Diagnostic accuracy of presenting symptoms in predicting SARS-CoV-2-positivity

ABSTRACT

BackgroundClinical characterisation studies of children with laboratory confirmed severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection have enabled better understanding of paediatric coronavirus disease 2019 (COVID-19). However, to our knowledge, there have been no studies evaluating the predictive value of clinical symptoms for paediatric COVID-19. This gap in evidence is significant as frontline healthcare settings use clinical symptoms to determine infection control policy and patient cohorting measures.Despite studies showing that children have a mild illness, the risk of children with comorbidities and other clinical vulnerabilities contracting COVID-19 remains poorly defined.Objectives• To determine the diagnostic accuracy of presenting clinical symptoms in predicting SARS-CoV-2 positivity.• To assess the co-relation of demographic characteristics, clinical co-morbidities and vulnerabilities with SARS-CoV-2 positivity.MethodsRetrospective single-centre observational study of children with suspected COVID-19 who underwent SARS-CoV-2 reverse transcription polymerase chain reaction (RT-PCR) testing on admission to a specialist children’s hospital in the United Kingdom. Data was collected from electronic patient records from 17/03/2020 to 14/04/2020.Results210 children, aged 0–16 years, median age of 2.5 years (Interquartile range 0.7–7.0) were included. 29 (14%) were SARS-CoV-2-positive. A higher proportion of SARS-CoV-2-positive children were male (62% vs 50%, p=0.24) and at the extremes of age (below 6 months and above 10 years).SARS-CoV-2 positive group had a higher incidence of diarrhoea, vomiting, abdominal pain and seizures whereas cough was seen more frequently in the SARS-CoV-2 negative group. Proportions of children presenting with any respiratory symptoms was similar in both SARS-CoV-2 positive and negative groups. Sensitivity of clinical symptoms individually or in combination in predicting SARS-CoV-2 positivity was 70% or lower. Specificity for predicting SARS-CoV-2 positivity was more than 80% for wheeze, abdominal pain, diarrhea and vomiting, reduced feeds, fever, lethargy, rash and headache.There were significantly more children of Black, Asian and Minority ethnic (BAME) groups in the SARS-CoV-2-positive group (72% [21/29] vs 46% [84/181], p=0.009).Proportions of children with comorbidities in both SARS-CoV-2-positive and negative groups were similar;48% (14/29) vs 54% (97/181) respectively (p=0.59). The commonest comorbidity in this study cohort were respiratory conditions, followed by neurological and oncological conditions.The proportion of children who tested SARS-CoV-2 positive was similar between the clinically vulnerable and non-vulnerable groups. (13.5% [7/52] vs 13.9% [22/158];p=0.85).ConclusionsIn our study cohort, there was no single symptom or cluster of symptoms predictive of a positive SARS-CoV-2 test. The only association that we found with SARS-CoV-2 positivity was belonging to a BAME group. Having comorbidities or being clinically vulnerable did not increase the likelihood of a children being SARS-CoV-2 positive.As presenting symptoms are not accurately predictive of SARS-CoV-2 detection, our study suggests that cohorting children based on clinical symptoms alone, could potentially increase the risk of transmission of infection within healthcare facilities. Until rapid and accurate point-of-care testing is widely available, a shift of emphasis from symptom-based cohorting towards measures such as physical distancing and use of face coverings, will enable better protection.