The first European COVID-19 case was reported on January 24, 2020, in France, with subsequent cases confirmed in Germany and Finland days later [1]. On March 11, 2020, the World Health Organization (WHO) declared that the spread of the novel coronavirus had exceeded the threshold of a pandemic [2] and, on March 13, 2020, the WHO declared Europe as the global epicenter, when their caseload and deaths exceeded the combined caseload in the rest of the world [1] (See Figure 1). The European Union (EU) closed all external borders on March 17, 2020 [1]. Although the EU coordinated the COVID-19 response between member countries, individual governments enacted separate national policies and made individual decisions regarding border closure and quarantine measures [3]. COVID-19 caseloads decreased for most European countries after peaking in April and May [4].

At present, European countries are experiencing a second wave of COVID-19 [5-11]. The WHO has warned that the death counts in Europe could surpass the peak observed in April 2020 [12]. Nations worldwide are struggling to control COVID-19 transmission by imposing social isolation and economic restrictions, with leaders reluctant to shut down businesses and quarantine citizens again [13,14]. As of February 9, 2021, the WHO reported 106,125,682 confirmed COVID-19 cases and 2,320,497 deaths worldwide [15]. Collectively, 33,534,153 COVID-19 cases have been reported in the EU and the United Kingdom, which have resulted in 740,733 deaths [4].

The World Bank (WB), a global partnership dedicated to reducing poverty and increasing sustainable prosperity in developing nations, divides the world into regions based on shared geographical, development, and cultural or historical features [16]. The Global SARS-CoV-2 Surveillance Project: Policy, Persistence, & Transmission provides surveillance data [17] based on these WB-defined regions. The focus of this study is on the spread of COVID-19 specifically within the Western European region, including Albania, Andorra, Austria, Belarus, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Greenland, Hungary, Iceland, Ireland, Isle of Man, Italy, Latvia, Liechtenstein, Lithuania, Luxembourg, Moldova, Monaco, Montenegro, the Netherlands, Norway, Poland, Portugal, Romania, San Marino, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Ukraine, the United Kingdom, and Vatican City.

Public health surveillance informs policy on “flattening the curve” of COVID-19 spread [17,35-37]. Epidemiologists have utilized various modeling techniques to forecast COVID-19 case numbers and attributed deaths [38-42]. The European Center for Disease Control, the WHO, and the Center for Systems Science and Engineering at Johns Hopkins University have developed tracking tools [11,38]. Although helpful, these static metrics are limited by incomplete case ascertainment and data contamination [17,36]. Existing surveillance is a proxy for the true COVID-19 caseload because public health surveillance systems tend to pick up the most severe cases [43,44], which is especially problematic when tracking SARS-CoV-2 infections because most carriers are asymptomatic or presymptomatic or may have mild symptoms [45-48]. Therefore, public health surveillance that can control for these limitations are needed. Moreover, metrics that detect the speed of transmission of the novel coronavirus, shifts in the pandemic, and acceleration of the speed and persistence of COVID-19 based on prior infections are needed to supplement existing measures.

Ideally, the development of a more advanced methodology for tracking and estimating COVID-19 transmission in regions within Europe will allow for a more reliable analysis of which policies are effective and what other factors may be associated with transmission rates. Public health departments, in addition to several universities and media outlets, are tracking COVID-19 metrics by using raw data, including the number of new cases, diagnostic tests, positive results, transmission rates and deaths, in addition to other measures such as local hospital capacity [4,49-57]. To remove temporal effects, many surveillance systems have shifted to 7-day moving averages to counter the dearth of reporting during holidays and weekends. Although moving averages temper volatility of data and testing or reporting affects, surveillance is still limited by missing cases. General public health surveillance is helpful and provides a proxy of the pandemic, but surveillance data are still limited by significant bias due to undercounts, reporting delays, testing errors, dearth of testing, asymptomatic carriers, and other types of data contamination. In fact, surveillance systems are predicated on the fact that they tend to include only the more severe cases, whereas mild cases and undiagnosed infections and deaths are excluded [43,44].

To that end, the objective of our study is to use a longitudinal trend analysis study design in concert with Dynamic Panel Modeling and Method of Moments to correct for existing surveillance data limitations [17,36]. Specifically, we will measure significant weekly shifts in the increase, decrease, or plateaued transmission of COVID-19. We will also measure the underlying causal effect from the previous week that persists through the current week, with a 7-day persistence rate to explain a clustering-declustering effect. The 7-day persistence represents an underlying disease transmission wave, wherein a large number of transmissions 7 days ago that resulted in a large number of infections today then echoes forward into a large number of new transmissions and, hence, a large number of new cases 7 days forward. An example of the 7-day lag would be large sporting events in the United Kingdom that drew huge crowds weekend after weekend even after new COVID-19 cases were confirmed in the country. Other potential “superspreader” events such as the exportation of COVID-19 cases from a popular ski town in the Austrian Alps back in March 2020 [58], would certainly contribute to this persistence effect as well. In summary, we will measure negative and positive shifts in the transmission of SARS-CoV-2 or the acceleration or deceleration rates. Our surveillance metric will provide public health surveillance data to inform governments in decision-making regarding disease control, mitigation strategies, and reopening policies as they continue to manage this unprecedented situation.

Regression results are presented for 39 European countries in Table 1. Weekly surveillance data presented in Tables 2-6 are based in part on these regressions. Data for 44 European countries were collected, but data for 5 countries were excluded in the regression analysis due to missing data. The regression Wald statistic is significant (X²₈=4980; P<.001). The Sargan test was not significant, failing to reject the validity of overidentifying restrictions (X²₅₁₁=39; P=.39).

The coefficient for the 7-day lag was positive and statistically significant (0.90, P<.001), indicating the number of infections 7 days prior to the study had a positive relationship that echoed forward 7 days later. The shift parameter 14 days ago was negative and statistically significant (coefficient –0.30, P<.001), suggesting that exogenous shift events had a negative effect on total case numbers (Table 1).

The lagging indicators and shift parameters suggested recent changes in disease transmission in Europe between November 30, 2020, and January 17, 2021. The shift in the most recent 14 days, or 2 weeks, was negative and statistically significant (P<.001). The coefficient for weekend was negative and statistically significant (–2.1, P<.02), as shown in Table 1.

Tables 2-6 display static and novel dynamic surveillance measures for the weeks of January 4-10, 2021, and January 11-17, 2021. Information pertaining to the prior weeks can be found in Tables S1-S8 of Multimedia Appendix 1. Static measures include the number of new cases during the first day of a given week, cumulative cases, the 7-day moving average of new cases, rate of infection, new deaths during the first day of a given week, cumulative deaths, the 7-day moving average of new deaths, and the rate of deaths (see Tables 2 and 3). The dynamic measures include a temporal element to better understand how past cases affect the present ones and how present cases affect the future ones. Dynamic measures (see Tables 4 and 5) include (1) speed—the number of new observed COVID-19 cases per day per 100,000, averaged over a week; (2) acceleration—the change in speed from the prior week to the current week; (3) jerk—the week-over-week change in acceleration as a function of time over the course of 2 weeks between January 4 and 17, 2021; and (4) the 7-day persistence effect on speed—the average of the number of new cases per day in a given week that are statistically attributable to new cases reported 7 days earlier.

Static measures in Europe for the week of January 4-10, 2021, are presented in Table 2 and those for the week of January 11-17, 2021, are presented in Table 3. New European cases slightly decreased from 158,741 to 152,064 during the first day of each week, with only cumulative cases increasing from 22,507,271 to 23,890,761, which is a weekly increase of 1,383,490 from January 10 to January 17, 2021. Cumulative deaths due to COVID-19 in Europe reached 558,494 by January 17, 2021. The 7-day moving average of new cases totaled 208,581 in the first week and 177,288 in the second week.

France, Germany, Italy, Spain, and the United Kingdom had the largest 7-day moving averages for new cases of infection at 18,269, 20,787, 17,292, 17,442, and 59,809 during the week of January 4-10, 2021. In the second week (January 11-17, 2021), 7-day moving averages increased to 18,318 and 28,829 for France and Spain, respectively. The 7-day moving average for Germany, Italy, and the United Kingdom decreased to 17,245, 14,969, and 46,338, respectively. The infection rate per 100,000 people during the week of January 4-10, 2021, was the highest in Ireland and the United Kingdom at 139.5 and 81.06, respectively. The Czech Republic, Portugal, and Lithuania reported the next highest rates at 78.90, 73.59, and 54.81 per 100,000 population. These 5 countries with the highest infection rates reported a change for the week of January 11-17, 2021, thereby also changing the ranking and magnitude of rates. The top 5 countries by infection rate in week 2 were Portugal at 101.85, Ireland at 59.66, Andorra at 58.24, France at 57.31, and the United Kingdom at 56.94 per 100,000 population.

During the week of January 4-10, 2021, the highest death rates were reported in Latvia, Slovakia, and the Czech Republic at 1.64, 1.50, and 1.28 per 100,000 population. The following week, the European countries with the highest death rates were Slovenia, Portugal, and the Czech Republic at 1.92, 1.49 and 1.15 per 100,000 population.

Tables 4 and 5 and Figure 2 (data sourced from The Global SARS-CoV-2 Surveillance Project [62]) display the dynamic metrics that offer a more temporal view of these data. Novel metrics are also displayed in Multimedia Appendices 2-4. From the week of January 4-10 to the week of January 11-17, 2021, in Europe, the dynamic measures of speed decreased (37.72 to 33.02 per 100,000), acceleration decreased (0.39 to –0.16 per 100,000), and jerk increased (–1.30 to 1.37 per 100,000). Speed was the highest and decreasing in Ireland (132.29 to 72.66 per 100,000), the Czech Republic (120.97 to 77.36 per 100,000), and Slovenia (97.54 to 67.64 per 100,000) during both weeks. Acceleration was the highest in Portugal, Ireland, and Slovakia in the week of January 4-10, 2021, at 5.77, 5.57, and 4.59 per 100,000 population, respectively. Only Ireland had a positive jerk during this time. Andorra, France, and Portugal had the largest acceleration rates during the week of January 4-10, with reported increases to 8.32, 4.70, and 4.04 per 100,000 population, respectively. Jerk was the highest in Serbia, Ireland, and Albania during the week of January 4-10, 2021, at 5.73, 1.38, and 0.67 per 100,000 population. Andorra, France, and Germany reported the highest jerk rates per 100,000 in the week January 11-17, 2021, at 9.24, 9.05, and 3.16, respectively.

The 7-day persistence difference in Table 6 demonstrates the changes in 7-day persistence for the top 4 European countries between January 4-10 and January 11-17, 2021, suggesting an underlying shift that significantly increased persistence for some countries but significantly decreased persistence for other countries during the week of January 11-17. Only two of the countries were in the top 4 for both weeks (ie, the Czech Republic and United Kingdom). Lithuania and San Marino were in the top 4 for the first week, but Lithuania decreased to 41.01 in the second week. San Marino slightly increased from 49.73 to 50.49, but did not make it into the top 4 for the second week.

Among the top 5 countries by population (Table 7), Germany and France remained relatively stable, and the United Kingdom had the highest indicators of cause for alarm, with positive increases in speed and persistence but slightly negative decreases in acceleration and jerk. Smaller countries such as the Czech Republic, Ireland, Andorra, and Portugal reported higher positive increases in speed, acceleration, jerk, and persistence.

Thus far, European COVID-19 infection surveillance has depended on static metrics with limited insight into longitudinal pandemic progression. Dynamic metrics provide an additional lens for surveillance that better captures the evolving prevalence of disease. After combining static and dynamic metrics, some European countries stand out as with the highest risk for uncontrolled growth. These high-risk countries must maintain transmission mitigating policies if they are to protect their citizens and the citizens of neighboring countries.

Europe, as a region, is still experiencing high COVID-19 case rates, but these appear to be trending downward as the region emerges from its second wave. The 7-day moving average of new cases showed a substantial decrease from 208,581 to 177,288 between January 4 and January 17, 2021. The 7-day moving average of COVID-19–related deaths, however, increased from the week of January 4-10 to the week of January 11-17, 2021. Speed of transmission in the region decreased and acceleration shifted from positive to negative from week 1 to week 2, suggesting that case rates may continue to trend downward in coming weeks. This shift in acceleration implies that the speed was increasing at the beginning of the study period but entered a downward trajectory by the end. However, jerk shifted from negative to positive during these two weeks, indicating that the downward trend in acceleration was slowing toward the end of the study period. Interventions and continued precautions will be necessary to maintain a decreasing 7-day moving average of new cases and to continue the downward trajectory of speed and acceleration.

Infection rates show the countries that were the hardest hit at the time of data collection. The top 5 most populous countries in Europe are Germany, the United Kingdom, France, Italy, and Spain. Unsurprisingly, these 5 countries also had the largest 7-day moving averages for new infections during the study period. The United Kingdom had the second highest infection rate per 100,000 people during week 1 of the data collection, along with the largest 7-day moving average of new cases. This finding indicates that the United Kingdom may be at risk of increasing transmission, but the infection rate per 100,000 people decreased from 81.06 during week 1 to 56.94 in week 2, which is reassuring. Both the speed of virus transmission and acceleration in the United Kingdom decreased over the recorded period as well, but jerk actually increased from –0.49 to 0.50, and the country’s 7-day persistence was the fourth highest in Europe during both weeks, indicating that the United Kingdom does need to stay vigilant and ensure proper enforcement of policies to reduce transmission in order to avoid another outbreak.

France and Germany both reported increases in acceleration over the two weeks, and the jerk transitioned from a negative value to a positive value, putting both countries at risk of experiencing increased growth in the coming weeks. Additionally, Spain had increasing speed and jerk, and its 7-day persistence effect increased over the 2-week period, indicating an increase in forward echoes of COVID-19 cases present in the country. Fortunately, Italy’s speed and acceleration both decreased over the recorded period, and the jerk was negative during both weeks, implying that mitigation strategies are currently proving to be effective in Italy—the country that was initially one of the hardest hit. These 5 most populous countries are responsible for a very significant portion of the total cases in the European region, and they will likely require regional policy coordination for optimal control of virus transmission.

Some smaller countries in the region have also demonstrated dynamic metrics that warrant concern, such as Andorra. The speed increased from week 1 to week 2, and jerk and acceleration both dramatically increased from negative to positive values, indicating that more intense restrictions are likely necessary to slow the spread. Ireland had the highest infection rate per 100,000 people during week 1 of data collection and the second highest infection rate during week 2. Additionally, Ireland had the largest speed and jerk and the second largest acceleration during week 1, thereby increasing concern for a potential future outbreak in the country. However, all of these dynamic metrics decreased dramatically during week 2 (January 11-17, 2021), with acceleration and jerk actually transitioning to negative values, supporting the idea that Ireland’s mitigation strategy is proving to be effective, at least during the time period in question.

Portugal was also at high risk of increased transmission, with a transmission rate per 100,000 people in the top 5 countries of the region during both weeks. With respect to novel dynamic metrics, Portugal had the largest positive acceleration in the region during week 1 and the third largest in week 2. Additionally, the country’s jerk increased from a negative value to a positive value and the 7-day persistence effect almost doubled from week to week. This finding indicates that Portugal should consider implementing new policies to reduce transmission and specifically to restrict the evolution of superspreader events, given the increase in 7-day persistence and the fact that Portugal had the second highest death rate per 100,000 people in the region during week 2 (January 11-17, 2021). Residents of Portugal were not only highly likely to contract COVID-19 during this time period, but they were also more likely to die of the disease than residents of most other European countries.

Although some European countries showed signs of uncontrolled growth for the near future, many demonstrate decreasing dynamic metrics that provide reassurance that transmission is being controlled appropriately. However, based on these results, countries with increasing dynamic metrics that are most at risk of outbreaks include Andorra, Portugal, and Spain. Fortunately, Andorra’s population is relatively small for the region, potentially insulating regional policy makers and agencies from an overwhelming surge in COVID-19 cases. In contrast, Spain and Portugal are relatively large countries. Their caseloads and positive dynamic metrics suggest that these two countries would require substantial effort to control the COVID-19 spread. Regional coordination would be essential given the size of these countries from a population and economic perspective. Additionally, some countries such as the Czech Republic have very high 7-day persistence effects but decreasing speed and acceleration, indicating that the overall transmission in the country may be decreasing, but focused policy targeted toward preventing superspreader events may be helpful.

Europe experienced a surge in COVID-19 transmission due to the second wave of the pandemic [11,63-65]. Because infection rates had significantly increased across Europe, many governments imposed strict lockdowns shutting down European economies again. Since SARS-CoV-2 cases were first reported in Europe earlier in 2020, COVID-19–related research has kept pace and, consequently, fewer deaths have been reported [61]. The virus is still just as contagious and deadly, but targeted therapies have resulted in attenuation of death rates across countries [61].

Data are limited by granularity and collection method. Data were collected at the country level, which precludes local analysis of surveillance trends. Moreover, data collection mechanisms differ by country and may even differ by region within a given country. These different methods lead to weekend effects, missing data points, and other contamination.

This study is part of a broader research program at Northwestern Feinberg School of Medicine, The Global SARS-CoV-2 Surveillance Project: Policy, Persistence, & Transmission. This research program developed novel surveillance metrics to include speed, acceleration, jerk, and 7-day persistence at the country level [17,66]. We have also derived surveillance metrics for all global regions.

Static and dynamic public health surveillance tools provide a more complete picture of the progression of the COVID-19 pandemic across countries and regions. Although static measures, including infection rates and death rates, capture data at a given point in time, they are less successful in assessing population health over a period of weeks or months. By including speed, acceleration, jerk, and 7-day persistence, public health officials may design policies with an eye to the future. According to surveillance data, all countries in Europe that were at the highest risk during the second wave of the COVID-19 pandemic shared a number of characteristics. The United Kingdom, Spain, and Portugal demonstrated high infection rates, jerk, and 7-day persistence rates. Looking ahead, policy makers in these countries and the region at large should be concerned about growth in the already substantial number of COVID-19 cases over the short term. Given the substantial 7-day persistence rates in large countries such as the United Kingdom, Spain, and the Czech Republic, it is imperative that efforts be made to target superspreader events. Analysis of subsequent surveillance data using both static and dynamic tools can help confirm the efficaciousness of new policies.