Populational Level Driving Force of COVID-19 Epidemic and Fatality: A Global Analysis

Method We scoped data related to reported epidemic durations, incidences, fatalities, and epidemic risk factors of the studied countries. Disease Development Speed (DDS), Population-Level Incidence (PLI), and Case-Fatality Rate (CFR) were calculated to assess the COVID-19 pandemic globally. The Spearman rank correlation was applied to further explore the relationship among DDS, PLI, CFR, and their inuencing factors.


Conclusion
Nationwide development of the COVID-19, its incidence and fatality suggested regional similarity in the world. National population structure, human resources of medical staff, structure of national health expenditure, and the sanitation facility were revealed as vital risk factors for the COVID-19 in epidemiology.

Background
Globally, SARS-CoV-2 has infected more than 30 million people and caused almost one million deaths until 19th Sep 2020. The incidence and fatality rate of this world pandemic were widely reported with regional disparities [1,2]. Previous studies indicated that age and co-morbidities were key risk factors for the incidence and fatality of the COVID-19 pandemic [3][4][5][6][7]. Meanwhile, other risk factors might also include the urbanisation, medical capacities, public health measures adopted and the prevalence of underlying diseases [8][9][10][11][12]. However, limited numbers of studies explored these risk factors of the incidence and fatality rate of the COVID-19 pandemic at a population level. Therefore, this research analysed the correlation among incidence, fatality and their risk factors. Additionally, we investigated the development speed of the pandemic in each country. This research has a wider implication providing evidence-based suggestion in disease control of the COVID-19.

Data collection and strati cation
Countries with over 1,000 and over 10,000 COVID-19 cases until 19th Sep 2020 were included for analysis separately. According to previous researches [1,[8][9][10][11][12][13], we included 21 measures subjected to 4 types in this study: population, medical capacity, national health structure, and prevalence of the underlying disease. Data in this study were obtained from the World Health Organization (WHO), World Bank and the Institute for Health Metrics and Evaluation (IHME) [14][15][16]

Statistical analysis
The skewed variables were presented as median value (IQR: 25th − 75th percentiles), and the categorical variables were presented as numbers and percentages. Spearman's rank correlation coe cient was applied to assess the correlation between outcome indicators (DDS, PLI and CFR) and variables of risk factors. Simple linear regression analysis was used to identify possible correlations between variables. All analyses were carried out in RStudio (Version 1.2.1335).

Result
Overall, 216 countries and regions globally had reported COVID-19 cases by 19th Sep 2020. 69.9% (151) countries reported over 1,000 cases, 39.4% (85) countries reported over 10,000 cases, and 13.0% (28) countries reported over 100,000 cases nationwide. The median of the epidemic duration worldwide was 167 days (IQR: 158-175 days; Range: 110-221 days). Besides, the medians of DDS, PLI, and CFR The medians of CFRs in AMRO and EURO countries were both higher than 0.02, while the medians of CFRs in the rest 4 regions were relatively lower. The CFRs in EURO countries disparate most signi cantly.
In contrary, CFRs of countries in WPRO regions distributed most evenly among the 6 regions.
The Correlation between Risk Factors, DDS, PLI, and CFR  Overall, the DDS had strong positive correlation with PLI, but very weakly positive correlation with the CFR. Besides, the DDS also had strong positive correlation with the number of reported deaths. It was very likely that the PLI had no correlation with the CFR, but a moderately positive correlation with the number of reported deaths. Moreover, the CFR had moderately positive correlation with the number of reported death.

Discussion
Among countries entering Stage III, higher CFRs were mostly seen in European countries (i.e. France, Italy, and Spain), while lowest CFRs were found in some Asian countries (i.e. Singapore and Thailand). Some low-CFR countries shared similar control measures and medical system characteristics, including mass tests, strict social distancing measures, and free medical services for diagnosed COVID-19 patients [18][19]. A brief description of policy in different countries was shown in Supplement 5.
Of note, among countries with DDS over 0.05, substantial differences in CFRs were observed (CFR > 0.10: France, the United Kingdom, Italy, and Belgium versus CFR < 0.05: Russia, Turkey, Peru, and Germany). In addition to delayed social distancing measures [20][21][22], other in uencing factors include higher elderly population proportions, more stringent symptom-based testing policies (i.e. in Italy and United Kingdom) [23][24], insu cient medical resources, and exceeded surge capacity due to rapidly increasing COVID patients in the former group of countries [25][26]. Germany and Turkey had lower CFRs mainly because they adopted extensive testing strategies, effective social distancing measures, and well-prepared health system [27][28]. Russia also stood out with a rather low CFR partly due to their 'early detection and early treatment' strategy implemented nationally as well as the relatively lower prevalence of co-morbidities in Russia [29]. Experiences in these countries provide important implications for other countries to control the COVID-19 pandemic.
Besides, higher urbanization might expect a higher PLI and maybe a higher DDS. Urban dwellers may have frequent social activities in dense places which is more di cult to control, but they are more likely to access better medical facilities. Population density was not signi cantly associated with higher incidence and DDS. This may be raised due to the national data only represented country-level population density, whereas urban population density is much higher. More granular data are however warranted to con rm these associations. In terms of correlations of PLIs and CFRs, the elderly (age > 65) proportion, the cardiovascular diseases and diabetes mellitus were found positively correlated. These ndings are consistent with earlier evidences suggesting that the elderly and individuals with underlying conditions were more likely to be infected by COVID-19. The virus was especially fatal among these sub-groups [30].
The limitation of this report is lacking of deep discussion on potential confounding by multivariable analyses. As limited by the manuscript type, the multivariable linear regression model is demonstrated in the Supplement 6.

Conclusion
In the population level, regional similarity was found among DDSs, PLIs, and CFRs of different countries with same epidemic durations. Total population and urban population nationwide, human resources of medical staff (doctor and nurse), the UHC coverage and out-of-pocket expenditure in health care, the population with basic handwashing facilities at home, and the national prevalence of respiratory infections & tuberculosis were key risk factors for the epidemic of the COVID-19, in its incidence, fatality, and disease development.