To our knowledge, this is the largest study documenting symptoms and duration of ‘long COVID’ in a national cohort comparing RT-PCR-verified SARS-COV-2 infected children to a control group of randomly selected children, who have not been tested positive for SARS-CoV-2.
During the COVID-19 pandemic, governmental interventions have been introduced in order to reduce transmission of SARS-CoV-2. Quarantine regimes, increased hand hygiene, school and day-care lock down, closure of sports and leisure activities and social distancing are among the numerous interventions. Reports have raised concern on the negative impact of these social implications on children’s mental health . Bearing this in mind, it is crucial to compare reported ‘long COVID’ symptoms in previously SARS-CoV-2 positive children to a cohort representative of the background population. Describing ‘long COVID’ in a paediatric population without a control group could otherwise overestimate ‘long COVID’ symptoms.
In the present study, children included in the SARS-CoV-2 group and the control group did not differ in age, gender or pre-existing chronic diseases. SARS-CoV-2 positive children as well as children in the control group reported a high prevalence of symptoms lasting > 4 weeks. These symptoms could potentially originate from SARS-CoV-2 infection, non-SARS-CoV-2 infections or be symptoms reflecting psychological and social consequences of the pandemic. Unfortunately, the present study lacks data on other contemporary viral infections in the children included. However, we would expect an equal distribution of non-SARS-CoV- 2 viral infections in the two groups. Our analysis showed that SARS-CoV-2 positive children presenting with symptoms lasting > 4 weeks reported a higher sense of well-being compared to children with symptoms > 4 weeks who have never been tested positive for SARS-CoV-2. As such, the prevalence of symptoms in the SARS-CoV-2 positive group cannot be assigned to psychological sequelae of social restrictions. At the same time, it can be speculated that the observed lower WHO-5 score and high prevalence of symptoms in the control group could reflect the implications of social restrictions and psychological consequences of the pandemic.
Despite a high prevalence of reported symptoms lasting > 4 weeks in the 6–17-year-old children in the SARS-CoV-2 positive group (28.0%), only a residual percentage difference of 0.8% is seen after subtraction of the corresponding symptom prevalence in the control group (27.2%). It may be speculated that the 0.8% could be a more reliable estimate of the true ‘long COVID’ prevalence in these SARS-CoV-2 positive children. However, it should be kept in mind that this mathematical calculation does not take into accountability difference in type of symptoms reported as well as the burden of symptoms between the two groups. Previous paediatric studies found that 8–58% of SARS-CoV-2 positive children experience ‘long COVID’ symptoms [13, 15,16,17, 21,22,23,24,25,26]. However, none of these studies included a control group, resulting in a risk of overestimating the prevalence of ‘long COVID’. Three studies included a control group and documented a much lower prevalence of ‘long COVID’ symptoms [19, 20, 30]. Symptoms lasting > 4 weeks in controls vs cases were reported to be 1.7% vs 4.6% , 0.9% vs 4.4%  and 53.4% vs 66.5% , respectively. These low prevalences of ‘long COVID’ are comparable to our findings.
In accordance with previous studies [19, 23, 26], the present study documented that the age distribution of symptoms differed with older school children being more frequently affected compared to younger school and pre-school children.
Comparing the responses from SARS-CoV-2 infected children to the control group, we found that the most common statistically significant ‘long COVID’ symptoms were fatigue, loss of smell and loss of taste and to a lesser extend muscle weakness, chest pain, dizziness and respiratory problems. These symptoms are also commonly reported as ‘long COVID’ symptoms in other studies . Especially fatigue and loss of smell and taste  have been reported as frequent symptoms among SARS-CoV-2 positive children. Fatigue has been reported in up to 85% of SARS-CoV-2 positive children with an illness duration of > 4 weeks [14,15,16, 19, 20, 22]. In comparison, median duration of fatigue after Epstein-Barr virus infection is 15.5 days .
Concentration difficulties, headache, muscle and joint pain, cough, nausea, diarrhoea and fever have previously been described as ‘long COVID’ symptoms of SARS-CoV-2 infection in children in non-controlled trials [24, 34]. However, we found that these symptoms were statistically more significant in the control group. Our study also documented that children in the control group had a lower WHO-5score compared to SARS-CoV-2 positive children. Therefore, it should be considered whether concentration difficulties, headache, muscle and joint pain and nausea could be symptoms reflecting the negative impact of the social implications of the pandemic on children’s mental and physical health. The reason for reports of more frequent fever occurrence in a randomly selected control group can only be speculated, since questions on specific method of temperature evaluation or accompanying symptoms were not included. In general, the first year of the pandemic has had a remarkably low incidence of otherwise high incidence infections, such as respiratory syncytial virus and influenza virus [35, 36]. We would therefore have expected a low occurrence of fever.
The duration of ‘long COVID’ symptoms is an important issue to address with implications for the children and families. In our paediatric cohort, most children recovered within a maximum of 1–5 months. Recovery time of 2 weeks–3 months has previously been described in paediatric studies [15, 19, 24].
There are significant limitations to this observational study. Firstly, our questionnaire was not validated in a larger pilot study. Moreover, data relies on participants’ retrospectively parent- or self-reported symptoms with a risk of recall bias. However, SARS-CoV-2 is an infection with increased focus in society. As such, participants have possibly been extremely aware of symptoms during their infection, and therefore, recall bias is assumed to be minimal. However, it can also be speculated that children in the control group, as well as the SARS-CoV-2 group who experienced symptoms, were more eager to respond to the questionnaire than those having mild symptoms or being asymptomatic. This could lead to selection bias and result in an overestimation of reported symptoms. The study may also suffer from non-response bias where non-symptomatic controls may be under-represented. The questionnaire was distributed and had to be answered online, which is likely to select participants with a high socio-economic background, who have a lower risk of poor outcomes following disease.
Secondly, the control group might include children who have had SARS-CoV-2 infection without having undergone testing. This is a risk since children have no or only few symptoms of acute COVID-19. It has previously been reported that 25–36% of SARS-CoV-2 positive children are asymptomatic [24, 37]. Moreover, parents might be more reluctant to let children undergo testing due to the invasive and unpleasant procedure. Children in the control group have never been tested positive for SARS-CoV-2. Nevertheless, seroprevalence data suggest an infection rate of 2–3 times higher than corresponding PCR results in the paediatric population . In March 2021, a Danish nationwide seroprevalence surveillance study of 530 randomly selected children 12–18 years old documented that 8.1% had positive SARS-CoV-2 IgG (unpublished national data). It could be assumed that children below 12 years also account for an 8.1% seroprevalence. The potential presence of children with previous SARS-CoV-2 infection in the control group would underestimate the symptoms of ‘long COVID’.
Thirdly, the number of children answering the questionnaire was higher among SARS-CoV-2 positive children than in the control group (44.9% versus 21.3%). The higher response rate in the group of children with previous SARS-CoV-2 infection could be expected since these children and families might have higher awareness on potential ‘long COVID’ symptoms, and a potential increased desire for knowledge about post-acute sequelae of SARS-CoV-2. This issue might result in an overestimation of ‘long COVID’. However, in both groups, the children who answered the questionnaire were an equal age distribution of the children who received the questionnaire.
The strengths of this study are primarily the large sample size and inclusion of a control group. Additionally, the Danish Health Data Authority has a nationwide coverage, and all Danish citizens have universal tax-funded health insurance. Therefore, our study population did not rely on access to health-care services, and RT-PCR tests for SARS-CoV-2 are free of charge in Denmark minimizing selection bias. The infection status of all SARS-CoV-2 positive children has been established by nationally validated RT-PCR, eliminating major misclassification of infection status. Moreover, the results of the SARS-CoV-2 test are linked to the unique personal identification number registered in the Danish Civil Population Register  and the national microbiology database making it a complete cohort of SARS-CoV-2 infected children below 18 years from the start of the pandemic in Denmark until 19th of March 2021.
The control group is not nationwide but consists of randomly selected children aged 0–17 years from five municipalities that include both larger cities as well as rural areas spread widely across Denmark. As such, the control group is considered to represent the general Danish population of children. The strengths mentioned above increase the generalizability of our results. However, it is important to consider that an observational study based on a questionnaire with subjective responses and without objective examination of the children can only provide us with one piece of the puzzle. Further studies are needed to increase the knowledge of ‘long COVID’ in the paediatric population.
In conclusion, to date, this study is the largest study on symptoms and duration of ‘long COVID’ in SARS-CoV-2 positive children also including a control group. It provides new evidence of ‘long COVID’ in children, documenting that ‘long COVID’ is primarily seen in older school children. Despite the high prevalence (12–51%) of reported long-lasting symptoms in the SARS-CoV-2 children, the true prevalence of ‘long COVID’ seems a lot lower, maybe as low as 0.8%. The most common ‘long COVID’ symptoms are fatigue, loss of smell and loss of taste, dizziness, muscle weakness, chest pain and respiratory problems. These symptoms cannot be assigned to psychological sequelae of social restrictions. Symptoms such as concentration difficulties, headache, muscle and joint pain and nausea may be related to other factors than SARS-CoV-2 infection. In most cases, ‘long COVID’ symptoms resolve within 1–5 months.