Never before have Americans relied so heavily on public health figures and politicians to share reliable information about a health threat. Public health officials and epidemiologists have become household names. The public relies on these reputable figures to make science easily digestible, and there is a broad assumption that these officials are speaking from an informed place of expertise. However, many public figures have not done due diligence in vetting studies before making claims about the risks of long COVID. Here we highlight concerns about public messaging on long COVID, specifically in kids, through the lens of the patient community.
Concern: Dissemination of flawed studies about long COVID in kids promotes misleading information about the risks of the disease. The downstream effect of misinformation includes:
- underdiagnosis of long COVID in kids.
- lack of access to treatments.
- obstacles to accommodations in schools.
- limited COVID prevention measures targeting children.
First, we need to define long COVID.
According to The World Health Organization: “Long COVID is an umbrella term for a spectrum of health conditions that occur after a COVID infection. Symptoms of post COVID-19 condition can persist from the initial illness, or begin after recovery. Symptoms may come and go or relapse over time. It is usually diagnosed three months after COVID-19. This allows the healthcare provider to rule out the normal recovery process after illness. The symptoms and effects last for at least two months. We are still learning about COVID-19 and post COVID-19 conditions.” (The authors choose to use the WHO definition of long COVID because of historical behavior of post-infectious illnesses, and because the WHO model can be applied to studies done throughout the world.)
Case Study: Your Local Epidemiologist (Katelyn Jetelina, Phd, MPH) has a large following of parents, professionals, and providers who rely on her easily digestible analysis of COVID. Her Substack has over 400,000 subscribers. In her own words, “The purpose of this newsletter is to provide a direct line of ‘translated’ public health science to the local, national, and international community.” With this in mind, we bring attention to her recent long COVID Series, specifically the article on Long COVID in Kids.
Jetelina cites seven studies with control groups that she believes are “rigorous enough to mention because they compare symptoms among children with a COVID infection to symptoms among children without a COVID infection.” This statement, and her following analysis, is based on a faulty premise:
Long COVID and post-infectious illnesses are syndromes. Syndromes are defined as a group of symptoms that together are characteristic of a specific disease, disorder or the like. This is an important distinction because the symptoms are concurrent, they are not isolated. The interplay, the cascading symptoms, the relationship between them is what illuminates the illness. Long COVID symptoms occur together and should be studied together to derive meaning. Studies of long COVID and other syndromes should consider the following: co-occurring symptoms, temporal (waning and recurring) symptoms, severity and duration of symptoms, and timing and pattern of symptoms.
We will use the studies cited by Jetelina to highlight how the use of control groups ignores the factors that define long COVID. When control groups are prioritized in studying a syndrome over holistic and inclusive factors, data can be skewed. Variables and their relationships must be defined and included or findings will be distorted and unreliable.
Additionally, there are multiple confounding factors in reliance on case/control study designs particular to long COVID in kids. The control group is usually defined as a population with no history of acute covid and negative antibody tests. The following factors make for a high likelihood of contaminated controls:
- Research has shown weak antibody response in cases of long COVID.
- Reliance on COVID antibodies to inform prior infection is flawed because children may serorevert quickly, which is to stop the production of antibodies.
- False negatives on PCR tests are more common in children. PCR tests frequently miss COVID infections in children.
- While some parts of the United States did employee surveillance testing of schools, most of the US did not. Overall, children have been vastly under-tested, due principally to shortage of tests in the US, especially early in the pandemic and during peak COVID surges.
Because a child is more likely to test negative on both PCR and antibody tests, they could be experiencing long COVID but grouped with the controls, making their symptoms classify as “normal.” Using a control group when studying long COVID in kids is assuming the controls have never had COVID, which is impossible to prove or determine. Under the circumstances of this virus and the unreliability of testing, using a control group can arguably result in determining that symptoms that are debilitating and chronic are “normal.”
Comparing long COVID to control groups is a simplistic and naive approach to studying a complicated disease with constantly changing variables: emerging variants, repeat infections, and underlying health conditions. To date there are over 200 documented symptoms associated with long COVID and new information emerging with each new surge.
In contrast to Jetelina’s assertion that control groups equal rigor, the following explains best practices for research on syndromes through the words of researchers and summaries of their perspective:
Co-occurring symptoms: The studies of syndromes must focus on the constellation of symptoms as well as when and how they occur together. Comparing one symptom in a subject to the same symptom in the general population is only informative if the symptom is very specific. For example, if a study on measles isolated the fever and rash as symptoms, the control group would contaminate the results, given that rashes and fevers are quite common in children for a host of reasons.
Dr. Hisham Ziauddeen, MCRPsych PhD, implores researchers to avoid isolating symptoms. “Often the co-occurrence of a particular set of symptoms and signs indicates that a particular kind of dysfunction is occurring in an organ system. Isolating symptoms in subjects to compare with controls dilutes the roles the symptoms may play in the subject. If you compare the least specific symptoms with a poorly selected control group (e.g. people with other ongoing illnesses), your signals will be poor.”
Temporal (waning and recurring) symptoms: Post-infectious syndromes are often characterized by symptoms that come and go. They are unpredictable and can lapse for days or weeks before returning. And in between, different symptoms take their place.
Dr. Deepti Gurdasani, Clinical Epidemiologist, points out on a detailed twitter thread how this occurs in long COVID: “Neurological symptoms with long COVID become more prominent over time, as does fatigue. The escalating and cascading symptoms are what is unique about these symptoms, not the symptom alone without context. The control group studies do not consider or document patterns of symptoms. Symptoms that appear together, or as cascading clusters of symptoms, are what define syndromes. To study a syndrome without this documentation is to catalog symptoms without giving them the appropriate weight.”
Severity of symptoms (compared to patient’s baseline): To understand severity, patient testimony is key to establishing a relevant baseline.
Duration of symptoms: Are they brief, indicating a potential non-COVID cause? Or are they chronic?
Timing of symptoms: To offer insight into long COVID, studies must consider the timeframe of infection, recovery, and onset of persistent symptoms, which cannot be informed by controls. Dr. Gurdasani explains that studies that rely on controls “don’t place importance on the timing of symptoms. If symptoms appeared after a COVID infection that were not present before – and persist the acute infection typically should resolve – this is crucial information to document. The absence of symptoms pre-COVID, and their persistence after COVID is the signal, not the symptom itself.”
Pattern of symptoms: Similar to timing, asking robust questions about patterns surrounding symptoms is critical to proper data analysis. Dr. Gurdasani uses PMS to illustrate the importance of documenting patterns in studies of long COVID. “Premenstrual syndrome is a constellation of symptoms along with timing (near the menstrual cycle). If one measures the symptoms in the general population, symptoms like bloating, constipation, breast tenderness are quite common. The pattern and when it occurs is important. If you did a case control comparison on symptoms within premenstrual syndrome, without considering the time association with menstruation, or combinations of symptoms, you’d come to the conclusion it didn’t exist.” She goes on to point out that prevalence of clinical syndromes have never been studied without pattern and timing factored in.
Having established that control groups are not appropriate for long COVID studies at this time and best practices for the study of syndromes, we examined the studies Jetelina cited, and found additional causes for concern.
Examination of Studies Jetelina Cites:
Illness durations and symptom profile in symptomatic UK school-aged children tested for SARS-COV-2 (The Zoe App Study) This study sought to determine the duration of long COVID in kids using data from the UK COVID Symptoms Study. Parents reported information on kids (ages 5-17) who were symptomatic for COVID via the Zoe App. This study concluded that 1-8% of children had COVID symptoms after 58 days.
Flaws and concerns: The study is conflating acute covid symptoms with long COVID. The World Health Organization defines long COVID in clear terms, in line with how other syndromes are defined “…usually 3 months from the onset of COVID-19 with symptoms and that last for at least 2 months and cannot be explained by an alternative diagnosis.”
In addition to mislabeling acute illness as long COVID, this study contains several additional limitations and flaws that make it unreliable in concluding prevalence or duration of long COVID in kids:
- The study excluded some of the most common long COVID symptoms (brain fog, low mood, post exertional malaise). These were later added to the app in November 2021, but that data was not included in the study’s analysis.
- Symptoms that lapsed for longer than a week were eliminated from the analysis. A common characteristic of long COVID (and most post-infectious syndromes) are intermittent and relapsing symptoms. Eliminating these is eliminating a characteristic of the syndrome. (See section above re: Duration of symptoms)
- When a parent stopped reporting on the app, this was interpreted by the study’s authors as a “recovery.” This is an assumption the authors should not have made. It ignores other likely reasons a parent may quit reporting: illness, forgotten passwords/technical issues, responsibility of caring for sick children, to name a few. The assumption that cessation of reporting correlates to recovery is one of bias.
Others had the same concerns about this study. In fact, a critique of this study was published in the Lancet, authored by thirteen researchers who echoed our concerns listed above.
This study was widely cited by people arguing against COVID protections in schools. The assertion that only “1-8% of children had COVID after 58 days” is especially harmful with regards to formulating appropriate public health guidance.
The Ciao Corona/Zurich Study: Jetelina/Your Local Epidemiologist referenced a research letter that surveyed parents to track students’ symptoms. The students agreed to repeated antibody tests during the survey period. The study concluded that 4% of seropositive positive children reported symptoms lasted longer than 12 weeks.
Flaws and concerns: There are multiple flaws in the study that without context can lead to incorrect interpretation:
- The date of infection for those with positive antibodies was unknown. This could be the difference between recovering from acute COVID infection and lingering symptoms, or chronic symptoms.
- Serology tests are not sensitive to antibodies in kids. This means that students who were put in the “no COVID” control group possibly had COVID at some point and didn’t retain or produce antibodies.
- The survey only included seven symptoms, excluding some of the most common symptoms seen in long COVID and most post-infectious syndromes.
- The survey does not include severity of symptoms. “Tiredness” and “difficulty concentrating” are common complaints among children. The difference in severity and the difference in a student’s pre-COVID baseline is what is relevant, but not asked here.
Mental Health of Adolescents in the Pandemic: Long COVID 19 or Long Pandemic Syndrome? This is a preprint of a study that sought to determine “long COVID infection related symptoms from pandemic related symptoms” via a survey of kids in 14 schools. This study used students who tested negative for COVID antibodies as a control group. Participants completed a survey of common long COVID symptoms. The authors of the study interpreted their findings: “The lack of differences comparing the reported symptoms between seropositive and seronegative students suggests that long-COVID19 might be less common than previously thought and emphasizing the impact of pandemic-associated symptoms regarding the well-being and mental health of young adolescents.”
First, this study is based on a false premise of a valid negative control group. Secondly, the conclusion that a significant portion of long COVID symptoms or syndromes are caused by pandemic induced stress lends itself to the psychologization of a mass-disabling event, rather than attempting to explore the validity of a disabling syndrome.
This is not new in post-infectious studies. This is an ongoing trend that has caused incredible harm to people with post-infectious syndromes. Patients with Lupus, AIDS, MS, ME/CFS, Lyme and stomach ulcers are the most common to fall victim to this kind of dismissal.
Flawed Studies like this lead to parents, doctors, employers, teachers, and insurance providers minimizing or questioning the very existence of disabling chronic disease. This historically has resulted in denial of treatment, loss of healthcare coverage, loss of employment, and denial of accommodations for people with post-infectious illness. The implications of this trend with long COVID are hard to quantify.
Long COVID Symptoms in a Prospective Cohort of Exposed and Infected Children and Adolescents and Their Parents One Year After SARS-CoV-2 Infection This is a preprint article on a study conducted by University Children’s Hospital Germany. The study aimed to examine the impact of the family on persisting COVID symptoms in families one year after infection. A survey of 341 homes with either a COVID infection (confirmed by lab tests) and their “exposed but not infected” household members.”
The study relies upon “exposed but not infected” subjects for the control group. As we’ve already established, negative antibody tests do not guarantee absence of infection. It would be simple enough to conclude that this study is flawed based on that false premise. But that would be a disservice to the post-infectious community that is further harmed by the assumptions the authors lead with.
Jetelina/Your Local Epidemiologist summarizes in her Substack article: “As the authors point out, this relationship (symptoms in “exposed” family members) could be due to a number of things. For example, shared genetics could predispose parents and children to severe disease and long COVID. It could also be due to behavior. For example, parents who experience severe disease may be more aware of pediatric symptoms related to COVID. Parental behavior, like stress, has also been shown to significantly influence child symptoms in a range of other health conditions.” (italics added by authors)
Here we highlight the dangerous and outdated assumptions both within the study’s discussion and the harm in Jetelina’s own words and analysis:
- From the study, “Parents’ perceptions of their own symptoms may have influenced their perception or reporting of their 22 children’s symptoms.” This is a harmful statement that calls parents’ observations into question. It frames their answers as untrustworthy and infantilizes the parent who is chronically ill. The insinuation that a parent with an illness is incapable of evaluating and reporting on their child’s symptoms is based in ableism and abuse of power. This casts parents who are ill/disabled as unreliable witnesses to their childrens’ medical symptoms. This characterization is used to cast doubt on whether disabled/chronically ill people are fit to parent. It’s also an assumption that serves as an obstacle to children getting diagnosed and treated for post infectious illness.
- Researchers suggest it could also be “behavior.” The implications of this – and Jetelina’s use of it – needs to be addressed. Persistent pain and fatigue will cause any person, child or adult, to act differently than they did when symptoms were not present. When it’s labeled as a change in behavior, the implication is that it’s by choice. Statements based purely on conjecture that can not be proven (or disproven, leaving children and parents in an impossible position) are dangerous. Consider the ramifications for childrens’ mental health if their illness is assigned to behavior: in school, in the home. Addressing an illness as a behavioral issue is to hold a child personally responsible for biological, environmental, and social mechanisms beyond their control. It is to inflict untold harm on a sick child, both physically and emotionally.
- “Parental behaviors have been shown to significantly influence symptoms” is a trope that has been disproven many times, but not until great damage is done first. “Refrigerator Moms” were blamed for their child’s autism, accused of being unloving and inattentive in the 1940’s, a characterization and bias that persists to this day, despite being dis-proven. In the 1930’s the concept of the “Schizophregenic Mother” was born. Mothers of schizophrenic children were blamed for being “rejecting” or “too overprotective.” A third theory blamed mothers for imparting their own mental health issues onto their children by way of mothering. This persisted for 40 years. Blaming parents for conditions and illnesses that are hard to define and diagnose is steeped in misogyny and a result of medicine’s aversion to the unknown. These theories are roundly embraced when medical understanding lags. Blaming parents and children, as these selective quotes in Jetalina’s article does, is not an accident of a sloppy study. It is a historical response to an unknown.
History is important here. Framing post-infectious symptoms as behavioral, or anchored in skewed perceptions, has resulted in denial of care for millions. The portrayal of those who experience chronic illness as unreliable narrators who are simply “more aware” of symptoms is purely speculative and biased. One could just as easily see these parents as people with valuable lived experience, and perhaps this lived experience resulted in early identification of their children. The authors made an editorial decision, and Jetalina echoed in her selection of the quote.
In addition to the selection of flawed (and mostly non-peer reviewed) studies, the language used by Jetelina/Your Local Epidemiologist in her article on Long COVID in Kids summarizes the risks in simplistic terms that belie the complicated and evolving reality of long COVID in kids.
For the children and families who are dismissed by doctors, who are confused by symptoms that linger for months or years, this is more than a distorted analysis by a popular pundit. It has a real life impact on diagnosis and treatment and school accommodations. Jetelina’s analysis is entirely based on studies she incorrectly selected as rigorous. The studies highlighted in her Long COVID in Kids article were poorly designed, misleading, and some relied heavily on damaging and distorted parameters.
Our concern extends beyond Jetalina’s popular Substack articles. These same flawed studies have been quoted and cited frequently to argue for less COVID safety protections and to minimize the threat of long COVID, especially in children. The damage of imprecise long COVID public messaging will echo for years to come. What is printed in news outlets, tweeted by public health officials, and quoted in the media will be remembered by parents, doctors, teachers, employers, and insurance providers. It will be absorbed by anyone in a position to determine whether an illness is deserving of treatment, accomodations, and coverage. Narrative creates behavior.
Public communication and analysis of studies needs to be through the lens and with an understanding of post-infectious syndromes. In other words, studies need to be done in consultation with those who have long COVID or other post-infectious syndromes. To do so without input from the community and those who study and treat syndromes is to forgo a baseline understanding of what to look for and how to accurately assess the illness.
Narrative drives behavior. An accurate narrative provides valuable guidance. However, an inaccurate, bad narrative can be impossible to correct and quick to spread by people with unsavory motives. It can result in cautious parents putting their children at risk, and ultimately can result in more children disabled by long COVID.
Public personalities who use their credentials to communicate about the risks of long COVID need to carry the full weight of that responsibility. The consequences of mishandled and misunderstood studies reverberate in ways physical, emotional, and financial.
We, the post-infectious patient population, call on all public health pundits and personas of good faith to show restraint and responsibility when communicating about long COVID in kids. We request that all abide by the principle of “nothing about us without us.”
We’ve lived with the downstream effects of past bad public health guidance. We know the nuances of language about our community, how language can help, and we know how it can destroy. We live with the shrapnel of past sloppy health communication. Our bodies are witnesses to the history. Consult us, include us, know us. Your analysis will be more accurate and the health of the public will be better for it.
Gretchen Kelly, Myalgic Encephalomyelitis/, affiliations: Long Covid Families Board Member, People’s CDC
Megan Carmilani, Myalgic Encephalomyelitis/, affiliations: Founder of Long Covid Families, People’s CDC
Megan Fitzgerald, PhD, Long COVID, Doctor of Neuroscience, affiliations: Long Covid Families, Patient Led Research