ORIGINAL RESEARCH published: 08 July 2021 doi: 10.3389/fpubh.2021.633123 Monitoring and Analysis of COVID-19 Pandemic: The Need for an Empirical Approach Martí Català 1,2 , Miquel Marchena 2 , David Conesa 2 , Pablo Palacios 2 , Tomas Urdiales 2 , Sergio Alonso 2 , Enrique Alvarez-Lacalle 2 , Daniel Lopez 2 , Pere-Joan Cardona 1,3,4 and Clara Prats 1,2* 1 Comparative Medicine and Bioimage Centre of Catalonia, Fundació Institut d’Investigació en Ciències de la Salut Germans Trias i Pujol, Badalona, Spain, 2 Department of Physics, Universitat Politècnica de Catalunya (UPC-BarcelonaTech), Barcelona, Spain, 3 Experimental Tuberculosis Unit, Fundació Institut d’Investigació en Ciències de la Salut Germans Trias i Pujol, Universitat Autònoma de Barcelona, Badalona, Spain, 4 Centro de Investigación Biomédica en Red de Enfermedades Respiratorias, Madrid, Spain Edited by: The current worldwide pandemic produced by coronavirus disease 2019 (COVID-19) Zisis Kozlakidis, has changed the paradigm of mathematical epidemiology due to the high number of International Agency For Research On unknowns of this new disease. Thus, the empirical approach has emerged as a robust Cancer (IARC), France tool to analyze the actual situation carried by the countries and also allows us to predict Reviewed by: Majid Taati Moghadam, the incoming scenarios. In this paper, we propose three empirical indexes to estimate Iran University of Medical the state of the pandemic. These indexes quantify both the propagation and the number Sciences, Iran Sufia Islam, of estimated cases, allowing us to accurately determine the real risk of a country. We East West University, Bangladesh have calculated these indexes’ evolution for several European countries. Risk diagrams *Correspondence: are introduced as a tool to visualize the evolution of a country and evaluate its current Clara Prats risk as a function of the number of contagious individuals and the empiric reproduction clara.prats@upc.edu number. Risk diagrams at the regional level are useful to observe heterogeneity on Specialty section: COVID-19 penetration and spreading in some countries, which is essential during This article was submitted to deconfinement processes. During the pandemic, there have been significant differences Infectious Diseases - Surveillance, Prevention and Treatment, seen in countries reporting case criterion and detection capacity. Therefore, we have a section of the journal introduced estimations about the real number of infectious cases that allows us to Frontiers in Public Health have a broader view and to better estimate the risk. These diagrams and indexes have Received: 24 November 2020 been successfully used for the monitoring of European countries and regions during the Accepted: 25 May 2021 Published: 08 July 2021 COVID-19 pandemic. Citation: Keywords: COVID-19, risk indexes, risk diagram, epidemic monitoring, COVID-19 outbreak Català M, Marchena M, Conesa D, Palacios P, Urdiales T, Alonso S, Alvarez-Lacalle E, Lopez D, Cardona P-J and Prats C (2021) INTRODUCTION Monitoring and Analysis of COVID-19 Pandemic: The Need for an Empirical The current pandemic produced by coronavirus disease 2019 (COVID-19) is strongly impacting Approach. the world. With more than 150 million confirmed cases and more than 3 million reported deaths, Front. Public Health 9:633123. the pandemic has been a worldwide tragedy, with consequences impacting far beyond these doi: 10.3389/fpubh.2021.633123 numbers. In addition to the health disaster in all the countries in the world, the control measures Frontiers in Public Health | www.frontiersin.org 1 July 2021 | Volume 9 | Article 633123 Català et al. Monitoring COVID-19 Pandemic With an Empirical Approach had important consequences, not only the expected and looking for an effective and proper communication of the socioeconomic derivatives but also emotional (1, 2), educational epidemic situation in a certain country, we have built a discrete (3, 4), or cultural (5, 6) consequences, to cite but a few. scale that assesses the level of incident cases. These indexes are Therefore, this emergency situation has required constant adapted to COVID-19 but can also be used in any pandemic monitoring at multiple levels—from the city neighborhood or epidemic. tracking of local outbreaks to a global continental perspective for socioeconomical decisions coordinated at the interstate level. Empiric Indexes Different political actors need different pieces of information to Empiric Reproduction Number take decisions regarding mobility, schools, or the redirection of Classically, epidemiology uses the effective reproduction number health resources, among others. (Rt ) (9) to measure the velocity at which the epidemic is Unfortunately, the spreading dynamics of the severe acute propagated during an outbreak. It is a measure of the mean respiratory syndrome coronavirus 2 (SARS-CoV-2) is largely number of new infections caused by an infectious individual. unknown and certainly not sufficiently characterized to develop Let R0 be the value of Rt before the epidemic starts, that is, at mechanistic models that properly predict its propagation in the t = t0 . To compute these parameters, SIR and SEIR (susceptible- medium term. For example, there is uncertainty in the literature exposed-infectious-removed) models (13) are traditionally used. regarding the influence of temperature and humidity in its However, these models are difficult to address COVID-19 transmission, with reports indicating both a very small (7) and pandemic due to the high number of unknowns about inherent a relatively large (8) effect. The apparent clustering behavior parameters (14). In addition, classical SEIR models are driven (9) of the transmission adds an important layer of uncertainty by susceptible population availability, while the evolution of regarding under which conditions the virus propagates optimally. this pandemic is mainly governed by the control measures like This renders complicated mechanistic models of propagation confinement or social conscience regarding the hygiene rather useless in the sense of providing useful quantitative information. than by susceptible population. Relevant indicators for policy makers must come from empirical Other methods to calculate Rt are also available based on epidemiological models of the cumulative cases and fatalities of the estimation of the generation time between two correlated the pandemic. infections and the probability of infection along the disease of One of these indicators is the effective reproduction number, an individual (11). The lack of complete knowledge of such Rt , that is normally assessed by means of an SIR model (i.e., a factors at the beginning of the epidemic suggested to assume a compartment model based on Susceptible-Infectious-Recovered naiver description. flows) (10) or likelihood-based estimation procedures based on We propose an empiric definition of the propagation rate (ρt ), the generation time interval method (11). These kinds of models which is defined as the number of newly infected in the last 3 days require previous parametrization. In this sense, we propose a divided by the number of newly infected during 3 days τ days ago: more transparent and empiric way to characterize the spreading of the epidemic that we call ρ t . This index measures the nc (t − 2) + nc (t − 1) + nc (t) ratio between new cases at an interval of 5 days. It is thus ρt (t − 1, τ ) = nc (t − 2 − τ ) + nc (t − 1 − τ ) + nc (t − τ ) a parametrization-free parameter that we will show is closely related to Rt . When combined with the evaluation of active cases, where nc(t) is the number of new cases at time t, and τ is the it provides an empirical quantification of the epidemiological risk incubation period, which in COVID-19 case is estimated to be in a given region. around 5 days (15, 16). Furthermore, 5 days also correspond The manuscript is structured as follows. First, the Methods to the average generation time (time between generations) (17), describe the empirical indexes used on daily tracking of giving rise to a simple first-order approximation to the effective epidemics (12). Then, the Results section shows that they are good reproduction number Rt . Then, similarly to the use of Rt , if ρt > short-term predictors, allowing a proper evaluation of the state of 1, the epidemic is growing because the number of new cases today the epidemic. is bigger than the number of new cases 5 days ago. Otherwise, the incidence of new cases is lower and the epidemic is reducing. METHODS When ρt = 1, the epidemic is not growing nor reducing and the new number of cases is maintained. The reported data from government sources about the pandemic Propagation rate is very sensitive to noise effects. Thus, at are large and normally not unified in their criteria. They must initial and final stages, when the number of new cases is small, be properly assessed and curated to obtain useful and truthful the behavior of ρt does not represent the reality. Besides, the information, especially about the trend of its spreading. Our temporal evolution of ρt is subject to human reporting data main aim is to analyze whether the situation is improving or effects, such as the weekend effect (12). To address these issues, getting worse, so that it can be used by policy makers when we define ρT as the moving average of ρt for T days: deciding different socioeconomic measures. To address this issue, we have developed or adapted three indexes to compare different T−1 situations and evaluate the resurgence risk: an empiric estimation 1 X 2 of the reproduction number, an index of the contagious pool ρT (t, τ ) = ρt (t + i, τ ). T and a risk index of the effective potential growth. Moreover, i=− T−1 2 Frontiers in Public Health | www.frontiersin.org 2 July 2021 | Volume 9 | Article 633123 Català et al. Monitoring COVID-19 Pandemic With an Empirical Approach In the following, we will set T = 7 days in order to avoid the In fact, this product provides, under constant conditions, an weekend effect (18). Note that this definition is only valid for odd order of magnitude of the expected number of new cases that will values of T. Otherwise, one would find non-integer values of t, be diagnosed (i.e., that will be reported) for the next 14 days per which is the time variable in days and must be an integer. 105 inhabitants. However, EPGREP is a magnitude that changes over time, so it can be used for evaluating the risk associated with The 14-Day Attack Rate: A Measure of Contagious this potential growth at any moment during the epidemic. This People index was used to decide which Catalan Sanitary Regions were Parameters ρt and ρ7 are useful to identify if an epidemic is deconfined, among other criteria. growing; however, it is not the same to obtain a ρt bigger than If we want to evaluate the risk based on the estimated pool of 1 with a large amount of potentially contagious individuals or, on contagious population, A14,EST , we obtain the expression: the contrary, if the fraction of potentially contagious individuals is small. The number of contagious individuals is a difficult quantity to estimate since contagious people are not necessarily EPGREP (t) EPGEST (t) = . detected. An index commonly used to follow active cases in DR (t) COVID-19 is the 14-day attack rate (i.e., new cases of last 14 days per 105 inhabitants, A14 ) (19, 20), which is defined as follows: The Biocom-Cov Scale Popular language often uses sea-related vocabulary to describe N (t) − N (t − 14) A14 (t) = · 105 , the dynamics of COVID-19 in a region or country: first wave to population refer to the first peak, secondary waves to describe subsequent outbreaks, or tsunami to refer to a totally uncontrolled outbreak, where N is the number of cumulative reported cases in a among others. The Douglas Scale is a discrete way of classifying country, and population is the population of the country or region the situation of the sea that considers, among others, the height under consideration. of the waves. We propose a discrete way of classifying the Nevertheless, the reported cases criterion is very different situation with regard to daily new cases in what we name Biocom- through the countries due to many facts: type of test reported, Cov scale. reporting frequency, update of reported data temporal series, Looking at orders of magnitude, 200 active cases per 100,000 number of available tests, percentage of diagnosed cases, biased inhabitants pose an impossible challenge, while 20 active cases subpopulations that are over/underdiagnosed, etc. (21). Thus, the can be dealt with by public health officials if they are properly number of reported cases is not as representative as one would found and the structure of test and trace is in place. Assuming expect. As for the reported deaths, there is also variability among that active cases are well-represented by A14 , corresponding countries but at lower levels (22). Then, it is possible to calculate average daily new cases would be 200/14 = 14.3 daily new cases. the diagnosis rate (DR) from these data, which allows us for the Then, the 14 daily new cases are placed as the threshold for the estimation of the real number of cases (Nest ) (22). This estimation highest level. Here, 100 active cases per 100,000 inhabitants are agrees with different seroprevalence testing done afterward (23). a highly problematic situation from the control perspective. This We can define the estimated 14-day attack rate (A14,EST ) as: situates another important threshold at 7–8 daily new cases per 100,000 inhabitants. Similarly, five daily new cases per 100,000 NEST (t) − NEST (t − 14) should count as rather high situation, and two daily cases (around A14,EST (t) = · 105 . Population 30 active cases) should be the limit of moderate. With these ideas in mind, we build the scale shown in Table 1, which gives a Assuming a constant diagnosis rate, this equation can complete and accurate picture of the situation. be simplified: A14 (t) A14,EST (t) = . DR (t) TABLE 1 | Biocom-Cov scale to assess the epidemic degree of a region. By symmetry, we will name the 14-day attack rate evaluated with Pandemic degree Daily new incident cases per 105 inhabitants reported data as A14,REP . 0 0 Effective Potential Growth 1 0–0.1 The effective potential growth (EPG) is a risk index that evaluates 2 0.1–0.5 the potential epidemic growth at short term. It is defined as 3 0.5–1.25 the product between the mean propagation rate of the last 7 4 1.25–2 days (ρ7 ), which reflects the velocity at which the epidemic is 5 2–3 spreading, and the 14-day attack rate (A14,REP ), which accounts 6 3–5 for the contagious population that could propagate at that rate: 7 5–8 8 8–14 EPGREP (t) = ρ7 (t) · A14,REP (t) . 9 >14 Frontiers in Public Health | www.frontiersin.org 3 July 2021 | Volume 9 | Article 633123 Català et al. Monitoring COVID-19 Pandemic With an Empirical Approach FIGURE 1 | Index ρt evolution and periodicity. (A) In blue dots, evolution of ρt in Europe; black thick line, 7-day moving average of ρt (ρ 7 ). (B) Fourier transform of ρt . The peak at frequency 1/7 days−1 is pointed out. (C) Seven-day moving average of ρt (ρ 7 ) for different European countries: Spain (red), Germany (green), Italy (blue), and United Kingdom (purple). Data were obtained from the European Center for Disease Prevention and Control (24) on May 25, 2020. RESULTS is being spread; the higher, the worse. Index A14 provides a way to quantify active cases, i.e., it is an indicator of the contagious ρt , an Indicator of Contagious Velocity people that can spread the virus at the velocity ρ 7 . Finally, EPG The evolution of the empiric reproduction number, ρt , is crucial evaluates the risk that results from both indexes. A high ρ 7 with to evaluate the dynamics of the pandemic. The number of a very low A14 does not represent high risk, and this is reflected new cases increases until ρt gets smaller than 1. A stable ρt by low values of EPG. We consider two types of EPG, EPGREP below 1 is needed to reduce the new number of cases. The and EPGEST evaluated, respectively, with the data reported by blue dots in Figure 1A show the weekend effect in ρt at the countries and with the estimated real incidence. European level caused by a weekend delay in data recording. Figure 2 shows the values of several variables and indexes Therefore, oscillations with a 7-day periodicity are observed. (including ρ 7 , A14,REP , A14,EST , EPGREP , and EPGEST ) for Figure 1B shows the Fourier transform of ρt , where the 7- different countries on May 22, 2020. At that date, countries at the day oscillation period is clearly identified. Therefore, the 7- highest risk according to EPGREP were Perú, Brazil, and USA. If day moving average in ρt is necessary to study the epidemic we look at the risk given by EPGEST , the most worrying situation dynamics, given by ρ7 . Figure 1A shows the comparison between was that of Brazil followed by Sweden, UK, USA, and Perú. A daily ρt and the averaged ρ7 . The comparison of ρ7 through comparison between the estimated and reported EPG is useful in various countries allows for the exploration of differences in determining which countries are underreporting. COVID-19’s dynamics among countries. Figure 1C shows the Figure 2 also incorporates the Biocom-Cov degree of each evolution of ρ7 in Germany, Italy, Spain, and United Kingdom. country. As shown, this scale immediately facilitates a good Italy was the first European country where the pandemic started, visualization of the country situation beyond the need for other while the fastest pandemic control (i.e., decrease in ρ7 ) was in precise numerical indicators. In order to bypass the weekend Germany. Spain followed a control dynamics similar to the one effect, we assign the Biocom-Cov degree looking at the average in Italy, even slightly faster. Italy, Germany, and Spain have of daily new cases in the last week. managed to stabilize ρ7 below 1. Nevertheless, ρ7 is still sensitive Time evolutions of A14,EST , EPGEST and EPGEST are shown to fluctuations in data, which is the reason why it increases in Figure 3 for the five European countries with the highest above 1 in two short periods for Spain. This country experienced number of total reported cases (UK, Spain, Italy, France, and some problems with data reporting the last weeks, and a couple Germany). Figure 3A shows that Italy was the first country in of spikes in new data masks the real decreasing global trend. Europe to reach the peak of contagious cases. Spain is the country United Kingdom shows a slower control of the epidemic, with that arrived at the highest number of contagious people per 105 a ρ7 that remained around 1 for several weeks, and that finally inhabitants, nearly doubling that of Italy (217 vs. 124), which was dropped below 1. the country with the second highest incidence. UK curve shows that this country has transited a plateau rather than a peak, thus Reported and Estimated Risk Indexes illustrating the delay in controlling the epidemic. Risk indexes ρ 7 , A14 , and EPG (on their reported and estimated Figure 3B shows the evolution of EPGREP in these countries. versions) have been used in the daily tracking of the epidemic Spain was the one that reached the highest risk on March 28, in European and several non-European countries. Index ρ 7 2020, when it had an EPGREP of 214. This index also provides provides a quantification of the velocity at which the epidemic similar risk levels achieved by the other four countries, which at Frontiers in Public Health | www.frontiersin.org 4 July 2021 | Volume 9 | Article 633123 Català et al. Monitoring COVID-19 Pandemic With an Empirical Approach FIGURE 2 | Table with last indexes and reported cases value as of May 22, 2020. Left to right columns are: country, cumulative reported cases, number of total cases per 105 inhabitants (attack rate), cumulative number of reported deaths, number deaths per 105 inhabitants, reported number of new cases last 14 days (active cases), reported active cases per 105 inhabitants (14-day attack rate), estimated number of new cases for the last 14 days (active cases), estimated active cases per 105 inhabitants (14-day attack rate), 7-day moving average empiric reproduction number (ρ7 ), effective potential growth of reported cases (EPGREP ), estimated effective potential growth (EPGEST ), and Biocom-Cov degree. Each column has its own color scale as seen at the bottom of the figure. Data were obtained from the European Center for Disease Prevention and Control (24) and World Health Organization (25) on May 23, 2020. FIGURE 3 | Indices evolution among time. (A) Evolution of reported 14-day attack rate per country. (B) Effective Potential Growth evolution per country computed with reported cases. (C) Effective Potential Growth evolution per country computed with estimated cases. Blue lines for Spain, green for Italy, red for France, cyan for United Kingdom, purple for Germany, and black for Europe. Data were obtained from the European Center for Disease Prevention and Control (24) on May 16, 2020. the same time would be similar to the one reached by Europe but at different moments, slightly higher (i.e., worse) than the as a whole. Moreover, if we compare the same countries using one shown by European average. Risk reduction in Spain is faster the estimated EPGEST (Figure 3C), we observe that Germany has than the one observed in Italy; in fact, they are in a similar level been in a better situation than the one reflected by reported data now. Germany and France show a similar EPGREP but totally when compared with other countries. According to this index, different estimated EPGEST . The number of reported deaths in Italy, UK, and France have had a similar level of estimated risk this pair of countries is quite different, with a six-fold increase Frontiers in Public Health | www.frontiersin.org 5 July 2021 | Volume 9 | Article 633123 Català et al. Monitoring COVID-19 Pandemic With an Empirical Approach FIGURE 4 | Risk diagrams for different countries using reported and estimated cases. Each point represents a different day. Starting and final days are marked by date. Color code depends on Effective Potential Growth (EPG). Two color codes are used: one for reported (top right bar) and another for estimated (bottom bar). Using reported cases: (A) Germany, (B) Spain, and (C) France risk diagrams. Using estimated cases: (D) Germany, (E) Spain, and (F) France risk diagrams. Note that x-scales are different, but plots can be compared through the color background. between them. This leads to the estimation of a higher diagnosis velocity of propagation drops down, but, since it is still above 1, rate in Germany than the one in France (25 vs. 7%) (22). From the attack continues to increase. The inflection point is achieved this analysis, it is important to note that the time scale in the when ρ7 crosses the threshold of 1. At that moment, the number reduction of the number of active cases is larger than the time of new infected cases starts to decrease; meanwhile, ρ7 remains scale observed during the growing phase. below 1. Then, the curve moves toward the green zone. Analyzing case-by-case, we see that Spain was the country Risk Diagrams: A Tool to Evaluate Risk that was in a worse situation since they went further in the risk The risk diagram is a tool to visualize the evolution of a zone, where there are more than 214 cases per 105 inhabitants region or country in terms of ρ7 (y-axis), A14 (x-axis), and expected for the next 14 days. Looking at the estimated diagrams, EPG (background color) with either reported or estimated we see that Germany was the country with a better estimated real data. Figure 4 shows the risk diagrams for France, Spain, and situation, since they did not reach the danger zone. Germany. In the upper part, risk diagrams using reported The analysis of a full country by using only this risk diagram data (Figures 4A–C), in the bottom part, risk diagrams using could lead to a misleading visualization of the real situation. In estimated data (Figures 4D–F). We consider a risk situation (red) these figures, we are studying the situation of these countries for EPGREP >100 and EPGEST >1,000. by considering them as a whole. However, the situation in The color code is related to the ability of a country or region to the different regions of each country could be very different, do contact tracing, setting at 1,000 estimated real cases the red as and a deeper analysis must be done. The fragmentation of the the threshold where it is impossible to carry it out. The maximum country into several regions allows us to better understand the number of daily PCR tests per 105 people performed sustainably situation as well as how the propagation has occurred. This has been of the order of 50–100 (26) in affected countries. At this information is crucial for policymakers to properly develop novel level of testing, it would take between 10 and 20 days to process all strategies during confinement and deconfinement. For instance, active cases, which is precisely the time it takes for infected people the regional variability for the cases of Spain is shown in Figure 5. to get seriously sick or die. Unless the infrastructure is scaled up COVID-19 risk diagrams are updated daily at Català et al. (12). dramatically in the future, 1,000 active cases are impossible to test and trace nowadays. Risk Diagrams as a Tool to Detect The general dynamics along the risk diagram is quite similar Uncontrolled Outbreaks for all countries. At the beginning of the pandemic, the attack Most European countries have overcome an initial peak and then rate is low while the propagation velocity is high (ρ7 > 1). When entered a long tail that is expected to finish when herd immunity restrictions and physical distancing measures are applied, the is achieved or an efficient vaccine is available. Meanwhile, the Frontiers in Public Health | www.frontiersin.org 6 July 2021 | Volume 9 | Article 633123 Català et al. Monitoring COVID-19 Pandemic With an Empirical Approach FIGURE 5 | Risk diagrams for Spain and some of its regions on July 17, 2020. (A) Aragon, (B) Catalonia, (C) Community of Madrid, (D) Community of Navarra, and (E) Spain risk diagrams. Each circle corresponds to a day, last day is marked with a black filled circle. (F) Spain region situation, where each dot is the position for each Spanish region on July 17, 2020 (see legend), and filled black circle is the situation of Spain. Background color depends on the Effective Potential Growth (EPG) risk. Red marks EPG = 100 and green is for EPG = 0; there is a linear degradation between both. Zones with an EPG higher than 100 are also marked in red. Data are obtained from Datadista (27) and Instituto de Salud Carlos III (28) on July 18, 2020. main concern of regions and countries is the early detection community transmission and the need for restrictions in mobility of outbreaks and their evaluation from the risk perspective, so or social interactions. that physical distancing measures or mobility restrictions can be imposed, if necessary. The main threshold between control and DISCUSSION uncontrol is the presence of community contagions, i.e., the loss of contagious chain traceability. We have shown that ρ7 and A14 are good indicators for assessing In risk diagrams, the loss of traceability is represented by epidemiological risk in regions or countries. They can be replaced a red zone, where control by test and trace is not possible. by alternative ways of measuring spreading rate and contagious Figure 6 shows the usefulness of risk diagrams to distinguish potential, but an indicator for each one must be considered if local controlled outbreaks from uncontrolled ones. In this figure, the risk is to be evaluated properly. In particular, the proposed we show controlled and uncontrolled outbreaks that start with an way to assess ρ7 is very sensitive to changes in the transmission initial increase in ρ7 up to 2. First, we can observe two outbreaks dynamics, which can be especially useful to detect those changes. in the control zone as a certain increase in ρ7 that is not followed A slight change in this methodology, applying the 7-day moving by an increase in active cases (Tarragona province). Therefore, average to the cases (nc) instead of the ρ7 , provides a more this is observed as a simple up–down trajectory of the plot in the stable evaluation of the transmission rate, which is less affected low risk zone. Contrarily, an outbreak that is not well-delimited by artifacts such as changes in the diagnosis protocols or and immediately controlled drags the curve to the right (Lleida). underreporting of holidays. While this dynamics occurs in the yellow zones, control with We have proposed the EPG index as a simple way to soft measures is possible (in this case, a slowing down of the de- account for both factors. During the growth phase (pre- escalation process). When a new increase in new cases leads the peak) of the epidemic, EPG is used to track the dynamics curve toward the red zone, it strongly suggests the presence of of the epidemic, and when it increases above a threshold, Frontiers in Public Health | www.frontiersin.org 7 July 2021 | Volume 9 | Article 633123 Català et al. Monitoring COVID-19 Pandemic With an Empirical Approach FIGURE 6 | Controlled and uncontrolled outbreaks in the risk. (A) Tarragona province: Two consecutive increases in ρ7 occur in the green zone and can be controlled with simply test and trace. (B) Lleida province: An initial increase in ρ7 occurs in the green-yellow zone and can be controlled through a slowing down of de-escalation process, while a second increase in ρ7 enters the red zone and becomes an uncontrolled outbreak. Data are obtained from Datadista (27) and Instituto de Salud Carlos III (28) on October 23, 2020. EPG can indicate the need for new control measures. the appearance of a more transmissible variant or to the increase Nevertheless, we consider the main focus of this index to in the vaccination coverage among a certain population can be be the management of the deconfinement process. It is easily detected using such an empiric approach, since it lacks a essential that de-escalation phases take into account the relative mechanistic hypothesis to be revised. epidemiological risk of the region or country in the context of the health system robustness and operability, together DATA AVAILABILITY STATEMENT with the capability to incorporate contact tracing strategies that avoid new outbreaks. In addition, an increase in EPG The original contributions presented in the study are included also can be used as an alarm symptom when looking for in the article/supplementary material, further inquiries can be secondary outbreaks. directed to the corresponding author. Risk diagrams are a good way to visualize the situation and dynamics of countries in this sense. Its color scale is based on AUTHOR CONTRIBUTIONS EPG values and its relation with the ability to trace given by the testing infrastructure typical in European countries. This scale, CP, P-JC, and DL conceived the study. MC, MM, DC, PP, and however, can be particularly adapted to each country, considering TU wrote the codes and prepared tables and figures. MC, MM, the level of incidence that the country can assume with the local SA, EA-L, DL, P-JC, and CP analyzed the results, discussed the ability to test and trace. Finally, those countries with a diagnostic implications, and proposed reformulations of the indexes. MC, level below 10% should try to incorporate estimations on the MM, PP, SA, EA-L, and CP prepared the draft version. All authors management of the epidemic (for instance, using EPGEST instead read and approved the final version. of EPGREP ). Most important limitations of any empirical approach to FUNDING the COVID-19 pandemic are related to the diagnosis effort and bias (22). An insufficient diagnostic effort may affect not CP, P-JC, and MC received funding from La Caixa Foundation only an underreporting of cases but also an underreporting (ID 100010434), under agreement LCF/PR/GN17/50300003. of deaths. In addition, holiday periods that modify the basic P-JC received funding from Agència de Gestió d’Ajuts structure of 5 working days plus 2 holidays per week can generate Universitaris i de Recerca (AGAUR), Grup Unitat de Tuberculosi artifacts on the observed data. In any case, and in conclusion, Experimental, 2017-SGR-500. CP, DL, SA, and MC received the use of empirical indexes like EPG and the risk diagrams funding from Ministerio de Ciencia, Innovación y Universidades can help with the monitoring of the COVID-19 epidemic and and FEDER, with the project PGC2018-095456-B-I00. This work to address relevant questions, for example, the classification also has been partially funded by the European Commission-DG analysis of the evolution of the cases or the appearance of new Communications Networks, Content and Technology through outbreaks. In particular, any change in historical trends due to the contract LC-01485746. Frontiers in Public Health | www.frontiersin.org 8 July 2021 | Volume 9 | Article 633123 Català et al. Monitoring COVID-19 Pandemic With an Empirical Approach REFERENCES 17. Ganyani T, Kremer C, Chen D, Torneri A, Faes C, Wallinga J, et al. Estimating the generation interval for coronavirus disease (COVID-19) based 1. Pedrosa AL, Bitencourt L, Fróes ACF, Cazumbá MLB, Campos RGB, de Brito on symptom onset data, March 2020. Eurosurveillance. (2020) 25:2000257. SBCS, et al. Emotional, behavioral, and psychological impact of the COVID-19 doi: 10.2807/1560-7917.ES.2020.25.17.2000257 pandemic. Front Psychol. (2020) 11:566212. doi: 10.3389/fpsyg.2020.566212 18. Prats C, Alonso S, Álvarez-Lacalle E, Marchena M, López D, Català M, et al. 2. Banerjee D, Kosagisharaf JR, Sathyanarayana Rao TS. “The dual pandemic” of Analysis and prediction of COVID-19 for EU-EFTA-UK and other countries. suicide and COVID-19: a biopsychosocial narrative of risks and prevention. Barcelona: Universitat Politècnica de Catalunya (2020). Available online at: Psychiatry Res. (2021) 295:113577. doi: 10.1016/j.psychres.2020.113577 http://hdl.handle.net/2117/186009 (accessed April 27, 2020) 3. Buonsenso D, Roland D, De Rose C, Vásquez-Hoyos P, Ramly B, Chakakala- 19. European Center for Disease Prevention and Control (ECDC). COVID-19 Chaziya JN, et al. Schools closures during the COVID-19 pandemic: Surveillance Report. Available online at: https://covid19-surveillance-report. a catastrophic global situation. Pediatr Infect Dis J. (2021) 40:e146–50. ecdc.europa.eu/ (accessed May 21, 2020). doi: 10.1097/INF.0000000000003052 20. Backer JA, Klinkenberg D, Wallinga J. Incubation period of 2019 novel 4. Viner RM, Bonell C, Drake L, Jourdan D, Davies N, Baltag V, et al. coronavirus (2019-nCoV) infections among travellers from Wuhan, Reopening schools during the COVID-19 pandemic: governments must China, 20–28 January (2020). Eurosurveillance. (2020) 25:2000062. balance the uncertainty and risks of reopening schools against the clear doi: 10.2807/1560-7917.ES.2020.25.5.2000062 harms associated with prolonged closure. Arch Dis Child. (2020) 106:111–3. 21. Morris C, Reuben A. Can You Compare Different Countries? British doi: 10.1136/archdischild-2020-319963 Broadcasting Corporation (2020). Available online at: https://www.bbc.com/ 5. Moghadam MT, Taati B, Paydar Ardakani SM, Suzuki K. Ramadan fasting news/52311014 (accessed November 24, 2020). during the COVID-19 pandemic; observance of health, nutrition and exercise 22. Català M, Pino D, Marchena M, Palacios P, Urdiales T, Cardona P- criteria for improving the immune system. Front Nutr. (2021) 7:570235. J, et al. Robust estimation of diagnostic rate and real incidence of doi: 10.3389/fnut.2020.570235 COVID-19 for European policymakers. PLoS ONE. (2021) 16:e0243701. 6. Atalay S, Solmazer G. The relationship between cultural value orientations doi: 10.1371/journal.pone.0243701 and the changes in mobility during the covid-19 pandemic: a national-Level 23. Instituto de Salud Carlos III. Estudio eNE-COVID19: Primera Ronda Estudio analysis. Front Psychol. (2021) 12:578190. doi: 10.3389/fpsyg.2021.578190 Nacional de Sero-Epidemiología de la Infección por SARS-CoV-2 en España. 7. Jüni P, Rothenbühler M, Bobos P, Thorpe KE, da Costa BR, Fisman DN, (2020). Available online at: https://www.mscbs.gob.es/ciudadanos/ene- et al. Impact of climate and public health interventions on the COVID- covid/docs/ESTUDIO_ENE-COVID19_PRIMERA_RONDA_INFORME_ 19 pandemic: a prospective cohort study. CMAJ. (2020) 192:E566–73. PRELIMINAR.pdf (accessed June 23, 2021). doi: 10.1503/cmaj.200920 24. European Centre for Disease Prevention Control (ECDC). Coronavirus 8. Wu Y, Jing W, Liu J, Ma Q, Yuan J, Wang Y et al. Effects of Disease 2019 (COVID-19) in the EU/EEA the UK. European Centre for Disease temperature and humidity on the daily new cases and new deaths Prevention and Control (2020). Available online at: https://www.ecdc.europa. of COVID-19 in 166 countries. Sci Total Environ. (2020) 28:139051. eu/en/cases-2019-ncov-eueea (accessed June 23, 2021). doi: 10.1016/j.scitotenv.2020.139051 25. World Health Organization. Coronavirus Disease (COVID-2019) Situation 9. Riou J, Althaus CL. Pattern of early human-to-human transmission Reports. Available online at: https://www.who.int/emergencies/diseases/ of Wuhan 2019 novel coronavirus (2019-nCoV), December novel-coronavirus-2019/situation-reports (accessed November 24, 2020). 2019 to January 2020. Eurosurveillance. (2020) 25:2000058. 26. Català M, Cardona PJ, Prats C, Alonso S, Álvarez-Lacalle E, Marchena doi: 10.2807/1560-7917.ES.2020.25.4.2000058 M, et al. Analysis and prediction of COVID-19 for EU-EFTA-UK and 10. Dietz K, Heesterbeek JA, Tudor DW. The basic reproduction ratio for sexually other countries. Barcelona: Universitat Politècnica de Catalunya (2020). transmitted diseases part 2. Effects of variable HIV infectivity. Math Biosci. Available online at: http://hdl.handle.net/2117/188444 (accessed May (1993) 117:35–47. doi: 10.1016/0025-5564(93)90016-4 20, 2020). 11. Wallinga J, Teunis P. Different epidemic curves for severe acute respiratory 27. Datadista. Datasets Relacionados con la Incidencia de la COVID-19 en España. syndrome reveal similar impacts of control measures. Am J Epidemiol. (2004) (2020). Available online at: https://github.com/datadista/datasets/tree/master/ 160:509–16. doi: 10.1093/aje/kwh255 COVID%2019 (accessed May 25, 2020). 12. Català M, Cardona PJ, Prats C, Alonso S, Álvarez-Lacalle E, Marchena M, et al. 28. Instituto de Salud Carlos III. Situación de COVID-19 en España. Available Analysis and prediction of COVID-19 for EU-EFTA-UK and other countries. online at: https://cnecovid.isciii.es/covid19/resources/agregados.csv (accessed Barcelona: Universitat Politècnica de Catalunya (2020). Available online at: May 18, 2020). https://upcommons.upc.edu/handle/2117/110978 (accessed May 25, 2020). 13. Heffernan smith JR, Wahl L. Perspectives on the basic reproduction ratio. J R Conflict of Interest: The authors declare that the research was conducted in the Soc Interface. (2005) 2:281–93. doi: 10.1098/rsif.2005.0042 absence of any commercial or financial relationships that could be construed as a 14. Weston S, Frieman MB. COVID-19: knowns, unknowns, and questions. potential conflict of interest. Msphere. (2020) 5:e00203-20. doi: 10.1128/mSphere.00203-20 15. Bai Y, Yao L, Wei T, Tian F, Jin DY, Chen L, et al. Presumed Copyright © 2021 Català, Marchena, Conesa, Palacios, Urdiales, Alonso, Alvarez- asymptomatic carrier transmission of COVID-19. JAMA. (2020) 323:1406–7. Lacalle, Lopez, Cardona and Prats. This is an open-access article distributed doi: 10.1001/jama.2020.2565 under the terms of the Creative Commons Attribution License (CC BY). The use, 16. Lauer SA, Grantz KH, Bi Q, Jones FK, Zheng Q, Meredith HR, et al. distribution or reproduction in other forums is permitted, provided the original The incubation period of coronavirus disease 2019 (covid-19) from publicly author(s) and the copyright owner(s) are credited and that the original publication reported confirmed cases: estimation and application. Ann Intern Med. (2020) in this journal is cited, in accordance with accepted academic practice. No use, 172:577–82. doi: 10.7326/M20-0504 distribution or reproduction is permitted which does not comply with these terms. Frontiers in Public Health | www.frontiersin.org 9 July 2021 | Volume 9 | Article 633123