Quantifying SARS-CoV-2 spread in Switzerland based on genomic sequencing data

Sarah Nadeau, Christiane Beckmann, Ivan Topolsky, Timothy G. Vaughan, Emma B. Hodcroft, Tobias Schär, Ina Nissen, Natascha Santacroce, Elodie Burcklen, Pedro Ferreira, Kim Philipp Jablonski, Susana Posada-Céspedes, Vincenzo Capece, Sophie Seidel, Noemi Santamaria de Souza, Julia M. Martínez-Gómez, Phil F. Cheng, Philipp P. Bosshard, Mitchell P. Levesque, Verena Kufner, Stefan Schmutz, Maryam Zaheri, Michael Huber, Alexandra Trkola, Samuel Cordey, Florian Laubscher, Ana Rita Gonçalves, Karoline Leuzinger, Madlen Stange, Alfredo Mari, Tim Roloff, Helena M. B. Seth-Smith, Hans H. Hirsch, Adrian Egli, Maurice Redondo, Olivier Kobel, Christoph Noppen, Niko Beerenwinkel, Richard A. Neher, Christian Beisel, Tanja StadlermedRxiv 2020 · DOI

Abstract

Pathogen genomes provide critical insights into their evolution and epidemic spread. In this study, we sequenced 1,439 SARS-CoV-2 genomes from Switzerland, representing 3-7% of all confirmed cases per week during the first half of 2020. Using these data, we demonstrate that no single lineage became dominant, suggesting that the virus did not evolve toward lower virulence during this period. On an epidemiological level, we find no evidence of cryptic transmission before the first confirmed case. Our analysis identifies numerous early viral introductions from Germany, France, and Italy, followed by a resurgence of introductions from Germany and France as borders reopened. By quantifying non-traceable infections, the effective reproductive number ($R_e$), and the degree of undersampling, we provide a framework for using genomic data to monitor and respond to infectious disease outbreaks.

Introduction

Standard epidemiological data, such as daily case counts, are essential for tracking an epidemic but are limited in their ability to reveal the origins and movement of a virus within a population. Switzerland, centrally located and highly connected within Europe, faced a complex landscape of viral introductions and local transmission. This study utilizes genomic sequencing to address three critical questions: whether the virus evolved toward lower virulence during the summer of 2020, whether undetected "cryptic" transmission occurred before the first confirmed case, and the extent to which local spread was driven by international travel versus community transmission.

Methodology

The research team utilized phylogenetic analysis to reconstruct the "family tree" of the virus based on genetic mutations.

  • Phylogenetic Trees: A dated maximum-likelihood tree was constructed using 1,439 Swiss sequences integrated with global datasets from GISAID, specifically filtered for genetic similarity to Swiss strains and high-travel regions.
  • Modeling Introductions: To account for incomplete genomic data, the authors employed two boundary assumptions: a "maximum introductions" model and a "minimum introductions" model. This provided a robust range for estimating the influx of new transmission chains.
  • Phylodynamic Modeling: Bayesian methods were used to co-infer the effective reproductive number ($R_e$) and the sampling proportion, allowing for a comparison between genomic estimates and clinical case data.

Key Insights and Results

Stability of Viral Virulence

While the dominant viral clades shifted over time (moving between 20A and 20B), the study found no evidence that any specific lineage evolved toward lower virulence. The observed "summer lull" in hospitalization and mortality rates was likely driven by demographic shifts—specifically, the virus spreading among younger, healthier populations—rather than a fundamental change in the virus's biology.

Border Dynamics and Transmission

The genomic data closely mirrored political developments. Viral introductions plummeted following the mid-March border closures and surged immediately after borders reopened in mid-June. By July, local transmission accounted for over 80% of cases. However, as summer travel peaked in August, the proportion of cases stemming from new introductions rose significantly, accounting for 10% to 60% of new cases depending on the model used.

The "Invisible" Window and Contact Tracing

The study identified a critical ~10-day window between an index case entering the country and being confirmed. During this period, contact tracers are effectively "blind" to the new transmission chain. The authors estimate that at least 25% of new infections during the summer could have been prevented if contact tracing had been able to identify all cases confirmed more than 10 days after the initial introduction.

Early Surveillance Success

Unlike some other European regions, Switzerland showed no evidence of cryptic transmission before its first confirmed case on February 24, 2020. This suggests that early surveillance and testing protocols were effective in detecting the initial arrival of the virus.

Conclusion and Limitations

The study demonstrates that genomic sequencing can accurately estimate the effective reproductive number ($R_e$), with results closely matching clinical data. However, the authors acknowledge that these findings are subject to sampling bias; even at a 3-7% sampling rate, many transmission chains may remain undetected. Furthermore, the geographic origin of introductions is dependent on the sequencing efforts of neighboring countries. Integrating genomic data with granular epidemiological metadata remains a priority for developing real-time public health intervention dashboards.

Summary updated

Abstract

Abstract Pathogen genomes provide insights into their evolution and epidemic spread. We sequenced 1,439 SARS-CoV-2 genomes from Switzerland, representing 3-7% of all confirmed cases per week. Using these data, we demonstrate that no one lineage became dominant, pointing against evolution towards general lower virulence. On an epidemiological level, we report no evidence of cryptic transmission before the first confirmed case. We find many early viral introductions from Germany, France, and Italy and many recent introductions from Germany and France. Over the summer, we quantify the number of non-traceable infections stemming from introductions, quantify the effective reproductive number, and estimate the degree of undersampling. Our framework can be applied to quantify evolution and epidemiology in other locations or for other pathogens based on genomic data. One Sentence Summary We quantify SARS-CoV-2 spread in Switzerland based on genome sequences from our nation-wide sequencing effort.
BibTeX
@misc{nadeau2020quantifying,
  author = {Sarah Nadeau and Christiane Beckmann and Ivan Topolsky and Timothy G. Vaughan and Emma B. Hodcroft and Tobias Schär and Ina Nissen and Natascha Santacroce and Elodie Burcklen and Pedro Ferreira and Kim Philipp Jablonski and Susana Posada-Céspedes and Vincenzo Capece and Sophie Seidel and Noemi Santamaria de Souza and Julia M. Martínez-Gómez and Phil F. Cheng and Philipp P. Bosshard and Mitchell P. Levesque and Verena Kufner and Stefan Schmutz and Maryam Zaheri and Michael Huber and Alexandra Trkola and Samuel Cordey and Florian Laubscher and Ana Rita Gonçalves and Karoline Leuzinger and Madlen Stange and Alfredo Mari and Tim Roloff and Helena M. B. Seth-Smith and Hans H. Hirsch and Adrian Egli and Maurice Redondo and Olivier Kobel and Christoph Noppen and Niko Beerenwinkel and Richard A. Neher and Christian Beisel and Tanja Stadler},
  title = {{Quantifying SARS-CoV-2 spread in Switzerland based on genomic sequencing data}},
  year = {2020},
  doi = {10.1101/2020.10.14.20212621},
  howpublished = {medRxiv},
}
Figure