Understanding the role of children in the spread of the coronavirus disease is a relevant issue as most countries are at this moment elaborating their de-isolation plans, and the re-opening of schools has a central role. The question is relevant because understanding the capacity of infected children to infect others will influence the policy regarding social isolation, which has a profound impact in all societies. This can be a decisive factor to ensure the effectiveness of a country’s response to the pandemic. To that purpose, several recently published studies focus to understand to what extent age should be considered a parameter to predict potential contagiousness of the new coronavirus SARS-CoV-2.

In the article “Is SARS-CoV-2 viral load lower in young children than adults? Jones et al provide evidence that it is (in spite of their claims to the contrary),” McConway and Spiegelhalter disprove the conclusion presented in the preprint article (i.e. not peer-reviewed), “An analysis of SARS-CoV-2 viral load by patient age,” by Jones et al. In this article, the authors aimed to define how contagious children can be by analyzing their viral loads and hence their contagious potential. They conclude that children and adults have similar viral loads which, in theory, make both subgroups potentially equally responsible for the spread of the virus. 

McConway and Spiegelhalter argue that the findings in Jones et al. are based on biased data interpretation, undefined age categories and inappropriate statistical analysis, which lead to an unjustified conclusion. According to McConway and Spiegelhalter, there are three major statistical challenges presented in Jones et al’s study. First, their analysis did not in fact focus on answering the primary question of interest (viral load difference between adults and children). Instead of comparing two well-defined groups specifically, they proceed with a general comparison among all the age groups in both categories (adults and children). 

The second problem is with the variable “age.” According to McConway and Spiegelhalter, “they assumed age as a categorical variable, rather than continuous or ordered.” The appropriate scientific assessment of continuous variables, such as age, requires fitting a regression line of some type to yield statistically meaningful results. By assuming that age is a categorical variable, Jones et al. inadvertently chose the inappropriate statistical approach to evaluate the correlation between age and viral load.

The final problem with the analyis outlined by McConway and Spiegelhalter is that “[Jones et al.] carried out an unnecessary number of multiple comparisons.” Given that the main conclusion of this study focuses on the viral loads in children relative to adults, multiple comparisons appeared redundant and flawed. The authors tested a large number of hypotheses between multiple age-groups categories which may make it “extremely difficult for any particular comparison to stand out.” In other words, using this comparative approach, it is not possible to know for certain whether the lack of difference between children and adults is in fact biological meaningful or a result of flawed methodology. McConway and Spiegelhater conclude that “[e]ssentially their inappropriate statistical analysis meant that, in spite of initially finding a statistically significant difference between subgroups, they made it disappear by doing so many additional and uninteresting comparisons.”

The assumption made by Jones et al’s article has serious implications as they point to the potential risks of unlimited re-opening of schools and kindergartens due to the possibility of children being as infectious as adults. However, McConway and Spiegelhater argue that the lack of scientific rigor and the limited samples used to make these inferences result in an inappropriate analysis and a dangerous generalization. Therefore, McConway and Spiegelhater’s analysis emphasize that the conclusions from Jones et al. should not guide policies, especially when they are meant to underpin public responses for an unprecedented crisis.

Reference:

McConway, K; Spiegelhater, D. (2020)  Is SARS-CoV-2 viral load lower in young children than adults? Jones et al provide evidence that it is (in spite of their claims to the contrary). Medium.  Available in: https://medium.com/@d_spiegel/is-sars-cov-2-viral-load-lower-in-young-children-than-adults-8b4116d28353   Accessed in 30 Mai 2020

 Jones, T. C.; Mühlemann, B. ; Veith, T. ;  Zuchowski, M.; Hofmann, J.; Stein, A.; Edelmann, A.; Max, V. C.; Drosten, C.; (2020)  An analysis of SARS-CoV-2 viral load by patient age in: https://zoonosen.charite.de/fileadmin/user_upload/microsites/m_cc05/virologie-ccm/dateien_upload/Weitere_Dateien/analysis-of-SARS-CoV-2-viral-load-by-patient-age.pdf  Accessed in 30 Mai 2020

Image Source: Shutterstock