Letter to the Editor regarding the recent contribution by Roussel et al., SARS-CoV-2: Fear versus data

Dear Editor,

The recent article, SARS-CoV-2: Fear versus data by Roussel et al. [1] is one of many publications about SARS-CoV-2 that have received significant attention, including from the public. The Roussel et al. article was specifically cited in a public letter from Sucharit Bhakdi, Emeritus Head of the Institute for Medical Microbiology and Hygiene at the Johannes-Gutenberg-Universität in Mainz, Germany, to the Chancellor of the Federal Republic of Germany.

Good scientific practice calls for prudent investigation of publications that receive such public attention. I read the Roussel et al. article preprint [1] carefully, and several aspects of the publication caused me concern. I believe the following aspects should have been considered and addressed by the authors:

• 1.The highlights section states that “The problem of SARS-CoV-2 is probably overestimated, as 2.6 million people die of respiratory infections each year compared with less than 4000 deaths for SARS-CoV-2 at the time of writing.”

  This statement compares a virus (SARS-CoV-2) that was first described in the scientific literature less than two months before the reference date of the publication data (March 2, 2020) with an annual figure for other diseases (respiratory infections). Furthermore, SARS-CoV-2 is commonly accepted to be in an exponential growth phase, with the projected numbers of cases and deaths associated with this virus expected to rise by several orders of magnitude [2]. There is no evidence to show the numbers presented by the authors in Table 2 are comparable to this situation.

• 2.
  The data displayed in Table 3 are outdated by the time of publication and have changed significantly. This has two implications regarding the validity of the conclusions from the study:

  • 1.
    It raises concerns regarding the comparability between SARS-CoV-2 and the other coronaviruses discussed in the paper.
  • 2.
    The P-values for the Chi-squared test presented by the authors should be recalculated with more recent data. At the time of writing this letter, the case fatality rate in Germany is 0.78% compared with the 0.0% as of March 2 presented by the authors. Rates in Austria (1.00%, previously 0.0%), Belgium (3.97%, previously 0.0%), and Canada (1.07%) have also increased. These changes will greatly affect the P-valueobtained by the authors. As the authors based their conclusions on the non-significant difference between SARS-CoV-2 and other coronaviruses, the conclusions need to be re-evaluated depending on the outcome.
• 3.
  The authors did not address the legitimacy of applying a Chi-squared test. The use of the Chi-squared test in this context is questionable for the following reasons:

  • 1.
    It is not clear how a Chi-squared test is applied in the given setting (what is the null hypothesis?); this needs to be explained in more detail.
  • 2.
    The authors compare a geographical distribution (countries) with a disease distribution (types of coronaviruses).
  • 3.
    Disease prevalence varies significantly between countries, as does prevalence of different coronaviruses. Why did the authors not perform normalization?
    These concerns about the validity of the study results and the timely relevance for public discussion and decision-making on this topic indicate that common scientific standards should be applied to such research to enable sound decision-making in the medical field.

Declaration of Competing Interest

None