The problem with a severely flawed study that has been reported by mainstream media as fact. The end result: people suffering from long COVID are failed, children are failed, parents are failed, and everyone else is misled once again.

First published in August 2021.

Rather disappointed by the reporting of what is a heavily flawed study in the media to suggest that long COVID is rare in children. There are many issues with this study that I am sure long COVID researchers and patients will flag, but here is my analysis.

1. Let’s look at the context of the study. It is proxy reporting for children by parents through the Zoe symptom tracker. The study acknowledges that those using the app are more likely to be white and higher socio-economic status, both associated with reduced risk of poor outcomes.

2. Even among those sampled and found to be PCR positive, only 24% appear to have had information deemed complete enough for analysis. Are those who were deemed to have more complete information logged likely to have been different from everyone using the app? Very possibly.

3. The app relies on parents proxy logging symptoms themselves, with some limited information through direct questions. In fact, the direct question list seems to have been revised in November 2020, but for some reason these were not included in analysis.

Illness duration and symptom profile in symptomatic UK school-aged children tested for SARS-CoV-2. | The Lancet

4. Any long COVID researcher, or person with lived experience looking at the list would immediately see that the list excludes important symptoms that are quite common in those with long COVID, including children – particularly neuro-cognitive symptoms.

Illness duration and symptom profile in symptomatic UK school-aged children tested for SARS-CoV-2. | The Lancet

5. In fact, in a later analysis (not the main analysis) that does look at when broader questions were asked systematically, symptoms like brain fog were found in 6% of younger children and 11% of older children. So, common known symptoms with long COVID were not directly assessed.

This is problematic. Any long COVID researcher who has done surveys, and included an ‘other’ descriptive option for free text knows that symptoms are often underreported unless they are asked for specifically.

6. A key limitation of this study is the lack of understanding of the relapsing and remitting nature of the disease which makes me seriously question whether there was adequate patient involvement, because symptoms were deemed to end if there was a more than 1 week gap between them.

7. There is extensive research now that shows that children (and adults) can have longer asymptomatic periods in the middle, and may even be largely asymptomatic at the start of infection for weeks before later symptoms occur.

8. Even among the few parents that participated in the study, many stopped logging symptoms after some time, including those with prolonged symptoms.

This was taken to mean symptoms had resolved.

Illness duration and symptom profile in symptomatic UK school-aged children tested for SARS-CoV-2. | The Lancet

We know from ONS data and long COVID research that significant number of children have symptoms even lasting for more than a year. ONS estimates 9,000 children with this currently. Would anyone expect parents to keep logging symptoms on an app for long periods of time with a sick child?

Duration was calculated using this anyway, which will very likely not just underestimate the incidence of long COVID but also the duration. In fact, some of this doesn’t even pass basic scrutiny. Neuro-cognitive symptoms have been found the most persistent with long COVID.

Illness duration and symptom profile in symptomatic UK school-aged children tested for SARS-CoV-2. | The Lancet

This study suggests brain fog symptoms lasted for a median of 1-3 days. Once again, I want to stress how relapsing and remitting these symptoms are and any person with lived experience could have really contributed to improving methodology here.

It is no surprise that the study estimates are out of line with the ONS long COVID study – which is nationally representative, asked a wider range of questions, and considered in a better way the relapsing and remitting nature of the disease – also engaging with researchers and patients.

I also want to question the characterisation of “rare”. Even if one believes the results are valid, and 4.4% of cases develop long COVID – let’s extrapolate this just from last week:

If one even considers an average 300 per 100,000 population per week incidence in the under 18s (this rate is over 600 per 100,000 population per week in the 10-19 year olds), that’s 42,000 under 18 cases in one week and around 1,800 cases of long COVID in a week in children.

Is this acceptable? No.

It is vital that long COVID research includes patients, and engages with their real experiences, also to improve the design of the research, so its limitations, and validity can be understood in the context of what we know about this syndrome.

While this study can be deemed an exploratory analysis of data that are available, this has not actually be designed as an adequate survey or evaluation of long COVID – by not assessing symptoms systematically or accounting for the nature of the illness.

Adding to this the high levels of missing data (only 24% participated, and many didn’t have complete information – with assumptions of completeness made that are almost certainly not accurate) and the lack of representativeness, it brings the validity of the results into serious question.

Also, parents with children with long COVID, or even in households with long COVID (unsurprisingly, there are households with long COVID) are very unlikely to be motivated to report even common symptoms (especially when many are not assessed directly) for long periods of time.

Surveys of long COVID that have been patient led, or patient informed, have worked very hard to get past many of these biases, as far as possible, because without these, it is very hard to quantify the incidence of a syndrome that is relapsing and remitting.

At the very least, comparisons with other studies in children would have revealed that many common symptoms weren’t even assessed, which almost certainly led to underreporting here. The authors themselves have highlighted brain fog – which was common but not included.

I honestly really worry about how these flawed studies that, ultimately, end up minimising the real experiences of children and parents impact those who struggle with this day-to-day. They have been gaslit extensively, having to fight for their suffering to just be recognised.

They are suffering from a new disease we don’t understand or have treatment for. The very least we can do is to design research well to capture their reality, working with them to improve their lives, rather than minimise their experiences, and add to their suffering. 


Dr Deepti Gurdasani, Senior Lecturer in Epidemiology, Statistical Genetics, Machine Learning, Queen Mary University of London.


Going Further:

  • Data collection for this study was completed before the Delta variant arrived in the UK.

[This piece was first published as a Twitter thread and turned into the above article on 7 August 2021 with the purpose of reaching a larger audience. It has been minorly edited and corrected, and published with the author’s consent. | The author of the tweets writes in a personal capacity.]

(Cover: Adobe Stock/Marina Andrejchenko. / Licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.)

Creative Commons License