Brian Tarran, Significance Magazine Editor

COVID-19, it is fair to say, became a much bigger story much more quickly than we were expecting. At the outset, there was hope the outbreak could be contained, that cases would be limited in number and geographical spread, and that COVID-19 would ultimately prove to be another warning—like SARS and MERS before it—that the world needed to wake up to the very real threat of a future pandemic.

But, as the days and weeks went by and the numbers of cases and affected countries kept growing, it became apparent a global public health crisis was upon us. And the question we found ourselves asking was, “What do we do about it?”

Significance is a statistics magazine, one dedicated to introducing ideas about statistics and statistical thinking to a nonexpert audience, so we knew we had something to contribute. Statistics was both telling the story of the pandemic—by recording the trajectory of the disease—and shaping the story—by informing the epidemiological models decision-makers were using to try to safeguard public health and helping the public understand the various related issues.

With statistics so deeply embedded in this crisis, we thought Significance could and should offer statisticians and data scientists a platform to share their perspectives, analyses, and insights. So, we decided to do something we had not done before: We issued an open call for contributors, asking statisticians and data scientists “to help us explain the statistics of COVID-19.”

This was not an idea arrived at immediately or easily. We actually commissioned our first COVID-19 article in late January, on the day the World Health Organization (WHO) issued its eighth coronavirus “situation report.” At that time, people were still referring to the outbreak informally as the “Wuhan coronavirus.” There were only 4,593 confirmed cases globally—all but 56 of which were in China.

The First Article

Steven E. Rigdon and Ronald D. Fricker Jr. were the authors of an upcoming new book, Monitoring the Health of Populations by Tracking Disease Outbreaks: Saving Humanity from the Next Plague, so we invited them to write about the role of statisticians/epidemiologists at this stage of a new outbreak. They accepted and delivered a first draft a few days later.

The first version of their article concluded on a note of optimism—that maybe the coronavirus would be successfully contained and eradicated. But about six weeks later, as we were getting ready to send the article to print in our April issue, the conclusion was rewritten. “Unfortunately, containment of the virus no longer seems achievable,” they said.

Fricker and Rigdon’s article was sent to print in the same week the WHO declared a pandemic. By the end of that week—March 13—the number of confirmed cases reported by the WHO was more than 132,000, with 51,000 outside China, spread across 122 countries. By March 20, when the article was first published online, confirmed cases had increased to 243,000. By April 1, roughly the date when our magazine would have started mailing, cases had more than tripled: 823,626 people were known to have been infected by the new coronavirus. The tally now stands at more than 30 million.

The huge increase in confirmed cases we saw over the course of the commissioning, writing, editing, and publishing of Fricker and Rigdon’s article was terrifying. It also brought home the realization that a print publication like ours, with long production lead times and only six issues per year, could not hope to keep pace with the pandemic.

If we wanted to say something about COVID-19, to contribute to the conversation, a different approach would be needed.

A Different Approach Takes Shape

The idea for what would become our COVID-19 collection of articles started to take shape on March 16. Millions of people around the world were already in lockdown by that point, and UK citizens would find themselves in a similar situation just over a week later. At the time, it was hard to tear oneself away from the news or social media streams. The temptation was to keep scrolling and scrolling, looking for some morsel of good news among the torrent of bad. There were also so many questions in need of answers. Some of the questions were being asked out loud. Others would silently dominate a person’s thoughts: How dangerous is COVID-19? Will I catch it? Will I recover?

We knew we could not answer all the questions people might have. Indeed, much still remains unknown about COVID-19. However, the Significance editorial board quickly settled on a list of questions we thought statisticians could address, including the following:

• How do we model the spread of a virus?
• How do control measures change our models/predictions?
• What is the “case fatality rate,” and does it give an incomplete picture of a disease?

We decided we would focus on addressing questions related to the processes of disease modeling, data collection, and reporting, rather than trying to explain what was happening at a particular point in time in terms of cases, hospitalizations, and deaths. Developments were moving so quickly, and we had nowhere near the resources of a major media outlet, so it seemed foolish to try to keep up with the work being done by newspapers, websites, radio, and TV.

However, one advantage we knew we had was our network of readers—many of whom are expert statisticians and data scientists and were therefore working on various aspects of COVID-19. It seemed sensible to reach out to that network publicly by issuing an open call for contributors. We hoped this would connect us with people who had not contributed to Significance before, and it did. We also knew it would bring in fresh ideas for coverage—questions we had not thought to ask and topics we had not considered.

The call went out through our website on March 25. Almost immediately, responses came in. The editorial board quickly established a subgroup of members to review all COVID-19 submissions, not only to apply their usual “statistical sanity-check” to articles, but also to ensure consistency and continuity in approach. As much as possible, we wanted articles to link to and build on those previously published. The world’s knowledge of COVID-19 was accumulating over time, and our collection of articles would mirror that growth. We also decided articles would be published online first, with perhaps a selection of edited or updated versions appearing in print. Timeliness is everything in a fast-changing situation.

Articles Address Pressing Questions

Our first batch of online articles was published on April 9, addressing such questions as the following:

• What to make of the coronavirus mortality rate?
• How do epidemiologists know how many people will get COVID-19?
• How many people are infected with COVID-19?

One article also addressed the need for more coronavirus tests, while another described a method for pooling test specimens as a way to deal with a shortage of tests.

Since then, we have published 32 articles. Testing for COVID-19 has been a frequent topic of conversation within articles. Some authors have looked at problems specific to certain countries. For example, Sheila M. Bird, a member of the Significance editorial board and the Royal Statistical Society’s COVID-19 Task Force, has written frequently about the ways in which the UK government could and should improve its reporting of COVID-19 test results. We also had University College London’s Nathan Green explain the differences between tests for active and past infection and how the sensitivity and specificity of tests determine the percentages of true and false positive and true and false negative results.

The reporting of COVID-19 deaths has also animated contributors. Kathryn Leeming wrote about the sometimes week-long delays in England between a person’s death occurring and their death being reported and how this reporting lag was creating a somewhat muddied picture of the progression of the pandemic. Tied to that, Oliver Stoner and Theo Economou discussed ways in which statistical modeling might correct for this reporting lag. Meanwhile, from Colombia, B. Piedad Urdinola outlined the many reasons why mortality may be undercounted during the pandemic and, from the UK, Simon Briscoe compared and contrasted three measures of the COVID-19 death toll.

Contributors have also taken to discussing the ways in which data about the pandemic has been, or could be, visualized, recognizing that visualizations of cases, hospitalizations, and deaths are a regular feature of government briefings to the public and may even help sway politicians when deciding on policy responses.

Pandemic Focus Transitions

Early on in the collection’s existence, there was certainly more of a focus on exploring core concepts that would help readers make sense of the pandemic. Exponential growth, for example, explains how and why COVID-19 became such a big story so quickly, which is why we thought it useful to dedicate an article to this topic. But as spring moved into summer, attention shifted somewhat. Contributors started to reflect more on what we had learned so far about the pandemic, whether from the peer-reviewed literature or the weeks and months of data we had amassed about patients and patient outcomes.

Now, as we transition from summer to autumn—and with winter on the horizon—contributors are again refocusing. Recent (and, so far, unpublished) submissions explore the various risks of “reopening” our societies and economies and what it might take to manage COVID-19 in such a way as to keep things from spiraling out of control once again. Whether that is even possible, though, remains an open question. In the week this article was written, UK Prime Minister Boris Johnson warned that Britain had reached “a perilous turning point,” with cases and hospitalizations once again on the rise and government scientific advisers warning of a possible 50,000 cases per day by October unless transmission of the virus is brought into check.

There is a grim sense of déjà vu. In many ways, it feels like we have returned to that point in mid-March, when we found ourselves worrying and wondering what to do. The difference is, we have a plan of action now. The questions that dominate may have changed, but statisticians are still well-placed to answer many of them. If you would like to be part of this continuing conversation and help “explain the statistics of COVID-19,” find out how to contribute.

All articles are free to read.