Back

Predicting SARS-CoV-2 infections for children and youth with single symptom screening

Webster, R. J.; Reddy, D.; Harrison, M.-A.; Farion, K. J.; Willmore, J.; Foote, M.; Thampi, N.

2021-08-23 infectious diseases
10.1101/2021.08.19.21262310 medRxiv
Show abstract

Symptom-based SARS-CoV-2 screening and testing decisions in children have important implications on daycare and school exclusion policies. Single symptoms account for a substantial volume of testing and disruption to in-person learning and childcare, yet their predictive value is unclear, given the clinical overlap with other circulating respiratory viruses and non-infectious etiologies. We aimed to determine the relative frequency and predictive value of single symptoms for paediatric SARS-CoV-2 infections from an Ottawa COVID-19 assessment centre from October 2020 through April 2021. Overall, 46.3% (n=10,688) of pediatric encounters were for single symptoms, and 2.7% of these tested positive. The most common presenting single symptoms were rhinorrhea (31.8%), cough (17.4%) and fever (14.0%). Among children with high-risk exposures children in each age group, the following single symptoms had a higher proportion of positive SARS-CoV-2 cases compared to no symptoms; fever and fatigue (0-4 years); fever, cough, headache, and rhinorrhea (5-12 years); fever, loss of taste or smell, headache, rhinorrhea, sore throat, and cough (13-17 years). There was no evidence that the single symptom of either rhinorrhea or cough predicted SARS-CoV-2 infections among 0-4 year olds, despite accounting for a large volume (61.1%) of single symptom presentations in the absence of high-risk exposures. Symptom-based screening needs to be responsive to changes in evidence and local factors, including the expected resurgence of other respiratory viruses following relaxation of social distancing/masking, to reduce infection-related risks in schools and daycare settings.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.