Illness duration and symptom profile in symptomatic UK school-aged children tested for SARS-CoV-2
Summary
Background
In children, SARS-CoV-2 infection is usually asymptomatic or causes a mild illness of short duration. Persistent illness has been reported; however, its prevalence and characteristics are unclear. We aimed to determine illness duration and characteristics in symptomatic UK school-aged children tested for SARS-CoV-2 using data from the COVID Symptom Study, one of the largest UK citizen participatory epidemiological studies to date.
Methods
In this prospective cohort study, data from UK school-aged children (age 5–17 years) were reported by an adult proxy. Participants were voluntary, and used a mobile application (app) launched jointly by Zoe Limited and King’s College London. Illness duration and symptom prevalence, duration, and burden were analysed for children testing positive for SARS-CoV-2 for whom illness duration could be determined, and were assessed overall and for younger (age 5–11 years) and older (age 12–17 years) groups. Children with longer than 1 week between symptomatic reports on the app were excluded from analysis. Data from symptomatic children testing negative for SARS-CoV-2, matched 1:1 for age, gender, and week of testing, were also assessed.
Findings
258 790 children aged 5–17 years were reported by an adult proxy between March 24, 2020, and Feb 22, 2021, of whom 75 529 had valid test results for SARS-CoV-2. 1734 children (588 younger and 1146 older children) had a positive SARS-CoV-2 test result and calculable illness duration within the study timeframe (illness onset between Sept 1, 2021, and Jan 24, 2021). The most common symptoms were headache (1079 [62·2%] of 1734 children), and fatigue (954 [55·0%] of 1734 children). Median illness duration was 6 days (IQR 3–11) versus 3 days (2–7) in children testing negative, and was positively associated with age (Spearman’s rank-order rs 0·19, p<0·0001). Median illness duration was longer for older children (7 days, IQR 3–12) than younger children (5 days, 2–9). 77 (4·4%) of 1734 children had illness duration of at least 28 days, more commonly in older than younger children (59 [5·1%] of 1146 older children vs 18 [3·1%] of 588 younger children; p=0·046). The commonest symptoms experienced by these children during the first 4 weeks of illness were fatigue (65 [84·4%] of 77), headache (60 [77·9%] of 77), and anosmia (60 [77·9%] of 77); however, after day 28 the symptom burden was low (median 2 symptoms, IQR 1–4) compared with the first week of illness (median 6 symptoms, 4–8). Only 25 (1·8%) of 1379 children experienced symptoms for at least 56 days. Few children (15 children, 0·9%) in the negatively tested cohort had symptoms for at least 28 days; however, these children experienced greater symptom burden throughout their illness (9 symptoms, IQR 7·7–11·0 vs 8, 6–9) and after day 28 (5 symptoms, IQR 1·5–6·5 vs 2, 1–4) than did children who tested positive for SARS-CoV-2.
Interpretation
Although COVID-19 in children is usually of short duration with low symptom burden, some children with COVID-19 experience prolonged illness duration. Reassuringly, symptom burden in these children did not increase with time, and most recovered by day 56. Some children who tested negative for SARS-CoV-2 also had persistent and burdensome illness. A holistic approach for all children with persistent illness during the pandemic is appropriate.
Funding
Zoe Limited, UK Government Department of Health and Social Care, Wellcome Trust, UK Engineering and Physical Sciences Research Council, UK Research and Innovation London Medical Imaging and Artificial Intelligence Centre for Value Based Healthcare, UK National Institute for Health Research, UK Medical Research Council, British Heart Foundation, and Alzheimer’s Society.
Introduction
- Xie Y
- Wang Z
- Liao H
- Marley G
- Wu D
- Tang W
,
- Sudre C
- Murray B
- Varsavsky T
- et al.
children are often asymptomatic (in 43–68% of cases) or have mild symptoms,
- Li B
- Zhang S
- Zhang R
- Chen X
- Wang Y
- Zhu C
and life-threatening illness and death from COVID-19 are rare.
- Bhopal SS
- Bagaria J
- Olabi B
- Bhopal R
,
- Ludvigsson JF
- Engerström L
- Nordenhäll C
- Larsson E
The pandemic has also resulted in the emergence of a new rare condition, termed multisystem inflammatory syndrome in children (MIS-C),
- Riphagen S
- Gomez X
- Gonzalez-Martinez C
- Wilkinson N
- Theocharis P
presenting at about 2–4 weeks after acute SARS-CoV-2 infection.
- Godfred-Cato S
- Bryant B
- Leung J
- et al.
Evidence before this study
COVID-19 in children is usually asymptomatic or manifests as a mild illness of short duration. Concerns have been raised regarding prolonged illness in children, with no clear resolution of symptoms several weeks after onset, as is observed in some adults. How common this might be in children, the clinical features of such prolonged illness in children, and how this might compare with illnesses from other respiratory viruses (and with general population prevalence of these symptoms) is unclear. A final database search was done in PubMed using the search terms ((“COVID-19” OR “SARS-CoV-2”) AND child*) on June 28, 2021.
Added value of this study
We provide a detailed description of the illness duration and symptom burden for COVID-19 in UK school-aged children (age 5–17 years). Our data, collected in a digital surveillance platform through one of the largest UK citizen science epidemiological studies, show that long illness duration after SARS-CoV-2 infection in school-aged children does occur, but is uncommon. Only a small proportion of children had illness duration beyond 4 weeks, and their symptom burden decreased over time. Almost all children had symptom resolution by 8 weeks, providing reassurance about long-term outcomes. Additionally, the symptom burden in children with what has been termed long COVID-19 was not greater than that in children with long illnesses due to causes other than SARS-CoV-2 infection.
Implications of all the available evidence
Our data confirm that COVID-19 in UK school-aged children is usually of short duration and low symptom burden. Some children do have longer illness duration, and our findings validate their experiences; however, most of these children usually recover with time. Our findings emphasise that appropriate resources will be necessary for any child with prolonged illness, whether due to COVID-19 or other illnesses. Our study provides crucial data to inform discussions about the effect and implications of the pandemic on health-care resource allocation.
- Nalbandian A
- Sehgal K
- Gupta A
- et al.
,
COVID-19 rapid guideline: managing the long-term effects of COVID-19 (NG188). Evidence reviews 2 and 3: prevalence.
Longitudinal data from the King’s College London COVID Symptom Study
- Menni C
- Valdes AM
- Freidin MB
- et al.
showed that 13·3% of adults with a positive test for SARS-CoV-2 had symptoms for at least 4 weeks (referred to as LC28) and 4·5% had symptoms for at least 8 weeks (LC56).
- Sudre C
- Murray B
- Varsavsky T
- et al.
Whether some children also experience prolonged illness duration,
COVID-19 rapid guideline: managing the long-term effects of COVID-19 (NG188). Evidence reviews 2 and 3: prevalence.
,
and if so, how this compares with other illnesses, is unknown.
- Varsavsky T
- Graham MS
- Canas LS
- et al.
,
Coronavirus (COVID-19) in the UK—lab-based testing and capacity, by test type.
Stay-at-home directives and school closures over winter resulted in unusually low circulation of viruses such as influenza and adenovirus in the UK;
Weekly national Influenza and COVID-19 surveillance report—week 11 report (up to week 10 data) March 18, 2021.
however, individuals who contracted other respiratory illnesses were often tested for SARS-CoV-2 due to symptom overlap.
We aimed to investigate illness duration and symptom prevalence, duration, and burden in UK school-aged children (age 5–17 years) testing positive for SARS-CoV-2, and similar data for symptomatic children testing negative.
Results
Coronavirus (COVID-19) in the UK—lab-based testing and capacity, by test type.
analyses were restricted to children with illness onset after Sept 1, 2020. 1734 (588 younger, 1146 older) children were proxy-logged on or before Jan 24, 2021, allowing LC28 to manifest.
- Sudre C
- Murray B
- Varsavsky T
- et al.
Similarly, 1379 (445 younger, 934 older) children had symptoms commencing on or before Dec 29, 2020, allowing LC56 to manifest.
- Sudre C
- Murray B
- Varsavsky T
- et al.
Details regarding the final cohort are presented in the table, and additional demographic details (including ethnicity and index of multiple deprivation decile) are in the appendix (pp 9–10).
Figure 1Study flowchart of inclusion and exclusion criteria
Overall number for the entire cohort of children is given first. Younger children=aged 5–11 years (UK primary school-aged children). Older children=aged 12–17 years (UK secondary school-aged children). Invalid result=PCR test or lateral flow test result proxy-reported as “failed test” or “still waiting”. Duration calculable=illness onset within defined timeframe of testing for SARS-CoV-2, and with defined endpoint. Regular logging=proxy-reporting at least once every 7 days during illness duration. Irregular logging=proxy-reporting with intervals of more than 7 days between proxy-reports during illness duration. Illness onset outside of study bounds=symptom onset before Sept 1, 2020, or after Jan 24, 2021. Presenting to hospital=either admitted to hospital or seen in the emergency department.
TableCharacteristics of school-aged children who tested positive for SARS-CoV-2, and the control cohort of children (matched 1:1 for age, gender, and week of testing) who tested negative for SARS-CoV-2
Data are n (%) or median (IQR). The cohort of children with positive SARS-CoV-2 testing is presented here both as younger and older groups, and for usual (ie, short) versus extended illness duration. Data refers to children with symptom onset between Sept 1, 2020, and Jan 24, 2021. Common paediatric comorbidities such as neurological or neurodisability disorders (eg, cerebral palsy) were not assessed. Presentation to hospital included presenting to the emergency department or admission to hospital.
Figure 2Prevalence of symptoms reported over the course of illness in younger (age 5–11 years, n=588) and older (age 12–17 years, n=1146) children testing positive for SARS-CoV-2
Data refers to children with symptom onset between Sept 1, 2020, and Jan 24, 2021.
Figure 3Median duration of each symptom and IQR in younger (age 5–11 years) and older (age 12–17 years) children
Data refers to children with symptom onset between Sept 1, 2020, and Jan 24, 2021. Black bars represent IQRs.
Figure 4Heat maps showing symptom duration in school-aged children (age 5–17 years) testing positive for SARS-CoV-2 in whom at least one symptom persisted for at least 28 days
n=77. Colour bar provides a percentage comparison. Data refers to children with symptom onset between Sept 1, 2020, and Jan 24, 2021.
- Sudre C
- Murray B
- Varsavsky T
- et al.
The median symptom burden in these children was 6 symptoms (IQR 4–8) during the first week, and 8 symptoms (6–10) over their entire illness. The most common symptoms in these 25 children over their entire illness were anosmia (21 children [84·0%]), headache (20 children [80·0%]), sore throat (20 children [80·0%]), and fatigue (19 children [76·0%]).
Older children were more likely to manifest LC28 than were younger children (59 [5·1%] of 1146 older children vs 18 [3·1%] of 588 younger children; p=0·046). This observation was not evident in children with LC56 (19 [2·0%] of 934 older children vs six [1·3%] of 445 younger children; p=0·52).
Regarding symptom reporting from the additional app questions added on Nov 4, 2020 (with a smaller cohort due to a shorter time frame), the most common symptom was rhinorrhoea (198 [39·1%] of 507 younger children and 439 [47·1%] of 932 older children) and then sneezing (143 [28·2%] of 507 younger children and 269 [28·9%] of 932 older children). These were also the most common symptoms observed in children with LC28 over their entire illness, with 33 (52·4%) of 63 children reporting each of rhinorrhoea and sneezing.
Proxy-reporting was assiduous for all children. Proxy-reporting density was higher in children testing negative for SARS-CoV-2 than for children testing positive (younger, 1·0, IQR 0·7–1·0 vs 0·9, 0·6–1·0; older, 1·0, 0·7–1·0 vs 0·8, 0·6–1·0), noting the shorter illness duration in children testing negative. Proxy-reporting until asymptomatic (ie, perseverance) was higher in children testing negative than in children testing positive (1674 [96·5%] of 1734 vs 1551 [89·4%] of 1734; p<0·0001).
Among children testing positive, symptom logging ceased in 183 children before an asymptomatic report (22 with illness duration ≥28 days, 161 with duration <28 days). In the 77 children with illness duration of 28 days or more, proxy-reporting continued until an asymptomatic report was received in 55 (71·4%). However, the remaining 22 children already had symptoms for at least 28 days and thus fulfil the definition of LC28. In children with illness duration of less than 28 days, an asymptomatic report was received for 90·3% (1496 of 1657). For the remaining 161 children, the median symptom burden at last report was 2 symptoms (IQR 1–3), and proxy-reporting usually ceased early in illness (in 11·3% of children with illness duration of <10 days, and 5·5% of children with illness duration of ≥10 days). Thus, we assumed that proxy-reporting cessation corresponded to illness resolution (ie, the child had recovered) and we calculated illness duration accordingly.
We considered the effect of this assumption. Excluding all children testing positive for whom a healthy report was not logged, the median illness duration in the remaining 1551 children was unchanged (6 days, IQR 3–11 days) with LC28 prevalence of 3·5% (55 of 1551), within the previously calculated confidence interval of 3·5–5·5, using data from the entire cohort. Excluding only the 161 children with assumed short-term symptoms, but including all children with LC28 (regardless of receipt of healthy report), LC28 prevalence was 4·9% (77 of 1573), again within the previously calculated confidence interval of 3·5–5·5.
In children testing negative, logging ceased in 60 children before a healthy report (four of 15 children with long illness duration and 56 of 1719 children with assumed short illness duration).
Discussion
- Sudre C
- Murray B
- Varsavsky T
- et al.
), with low symptom burden. Prolonged illness can occur but is infrequent (4·4% for LC28 and 1·8% for LC56) and lower than for adults (13·3% for LC 28 and 4·5% for LC56).
- Sudre C
- Murray B
- Varsavsky T
- et al.
We found age to correlate with illness duration overall and in children with illness duration of 28 days or more, consistent with our previous findings in adults.
- Sudre C
- Murray B
- Varsavsky T
- et al.
- Li B
- Zhang S
- Zhang R
- Chen X
- Wang Y
- Zhu C
In our cohort, 37·7% reported fever and 25·5% reported a persistent cough. Anosmia and dysosmia were common (39·6%), which was higher than in a small study of adolescents with mild to moderate COVID-19 in which anosmia was reported in 24·1% of participants,
- Kumar L
- Kahlon N
- Jain A
- Kaur J
- Singh M
- Pandey AK
noting that anosmia was a core symptom for testing access during our study period. Relevant to UK testing criteria, anosmia became more prevalent later in illness, at least in children with LC28.
- Sudre C
- Murray B
- Varsavsky T
- et al.
other adult studies have reported persistence of fatigue as 53·1% at 60 days
- Carfì A
- Bernabei R
- Landi F
and 52·3% at 10 weeks.
- Townsend L
- Dyer AH
- Jones K
- et al.
- Abu-Arafeh I
- Razak S
- Sivaraman B
- Graham C
and 66% of children aged 5–15 years reported headaches over the previous year.
Regarding fatigue, a study of 2936 children found that 4·4% had “more than a few days of disabling fatigue”.
- Farmer A
- Fowler T
- Scourfield J
- Thapar A
In one study, prevalence of chronic fatigue syndrome (here defined as “disabling fatigue lasting more than 3 months…with no other cause”) was 1% in children aged 11–16 years;
- Crawley EM
- Emond AM
- Sterne JAC
and in the Avon Longitudinal Study prevalence of “chronic disabling fatigue” (here defined as “fatigue lasting more than 6 months and associated with absence from full-time school or had prevented…activities”) was 1·5% in children aged 13 years and 2·2% in children aged 16 years.
- Norris T
- Collin SM
- Tilling K
- et al.
Considering fatigue after viral infection, median illness duration after Epstein–Barr virus infection in symptomatic university students was 10 days, with a median duration of fatigue of 15·5 days.
- Balfour HH
- Odumade OA
- Schmeling DO
- et al.
- Duarte-Salles T
- Vizcaya D
- Pistillo A
- et al.
However, symptoms were only reported as present or absent at 30 days in both groups, preventing more granular comparisons.
Weekly national Influenza and COVID-19 surveillance report—week 11 report (up to week 10 data) March 18, 2021.
,
Weekly national seasonal respiratory report—week ending March 28, 2021—week 12.
,
Weekly influenza activity in Wales report.
except for the rhinovirus peak commonly observed in September (return-to-school).
Weekly national Influenza and COVID-19 surveillance report—week 11 report (up to week 10 data) March 18, 2021.
With the relaxation of social distancing in the UK, these illnesses might return towards their usual higher prevalences. Our data emphasise that other childhood illnesses might also have protracted burdensome courses, requiring consideration in post-pandemic service planning.
In our cohort, attentional problems, memory complaints, and anxiety were not reported, and cases of low mood and irritability were consistent with previous school-aged population data. Our data do not support anecdotal reports of weakness and seizures as being common in children with COVID-19 of any duration, and no severe neurological symptoms were reported. However, any persistent illness can have adverse mental health outcomes and affect school attendance.
- Hysing M
- Elgen I
- Gillberg C
- Lie SA
- Lundervold AJ
Mental health data could not be proxy-reported, limiting our ability to assess mental health comprehensively in children during the pandemic, and we did not collect data regarding school attendance.
- Varsavsky T
- Graham MS
- Canas LS
- et al.
compared with the general UK population. Furthermore, we cannot characterise regional variability because geographic information was unavailable for many participants. These issues might cause selection bias in our study.
All data related to prevalence of ongoing symptoms following coronavirus (COVID-19) infection in the UK: 1 April 2021.
These figures decreased compared with previous ONS estimates in January, 2021 (eg, from 12·9% to 9·8% in children aged 2–11 years).
All data related to prevalence of ongoing symptoms following coronavirus (COVID-19) infection in the UK: 1 April 2021.
,
Updated estimates of the prevalence of long COVID symptoms.
The ONS reported a control group (never symptomatic, never tested, never self-isolated, and never a contact of anyone testing positive) with baseline pooled symptom rates of 2% in children aged 2–11 years and 1·7% in children aged 12–16 years.
All data related to prevalence of ongoing symptoms following coronavirus (COVID-19) infection in the UK: 1 April 2021.
,
Updated estimates of the prevalence of long COVID symptoms.
There is limited published detail of ONS methodology.
All data related to prevalence of ongoing symptoms following coronavirus (COVID-19) infection in the UK: 1 April 2021.
), which is more consistent with our results. Additionally, ONS estimates are made using current and recalled data collected in the first week of each month, whereas our app-based data are collected in real-time. Our data concord with a small Australian study reporting follow-up of 151 children with SARS-CoV-2 (median age 3 years): 12 had symptoms 3–8 weeks after initial presentation (most commonly cough or fatigue, or both), and all returned to baseline health by 3–6 months.
- Say D
- Crawford N
- McNab S
- Wurzel D
- Steer A
- Tosif S
- Sudre C
- Murray B
- Varsavsky T
- et al.
,
- Drew DA
- Nguyen LH
- Steves CJ
- et al.
Our census points allowed capturing of all children with illness duration of at least 8 weeks who presented before the UK peak positive specimen date. By ensuring that symptoms concorded with testing, we could attribute symptoms to SARS-CoV-2 infection defined by contemporaneous test results. We avoided bias from limited test availability early in the pandemic by restricting analyses from Sept 1, 2020, but we acknowledge there were still some access issues.
- Graham MS
- May A
- Varsavsky T
- et al.
Specifically, to be eligible for PCR testing, individuals were required to have fever, cough, anosmia, or a combination of these symptoms, criteria that were largely informed by adult symptomatology, which might miss some paediatric manifestations of COVID-19 (eg, abdominal pain
- Li B
- Zhang S
- Zhang R
- Chen X
- Wang Y
- Zhu C
, reported in 27·8% of our younger children). Free-text data did not suggest common symptoms unique to children; qualitative analysis was not undertaken given its ad hoc collection and potential bias from additional direct questions after Nov 4, 2020 (ie, once directly asked, a symptom was unlikely to be free-text reported). Additionally, we did not ask specifically about MIS-C. Only 74·5% of children testing positive and 46·4% of children testing negative were reported to have fever, cough, anosmia, or a combination of these symptoms. We do not know why the remaining children were tested. The positive predictive value of any symptom varies according to illness prevalence, and here is clearly subject to the pandemic dynamics.
- Sudre CH
- Keshet A
- Graham MS
- et al.
However, nearly a quarter of symptomatic children testing positive for SARS-CoV-2 during the UK’s second wave did not report any of these symptoms, suggesting that UK testing policy needs reconsideration.
- Menni C
- Valdes AM
- Freidin MB
- et al.
Socioeconomic demographics of our study’s participants, along with UK general population data, are presented in the appendix (pp 9–10); beyond this we cannot comment on the characteristics of proxy-reported versus non-reported children. We did not have information on the relationship of the contributor to the proxy-reported child, which could influence reporting. For example, an unwell contributor might be too ill to proxy-report for a child; however, our high proxy-reporting density and perseverance of all symptomatic children suggest that this was uncommon. Current or previous symptoms experienced by contributors might also influence their proxy-reporting.
Our national study provides the first detailed description of COVID-19 in symptomatic school-aged children. Although uncommon, a small proportion of children have prolonged illness duration and persistent symptoms. Our LC56 data provide reassurance regarding their long-term outcomes. The symptom burden in children testing negative for SARS-CoV-2 but with long illness duration emphasises that allocation of appropriate resources will be necessary for any child with prolonged illness, whether from SARS-CoV-2 infection or other illness. Our study provides timely and crucial data about the effect and implications of the pandemic on UK paediatric health-care resource allocation.
EM, CHS, LSC, and ELD did the analyses. EM, ELD, LSC, CHS, MAb, SSB, and RCH worked on conceptualisation and methodology. BM, EK, LC, JD, MM, and CHS did data extraction and curation. EM, SSB, RCH, AH, MAb, and ELD wrote the manuscript. CH, SS, KR, and JCP developed the data collection system. TDS, SO, and CJS conceived the COVID Symptom Study and obtained funds. MM, SO, CJS, and ELD coordinated this research. BM, EK, LC, JD, MM, EM, CHS, LSC, and ELD verified the data of this study. All authors critically reviewed the manuscript. All authors had access to the COVID Symptom Study dataset, which is also accessible to researchers in the public interest. All authors accept responsibility for the decision to submit for publication.