key: cord-0722915-e1zgqqqu
authors: Billah, M. A.; Miah, M. M.; Khan, M. N.
title: Reproductive number of COVID-19: A systematic review and meta-analysis based on global level evidence
date: 2020-05-26
journal: nan
DOI: 10.1101/2020.05.23.20111021
sha: 0d0f7eed67d2f1cbf26b7a4c8252b206fb7d7c9f
doc_id: 722915
cord_uid: e1zgqqqu

Background The coronavirus (COVID-19) is now a global concern because of its higher transmission capacity and associated adverse consequences including death. The reproductive number of COVID-19 provides an estimate on the possible extent of the transmission. This study aims to provide the average reproductive number of COVID-19 based on available global level evidence. Methods We searched three databases (PubMed, Web of Science, and Science Direct) to find studies reported the reproductive number of COVID-19. The searches were conducted using a pre-specified search strategy that includes keywords of COVID-19 and its reproductive number related terms, which were combined using the Boolean operators. Narrative synthesis was used to explain the studies included and the meta-analysis was used to estimate the average reproduction number of COVID-19. Results Total of 30 studies included in this review whereas 24 of them were included in the meta-analysis. The average estimated reproductive number was 2.70 (95% CI, 2.21-3.30). We found evidence of very high heterogeneity (99.5%) of the reproductive numbers reported in the included studies. The highest reproductive number was reported for Diamond Princes Cruise Ship, Japan (14.8). In the country-level, the higher reproductive number was reported for France (R, 6.32, 95% CI, 5.72-6.98) following Germany (R, 6.07, 95% CI, 5.51-6.69) and Spain (R, 5.08, 95% CI, 4.50-5.73). We also found estimation models, methods, and the number of cases considered to estimate reproductive number were played a role in arising the heterogeneity of the estimated reproductive number. Conclusion The estimated reproductive number indicates an exponential increase of COVID-19 infection in the coming days. Comprehensive policies and programs are important to reduce new infections as well as the associated adverse consequences including death.

Knowing the accurate reproductive number of COVID-19, which means the capability of transmission per primary infected person to the secondarily infected persons, is significant for various reasons: to assess epidemic transmissibility and to predict the future trend of spreading [18] . These are important to reduce new infections and to design control measures such as social distancing [19] and to know the duration of keeping control measures [5] .

Moreover, it also helps to develop an effective epidemiological mathematical models considering possible transmission ways, such as, droplets and direct contacts with infected people, which are important to know the risk population and the appropriate epidemiologic parameters [20, 21] .

Various researchers worldwide estimated reproductive number of COVID-19. However, these were not consistent and measurement procedures and methods were different across the studies [20, 22] . The reproductive number estimated was also found different across the countries, stages of infection, and the preventive measures applied [23] . Another important source of variation of estimated reproductive number was type of reproductive numbers considered [20] . Of the three reproductive numbers estimated, namely basic reproductive indicates new infection will increase [24, 25] .

Considering the higher variability of the reproductive number estimated and its underlying importance, in this study, an attempt has been made to summarize available reproductive number of COVID-19 to give an average estimate. If applicable, sources of variations of the estimated reproductive number have also been addressed. Findings will help policymakers to know about the possible increase of COVID-19's patients and take policies and programs accordingly.

Literature searches were conducted in three databases on April 10, 2020: PubMed, Web of Science, and Science Direct. The pre-specified search strategies were used to search . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020 . . https://doi.org/10.1101 databases. We developed search strategies consisting of virus-specific (corona virus, coronavirus, SARS-CoV-2, COVID-19, nCoV-2019) and reproductive number related (reproduction number, transmissibility) keywords that were combined using the Boolean operators (AND, OR) . Additional searches were conducted in the reference list of the selected articles, and the relevant journal's websites.

Studies meet the following inclusion criteria were included: wrote in the English language, related to COVID-19, and presented the reproductive number of COVID-19. We did not apply any time restriction, i.e. all studies from the onset of COVID-19 to the date of conducting formal search were included. Studies that did not meet these criteria were excluded.

Two authors (MAB, MMM) extracted information by using a pre-designed, trailed, and modified data extraction sheet. The extracted information includes: year of publication, study's location, model used to estimate the reproductive number, time period for when the reproductive number was estimated, number of cases considered to estimate the reproductive number, assumption(s) that was/were set to a calculate the reproductive number, intervention strategy, types of reproductive number, and the estimated reproductive number with its 95% confidence interval (CI). The corresponding author (MNK) solved any disagreement on information extraction.

The information recorded were mostly dichotomous in nature where the numerical reproductive number was reported in all selected studies. We, therefore, used both narrative synthesis and meta-analysis to summaries findings from retrieved studies. Narrative synthesis . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 26, 2020. 

Total of 134 studies included, 130 of them were extracted from three databases searched ( Figure 2 ). Of these, 102 studies were excluded through title and abstract screening leaving 32 studies for full-text review. A total of 30 of them were finally included in this study and 24 of them were included in meta-analysis.

. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 26, 2020 . . https://doi.org/10.1101 Majority of the studies selected were conducted in China (8) [18, [26] [27] [28] [29] [30] [31] [32] and its province (6) [33-38]. The remaining studies were conducted in Japan (4) [19, [48] [49] [50] .

Of the 30 studies included in this review reported different reproductive numbers ( high-heterogeneity (99.5%) ( Figure 3 ). However, we did not found any evidence of publication biases that were assessed Trim and Fill estimate (result are not shown). The subgroup analyses to address was used to address heterogeneity. We found study's characteristics, such as types of reproductive number estimated, countries for which the reproductive number estimated, models and methods used to estimate the reproductive number, and the number of cases used to estimate reproductive number, played a significant role of arising such heterogeneity ( . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . .

It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . .

It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . https://doi.org/10.1101/2020.05.23.20111021 doi: medRxiv preprint . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . The results of the narrative synthesis are presented in . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . .

It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020 . . https://doi.org/10.1101 increasing new infection exponentially as in this period, an infected person is usually confounded in the community with other people. This risk is further increased significantly for the countries where population density are higher [59] .

This study also found evidence of the very high (99.5%) heterogeneity of the estimated reproductive number. Along with the factors described above, characteristics used to estimate reproductive numbers are important for such heterogeneity. For instance, the reproductive number found higher if it was estimated for the countries where no restriction was applied or restriction was applied in delayed. The forms of restrictions were control people's movement, personal hygiene, and wearing mask [10, 60] . These implications act to control virus transmission from an infected to the susceptible and reduce the new infections. These also affect the average transmissibility of COVID-19 within the specific population and settings [61, 62] .

Estimation models, assumptions applied, and estimating processes were empirical sources of variability of the estimated COVID-19's reproductive numbers [63] . For instance, studies included in this analysis were followed assumption of generation time (which is followed by the gamma distribution) or serial interval (which is followed by the poison distribution) which is an important source of heterogeneity [64] [65] [66] . The reason of such difference is the underlying concept: generation time refers to the average time between transmission the virus from an infected person to the non-infected person whereas serial interval refers duration between onset of symptoms in an index case to the transmission in a secondary case [64, 65, 67] . Moreover, the estimated reproduction number generated by mathematical models is dependent on numerous decisions made by the researcher such as homogeneity or . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020 May 26, . . https://doi.org/10.1101 May 26, /2020 doi: medRxiv preprint 0 heterogeneity of the population considered; use a deterministic or stochastic approach and which distributions to be used to describe the probable values of parameters [57] . Another important source of the heterogeneity is the type of reproductive number calculated. A basic reproductive number is usually produce higher value as it is applied in the situation where no restriction was applied to prevent new infections [22] . In comparison, both net reproductive number (ܴ ሻ , and time dependent reproductive number (ܴ ௧ ሻ are calculated for situation where restriction(s) was/were applied to control new infection [25, 47] . Therefore, these provide lower estimated value which supports our narrative synthesis. However, higher timedependent reproductive number than basic reproductive number that found in summary analysis is conflicting with the usual understanding. However, studies presented timedependent reproductive were conducted for Diamond Princes Cruise Ship in Japan Italy, France, Germany, and Spain where very large outbreak of COVID-19 reported mainly for delayed in taking preventive measures. This could be the reason of distortion reported in this study.

This study was first of its kind that provides an estimation of reproductive numbers based on worldwide' literature. Moreover, we have considered the heterogeneity of the reproductive numbers estimated worldwide and explored the sources of heterogeneity across selected studies' characteristics. However, many other factors may explain the sources of heterogeneity of the reported reproductive number of COVID-19 worldwide. We did not explore these because of the lack of data. Combining three different reproductive numbers . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . https://doi.org/10.1101/2020.05.23.20111021 doi: medRxiv preprint may overestimate or under-estimate the actual reproductive number of COVID-19. However, our sub-group analysis across types of reproductive number overcomes this limitation.

The average estimated reproductive number was 2.70. We found evidence of higher heterogeneity of the reproductive number reported worldwide. There are numerous causes of such heterogeneity, however, study related characteristics were types of reproductive number estimated, countries for which the reproductive number estimated, methods and models used to estimate reproductive number, and the number of cases considered to estimate reproductive number. This analysis indicates a significant increase of COVID-19 infections in the coming days. Strengthing existing preventive measures as well as new policies and programs are important to reduce new infections.

The authors are grateful to the authors of the paper included in this review.

There is no conflict of interest.

This study does not have any institutional support.

This study does not require any ethical approval as the review is based on the published articles and the original data are anonymous.

. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020 . . https://doi.org/10.1101 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . https://doi.org/10.1101/2020.05. 23.20111021 doi: medRxiv preprint Supplementary file . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2020. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 26, 2020. (2.05, 3.17) 5.77 (4.57, 7.29) 6.07 (5.51, 6.69) 4.56 (2.27, 9.17) 6.32 (5.72, 6.99) 5.08 (4.50, 5.73) 3.53 (2.10, 5. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 26, 2020. . https://doi.org/10.1101/2020.05. 23.20111021 doi: medRxiv preprint 

Coronavirus Update (Live) -Worldometer. In: Worldometer [Internet

World Health Organization

SARS-CoV-2: The Path of Prevention and Control

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study

First two months of the 2019 Coronavirus Disease (COVID-19) epidemic in China: real-time surveillance and evaluation with a second derivative model

History and Recent Advances in Coronavirus Discovery

Epidemiology and cause of severe acute respiratory syndrome (SARS) in Guangdong, People's Republic of China

Isolation of a Novel Coronavirus from a Man with Pneumonia in Saudi Arabia

Preliminary epidemiological assessment of MERS-CoV outbreak in South Korea

Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan

A new coronavirus associated with human respiratory disease in China

Real-time tentative assessment of the epidemiological characteristics of novel coronavirus infections in Wuhan, China, as at 22

Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72 314 Cases From the Chinese Center for Disease Control and Prevention

Correlation between travellers departing from Wuhan before the Spring Festival and subsequent spread of COVID-19 to all provinces in China

Preliminary prediction of the basic reproduction number of the Wuhan novel coronavirus 2019 nCoV

Pattern of early human-to-human transmission of Wuhan

The type-reproduction number T in models for infectious disease control

Preparation for Possible Sustained Transmission of 2019 Novel Coronavirus: Lessons From Previous Epidemics

Perspectives on the basic reproductive ratio

Quantifying Transmission

An Introduction to Mathematical Modeling of Infectious Diseases

An introduction to infectious disease modelling

Real-Time Estimation of the Risk of Death from Novel Coronavirus (COVID-19) Infection: Inference Using Exported Cases

A Simulation on Potential Secondary Spread of Novel Coronavirus in an Exported Country Using a Stochastic Epidemic SEIR Model

The reproductive number of COVID-19 is higher compared to SARS coronavirus

Social distance and SARS memory: impact on the public awareness of 2019 novel coronavirus (COVID-19) outbreak. Infectious Diseases (except HIV/AIDS)

Transmission characteristics of MERS and SARS in the healthcare setting: a comparative study

Complexity of the Basic Reproduction Number (R0)

A rapid advice guideline for the diagnosis and treatment of 2019 novel coronavirus (2019-nCoV) infected pneumonia (standard version)

The basic reproduction number (R 0 ) of measles: a systematic review

Impact assessment of non-pharmaceutical interventions against COVID-19 and influenza in Hong Kong: an observational study. Epidemiology

Effect of nonpharmaceutical interventions for containing the COVID-19 outbreak in China. Infectious Diseases (except HIV/AIDS)

Adoption and impact of non-pharmaceutical interventions for COVID-19

Assessing heterogeneity in meta-analysis: Q statistic or I2 index?

Subtotal (I-squared = 97.8%, p = 0.000) China and overseas Subtotal (I-squared

Subtotal (I-squared

Subtotal (I-squared = 0.0%, p = 0.490)

South Korea Diamond Princes Cruise ship

Subtotal (I-squared = 99.2%, p = 0.000)

Subtotal (I-squared

Italy Subtotal (I-squared = 99.4%, p = 0.000) China Subtotal (I-squared

Subtotal (I-squared

Subtotal (I-squared = 36

GISAID data

Subtotal (I-squared = 99

SEIR method NGMA Majumder & Mandl

Subtotal (I-squared

Subtotal (I-squared = 99.5%, p = 0.000)

Subtotal (I-squared = 98

Subtotal (I-squared = 99.1%, p = 0.000)

Author Ki, 2020 Choi & Ki, 2020 Mizumoto & Chowell, 2020 Liu et al., 2020 Lai et al., 2020 Majumder & Mandl, 2020