Risk assessment for cross-border transmission of multi-country Mpox outbreaks in 2022

Abstract

Objectives

Global trade and travel have facilitated infectious disease transmission. In 2022, over a short time, cross-border Mpox (monkeypox) outbreaks were reported. Since, most countries are at risk of cross-border Mpox transmissions, in this study, we developed a real-time risk assessment model for the cross-border transmission of Mpox.

Methods

This model includes priori indicators related to the source area before the Mpox outbreak and posterior indicators derived from the quantitative data evaluation afterward. Based on transportation, this model can also be used to assess the global import risk of Mpox for specific countries and cities.

Results

European risk values displayed high levels between May and July 2022 and gradually decreased after July. After September 2022, risk values elevated in most countries and regions in the Americas. As for China, high importation risk cities were highly exposed to the United States and moderately exposed to Australia and Germany. Some cities were exposed to the potential risks from only one source country.

Conclusions

Dynamic surveillance of the cross-border spread of infectious diseases is essential. Importation risks vary widely across cities and regions, and developing risk prevention and control strategies specific to the traffic flow, medical care capabilities, and risk levels in the main source countries are essential.

Introduction

Accelerated globalization has contributed to the rapid spread of infectious diseases and continues to sound alarm bells regarding global health security [1]. Global trade and travel have facilitated infectious disease transmission, including the plague in the 14th century; smallpox in the 16th century; and severe acute respiratory syndrome (SARS), novel influenza viruses, and coronavirus disease 2019 (COVID-19) in the 21st century [2], [3], [4], [5]. Dynamic monitoring and assessment of the extent of an epidemic's global public health risk are important for prevention and control [6]. The highest risks tend to occur when novel diseases appear or when familiar diseases appear in novel geographic locations [7]. Assessing disease-related health risks thus necessitates understanding where they may arise and how they may be transmitted [8].

The Mpox (monkeypox) outbreak was declared a public health emergency of international concern (PHEIC) by the World Health Organization (WHO) on July 23, 2022 [9], [10], making it the seventh declaration of a PHEIC [11]. The Mpox virus was first isolated in 1958 from Mpox lesions in cynomolgus monkeys [12]. In 1970, nine months after the Democratic Republic of the Congo (DRC) declared smallpox eradicated, the Mpox virus was isolated for the first time from a child, and Mpox was recognized as zoonotic [13]. Although historically there have been sporadic cases of Mpox in non-endemic countries, the transmission chain is relatively clear [14]. Since the first case of Mpox was detected in the United Kingdom in early May 2022, Mpox cases have arisen in many continents and countries in a short time. As of October 17, 2022, WHO has reported 73,437 laboratory-confirmed cases and 29 deaths in 109 countries, areas, or territories, with tremendous impacts on human health and public health security [15].

Mpox cases in 2022 showed “non-classical” epidemiological and clinical features that differ from previous years, which makes cross-border transmission surveillance more challenging. [16]. Previously, Mpox outbreaks in non-endemic countries had clear transmission chains and only small outbreak areas, with infected patients having a clear history of animal exposure, recent travel to the outbreak area, or exposure to imported cases. However, in 2022, the Mpox outbreak rapidly spread to multiple countries and regions in a short time, causing global concern. There were insidious social transmissions of the outbreak with predominantly spread among the men who have sex with men (MSM) population. In addition, many cases had mild clinical manifestations and the symptoms were not historically classic of Mpox. For example, the first place of rash is not the face but the anus and genitals. Atypical clinical signs may also partially lead to misdiagnosis or underdiagnosis, increasing the difficulty of transmission chain tracing [17].

Most affected countries have only reported a few imported cases without community transmission, which are at risk of cross-border transmission of Mpox [16]. Therefore, countries with predominantly imported cases should have access to risk assessment indicators and tools to monitor the epidemic’s progression and preemptively develop plans. In this study, we developed a risk assessment index for Mpox cross-border transmission. This risk assessment model may provide evidence and methods for the prevention and control of cross-border transmissions of the current Mpox outbreak as well as other infectious diseases in the future.

Methods

To assess the risk of cross-border transmission of Mpox, we constructed an Mpox risk assessment model ($\mathit{MRA}:F\times D\times t$), where $F$ is the inbound flight dataset (Table S1 in the Supplement), $D$ is the publicly available WHO Mpox dataset (Table S2 in the Supplement), and $t$ is the evaluation time. Given inbound flight data ($f\in F$) and local epidemic data ($d\in D$), $\mathit{MRA}(f,d,t)$ describes the risk of epidemic imports from the region to China at time $t$. The larger the $\mathit{MRA}(f,d,t)$, the greater the risk of epidemic importation.

Inbound flights are the primary channel for cross-border infectious disease importation because of current COVID-19 epidemic prevention and control measures. Based on air travel flow data for May, June, and July 2022, we calculated the average daily inbound flights from countries worldwide to major cities in China.

Assuming that the epidemic time-series data observed in a region are $d\in D$, we defined the Local Epidemic Risk Index (LERI) as follows:

$$\mathit{LERI}\left(t\right)=R_{\mathit{prior}}\times R_{\mathit{posteriori}}\left(t\right)$$

Where $R_{\mathit{prior}}$ is a priori indicator related to the source area before the Mpox epidemic (e.g., the region’s sanitary status and state of sanitation facilities). $R_{\mathit{posteriori}}\left(t\right)$ is the risk index of the region at time $t$ assessed based on $d$, reflecting the posterior knowledge derived from quantitative measurements and assessments of the data after the occurrence of the epidemic.

Considering various initiatives and policies in response to the recent global COVID-19 pandemic, we chose the stringency index (Table S3 in the Supplement) derived from the Oxford COVID-19 Government Response Tracker to measure the ability of regions to prevent and control the spread of infectious diseases [18]. The stringency index records the strictness of “lockdown style” policies that restrict individual behavior, with values ranging between 0 and 100.

We denoted $R_{\mathit{posteriori}}\left(t\right)=\omega \bullet v(t)$, where $\omega \in R^n$ is the weight vector and $v(t)\in R^n$ is the n-dimensional vector associated with the epidemic, describing the risk level of the epidemic in a particular region in different dimensions, such as the number of infections and the rate of disease growth. We constructed a simplified but reliable epidemic transmission model to obtain $v(t)$ explicitly from the time-series data $d$. $P(t)$ denotes the number of infections in a region at time $t$, following the idea of the susceptible–exposed–infectious–recovery (SEIR) model. We expressed $P(t)$ as an ordinary differential equation as follows:

$$\frac{\mathit{dP}(t)}{\mathit{dt}}=\lambda (t)P(t)$$

$\lambda (t)$ is a pooled coefficient used to describe the change in the number of infections over time, considering the basic regeneration number, growth rate, and containment intervention. Considering the first-order approximation of $\lambda (t)$:

$$\lambda \left(t\right)=\mathit{at}+b$$

We could solve the above differential equation and obtain $P(t)$:

$$P\left(t\right)=P\left(0\right)e^{\frac{a}{2}{t}^2+\mathit{bt}}$$

Parameters $a$ and $b$ could be estimated from the time-series data ($d\in D$), and based on $\lambda \left(t^{*}\right)=0$, the inflection point of the epidemic could be deduced as follows:

$$t^{*}=-\frac{b}{a}$$

Here, the parameter $a$ describes the acceleration of the infection rate under the combined influence of various factors, which was difficult to estimate directly using the original SEIR model.

In addition to the above derivation that provided chronological information on the development of the epidemic, we were required to consider statistics related to the epidemic itself. Available studies suggested that the basic regeneration rate of Mpox was approximately 2.4 in the MSM population and 0.8 in the non-MSM population [19], [20]. To better characterize the population transmission of Mpox, we divided the infected individuals at time $t$ into MSM (denoted as $P_{\mathit{MSM}}(t)$) and non-MSM (denoted as $P_{\mathit{nonMSM}}(t)$) populations according to the proportion of case statistics provided by the WHO (Table S4 in the Supplement).

Denoting the total population of the source country as $N$ (Table S5 in the Supplement), we selected the following five-dimensional vector to describe the epidemic risk situation in the region:

$$v\left(t\right)=\left(\left|t^{*}-t\right|,\lambda \left(t\right),P_{\mathit{MSM}}(t),P_{\mathit{nonMSM}}(t),\frac{P(t)}{N}\right)$$

The first two dimensions incorporated the main time-series information (i.e., the assessed epidemic inflection points and change coefficient), and the last three dimensions incorporated the main epidemic-related vital statistics, namely the number of MSM infections, the number of non-MSM infections, and the infection rate. We utilized min-max normalization to linearly map the data values of each dimension into the interval [0,1] and calculated the risk index of the region ($R_{\mathit{posteriori}}\left(t\right)$). By multiplying the a priori knowledge of a region ($R_{\mathit{prior}}$) and the number of flights at time $t$, the risk of epidemic import from the region to China ($\mathit{MRA}(f,d,t)$) could be estimated. This risk assessment model is a non-static, real-time model that reflects trends over time, and changes in real-time data dynamically affect the results of the model.

All analyses were performed using RStudio software, R version 4.2.1, and the code used in this study is available online [21]. Data were analyzed from May to October 2022, and the results were calculated from 100 randomized experiments.

Results

Global distribution of the Mpox risk index

From January 1 to October 16, 2022, a total of 73,437 Mpox cases and 29 deaths involving 109 countries were reported by the WHO. Fifty-four WHO regions and countries reported a cumulative number of cases of 10 or less, of which 15 reported only one case. We selected 44 countries with a cumulative number of cases greater than 20 and produced a cluster analysis map of the risk index over time ( Fig. 1). Among these 44 countries, European countries had the highest proportion (26 countries in total), followed by those in the Americas (13 countries). Additionally, three and two countries in the Western Pacific and African regions, respectively, were included.

Fig. 1.

Heatmap showing risk index from May 7 to October 26, 2022. Forty-four countries with cumulative cases greater than 20 are shown. Each column denotes a specific date, and each row represents a country. The colors behind the row names correspond to different regions. The white dashed line marks three specific points in time: May 23, July 23, and September 23, 2022. MLT=Malta, DOM=Dominican Republic, NZL=New Zealand, COD=Democratic Republic of the Congo, FIN=Finland, HRV=Croatia, GHA=Ghana, SRB=Serbia, BRA=Brazil, PRI=Puerto Rico, CHL=Chile, PER=Peru, COL=Colombia, MEX=Mexico, LUX=Luxembourg, HUN=Hungary, NOR=Norway, POL=Poland, SVN=Slovenia, CZE=Czechia, AUT=Austria, AUS=Australia, GRC=Greece, ROU=Romania, NGA=Nigeria, BOL=Bolivia, GTM=Guatemala, ARG=Argentina, ECU=Ecuador, BEL=Belgium, CAN=Canada, IRL=Ireland, SWE=Sweden, ITA=Italy, DNK=Denmark, ISR=Israel, NLD=Netherlands, CHE=Switzerland, FRA=France, USA=United States of America, DEU=Germany, ESP=Spain, GBR=The United Kingdom, PRT=Portugal.

As shown in Fig. 1, between May and July 2022 in Europe, the United Kingdom, Spain, Germany, and Portugal showed high-risk values, while Belgium, Ireland, Italy, Sweden, Denmark, Israel, the Netherlands, Switzerland, and France showed medium-to-high-risk values. The risk values for these countries gradually decreased after July 2022. Among the countries in the Americas, the risk value for the United States started in mid-May, rose significantly at the end of July 2022, peaked in mid-September, and then showed a slow downward trend. Cluster analyses found that the time series of the risk index in the United States was closer to that of the European region. In addition, most countries or regions in the Americas, such as Brazil, Puerto Rico, Chile, Peru, Colombia, and Mexico, experienced elevated risk values after September.

Fig. 1 also shows three broad bands of cross-border transmission of multi-national Mpox outbreaks: 1) a high-risk period in Europe, 2) a high-risk period in the Americas, and 3) a “transition zone” in between these two phases. In the “transition zone”, the risk values in the European region gradually decreased, while the risk values in the Americas gradually increased. We selected three time points from the above three phases (i.e., May 23, July 23, and September 23) to demonstrate the global distribution of the Mpox risk index ( Fig. 2).

Fig. 2.

Global distribution of Mpox risk index on May 23, July 23, and September 23, 2022. The distribution of the Mpox risk index in the European region has been enlarged and is shown in the black frame on the right. A darker red color indicates a higher Mpox risk index, and lighter yellow color indicates a lower Mpox risk index. Countries in gray are estimated to have negligible risk within the selected time window. Panel A. shows the global distribution of the risk index on May 23, 2022, the top 10 countries in terms of risk value for Mpox on May 23, 2022, were Portugal, United Kingdom, Spain, Belgium, Germany, Netherlands, Ireland, Switzerland, France, and Italy; Panel B shows the global distribution of the risk index on July 23, 2022, the top 10 countries in terms of Mpox risk values were Spain, United States, Germany, United Kingdom, France, Brazil, Netherlands, Switzerland, Canada, and Portugal; Panel C shows the global distribution of the risk index on September 23, 2022, the top 10 countries in terms of the Mpox risk values were Mexico, United States, Brazil, Colombia, Chile, Puerto Rico, Peru, Spain, France, and Argentina.

The top 10 countries in terms of risk value for Mpox on May 23, 2022, were all in the European region (Fig. 2A). On July 23, 2022, the top ten countries in terms of risk value for Mpox included seven countries in the European region and three countries in the American region (Fig. 2B). On September 23, 2022, only two of the top ten countries at risk were in the European region, while the remaining eight were in the American region (Fig. 2C). This shows that during the period from May 23 to September 23, 2022, the hotspots of high-risk areas gradually shifted from Europe to the Americas. Detailed data and results for the countries or regions are provided in the Supplementary (Tables S6 and S7).

Assessment of Mpox importation risk

Our risk assessment model may be used to assess the risk of the global importation of Mpox in a particular country. Currently, most countries have reported only a few imported cases without community transmission. We used China as an example as it is at risk of cross-border transmission.

On September 23, 2022, for China, the importation risk from the United States was highest (mean importation risk=29.25, SD=2.86), with moderate stringency index scores of 25.99. Generally, the higher the stringency index score, the better the country’s capacity to prevent and control infectious diseases. Between countries with similar importation risk values (e.g., Australia, Germany, and Canada), a higher stringency index score may lead to relatively lower risks (i.e., Canada may have a lower risk than Australia and Germany). In the high-risk and low-protection capacity quadrants, European countries, such as Germany, the United Kingdom, the Netherlands, France, Italy, Belgium, Spain, Switzerland, and Ireland, made up the majority ( Fig. 3).

Fig. 3.

Importation risk as a function of the Stringency Index in China. The upper area of the horizontal black dashed line indicates higher importation risk, and the lower area indicates lower importation risk. The right area of the vertical black dashed line indicates higher capacity and the left area indicates lower capacity. The higher the Stringency index score, the better the country's capacity to prevent and control infectious diseases. The circle areas represent country populations. The circle colors correspond to the different regions. AUS=Australia, AUT=Austria, BHR=Bahrain, BEL=Belgium, CAN=Canada, CZE=Czech, DNK=Denmark, EGY=Egypt, FIN=Finland, FRA=France, DEU=Germany, GRC=Greece, GUM=Guam, HUN=Hungary, IND=India, IDN=Indonesia, IRN=Iran, ISR=Israel, ITA=Italy, JPN=Japan, JOR=Jordan, LUX=Luxembourg, NLD=Netherlands, NZL=New Zealand, NGA=Nigeria, POL=Poland, PRT=Portugal, QAT=Qatar, RUS=Russia, SAU=Saudi Arabia, SRB=Serbia, SGP=Singapore, ZAF=South Africa, KOR=South Korea, ESP=Spain, SWE=Sweden, CHE=Switzerland, THA=Thailand, PHL=the Philippines, TUR=Turkey, GBR=The United Kingdom, USA= The United States of America, ARE=United Arab Emirates.

Our risk assessment model can also assess the risk of Mpox importation into key port cities. For instance, we divided the 42 Chinese cities with international traffic into three clusters according to the magnitude of the importation risk value ( Fig. 4). Cluster 1 showed higher exposure risk and included the cities of Chinese Taipei, Shanghai Municipality, Hong Kong Special Administrative Region (HKSAR), Beijing Municipality, Guangzhou, Shenzhen, Chengdu, Xiamen, Tianjin Municipality, Changsha, Nanjing, Hangzhou, Chongqing Municipality, and Shenyang (Fig. 4A). Cluster 1 was highly exposed to the United States and moderately exposed to Australia and Germany (Fig. 4B). Some cities were exposed to the potential risks of only one source country. For instance, Changsha City, Chinese Tainan, Dongying, and Taoyuan were exposed to potential risks from the United States, Japan, South Korea, and Thailand, respectively. Detailed data and results are provided in the Supplementary Information (Tables S8 and S9).

Fig. 4.

Importation risk of key port cities in China. Panel A. The distribution of importation risk in China. The map's background color refers to the population density (people per km²). The area of circles represents importation risk. Panel B. The proportion of importation risks for specific Chinese key port cities. Stacked histograms showing importation risk, colored by different countries/regions. Cities in Cluster 1 showed high exposed risk, Cluster 2 showed moderate exposed risk, and Cluster 3 showed low exposed risk.

Discussion

Real-time surveillance of infectious diseases may be a major public health issue in countries with fragile health systems. Evidence suggests that the epidemiology of human Mpox in Africa has changed over the past 30 years [22], [23], with the number and magnitude of outbreaks increasing in the DRC and Nigeria [22], [24]. However, deficiencies remain in the quality of infectious disease surveillance in these two Mpox hotspots. The Mpox outbreak may evolve rapidly, and more cases are expected to be identified in non-endemic countries as surveillance coverage and efforts expand. Existing literature lacks the examination of cross-border transmission of Mpox. Cross-border infectious disease outbreaks may cause casualties and panic as well as affect economic development and social stability. Therefore, dynamic monitoring of cross-border spread and establishing risk assessment tools for the transmission of infectious diseases is essential.

To assess the risk of cross-border transmission, we developed a statistical real-time risk assessment model based on international flight information, epidemiological characteristics of infectious diseases, and demographic information of source countries. In the case of Mpox, we integrated a priori indicators related to source areas before outbreaks, (e.g., control and prevention strategies, health status, and sanitation status) and a posteriori knowledge derived from quantitative measurements and assessments of the data after outbreaks (e.g., number of infections, characteristics of MSM populations, and disease growth rates). These factors allowed for a more comprehensive understanding of the risk index of Mpox. In addition, based on air transportation data, our risk assessment model may be extended to assess the global importation risk of Mpox to specific countries or cities.

As globalization and social, economic, cultural, and population exchanges increase, the risk of infectious disease may increase, and preventing and controlling cross-border transmission of infectious diseases may become increasingly challenging [25], [26]. We identified three significant periods in multi-national Mpox outbreaks as of October 26, 2022. May to July was the high-risk period for the European region. This period was followed by a gradual decrease in risk values in European and a gradual increase in risk values in the Americas, primarily the United States and Brazil. From September 2022 onwards, the Americas entered a high-risk period. Owing to the promotion of vaccines and the application of public health measures [27], cases in Europe have shown a clear downward trend, and a similar downward trajectory has been observed in the Americas. However, the highest risks remain concentrated in the Americas and European countries. These findings may help improve the surveillance and early warning of cross-border transmission, promote joint prevention and control between countries, and protect the health and safety of people worldwide [28].

Prior to this outbreak, no Mpox cases had been reported in China. However, as of October 26, 2022, six cases have been reported in China, all of which were imported. Our study showed that 14 key port cities in China were at a higher risk of importing Mpox, including Chinese Taipei, Shanghai, HKSAR, and Chongqing Municipality. These cities were highly exposed to the United States and moderately exposed to Germany, which is consistent with the actual situation of the reported cases in China. Chinese Taipei reported its first imported case of Mpox on June 24, 2022, which entered the country via Taipei with a recent history of residence in Germany [29]. The second, third, and sixth cases were reported by Chinese Taipei, and all had recent residencies in the United States (July 12, August 6, and October 9, 2022, respectively). An additional imported case was reported in the HKSAR on September 6 [30], and the Chongqing Municipality reported its first imported case on September 16, 2022 [31]. These findings indicated that the importation risk index constructed in this study may provide practical guidance for the prevention and control of cross-border transmission of infectious diseases. Additionally, a high degree of diversity in import risks in different cities or regions may exist. Therefore, specific risk prevention and control strategies based on specific locations may be necessary, particularly regarding their traffic flow, level of medical care, and the level of risk of primary source countries.

The current study had several limitations. First, two dimensions of our selected five-dimensional vector v(t) were closely related to the epidemic process of infectious diseases. However, the number of cases in some countries was extremely small, with only a few sporadic cases in a few days, which may have been insufficient in reflecting the changing trend of the epidemic. Therefore, directly applying our risk index assessment model maybe be difficult. For such cases, we primarily considered epidemic-related vital statistics and minimized the influence of time-series-related factors. In addition, multiple routes of cross-border transmission of infectious diseases exist, even though Mpox is primarily transmitted human-to-human. The assessment model considered the risks associated with cross-border flights, while other modes, such as ground travel and animal and food transmission, were not included.

Conclusion

Dynamic surveillance of cross-border transmission of infectious diseases and the establishment of tools to assess the risk of transmission of infectious diseases is essential. The previously neglected zoonotic Mpox has become a global concern. This study found that the Mpox epidemic in Europe continues to decline and there are indications that the epidemic in the Americas may follow a similar trend. Cluster analysis found that the time series characteristics of the risk index in the United States are more similar to those in the European region. In addition, the risk of importation varies highly across cities and regions, requiring tailored risk prevention and control plans. Traffic flow, medical care level, and risk level in primary source countries may need to be considered. This study establishes a risk assessment model for cross-border transmission of Mpox outbreaks through publicly available data, which can provide ideas and evidence for future emergency surveillance of other infectious diseases. Not only is this important for early prediction and prevention, but also for controlling the spread of the epidemic.

Ethical approval statement

This modeling study did not require institutional review board review or approval because only simulated data and public-available data are used.

Funding

This study was supported by the grants from the Chinese Academy of Medical Sciences (CAMS) Innovation Fund for Medical Sciences (2021-I2M-1-044), and the National Social Science Fund of China (20&ZD201). The funder of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

Table S1

Supplementary material

.

Table S2

Supplementary material

.

Supplementary material

.

Supplementary material

.

Supplementary material

.

Supplementary material

.

Supplementary material

.

Supplementary material

.

Supplementary material

.