Estimation of time-varying reproduction numbers of COVID-19 in American countries with regards to non-pharmacological interventions

Introduction

On January 21, the first case of COVID-19 in the United States¹ was announced in a traveler who had arrived in the country from China 5 days earlier. Wuhan, China is where the health emergency originated² in December 2019. From late January, the North American continent started to see case reports in different states, with cases arriving in South America on February 26 through Brazil after passing through Central America and the Caribbean³. Today, America is the epicentre of the pandemic, with the USA and Brazil being the two largest countries with the most cases. In addition, Peru, Chile, Mexico, Canada, Colombia and Ecuador are among the top 25 countries with the highest incidence worldwide.

Since then, governments across the region have implemented a series of non-pharmacological measures to protect their citizens and contain the spread of COVID-19. All South American countries, including the French overseas region of French Guiana and the Falkland Islands, have reported the presence of coronavirus within their borders, and all took measures at different times of the outbreak.

Non-pharmacological measures of containment (mitigation and suppression) have been implemented from the index case on different dates in each country^2,4–6. This is based on projections of contagion and lethality resulting from SIR epidemiological models (susceptible, infected and recovered)⁷ with data taken from the behavior of the virus in Asia and Europe especially, and with approximations to the basic reproduction number (R₀)^8,9, effective reproduction number (R_e), and variable time estimation of R_e (R_t), among others. R₀ estimates the speed with which a disease can spread in a population and plays a crucial role in predicting the prevalence of infectious disease, in addition to better understanding an epidemic outbreak and preparing the corresponding public health response. R_t is the reproduction number at time t, which reveals the actual transmission rate of the virus at a given moment. This number will vary depending on the control protocols established in each country.

Mitigation measures are aimed at slowing down the spread of infection by progressively reducing R₀¹⁰. Suppression measures aim to reduce R₀ below 1, to progressively decrease the number of cases until they disappear. The greatest challenge that this strategy entails is that it needs to be maintained until the pandemic has disappeared or an effective treatment or vaccine is available^6,10,11. The estimation of R₀ and R_t in different scenarios allows a better prognosis of the trends of the global epidemic¹², in addition to evaluating the effectiveness in reducing transmission with the measures taken by the different countries to contain COVID-19¹¹.

While this article was being prepared, the R₀ and R_t values of some countries were published and modeled in three intervention scenarios; however, they only included the United States and Argentina from the American continent, limiting the comparative analysis for the characteristics of America¹³. Previous work has calculated the R₀ and R_t using the maximum likelihood method and sequential Bayesian method, and found that European and North American countries possessed a higher R₀ and unsteady R_t fluctuations, whereas some heavily affected Asian countries showed a relatively low R₀ and declining R_t , with a small number of patients in Africa and Latin America, yet with the potential risk of large outbreaks. Xu et al.¹³ who determined R_t as an indicator to measure the transmission of SARS-CoV-2 before and after interventions, showed that the change in the rate of transmission of SARS-CoV-2 is associated with reducing social interactions.

This study aimed to estimate R₀ and R_t in multiple countries of the American continent and evaluate their behavior from the first (index) case in each country, and the time of implementation of the chosen non-pharmacological containment measures.

Methods

Results

Basic reproduction number (R₀)

El Salvador, the Dominican Republic, Panama, and Peru have the lowest R₀; in contrast, the USA and Canada have the highest. Other countries have an R₀ around 1.4 (Figure 1).

Figure 1. Basic reproduction number by country.

Time-varying effective reproductive number (R_t)

Countries could be divided into three groups based on the varied behavior of R_t over time (Figure 2–Figure 4). The first group (Mexico, USA, Colombia and Brazil) started with a high R_t, which decreased post-intervention. In the second group (Canada, Argentina, Chile Peru, Panama and Dominican Republic), the intervention was performed at the moment when the R_t, is high and this then decreased slowly post-intervention. In the third group (Bolivia, Peru and Guatemala), the R_t is erratic and cannot be attributable to the intervention; the R_t increases and decreases without any explanation.

Analysis by regions (North, Central and South America) shows differences between regions and countries. The countries of North America have a low R_t, which increased over time. Post-intervention control is achieved by a progressive decrease in all cases in Canada, Mexico and the USA. The behavior of the Dominican Republic and the countries of Central America is similar. Excepting Bolivia, Guatemala and Peru, the countries of South America and Central America show a similar behavior to the first group; the R_t is high at the beginning and decreases suddenly after control measures. However, Peru had a considerable peak after intervention implementation, as did Chile and Panama. This is possibly related to the flexible measures of mass community isolation

Despite the differences between the first and second group, the R_t is high before the intervention in all countries and its decrease is achieved post-intervention, corroborating the purpose of these interventions in the mitigation and suppression of a pandemic.

Figure 2. Time-varying effective reproductive number by country in North America.

The dotted horizontal line shows the target of 1 at R_t, and the vertical line represents the start date of the containment measures decreed in each country. Although there is a difference in the days, there is also a difference in the size of the countries, both in surface area and number of inhabitants that would allow mobility and different behavior of virus spread.

Figure 3. Time-varying effective reproductive number by country in South America.

The dotted horizontal line shows the target of 1 at R_t, and the vertical line represents the start date of the containment measures decreed in each country. Although there is a difference in the days, there is also a difference in the size of the countries, both in surface area and number of inhabitants that would allow mobility and different behavior of virus spread.

Figure 4. Time-varying effective reproductive number by country in Central America.

The dotted horizontal line shows the target of 1 at R_t, and the vertical line represents the start date of the containment measures decreed in each country. Although there is a difference in the days, there is also a difference in the size of the countries, both in surface area and number of inhabitants that would allow mobility and different behavior of virus spread.

Discussion

The for the countries selected in this study was 1.2–1.8, which is much lower compared to China where it was reported to be 2.2, 2.6, 3.8, or even 6.47 according to different research^2,18. Our study findings agree with Xu et al.¹³ in the estimated for the United States and Argentina. The R_t for these countries is also identical to that found by Xu et al.

As in the study by Xu et al.¹³, there are some limitations to the current study. First, the variation in depends on the series interval. Here, we assume 7.5 days for the 13 countries due to limited information on the disease in different nations¹³. In addition, we noticed some errors in the John Hopkins University data when we cross referenced these with official country websites. We were able to correct these by searching for cases directly on the official website of each country. Another drawback is the diversity of RT-PCR (real-time reverse transcription polymerase chain reaction) tests used in the diagnostic process and their sensitivity, which may throw up false negatives, perhaps under-estimating the true impact of the pandemic in various countries. Although data were available from other countries, in the case of Ecuador, the report was not included, because cases were not confirmed by RT-PCR, which is a that s technical criteria.

The epidemic trends in different regions around the world are significantly different due to the great difference in the design and implementation of prevention and control measures⁵. Estimates of both R₀ and R_t in different countries will enable a better forecast of global epidemic trends.

In conclusions, our study shows that there is a close relationship between R_t and non-pharmacological interventions decreed by countries’ governments for the control of the COVID-19 pandemic. Additionally, we showed that there are immediate changes in the behavior of the R_t, and therefore in the progression of the outbreak, when the interventions are implemented closer to the index case for each country.

Data availability