Our online courses (all are free and have no ads):

Other useful resources

fast.ai in the news:

Doing Data Science for Social Good, Responsibly

The phrase “data science for social good” is a broad umbrella, ambiguously defined. As many others have pointed out, the term often fails to specify good for whom. Data science for social good can be used to refer to: nonprofits increasing their impact through more effective data use, hollow corporate PR efforts from big tech, well-intentioned projects that inadvertently result in surveillance and privacy invasion of marginalized groups, efforts seeped in colonialism, or many other types of projects. Note that none of the categories in the previous list are mutually exclusive, and one project may fit several of these descriptors.

Picture from a presentation given in 2018 by Sara Hooker, founder of non-profit Delta Analytics and an AI researcher at Google, on Why “data for good” lacks precision.

I have been involved with data science for social good efforts for several years: chairing the Data for Good track at the USF Data Institute Conference in 2017; coordinating and mentoring graduate students in internships with nonprofits Human Rights Data Analysis Group (for a project on entity resolution to obtain more accurate casualty conflicts in Syria and Sri Lanka) and the American Civil Liberties Union (one student analyzed covid vaccine equity in California and another analyzed disparities in school disciplinary action against Black and disabled students) during my time as director of the Center for Applied Data Ethics at USF; and now as a co-lead of the Data Science for Social Good program at Queensland University of Technology (QUT). At QUT, grad students and recent graduates partnered with non-profits Cancer Council Queensland (well known for their Australian Cancer Atlas) and FareShare food rescue organisation, which operates Australia’s largest charity kitchens. While data for good projects can be incredibly useful, there are also pitfalls to be mindful of when approaching data for social good.

Some Questions & Answers

I recently spoke on a panel at the QUT Data Science for Social Good showcase event. I appreciated the thoughtful, nuanced questions from the moderators, Dr. Timothy Graham and Dr. Char-lee Moyle, who brought up some of the potential risks. I want to share their questions below, along with an expanded version of my answers.

What ethical and governance considerations do you think not-for-profits should consider when starting to adopt data science?

  1. Be specific about the goals of the project and how different stakeholders will be impacted: A series of interviews with African data experts revealed that power imbalances, failure to acknowledge extractive practices, failure to build trust, and Western-centric policies were all prevalent. Even in “data for good” projects, the people whose data is accessed and shared may not reap the benefits that those who control the project do. Stakeholders such as government bodies and non-profits have significantly more power and leverage compared to data subjects. There are issues where data gathered for one goal ends up being repurposed or shared for other uses. While Western “notions of privacy often focus on the individual, there is growing awareness that collective identity is also important within many African communities, and that sharing aggregate information about communities can also be regarded as a privacy violation.”
  2. Center the problem to be solved, not a flashy solution. Sometimes machine learning practitioners have a solution searching for a problem. It is important to stay focused on the root problem and be open to “mundane” or even non-technical solutions. One data for good project used the records of 15 million mobile phone owners in Kenya to quantify the movements of workers who migrate for seasonal work to an area with malaria, and made recommendations to increase malaria surveillance in their hometowns when they return. As a journalist for Nature reported, “But it’s unclear whether the results were needed, or useful. Malaria-control officers haven’t incorporated the analyses into their efforts.” The excitement around “flashy” big data approaches contrasts with the lack of funding for proven measures like bed nets, insecticides, treatment drugs, and health workers.
  3. Take data privacy seriously. Be clear about how the data will be stored, who has access to it, and what will happen to it later. Ask what data is truly needed, and if there are less invasive ways to get this information. Note that the above example tracking Kenyan mobile phone owners raises risks around lack of consent, invasion of privacy, and risk of de-anonymization.
  4. Include the people most impacted, and recognize that their values may be different from those of both non-profits or academic stakeholders involved. A recent article from AI Now Institute recommended that “social good projects should be developed at a small scale for local contexts — they should be designed in consultation with the community or social environment impacted by the systems in order to identify core values and needs.” One example of differing values: Indigenous scholars highlighted that a set of open data principles developed primarily by Western scholars to improve data discovery and reuse created tension with Indigenous values. The FAIR principles, first developed at a workshop in the Netherlands in 2014 and elaborated on in this paper published in Nature, call for data to be findable, accessible, interoperable, and reusable. In response, Indigenous scholars convened to develop the CARE principles for Indigenous data governance, calling for collective benefit, authority to control, responsibility, and ethics, intended as a complement to the FAIR principles.
  5. Avoid answering the “wrong problem.” For instance, many European governments are currently using algorithmic approaches to justify austerity cuts. Arguments about reducing fraud often accompany these, even when fraud is minimal. Due to a biometric identity system in India, many poor and elderly people are no longer able to access their food benefits due to faded fingerprints, not being able to travel to scanners, or intermittent internet connections.

Do you think that data science for social good can increase the surveillance and control of disadvantaged groups or certain segments of society?

Many well-meaning projects inadvertently lead to increased surveillance, despite good intentions. Cell-phone data from millions of phone owners in over two dozen low- and middle- income countries has been anonymized and analyzed in the wake of humanitarian disasters. This data raises concern of the lack of consent of the phone users and risks of de-anonymization. Furthermore, it is often questionable whether the results are truly useful, as well as if they could have been discovered through other, less invasive approaches. One such project analyzed the cell phone data of people in Sierra Leone during an Ebola outbreak. However, this approach didn’t address how Ebola spreads (only through direct contact with body fluids) or help with the most urgent issue (which was convincing symptomatic people to come to clinics to isolate).

What do you think is the role of government and universities in supporting and incentivising the not-for-profit sector in adopting data science?

Academia and government have a big role to play. Often non-profits lack the in-house data science skill to take advantage of their data, and many data scientists who are searching for meaningful and impactful real-world problems to work on. We will also need the government to regulate topics such as data privacy to help protect those who may be impacted. It is important to recognize that privacy should not just be considered an individual right, but also a public good.

What are your thoughts around the development of ethical frameworks to guide data science – are they more than marketing tactics to increase trustworthiness and reputation of data science?

We need ethical frameworks AND regulation. Both are crucially important. Many people want to do the right thing, and having standardized processes to guide them can help. I recommend the Markkula Center Tech Ethics Toolkit, which includes practical processes you can implement in your organization to try to identify ethical risks BEFORE they cause harm. At the same time, we need legal protections anywhere that data science impacts human rights and civil rights. Meaningful consequences are needed for those who cause harm to others. Also, policy is the appropriate tool to address negative externalities, such as when corporations offset their costs and harms to society while reaping the profits for themselves. Otherwise, there will always be a race to the bottom.

What skills and training do you think the not-for-profits sector needs to embrace data science and what’s the best strategies for upskilling?

The people who are already working for an organization are best positioned to understand that organization’s problems and challenges, and where data science can help. Upskilling in-house talent is underutilized. Don’t feel that you need to hire someone new with a fancy pedigree, if there are people at your organization who are interested and eager to learn. I would start by learning to code in Python. Have a project from your not-for-profit that you are working on as you go, and let that project motivate you to learn what you need as you need to (rather than feeling like you need to spend years studying before you can tackle the problems you care about). One of our core missions with fast.ai is to train people in different domains to use machine learning for themselves, as they best understand the problems in their domain and what is needed. There are many myths that you need a super-elite background to use techniques like deep learning, but it’s not magic. Anyone with a year of coding experience can learn to use state-of-the-art deep learning.

Further Reading/Watching

Here are some additional articles (and one video) that I recommend to learn more on this topic:

Avoiding Data Disasters

Things can go disastrously wrong in data science and machine learning projects when we undervalue data work, use data in contexts that it wasn’t gathered for, or ignore the crucial role that humans play in the data science pipeline. A new multi-university centre focused on Information Resilience, funded by the Australian government’s top scientific funding body (ARC), has recently launched. Information Resilience is the capacity to detect and respond to failures and risks across the information chain in which data is sourced, shared, transformed, analysed, and consumed. I’m honored to be a member of the strategy board, and I have been thinking about what information resilience means with respect to data practices. Through a series of case studies and relevant research papers, I will highlight these risks and point towards more resilient practices.

Case study: UK covid tracking app

Data from a covid-symptom tracking app was used in a research paper to draw wildly inaccurate conclusions about the prevalence of Long Covid, the often debilitating neurological, vascular, and immune disease that can last for months or longer (some patients have been sick for 20 months and counting). The app suggested that only 1.5% of patients still experience symptoms after 3 months, an order of magnitude smaller than estimates of 10-35% being found by other studies.

How could this research project have gone so wrong? Well, the app had been designed for a completely different purpose (tracking 1-2 week long respiratory infections), didn’t include the most common Long Covid symptoms (such as neurological dysfunction), had a frustrating user-interface that led many patients to quit using it, and made the erroneous assumption that those who stopped logging must be fully recovered. The results from this faulty research paper were widely shared, including in a BBC article, offering false reassurance than Long Covid prevalence is much rarer than it is. Patients had been voicing their frustrations with the app all along, and if researchers had listened sooner, they could have collected a much higher quality and more accurate data set.

This research failure illustrates a few common issues in data projects:

  • The context of the data was not taken into account. The user-interface, the categories listed, the included features– these were all designed to record data about a short-term mild respiratory infection. However, when it was used for a different purpose (long covid patients suffering for months with vascular and neurological symptoms), it did a poor job, and led to missing and incomplete data. This happens all too often, in which data gathered for one context is used for another
  • The people most impacted (long covid patients) were ignored. They had the most accurate expertise on what long covid actually entailed, yet were not listened to. Ignoring this expertise led to lower quality data and erroneous research conclusions. Patients have crucial domain expertise, which is distinct from that of doctors, and must be included in medical data science projects. From the start of the pandemic, patients who had suffered from other debilitating post-viral illnesses warned that we should be on the lookout for long-term illness, even in initially “mild” cases.

Data is Crucial

Collecting data about covid and its long-term effects directly from patients was a good idea, but poorly executed in this case. Due to privacy and surveillance risks, I frequently remind people not to record data that they don’t need. However, the pandemic has been a good reminder of how much data we really do need, and how tough it is when it’s missing.

At the start of the pandemic in the United States, we had very little data about what was happening– the government was not tabulating information on cases, testing, or hospitalization. How could we know how to react when we didn’t understand how many cases there were, what death rates were, how transmissible the disease was, and other crucial information? How could we make policy decisions in the absence of a basic understanding of the facts.

In early March 2020, two journalists and a data scientist from a medication-discovery platform began pulling covid data together into a spreadsheet to understand the situation in the USA. This launched into a 15-month long project in which 500 volunteers compiled and published data on COVID-19 testing, cases, hospitalizations, and deaths in the USA. During those 15 months, the Covid Tracking Project was the most comprehensive source of covid data in the USA, even more comprehensive than what the CDC had, and it was used by the CDC, numerous government agencies, and both the Trump and Biden Administrations. It was cited in academic studies and in thousands of news articles.

A data infrastructure engineer and contributor for the project later recounted, “It quickly became apparent that daily, close contact with the data was necessary to understand what states were reporting. States frequently changed how, what, and where they reported data. Had we set up a fully automated data capture system in March 2020, it would have failed within days.” The project used automation as a way to support and supplement manual work, not to replace it. At numerous points, errors in state reporting mechanisms were caught by eagle-eyed data scientists notifying discrepancies.

This vision of using automation to support human work resonates with our interest at fast.ai in “augmentedML”, not “autoML”. I have written previously and gave an AutoML workshop keynote on how too often automation ignores the important role of human input. Rather than try to automate everything (which often fails), we should focus on how humans and machines can best work together to take advantage of their different strengths.

Speaking about AugmentedML vs. AutoML at ICML 2019
Speaking about AugmentedML vs. AutoML at ICML 2019

Data Work is Undervalued

Interviews of 53 AI practitioners across 6 countries on 3 continents found a pattern that is very familiar to many of us (including me) who work in machine learning: “Everyone wants to do the model work, not the data work.” Missing meta-data leads to faulty assumptions. Data collection practices often conflict with the workflows of on-the-ground partners, such as nurses or farmers, who are usually not compensated for this extraneous effort. Too often data work is arduous, invisible, and taken for granted. Undervaluing of data work leads to poor practices and often results in negative, downstream events, including dangerously inaccurate models and months of lost work.

Throughout the pandemic, data about covid (both initial cases and long covid) has often been lacking. Many countries have experienced testing shortages, leading to undercounts of how many people have covid. The CDC decision not to track breakthrough cases unless they resulted in hospitalization made it harder to understand prevalence of break-throughs (a particularly concerning decision since break-throughs can still lead to long covid). In September, it was revealed that British Columbia, Canada was not including covid patients in their ICU counts once the patients were no longer infectious, a secretive decision that obscured how full ICUs were. Some studies of Long Covid have failed to include common symptoms, such as neurological ones, making it harder to understand the prevalence or nature.

Data has Context

Covid is giving us a first-hand view of how data, which we may sometimes want to think of as “objective”, are shaped by countless human decisions and factors. In the example of the symptom tracking app, decisions about which symptoms were included had a significant impact on the prevalence rate calculated. Design decisions that influenced the ease of use impacted how much data was gathered. Lack of understanding of how the app was being used (and why people quit using it) led to erroneous decisions about which cases should be considered “recovered”. These are all examples of the context for data. Here, the data gathered was reasonably appropriate for understanding initial covid infections (a week or two of respiratory symptoms), but not for patients experiencing months of neurological and vascular symptoms. Numbers can not stand alone, we need to understand how they were measured, who was included and excluded, relevant design decisions, under what situations a dataset is appropriate to use vs. not.

As another example, consider covid testing counts: Who has access to testing (this involves health inequities, due to race or urban vs. rural), who is encouraged to get tested (at various times, people without symptoms, children, or other groups have been discouraged from doing so), varying accuracies (e.g. PCR tests are less accurate on children, missing almost half of cases that later go on to seroconvert), and making decisions about what counts as a “case” (I know multiple people who had alternating test results: positive, negative, positive, or the reverse– what counts as a positive case?)

Datasheet for an electrical component. Image from 'Datasheets for Datasets'
Datasheet for an electrical component. Image from 'Datasheets for Datasets'

One proposal for capturing this context is Datasheets for Datasets. Prior to doing her PhD at Stanford in computer vision and then co-leading Google’s AI ethics team, Dr. Timnit Gebru worked at Apple in circuit design and electrical engineering. In electronics, each component (such as a circuit or transistor) comes with a datasheet that lists when and where it was manufactured, under what conditions it is safe to use, and other specifications. Dr. Gebru drew on this background to propose a similar idea for datasets: listing the context of when and how it was created, what data was included/excluded, recommended uses, potential biases and ethical risks, work needed to maintain it, and so on. This is a valuable proposal towards making the context of data more explicit.

The People Most Impacted

The inaccurate research and incomplete data from the covid tracking app could have been avoided by drawing on the expertise of patients. Higher quality data could have been collected sooner and more thoroughly, if patients were consulted in the app design and in the related research studies. Participatory approaches to machine learning is an exciting and growing area of research. In any domain, the people who would be most impacted by errors or mistakes need to be included as partners in the design of the project.

The Diverse Voices project from University of Washington Tech Policy Lab involves academic papers and practical how-to guides.
The Diverse Voices project from University of Washington Tech Policy Lab involves academic papers and practical how-to guides.

Often, our approaches to addressing fairness or other ethics issues, further centralize the power of system designers and operators. The organizers of an ICML workshop on the topic called for more cooperative, democratic, and participatory approaches instead. We need to think not just about explainability, but about giving people actionable recourse. As Professor Berk Ustun highlights, when someone asks why their loan was denied, usually what they want is not just an explanation but to know what they could change in order to get a loan. We need to design systems with contestability in mind, to include from the start the idea that people should be able to challenge system outputs. We need to include expert panels of perspectives that are often overlooked, depending on the application, this could mean formerly or currently incarcerated people, people who don’t drive, people with very low incomes, disabled people, and many others. The Diverse Voices project from University of Washington Tech Lab provides guidance on how to do this. And it is crucial that this not just be tokenistic participation-washing, but a meaningful, appropriately compensated, and ongoing role in their design and operation.

Towards Greater Data Resilience

I hope that we can improve data resilience through:

  • Valuing data work
  • Documenting context of data
  • Close contact with the data
  • Meaningful, ongoing, and compensated involvement of the people impacted

And I hope that when our data represents people we can remember the human side. As AI researcher Inioluwa Deborah Raji wrote, “Data are not bricks to be stacked, oil to be drilled, gold to be mined, opportunities to be harvested. Data are humans to be seen, maybe loved, hopefully taken care of.”

Quote from AI researcher Inioluwa Deborah Raji
Quote from AI researcher Inioluwa Deborah Raji

SARS-CoV-2 Spike Protein Impairment of Endothelial Function Does Not Impact Vaccine Safety

My colleague Dr Uri Manor was a senior author on a study in March this year which has become the most discussed paper in the history of Circulation Research and is in the top 0.005% of discussed papers across all topics. That’s because it got widely picked up by anti-vaxx groups that totally misunderstood what it says. Uri and I decided to set the record straight, and we wrote a paper that explains that “SARS-CoV-2 Spike Protein Impairment of Endothelial Function Does Not Impact Vaccine Safety”. Unfortunately peer review has taken months, so it’s still not published. Therefore, we’ve decided to make the paper available prior to review below (as HTML) and here (as PDF).



Lei et al. [2021] showed the spike protein in SARS-CoV-2 alone was enough to cause damage to lung vascular endothelium. The authors noted that their results suggest that “vaccination-generated antibody and/or exogenous antibody against S protein not only protects the host from SARS-CoV-2 infectivity but also inhibits S protein-imposed endothelial injury”. We show that there is no known mechanism by which the spike protein impairment of endothelial function could reduce vaccine safety, and that vaccine safety data clearly shows that the spike proteins in vaccines does not reduce vaccine safety. Overall, we conclude that spike proteins encoded by vaccines are not harmful and may be beneficial to vaccine recipients.


COVID-19 has been widely understood to be a respiratory lung disease. However, there is now a growing consensus that SARS-CoV-2 also attacks the vascular system [Potus et al., 2020, Ackermann et al., 2020, Siddiqi et al., 2020, Teuwen et al., 2020]. Earlier studies of other coronaviruses have suggested that their spike proteins contributed to damaging vascular endothelial cells [Kuba et al., 2005].

Lei et al. [2021] created a pseudovirus surrounded by a SARS-CoV-2 crown of spike (S) proteins, but did not contain any actual virus, and found that exposure to this pseudovirus resulted in damage to the lungs and arteries of an animal model. They concluded that “S protein alone can damage vascular endothelial cells (ECs) by downregulating ACE2 and consequently inhibiting mitochondrial function”.

Lei et al. [2021] noted that their conclusions suggest that vaccine-induced antibodies “not only protects the host from SARS-CoV-2 infectivity but also inhibits S protein-imposed endothelial injury”. However, they did not tackle the question of whether the findings of EC damage from S protein might also have an unintended negative side effect of reducing vaccine safety.

Vaccine safety has become an important issue due to Vaccine-induced Immune Thrombotic Thrombocytopenia (VITT), also known as Vaccine-induced Immune Thrombocytopenia and Thrombosis, which has resulted in cases in recipients of the Oxford/AstraZeneca (AZ) and Johnson & Johnson (JJ) vaccines [Makris et al., 2021]. VITT refers to a rare combination of thrombosis (usually CVST) and thrombocytopenia which have been found in some patients 4 to 30 days after they receive their first AZ or JJ vaccine dose (and occasionally after their second dose).

Regulators have found that clots are extremely rare, and that the benefits of the vaccines outweigh the risks. However, the roll-out of the AZ and JJ vaccines have been restricted in many jurisdictions [Mahase, 2021]. In the UK, for instance, the Joint Committee on Vaccination and Immunisation (JCVI) recommend avoiding the AZ vaccine for those under 40 years old, based on “reports of blood clotting cases in people who also had low levels of platelets in the UK, following the use of Oxford/AstraZeneca vaccine.” [Public Health England, 2021]

With an Altmetric Attention Score of 3726 (as of May 23rd, 2021), Lei et al. [2021] has become the most discussed paper in the history of Circulation Research and is in the top 0.005% of discussed papers across all topics. By reading a random sample of social media posts that link to the paper, we found that the great majority of readers express a view that the paper shows that the vaccine is not safe, and that therefore people should not get vaccinated. This view has also been widely shared in blog posts, such as Adams [2021], which states, “Bombshell Salk Institute science paper reveals the covid spike protein is what’s causing deadly blood clots and it’s in all the covid vaccines (by design)” and concludes “The vaccines literally inject people with the very substance that kills them. This isn’t medicine; its medical violence against humanity”. Furthermore, some doctors are now publicly expressing concerns about vaccine safety, based on concerns about the impact of spike proteins. [Bruno et al., 2021]

Because Lei et al. [2021] did not explicitly discuss the relevance of its findings to vaccine safety, and because it has been widely cited as showing that vaccines are not safe, including by some doctors, we will examine whether its findings should result in pausing or stopping the vaccine rollouts.

Analysis of Current Data

To ascertain whether spike protein impairment of endothelial function reduces vaccine safety we can directly observe the results of vaccine use.

Overall Vaccine Safety

The vaccine with the most recorded cases of VITT is the AZ vaccine. The largest roll-out of the AZ vaccine is in England. The roll-out began in December 2020, and by the start of February 2021 over 10 million people had received at least one dose. By mid-April 2021, over 10 million people had received their second dose.

Public Health England publishes data on “Excess mortality in England”. This data shows that from March 20, 2020, until February 19, 2021, there were 101,486 excess deaths in England. From February 20, 2021 (two months after the start of England’s vaccine roll-out), until April 30, 2021 (the latest data available at writing), there have been no excess deaths in England.

As of May 5, 2021, there were 262 reported cases of VITT in the UK after the first dose of the vaccine, resulting in 51 deaths, and eight cases have been reported after a second dose [Medicines & Healthcare products Regulatory Agency, 2021]. 35 million people had received their first vaccination by this time. This is over half the population of the UK, and nearly all adults over 30 years old. Children and young adults in the UK will not be receiving the AZ vaccine, based on current guidelines.

Overall, with 51 deaths due to VITT, compared to 101,486 probably due to COVID-19, we can see that the overall impact of the vaccine is to greatly reduce deaths. Even if the spike proteins in vaccines resulted in reduced endothelial function (which, as we shall see shortly, they do not), the impact would clearly not be significant enough to result in the need to reduce or stop vaccine rollouts.

Comparative Analysis

All currently approved SARS-CoV-2 vaccines incorporate spike proteins. If the spike proteins in vaccines resulted in significantly reduced endothelial function, causing VITT, then we would expect to see reports of VITT in recipients of all the available vaccines. However, this is not the case. There are no reports of VITT in recipients of the Moderna or Pfizer vaccines.

It is unlikely that this is due to failure to identify VITT, since the particular combination of thrombosis and thrombocytopenia is very rare, and the issues around vaccine safety widely reported and discussed.

Furthermore, it is statistically unlikely. As of May 15, 2021, in the USA 156.2 million people had received at least one dose of SARS-CoV-2 vaccine, the vast majority of which were Pfizer and Moderna. Since each recipient’s vaccine VITT response is an independent binary event, we can model it with a binomial distribution. The UK VITT death rate is 0.0001%. If the spike proteins were the cause of VITT, we would expect the same death rate in the US, which would result in 183-273 deaths (99% confidence interval). However, we have seen zero reports of VITT in the US.

Mechanism of genetically-encoded spike protein vaccines

Lei et al. [2021] found that freely circulating, spike protein-decorated pseudovirus at a very high dosage (half a billion pseudovirus particles per animal) delivered directly to the trachea damages lung arterial endothelial cells in an animal model. Similarly, an extremely high concentration (4 micrograms per milliliter) of purified recombinant spike protein could damage human pulmonary arterial endothelial cells in vitro [Lei et al., 2021]. These extremely high concentrations were used to simulate what may happen during a severe case of COVID-19 infection, wherein humans may have what some have estimated to be as high as 1 to 100 billion virions in the lungs [Sender et al., 2020]. Given there are approximately 100 spike proteins per virion [Neuman et al., 2011], this means COVID-19 infections could in theory result in as many as 10 trillion spike proteins. In wild-type viruses, the spike protein is cleaved such that the S1 portion is released and can be free to circulate in the serum [Xia et al., 2020], where it could potentially interact with ACE2 receptors on the endothelium. Thus, in both the spike protein laboratory experiments described in Lei et al. [2021] and in severe COVID-19 cases, exceedingly large amounts of freely circulating spike protein are present.

Animal studies have been performed to measure the distribution of genetically-encoded vaccines and their protein products. In the intramuscular injection site, which is by definition where the maximum amount of payload (i.e. lipid nanoparticles-packed mRNA or adenovirus) will be present and, by extension, where the maximum amount of spike protein will be produced, the payload is undetectable within 24-72 hours in vivo and the protein is undetectable within 10 days at most, and closer to 4 days post-injection when using lower doses more similar to that given to patients. Animal studies show there is some dispersal of payload to distal regions of the body, but as expected the concentrations dramatically decrease from maximum concentration at the injection site (5,680 ng/mL) to much lower concentrations elsewhere, for example they found \>3000x lower concentrations (1.82 ng/mL) in the lung, and 10,000x lower concentrations in the brain (0.429 ng/mL) [Feldman et al., 2019].

Given only a fraction of the payload will be expressed and given that the measurements of mRNA do not necessarily distinguish between functional, full-length mRNA versus non-functional mRNA fragments, only a small fraction of the measured mRNA will be translated into spike protein. The distribution of actual spike protein throughout the body appears to follow an even steeper gradient — in vivo luciferase measurements in animals treated with mRNA vaccines show significant protein expression almost entirely confined to the site of injection [Pardi et al., 2015]. Note that the concentration given to patients is even lower than those used in these animal studies, and that the dispersion appears to drop off faster for lower doses. Overall, these data indicate relatively low, transient amounts of spike protein are produced by the vaccine, and the vast majority of spike protein produced is confined to the site of injection. Therefore, the concentration of freely circulating spike protein from vaccines available to the public is bound to be many orders of magnitude times lower than the amount used in Lei et al. [2021]. The impairment found in that study would not be expected from the relatively tiny, physiologically irrelevant amount of spike proteins found in a vaccine.

In order to be physiologically relevant to, let alone damaging to blood vessels, freely moving, soluble spike proteins would have to enter the circulatory system at high enough concentrations to bind and disrupt a significant number of ACE2 receptors on a significant number of vascular endothelial cells. As discussed above, measurements indicate that no significant amount of vaccine enters circulation. The confinement of the expressed spike protein away from the circulatory system prevents it from causing significant damage. In addition to the confined localization of expression, there is another safeguard preventing spike protein from accessing the vascular endothelium in any significant amount: The vaccine uses an engineered form of the spike protein that is fused to a transmembrane anchor. The transmembrane anchor allows the spike protein to appear on the surface or membrane of the cell, but it is held in place by the anchor. This prevents the vast majority of spike protein from drifting away while at the same time creates a fixed target for the immune system to recognize and develop antibodies against the spike protein [Corbett et al., 2020]. While there is a chance for the mRNA-expressing cells to release full spike protein upon destruction by immune cells, the amount released is only going to be a small fraction of that produced by the vaccine, and certainly at too low a level to be physiologically relevant.

In agreement with the mechanism-based estimates outlined above, Ogata et al. [2021] recently published empirical measurements of freely circulating spike protein produced by the vaccines using an ultra-sensitive SIMOA assay. Their measurements revealed the average spike protein levels to be less than 50 picograms per milliliter [Lei et al. 2021], which translates to 300 fM. In contrast, the dissociation constant for ACE2 is 15-40nM [Wang et al., 2020, Wrapp et al., 2020, Lan et al., 2020, Shang et al., 2020]. Thus, the femtomolar levels produced by the vaccines are approximately 100,000x lower than physiologically relevant concentrations, let alone pathological. Importantly, peak spike protein levels are reached within days after injection, and rapidly disappear to undetectable levels within 9 days of the first injection, and much lower to undetectable levels within 3 days after the second injection. At the same time, antibodies against spike protein are inversely correlated with circulating spike protein, supporting the hypothesis that anti-spike antibodies can quickly and effectively neutralize freely circulating spike protein.

Adenovirus Vector-Based Vaccines and VITT

Importantly, the endothelial damage described by Lei et al. [2021] is not the mechanism by which VITT occurs. VITT is an extremely rare and unique form of adverse event associated only with adenovirus vector-based vaccines. It is not caused by spike proteins targeting the endothelial cells, but rather due to induction of an immune response against platelet factor 4 (PF4) by adenovirus vector-based vaccines. PF4 is released by platelets and causes them to clump and form small clots and have a physiological role in stopping bleeding (hemostasis). Antibodies are not generated against self PF4, but have been described as a rare side effect of heparin, a commonly used blood thinner. In this condition termed heparin inducted thrombocytopenia (HIT), heparin binds to PF4, and the complex then stimulates an aberrant immune response. Antibodies to PF4 are generated, and these antibodies bind to PF4, and the resulting immune complex then binds to platelets and activates them. This releases more PF4, and a cycle ensues. Activated platelets in HIT form arterial and venous clots, and as platelets get consumed in the clots, the platelet count drops resulting in severe thrombocytopenia. This combination of clots and severely low platelets is unusual.

The reason VITT raised an alarm even with very few cases was the unique HIT like clinical presentation in the absence of any heparin exposure. The seriousness of the condition, and the need to use a blood thinner besides heparin also made it an important clinical and management issue. Studies now show that in VITT, the adenovirus vector-based vaccine is able to induce high levels of PF4 antibodies in 1 in 100,000 to 1 in 500,000 individuals much the same way as heparin does in patients with HIT [Greinacher et al., 2021a].

Greinacher et al. [2021b] assessed the clinical and laboratory features of VITT patients and found that the AZ vaccine “can result in the rare development of immune thrombotic thrombocytopenia mediated by platelet-activating antibodies against PF4”. The AZ vaccine contains the preservative EDTA, which can help human cell-derived proteins from the vaccine enter the bloodstream, binding to PF4 and producing antibodies Greinacher et al. [2021a]. Lab tests showed that “High-titer anti-PF4 antibodies activate platelets and induce neutrophil activation and NETs [neutrophil extracellular traps] formation, fuelling the VITT prothrombotic response” Greinacher et al. [2021a]. Although the JJ vaccine does not use EDTA, it is an adenovirus vector-based vaccine, which is a particularly inflammatory stimulating virus [Appledorn et al., 2008, S Ahi et al., 2011]. The lack of EDTA may result in less cases of VITT, but even without EDTA proteins from the vaccine can enter the bloodstream.

The hypothesis that the cause of VITT is due to acute inflammatory reactions to vaccine components independent of the spike protein in adenovirus vector-based vaccines is consistent both with the experimental results of Greinacher et al. [2021a], and is consistent with the observation that only the AZ and JJ vaccines (both of which are adenovirus vector-based vaccines) have been associated with VITT, whereas the Moderna and Pfizer vaccines (both of which are mRNA vaccines) have not been associated with VITT.

Although the hypothesis of Greinacher et al. [2021a] has not yet been fully confirmed, it is consistent with lab testing, empirical evidence, the extreme rarity of VITT, and mechanistic constraints. It is also possible that it is remediable, since it is not due to the nature of the vaccine itself, but specific to the particulars of the formulation.


Given these observations, we conclude the vaccines do not produce enough freely circulating spike protein to induce vascular damage via the ACE2 receptor destabilization mechanism described in Lei et al. [2021]. On the contrary, the extremely low, femtomolar levels of circulating spike protein induced by the vaccine are unlikely to have any physiological relevance to vascular endothelial cells, while still allowing the immune system to develop a robust immune response to spike proteins. The presence of anti-spike antibodies may in fact serve to protect vaccinated individuals against not only SARS-CoV-2 infection, but also against spike-protein induced damage to the vascular endothelium. We speculate that this protection against spike protein-induced damage may in part explain why COVID19 symptoms are much less severe in vaccinated individuals Rossman et al. [2021].

There is now a very large amount of empirical data available that clearly shows the benefits of all approved SARSCoV-2 vaccines are far greater than the risks of extremely rare side effects. The data also is not consistent with the hypothesis that VITT is due to spike proteins, since the Pfizer and Moderna vaccines are not resulting in any reports of VITT. The data is, however, consistent with the hypothesis that side effects are due to inflammatory reactions to vaccine components in adenovirus vector-based vaccines.

Overall, we conclude that all approved SARS-CoV-2 vaccines provide far more benefits than risks, and that the very rare risk of VITT from the AZ and JJ vaccines is not due to the spike proteins, which are a fundamental part of how the vaccines work, but is most likely due to specific details of the formulation of the vaccines.


Yuyang Lei, Jiao Zhang, Cara R. Schiavon, Ming He, Lili Chen, Hui Shen, Yichi Zhang, Qian Yin, Yoshitake Cho, Leonardo Andrade, Gerald S. Shadel, Mark Hepokoski, Ting Lei, Hongliang Wang, Jin Zhang, Jason X. J. Yuan, Atul Malhotra, Uri Manor, Shengpeng Wang, Zu-Yi Yuan, and John Y-J. Shyy. Sars-cov-2 spike protein impairs endothelial function via downregulation of ace 2. Circulation Research, 128(9):1323–1326, 2021. doi:10.1161/CIRCRESAHA.121.318902.

Francois Potus, Vicky Mai, Marius Lebret, Simon Malenfant, Emilie Breton-Gagnon, Annie C Lajoie, Olivier Boucherat, Sebastien Bonnet, and Steeve Provencher. Novel insights on the pulmonary vascular consequences of covid-19. American Journal of Physiology-Lung Cellular and Molecular Physiology, 319(2):L277–L288, 2020.

Maximilian Ackermann, Stijn E Verleden, Mark Kuehnel, Axel Haverich, Tobias Welte, Florian Laenger, Arno Vanstapel, Christopher Werlein, Helge Stark, Alexandar Tzankov, et al. Pulmonary vascular endothelialitis, thrombosis, and angiogenesis in covid-19. New England Journal of Medicine, 383(2):120 128, 2020.

Hasan K Siddiqi, Peter Libby, and Paul M Ridker. Covid-19–a vascular disease. Trends in Cardiovascular Medicine, 2020.

Laure-Anne Teuwen, Vincent Geldhof, Alessandra Pasut, and Peter Carmeliet. Covid-19: the vasculature unleashed. Nature Reviews Immunology, 20(7):389–391, 2020.

Keiji Kuba, Yumiko Imai, Shuan Rao, Hong Gao, Feng Guo, Bin Guan, Yi Huan, Peng Yang, Yanli Zhang, Wei Deng, et al. A crucial role of angiotensin converting enzyme 2 (ace2) in sars coronavirus–induced lung injury. Nature medicine, 11(8):875–879, 2005.

M Makris, S Pavord, W Lester, M Scully, and BJ Hunt. Vaccine-induced immune thrombocytopenia and thrombosis (vitt). Research and Practice in Thrombosis and Haemostasis, page e12529, 2021.

Elisabeth Mahase. Astrazeneca vaccine: Blood clots are “extremely rare” and benefits outweigh risks, regulators conclude. BMJ, 373, 2021. doi:10.1136/bmj.n931.

Public Health England. Jcvi advises on covid-19 vaccine for people aged under 40, May 2021. URL https://tinyurl.com/a8eud9a6.

Mike Adams. Bombshell salk institute science paper reveals the covid spike protein is what’s causing deadly blood clots, Jul 2021. URL https://tinyurl.com/52pncva7.

Roxana Bruno, Peter McCullough, Teresa Forcades i Vila, Alexandra Henrion-Caude, Teresa García-Gasca, Galina P Zaitzeva, Sally Priester, María J Martínez Albarracín, Alejandro Sousa-Escandon, Fernando López Mirones, et al. Sars-cov-2 mass vaccination: Urgent questions on vaccine safety that demand answers from international health agencies, regulatory authorities, governments and vaccine developers. Beaufort Observer, 2021.

Medicines & Healthcare products Regulatory Agency. Coronavirus vaccine - weekly summary of yellow card reporting, May 2021. URL https://tinyurl.com/8xwydmyf.

Ron Sender, Yinon Moise Bar-On, Avi Flamholz, Shmuel Gleizer, Biana Bernsthein, Rob Phillips, and Ron Milo. The total number and mass of sars-cov-2 virions in an infected person. medRxiv, 2020.

Benjamin W Neuman, Gabriella Kiss, Andreas H Kunding, David Bhella, M Fazil Baksh, Stephen Connelly, Ben Droese, Joseph P Klaus, Shinji Makino, Stanley G Sawicki, et al. A structural analysis of m protein in coronavirus assembly and morphology. Journal of structural biology, 174(1):11–22, 2011.

Shuai Xia, Qiaoshuai Lan, Shan Su, Xinling Wang, Wei Xu, Zezhong Liu, Yun Zhu, Qian Wang, Lu Lu, and Shibo Jiang. The role of furin cleavage site in sars-cov-2 spike protein-mediated membrane fusion in the presence or absence of trypsin. Signal transduction and targeted therapy, 5(1):1–3, 2020.

Robert A Feldman, Rainard Fuhr, Igor Smolenov, Amilcar Mick Ribeiro, Lori Panther, Mike Watson, Joseph J Senn, Mike Smith, rn Almarsson, Hari S Pujar, et al. mrna vaccines against h10n8 and h7n9 influenza viruses of pandemic potential are immunogenic and well tolerated in healthy adults in phase 1 randomized clinical trials. Vaccine, 37 (25):3326–3334, 2019.

Norbert Pardi, Steven Tuyishime, Hiromi Muramatsu, Katalin Kariko, Barbara L Mui, Ying K Tam, Thomas D Madden, Michael J Hope, and Drew Weissman. Expression kinetics of nucleoside-modified mrna delivered in lipid nanoparticles to mice by various routes. Journal of Controlled Release, 217:345–351, 2015.

Kizzmekia S Corbett, Darin K Edwards, Sarah R Leist, Olubukola M Abiona, Seyhan Boyoglu-Barnum, Rebecca A Gillespie, Sunny Himansu, Alexandra Schäfer, Cynthia T Ziwawo, Anthony T DiPiazza, et al. Sars-cov-2 mrna vaccine design enabled by prototype pathogen preparedness. Nature, 586(7830):567–571, 2020.

Alana F Ogata, Chi-An Cheng, Michaël Desjardins, Yasmeen Senussi, Amy C Sherman, Megan Powell, Lewis Novack, Salena Von, Xiaofang Li, Lindsey R Baden, and David R Walt. Circulating SARS-CoV-2 Vaccine Antigen Detected in the Plasma of mRNA-1273 Vaccine Recipients. Clinical Infectious Diseases, 05 2021. ISSN 1058-4838. doi:10.1093/cid/ciab465. ciab465.

Qihui Wang, Yanfang Zhang, Lili Wu, Sheng Niu, Chunli Song, Zengyuan Zhang, Guangwen Lu, Chengpeng Qiao, Yu Hu, Kwok-Yung Yuen, et al. Structural and functional basis of sars-cov-2 entry by using human ace2. Cell, 181 (4):894–904, 2020.

Daniel Wrapp, Nianshuang Wang, Kizzmekia S Corbett, Jory A Goldsmith, Ching-Lin Hsieh, Olubukola Abiona, Barney S Graham, and Jason S McLellan. Cryo-em structure of the 2019-ncov spike in the prefusion conformation. Science, 367(6483):1260–1263, 2020.

Jun Lan, Jiwan Ge, Jinfang Yu, Sisi Shan, Huan Zhou, Shilong Fan, Qi Zhang, Xuanling Shi, Qisheng Wang, Linqi Zhang, et al. Structure of the sars-cov-2 spike receptor-binding domain bound to the ace2 receptor. Nature, 581 (7807):215–220, 2020.

Jian Shang, Gang Ye, Ke Shi, Yushun Wan, Chuming Luo, Hideki Aihara, Qibin Geng, Ashley Auerbach, and Fang Li. Structural basis of receptor recognition by sars-cov-2. Nature, 581(7807):221–224, 2020.

Andreas Greinacher, Kathleen Selleng, Jan Wesche, Stefan Handtke, Raghavendra Palankar, Konstanze Aurich, Michael Lalk, Karen Methling, Uwe Völker, Christian Hentschker, et al. Towards understanding chadox1 ncov19 vaccine-induced immune thrombotic thrombocytopenia (vitt). Research Square, 2021a.

Andreas Greinacher, Thomas Thiele, Theodore E Warkentin, Karin Weisser, Paul A Kyrle, and Sabine Eichinger. Thrombotic thrombocytopenia after chadox1 ncov-19 vaccination. New England Journal of Medicine, 2021b.

DM Appledorn, A McBride, S Seregin, JM Scott, Nathan Schuldt, A Kiang, S Godbehere, and A Amalfitano. Complex interactions with several arms of the complement system dictate innate and humoral immunity to adenoviral vectors. Gene therapy, 15(24):1606–1617, 2008.

Yadvinder S Ahi, Dinesh S Bangari, and Suresh K Mittal. Adenoviral vector immunity: its implications and circumvention strategies. Current gene therapy, 11(4):307–320, 2011.

Hagai Rossman, Smadar Shilo, Tomer Meir, Malka Gorfine, Uri Shalit, and Eran Segal. Covid-19 dynamics after a national immunization program in israel. Nature medicine, pages 1–7, 2021.

Statistical problems found when studying Long Covid in kids

Summary: Statistical tests need to be paired with proper data and study design to yield valid results. A recent review paper on Long Covid in children provides a useful example of how researchers can get this wrong. We use causal diagrams to decompose the problem and illustrate where errors were made.



A recent review paper by Australian and Swiss doctors, How Common Is Long COVID in Children and Adolescents?, was widely discussed in the press, with 128 news stories from 103 outlets. The headlines were reassuring:

The paper in question does not actually say any of these things, but rather concludes that “the true incidence of this syndrome in children and adolescents remains uncertain.” However, the challenges of accurate science journalism are not the topic for our article today. Rather, we will describe a critical flaw in the statistical analysis in this review, as an exercise in better understanding how to interpret statistical tests.

A key contribution of the review is that it separates those studies that use a “control group” from those that do not. The authors suggest we should focus our attention on the studies with a control group, because “in the absence of a control group, it is impossible to distinguish symptoms of long COVID from symptoms attributable to the pandemic.” The National Academy of Sciences warns that “use of an inappropriate control group can make it impossible to draw meaningful conclusions from a study.” As we will see, this is, unfortunately, what happened in this review. But first, let’s do a brief recap of control groups and statistical tests.

Control groups and RCTs

When assessing the impact of an intervention, such as the use of a new drug, the gold standard is to use a Randomised Controlled Trial (RCT). In an RCT, a representative sample is selected, and randomly split into two groups, one of which receives the medical intervention (e.g. the drug), and one which doesn’t (normally that one gets a placebo). This can, when things go well, show clearly whether the drug made a difference. Generally, a “p value” is calculated, which is the probability that the effect seen in the data would be observed by chance if there was truly no difference between cases and controls (i.e. null hypothesis was true), along with a “confidence interval”, which is the range of outcomes that would be expected after considering random variation. If the p value is less than some number (often 0.05) the RCT is considered to be “statistically significant”. Without an RCT, it can be harder to distinguish whether two groups differ because of the intervention, or because of some other difference between the groups.

We can represent this analysis as a diagram like so:

Causal diagram for an RCT
Causal diagram for an RCT

This is an example of a (simplified and informal) causal diagram. The black arrows show the direct relationships we can measure or control – in this case, our selection of control group vs experimental group is used to decide who gets the drug, and we then measure the outcome (e.g. do symptoms improve) for each group based on our group selection. Because the selection was random (since this is an RCT), we can infer the dotted line: how much does taking the drug change the outcome? If the size of the control or experimental group is small, then it is possible that the difference in outcomes between the two groups is entirely due to random chance. To handle that, we pop the effect size and sample size into statistical software such as R and it will tell us the p value and confidence interval of the effect.

Because RCTs are the gold standard for assessing the impact of a medical intervention, they are used whenever possible. Nearly all drugs on the market have been through multiple RCTs, and most medical education includes some discussion of the use and interpretation of RCTs.

Control groups and observational studies

Sometimes, as discussed in The Planning of Observational Studies of Human Populations, “it is not feasible to use controlled experimentation”, but we want to investigate a causal relationship between variables, in which case we may decide to use an observational study. For instance, studying “the relationship between smoking and health”, risk factors for “injuries in motor accidents”, or “effects of new social programmes”. In cases like these, it isn’t possible to create a true “control group” like in an RCT, since we cannot generally randomly assign people, for instance, to a group that are told to start smoking.

Instead, we have to try to find two groups that are as similar as possible, but differ only in the variable under study – for instance, a group of smokers and a group of non-smokers that are of similar demographics, health, etc. This can be challenging. Indeed, the question “does smoking cause cancer” remained controversial for decades, despite many attempts at observational studies.

Researchers have noted that “results from observational studies can confuse the effect of interest with other variables’ effects, leading to an association that is not causal. It would be helpful for clinicians and researchers to be able to visualize the structure of biases in a clinical study”. They suggest using causal diagrams for this purpose, including to help avoid confounding bias in epidemiological studies. So, let’s give that a try now!

Structure of the Long Covid review

In How Common Is Long COVID in Children and Adolescents? the authors suggest we focus on studies of Long Covid prevalence that include a control group. The idea is that we take one group that has (or had) COVID, and one group that didn’t, and then see if they have Long Covid symptoms a few weeks or months later. Here’s what the causal diagram would look like:

Idealised causal diagram for Long Covid prevalence
Idealised causal diagram for Long Covid prevalence

Here we are trying to determine if COVID infection causes Long Covid symptoms. Since COVID infection is the basis of the Control group selection, and we can compare the Long Covid symptoms for each group, that would allow us to infer the answer to our question. The statistical tests reported in the review paper only apply if this structure is correct.

However, it’s not quite this simple. We don’t directly know who has a COVID infection, but instead we have to infer it using a test (e.g serology, PCR, or rapid). It is so easy nowadays to run a statistical test on a computer, it can be quite tempting to just use the software and report what it says, without being careful to check that the statistical assumptions implicitly being made are met by the data and design.

We might hope that we could modify our diagram like so:

Idealised causal diagram including testing
Idealised causal diagram including testing

In this case, we could still directly infer the dotted line (i.e “does COVID infection cause Long Covid symptoms?”), since there is just one unknown relationship, and all the arrows go in the same direction.

But unfortunately, this doesn’t work either. The link between test results and infection is not perfect. Some researchers, for instance, have estimated that PCR tests may miss half, or even 90% of infections. Part of the reason is that “thresholds for SARS-CoV-2 antibody assays have typically been determined using samples from symptomatic, often hospitalised, patients”. Others have found that 36% of infections do not seroconvert, and that children in particular may serorevert. It appears that false negative test results may be more common in children – tests are most sensitive when used for middle-aged men.

To make things even more complicated, research shows that “Long-COVID is associated with weak anti-SARS-CoV-2 antibody response.”

Putting this all together, here’s what our diagram now looks like, using red arrows here to indicate negative relationships:

Causal diagram including some confounders
Causal diagram including some confounders

This shows that test results are not just associated with COVID infection, but also with Age and Long Covid symptoms, and that the association between COVID infection and test result is not imperfect and not fully understood.

Because of this, we can’t now directly infer the relationship between COVID infection and Long Covid symptoms. We would first need to fully understand and account for the confounders and uncertainties. Simply reporting the results of a statistical test does not give meaningful information in this case.

In particular, we can see that the issues we have identified all bias the data in the same direction: they result in infected cases being incorrectly placed in the control group.

For more details about this issue, see the article Long Covid: issues around control selection, by Dr Nisreen Alwan MBE.

The problem of p-values

The review claims that “all studies to date have substantial limitations or do not show a difference between children who had been infected by SARS-CoV-2 and those who were not”. This claim appears to be made on the basis of p-values, which are shown for each control group study in the review. All but one study did actually find a statistically significant difference between the groups being compared (at p<0.05, which is the usual cut-off for such analyses).

Regardless of what the results actually show, p-values are not being used in an appropriate way here. The American Statistical Association (ASA) has released a “Statement on Statistical Significance and P-Values” with six principles underlying the proper use and interpretation of the p-value. In particular, note the following principles:

  • P-values do not measure the probability that the studied hypothesis is true, or the probability that the data were produced by random chance alone.
  • Scientific conclusions and business or policy decisions should not be based only on whether a p-value passes a specific threshold.
  • A p-value, or statistical significance, does not measure the size of an effect or the importance of a result.

A p-value is lower when there is more data, or a stronger relationship in the data (and visa versa). A high p-value does not necessarily mean that there is not a relationship in the data – it may simply mean that not enough data has been collected.

Because a p-value “does not measure the size of an effect or the importance of a result”, they don’t actually tell us about the prevalence of Long Covid. The use of p-values in studying drug efficacy is very common, since we do often want to answer the question “does this drug help at all”? But to assess what the range of prevalence levels may be, we instead need to look at confidence intervals, which unfortunately are not shown at all in the review.

Furthermore, we should not look at p-values out of context, but instead need to also consider the likelihood of alternative hypotheses. The alternative hypothesis provided in the review is that the symptoms may be due to “lockdown measures, including school closures”.

One of the included control group studies stood out as an outlier, in which 10% of Swiss children with negative tests were found to have Long Covid symptoms, many times higher than other similar studies. Was this because of the confounding effects discussed in the previous section, or was it due to lockdowns and school closures? Switzerland did not have a full lockdown, and schools were only briefly closed, reopening nearly a year before the Long Covid symptom tests in the study. On the other hand, Switzerland may have had a very high number of cases. Wikipedia notes that “the Swiss government has had an official policy of not testing people with only mild symptoms”, and has still recorded nearly 900 thousand cases in a population of just 8 million people.

In a statistical design, an alternative hypothesis should not be considered the null hypothesis unless we are quite certain it represents the normal baseline behaviour. But assuming that the symptoms found in the control group are due to pandemic factors other than infection is itself a hypothesis that needs careful testing and does not seem to be fully supported by the data in the study. It is not an appropriate design to use this as the base case, as was done in the review.

Conclusion and next steps

The problem with control group definition, incorrect use of statistical tests, and statistical design problems does not change the key conclusion of the review: “the true incidence of this syndrome in children and adolescents remains uncertain.” So, how to we resolve this uncertainty?

The review has a number of suggestions for future research to improve our understanding or Long Covid prevalence in children. As we’ve seen in this article, we also need to more carefully consider and account for confounding bias. It is often possible, mathematically, to infer an association even in more complex causal relationships such as we see above. However, doing so requires a full and accurate understanding of all of the relationships in the causal structure.

Furthermore, a more complete and rigorous assessment of confounders needs to be completed. We’ve only scratched the surface in this article on one aspect: bias in the control group. Bias in the “Long Covid symptoms” node also needs to be considered. For instance: are all Long Covid symptoms being considered; is there under-reporting due to difficulties of child communication or understanding; is there under-reporting due to gender bias; are “on again / off again” variable symptoms being tracked correctly; and so forth.

Whatever the solution turns out to be, it seems that for a while at least, the prevalence of Long Covid in children will remain uncertain. How parents, doctors, and policy makers respond to this risk and uncertainty will be a critical issue for children around the world.


Many thanks to Hannah Davis, Dr Deepti Gurdasani, Dr Rachel Thomas, Dr Zoë Hyde, and Dr Nisreen Alwan MBE for invaluable help with research and review for this article.

Medicine is Political

Experts warn that we are not prepared for the surge in disability due to long covid, an illness that afflicts between one-fourth and one-third of people who get covid, including mild cases, for months afterwards. Some early covid cases have been sick for 18 months, with no end in sight. The physiological damage that covid causes can include cognitive dysfunction and deficits, brain activity scans similar to those seen in Alzheimer’s patients, GI immune system damage, cornea damage, immune dysfunction, increased risk of kidney outcomes, dysfunction in T cell memory generation, pancreas damage, and ovarian failure. Children are at risk too.

As the evidence continues to mount of alarming long term physiological impacts of covid, and tens of millions are unable to return to work, we might expect leaders to take covid more seriously. Yet we are seeing concerted efforts to downplay the long-term health effects of covid using strategies straight out of the climate denial playbook, such as funding contrarian scientists, misleading petitions, social media bots, and disingenuous debate tactics that make the science seem murkier than it is. In many cases, these minimization efforts are being funded by the same billionaires and institutions that fund climate change denialism. Dealing with many millions of newly disabled people will be very expensive for governments, social service programs, private insurance companies, and others. Thus, many have a significant financial interest in distorting the science around long term effects of covid to minimize the perceived impact.

In topics ranging from covid-19 to HIV research to the long history of wrongly assuming women’s illnesses are psychosomatic, we have seen again and again that medicine, like all science, is political. This shows up in myriad ways, such as: who provides funding, who receives that funding, which questions get asked, how questions are framed, what data is recorded, what data is left out, what categories included, and whose suffering is counted.

Scientists often like to think of their work as perfectly objective, perfectly rational, free from any bias or influence. Yet by failing to acknowledge the reality that there is no “view from nowhere”, they miss their own blindspots and make themselves vulnerable to bad-faith attacks. As one climate scientist recounted of the last 3 decades, “We spent a long time thinking we were engaged in an argument about data and reason, but now we realize it’s a fight over money and power… They [climate change deniers] focused their lasers on the science and like cats we followed their pointer and their lead.”

The American Institute for Economic Research (AIER), a libertarian think tank funded by right wing billionaire Charles Koch which invests in fossil fuels, energy utilities, and tobacco, is best known for its research denying the climate crisis. In October 2020, a document called the Great Barrington Declaration (GBD) was developed at a private AIER retreat, calling for a “herd immunity” approach to covid, arguing against lockdowns, and suggesting that young, healthy people have little to worry about. The three scientists who authored the GBD have prestigious pedigrees and are politically well-connected, speaking to White House Officials and having found favor in the British government. One of them, Sunetra Gupta of Oxford, had released a wildly inaccurate paper in March 2020 claiming that up to 68% of the UK population had been exposed to covid, and that there were already significant levels of herd immunity to coronavirus in both the UK and Italy (again, this was in March 2020). Gupta received funding from billionaire conservative donors, Georg and Emily von Opel. Another one of the authors, Jay Bhattacharya of Stanford, co-authored a widely criticized pre-print in April 2020 that relied on a biased sampling method to “show” that 85 times more people in Santa Clara County California had already had covid compared to other estimates, and thus suggested that the fatality rate for covid was much lower than it truly is.

Half of the social media accounts advocating for herd immunity seem to be bots, characterized as engaging in abnormally high levels of retweets & low content diversity. An article in the BMJ recently advised that it is “critical for physicians, scientists, and public health officials to realize that they are not dealing with an orthodox scientific debate, but a well-funded sophisticated science denialist campaign based on ideological and corporate interests.”

This myth of perfect scientific objectivity positions modern medicine as completely distinct from a history where women were diagnosed with “hysteria” (roaming uterus) for a variety of symptoms, where Black men were denied syphilis treatment for decades as part of a “scientific study”, and multiple sclerosis was “called hysterical paralysis right up to the day they invented a CAT scan machine” and demyelination could be seen on brain scans.

However, there is not some sort of clean break where bias was eliminated and all unknowns were solved. Black patients, including children, still receive less pain medication than white patients for the same symptoms. Women are still more likely to have their physical symptoms dismissed as psychogenic. Nearly half of women with autoimmune disorders report being labeled as “chronic complainers” by their doctors in the 5 years (on average) they spend seeking a diagnosis. All this impacts what data is recorded in their charts, what symptoms are counted.

Medical data are not objective truths. Like all data, the context is critical. It can be missing, biased, and incorrect. It is filtered through the opinions of doctors. Even blood tests and imaging scans are filtered through the decisions of what tests to order, what types of scans to take, what accepted guidelines recommend, what technology currently exists. And the technology that exists depends on research and funding decisions stretching back decades, influenced by politics and cultural context.

One may hope that in 10 years we will have clearer diagnostic tests for some illnesses which remain contested now, just as the ability to identify multiple sclerosis improved with better imaging. In the meantime, we should listen to patients and trust in their ability to explain their own experiences, even if science can’t fully understand them yet.

Science does not just progress inevitably, independent of funding and politics and framing and biases. A self-fulfilling prophecy often occurs in which doctors:

  1. label a new, poorly understood, multi-system disease as psychogenic,
  2. use this as justification to not invest much funding into researching physiological origins,
  3. and then point to the lack of evidence as a reason why the illness must be psychogenic.

This is largely the experience of ME/CFS patients over the last several decades. Myalgic encephalomyelitis (ME/CFS), involves dysfunction of the immune system, autonomic systems, and energy metabolism (including mitochondrial dysfunction, hypoacetylation, reduced oxygen uptake, and impaired oxygen delivery). ME/CFS is more debilitating than many chronic diseases, including chronic renal failure, lung cancer, stroke, and type-2 diabetes. It is estimated 25–29% of patients are homebound or bedbound. ME/CFS is often triggered by viral infections, so it is not surprising that we are seeing some overlap between ME/CFS and long covid. ME/CFS disproportionately impacts women, and a now discredited 1970 paper identified a major outbreak in 1958 amongst nurses at a British hospital as “epidemic hysteria”. This early narrative of ME/CFS as psychogenic has been difficult to shake. Even as evidence continues to accumulate of immune, metabolic, and autonomous system dysfunction, some doctors persist in believing that ME/CFS must be psychogenic. It has remained woefully underfunded: from 2013-2017, NIH funding was only at 7.3% relative commensurate to its disease burden. Note that the below graph is on a log scale: ME/CFS is at 7%, Depression and asthma are at 100% and diseases like cancer and HIV are closer to 1000%.

Graph of NIH funding on log scale, from above paper by Mirin, Dimmock, Leonard
Graph of NIH funding on log scale, from above paper by Mirin, Dimmock, Leonard

Portraying patients as unscientific and irrational is the other side of the same coin for the myth that medicine is perfectly rational. Patients that disagree with having symptoms they know are physiological dismissed as psychogenic, that reject treatments from flawed studies, or who distrust medical institutions based on their experiences of racism, sexism, and mis-diagnosis, are labeled as “militant” or “irrational”, and placed in the same category with conspiracy theorists and those peddling disinformation.

On an individual level, receiving a psychological misdiagnosis lengthens the time it will take to get the right diagnosis, since many doctors will stop looking for physiological explanations. A study of 12,000 rare disease patients covered by the BBC found that “while being misdiagnosed with the wrong physical disease doubled the time it took to get to the right diagnosis, getting a psychological misdiagnosis extended it even more – by 2.5 up to 14 times, depending on the disease.” This dynamic holds true at the disease level as well: once a disease is mis-labeled as psychogenic, many doctors will stop looking for physiological origins.

We are seeing increasing efforts to dismiss long covid as psychogenic in high profile platforms such as the WSJ and New Yorker. The New Yorker’s first feature article on long covid, published last month, neglected to interview any clinicians who treat long covid patients nor to cite the abundant research on how covid causes damage to many organ systems, yet interviewed several doctors in unrelated fields who claim long covid is psychogenic. In response to a patient’s assertion that covid impacts the brain, the author spent an entire paragraph detailing how there is currently no evidence that covid crosses the blood-brain barrier, but didn’t mention the research on covid patients finding cognitive dysfunction and deficits, PET scans similar to those seen in Alzheimer’s patients, neurological damage, and shrinking grey matter. This leaves a general audience with the mistaken impression that it is unproven whether covid impacts the brain, and is a familiar tactic from bad-faith science debates.

The New Yorker article set up a strict dichotomy between long covid patients and doctors, suggesting that patients harbor a “disregard for expertise”; are less “concerned about what is and isn’t supported by evidence”; and are overly “impatient.” In contrast, doctors appreciate the “careful study design, methodical data analysis, and the skeptical interpretation of results” that medicine requires. Of course, this is a false dichotomy: many patients are more knowledgeable about the latest research than their doctors, some patients are publishing in peer-reviewed journals, and there are many medical doctors that are also patients. And on the other hand, doctors are just as prone as the rest of us to biases, blind spots, and institutional errors.

AP Photo/J. Scott Applewhite
AP Photo/J. Scott Applewhite

In 1987, 40,000 Americans had already died of AIDS, yet the government and pharmaceutical companies were doing little to address this health crisis. AIDS was heavily stigmatized, federal spending was minimal, and pharmaceutical companies lacked urgency. The activists of ACT UP used a two pronged approach: creative and confrontational acts of protest, and informed scientific proposals. When the FDA refused to even discuss giving AIDS patients access to experimental drugs, ACT UP protested at their headquarters, blocking entrances and lying down in front of the building with tombstones saying “Killed by the FDA”. This opened up discussions, and ACT UP offered viable scientific proposals, such as switching from the current approach of conducting drug trials on a small group of people over a long time, and instead testing a large group of people over a short time, radically speeding up the pace at which progress occurred. ACT UP used similar tactics to protest the NIH and pharmaceutical companies, demanding research on how to treat the opportunistic infections that killed AIDS patients, not solely research for a cure. The huge progress that has happened in HIV/AIDS research and treatment would not have happened without the efforts of ACT UP.

Across the world, we are at a pivotal time in determining how societies and governments will deal with the masses of newly disabled people due to long covid. Narratives that take hold early often have disproportionate staying power. Will we inaccurately label long covid as psychogenic, primarily invest in psychiatric research that can’t address the well-documented physiological damage caused by covid, and financially abandon the patients who are now unable to work? Or will we take the chance to transform medicine to better recognize the lived experiences and knowledge of patients, to center patient partnerships in biomedical research for complex and multi-system diseases, and strengthen inadequate disability support and services to improve life for all people with disabilities? The decisions we collectively make now on these questions will have reverberations for decades to come.