to COVID-19, with the male being the most prevalent. 10,12,18,21,22,[24][25][26][27][28]38 In China, a retrospective study analyzing factors associated with mortality due to COVID-19 demonstrated, through multivariate logistic regression analysis, 2.75 more chances of mortality in men than in women. 28 Approximate value was verified in a survey conducted in a hospital unit in Iran, in which the odds ratio of mortality was three times higher in men. 18 This higher mortality in men can be analyzed from some factors, such as higher prevalence of pre-existing diseases in this group (coronary, chronic lung diseases and DM), more frequent risk behaviors such as smoking and alcohol habits, occupational exposure and sexual, genetic and hormonal differences. 42 Additionally, women generally produce a more effective and adaptive immune response to viruses, which favors a less severe evolution of  Regarding the clinical manifestations of patients who progressed to death, the most frequent were dyspnea, fever, cough, 22,28 severe acute respiratory syndrome and consequent need for mechanical ventilation support. 18,23,26 This is justified due to the disease predominantly affecting the respiratory system, causing upper and/or lower airway infection. 44,45 Among the comorbidities associated with deaths due to COVID-19, cardiovascular diseases (CVD) and DM stood out. 18,20,23,[29][30][31] Patients with CVD have increased serum levels of angiotensin-converting enzyme 2, which binds to the Spike protein of SARS-CoV-2, which may contribute to more severe manifestations. 46 In diabetics, the higher concentration of glucose in monocytes can result in greater viral replication and production of pro-inflammatory cytokines, so they present a late hyperinflammatory response and a decrease in adaptive immunity. 47 Other studies have also shown other health conditions associated with deaths, such as neurological, 20,32 respiratory, kidney diseases 12,27,38 and obesity. 20,21,35 In addition to biological and clinical factors, other factors that influenced mortality due to COVID-19 were socioeconomic, environmental and spatial distribution of the disease in the area of residence. Research conducted in 209 countries around the world, 29 in which spatial analysis techniques were used, such as global and local Moran's Index, in addition to multivariate logistic regression, identified the association of economic factors and population density related to mortality from the disease, establishing that mortality rates were associated with low economic level and higher population density in low-and middle-income countries.
These data corroborate a study conducted in the United States, where it was found that income inequality and precarious housing conditions influenced the increase in mortality rates from the disease. 30 Still in this perspective, in research that covered countries on the American, European and Asian continents, researchers reported that socioeconomic status can affect mortality due to COVID-19, because, as the number of severe cases increases and the public health system is overloaded, patients who need intensive care may not be able to receive care and do not have financially resources to provide another source of care. 11 Socioeconomic and demographic risk factors, commuting route to work, environment, health status, and climate-related factors in the United States were associated with the COVID-19 mortality rate. It is also noteworthy that the concentration of benzene in the air showed a high correlation with death due to COVID-19, appearing as the main risk factor in 24% of the municipalities analyzed in this study. 1 These findings are possibly related to the fact that SARS-CoV-2 spreads through the air; thus, walking to work reduces social distancing, which increases the risk of infection and dissemination. As for the influence of benzene concentration on increased mortality, it occurs due to the possible bonds of suspended particles that spread in the air, so an air with more pollutant particles is more favorable in viral propagation. 1 Regarding the spatial distribution of COVID-19 deaths worldwide, until July 2, 2020, it was found that the highest concentration of deaths occurred in Yemen (27%), Western and Northern Europe (14% -19%) and North America (9% -12%). When employing Moran's analysis, a statistically significant spatial dependence relationship was found around countries/territories with high lethality due to the disease; thus, the areas considered to be at greatest spatial risk were North America and Western Europe. 29 A study that investigated the association of multiple factors, including demographic, social, economic, health care, child health and non-communicable diseases with COVID-19 lethality worldwide, employed multivariate logistic regression and Spearman's correlation, revealed that, in April 2020, the three countries with the highest mortality rates from the disease were Zimbabwe, with a rate of 21.4% (Africa), Algeria with a rate of 17.6% (Africa) and Italy, with a rate of 15.6% (Europe). 19 In addition to this, it was also observed that the polio vaccine immunization coverage was related to a lower mortality due to COVID-19.
Research carried out in India using GWR techniques showed that the geographical distribution of deaths reported by COVID-19 did not occur randomly and was related to underlying factors, including demographics, socioeconomic and environmental variations related to pollution between different territories, with emphasis on regions with higher mortality rates linked to environmental pollution by PM2.5 particles in the east of the country. 33 It is also noteworthy that, through spatial scan statistic, significant clusters of high relative risk were detected statistically significant for the occurrence of deaths, and priority locations for health interventions were indicated by management and health services and systems, namely the city of Manaus, in the state of Amazonas (Brazil) 34 , and in the states of Oaxaca, Yucatán and Sonora (Mexico). 35 This technique has been routinely used in several ecological studies carried out around the world dealing with communicable diseases, as it allows the identification of clusters, whether of low or high relative risk, of a given event in space and space, calculating radii in which values maximize the likelihood function related to the total number of cases observed. 49 Other significant clusters for mortality were identified through spatial analysis of the global and local Moran's Indexes in border counties in the Phoenix area of Arizona (USA), 30 state of West Bengal (India). 33 Clusters with high lethality values were found in regions of high morbidity, especially in the state of Grand-Est (France). 37 It is noteworthy that the global and local Moran methods incorporate information about the meaning of spatial patterns and identify spatial autocorrelation between ecological units of analysis. 50 In the case of the studies of this review that used them, it was possible to identify and visualize, through LISA Maps, areas with higher mortality and lethality due to COVID-19, considered as priority areas for interventions aimed at monitoring the disease. This index provides a unique value as a measure of spatial association for the entire data set. This general measure of spatial association in the data set ranges from -1 to +1, in which values close to zero are related to the lack of spatial autocorrelation considering the values of the objects and their neighbors. Values close to 1 indicate positive autocorrelation, and negative values indicate negative autocorrelation. 50 Thus, the use of geography in health through spatial analysis techniques in diseases is substantial, as it shows geographic patterns, detects spatial or spatial-temporal clusters of diseases and verifies their significance, pointing out which spatial correlations occur in the studied areas, as well as to make maps that allow visualizing disease mortality. Thus, in relation to COVID-19, in which spatial dissemination is an important factor, it assists in surveillance and control, by pointing out priority areas for necessary health and socio-spatial interventions. 29 Among the limitations identified in the studies analyzed, the fact that they were carried out from secondary data (collections in databases, public domain websites or electronic patient records) stands out initially. Thus, the possibility of underreporting of pre-existing cases, deaths and comorbidities stands out, which interferes with the reliable determination of the disease's morbidity and mortality indicators, in addition to making it difficult to verify other variables of the clinical and epidemiological context associated with deaths due to the disease. Another limitation cited in studies on the spatial distribution of deaths was the so-called ecological fallacy, emphasizing that these findings should not have their results reproduced at the individual level, since they are population studies with interpretation at the group level of the regions analyzed. 48

CONCLUSION
The studies analyzed demonstrated how some risk factors and geographic distribution affected COVID-19 mortality worldwide. Clinical, social and epidemiological conditions are crucial points in the outcome of the disease. Variables, such as age over 60 years, males and presence of comorbidities resulting from cardiovascular complications and DM, were predominantly associated with deaths due to COVID-19 and were listed as the main risk factors.
As for the spatial distribution of deaths, a heterogeneous distribution was observed in the different settings investigated. The use of different analysis techniques helped to identify geographic areas with higher population density, lower income, lower presence of sanitary sewage, higher environmental pollution and lower hospital capacity, associated with increased mortality.
These data indicate disparities in health, environmental and socioeconomic conditions that exist in different parts of the world. From this, the need to intervene on the identified risk factors emerges and to investigate, through additional studies, other factors associated with death in infected patients, as well as to know the spatial distribution of the disease in vulnerable territories, in order to contribute to the fight against the pandemic and to the elaboration of health strategies and policies that minimize mortality due to COVID-19.

ACKNOWLEDGMENTS
Coordination for the Improvement of Higher Education Personnel (CAPES -