1. Peter Doshi, senior editor,

2. Fiona Godlee, former editor in chief,

3. Kamran Abbasi, editor in chief

4. The BMJ, London, UK

5. Correspondence to: P Doshi Pdoshi{at}bmj.com

Data should be fully and immediately available for public scrutiny

In the pages of The BMJ a decade ago, in the middle of a different pandemic, it came to light that governments around the world had spent billions stockpiling antivirals for influenza that had not been shown to reduce the risk of complications, hospital admissions, or death. The majority of trials that underpinned regulatory approval and government stockpiling of oseltamivir (Tamiflu) were sponsored by the manufacturer; most were unpublished, those that were published were ghostwritten by writers paid by the manufacturer, the people listed as principal authors lacked access to the raw data, and academics who requested access to the data for independent analysis were denied.1234

The Tamiflu saga heralded a decade of unprecedented attention to the importance of sharing clinical trial data.56 Public battles for drug company data,78 transparency campaigns with thousands of signatures,910 strengthened journal data sharing requirements,1112 explicit commitments from companies to share data,13 new data access website portals,8 and landmark transparency policies from medicines regulators1415 all promised a new era in data transparency.

Progress was made, but clearly not enough. The errors of the last pandemic are being repeated. Memories are short. Today, despite the global rollout of covid-19 vaccines and treatments, the anonymised participant level data underlying the trials for these new products remain inaccessible to doctors, researchers, and the public—and are likely to remain that way for years to come.16 This is morally indefensible for all trials, but especially for those involving major public health interventions.

Unacceptable delay

Pfizer’s pivotal covid vaccine trial was funded by the company and designed, run, analysed, and authored by Pfizer employees. The company and the contract research organisations that carried out the trial hold all the data.17 And Pfizer has indicated that it will not begin entertaining requests for trial data until May 2025, 24 months after the primary study completion date, which is listed on ClinicalTrials.gov as 15 May 2023 (NCT04368728).

The lack of access to data is consistent across vaccine manufacturers.16 Moderna says data “may be available … with publication of the final study results in 2022.”18 Datasets will be available “upon request and subject to review once the trial is complete,” which has an estimated primary completion date of 27 October 2022 (NCT04470427).

As of 31 December 2021, AstraZeneca may be ready to entertain requests for data from several of its large phase III trials.19 But actually obtaining data could be slow going. As its website explains, “timelines vary per request and can take up to a year upon full submission of the request.”20

Underlying data for covid-19 therapeutics are similarly hard to find. Published reports of Regeneron’s phase III trial of its monoclonal antibody therapy REGEN-COV flatly state that participant level data will not be made available to others.21 Should the drug be approved (and not just emergency authorised), sharing “will be considered.” For remdesivir, the US National Institutes of Health, which funded the trial, created a new portal to share data (https://accessclinicaldata.niaid.nih.gov/), but the dataset on offer is limited. An accompanying document explains: “The longitudinal data set only contains a small subset of the protocol and statistical analysis plan objectives.”

We are left with publications but no access to the underlying data on reasonable request. This is worrying for trial participants, researchers, clinicians, journal editors, policy makers, and the public. The journals that have published these primary studies may argue that they faced an awkward dilemma, caught between making the summary findings available quickly and upholding the best ethical values that support timely access to underlying data. In our view, there is no dilemma; the anonymised individual participant data from clinical trials must be made available for independent scrutiny.

Journal editors, systematic reviewers, and the writers of clinical practice guideline generally obtain little beyond a journal publication, but regulatory agencies receive far more granular data as part of the regulatory review process. In the words of the European Medicine Agency’s former executive director and senior medical officer, “relying solely on the publications of clinical trials in scientific journals as the basis of healthcare decisions is not a good idea ... Drug regulators have been aware of this limitation for a long time and routinely obtain and assess the full documentation (rather than just publications).”22

Among regulators, the US Food and Drug Administration is believed to receive the most raw data but does not proactively release them. After a freedom of information request to the agency for Pfizer’s vaccine data, the FDA offered to release 500 pages a month, a process that would take decades to complete, arguing in court that publicly releasing data was slow owing to the need to first redact sensitive information.23 This month, however, a judge rejected the FDA’s offer and ordered the data be released at a rate of 55 000 pages a month. The data are to be made available on the requesting organisation’s website (phmpt.org).

In releasing thousands of pages of clinical trial documents, Health Canada and the EMA have also provided a degree of transparency that deserves acknowledgment.2425 Until recently, however, the data remained of limited utility, with copious redactions aimed at protecting trial blinding. But study reports with fewer redactions have been available since September 2021,2425 and missing appendices may be accessible through freedom of information requests.

Even so, anyone looking for participant level datasets may be disappointed because Health Canada and the EMA do not receive or analyse these data, and it remains to be seen how the FDA responds to the court order. Moreover, the FDA is producing data only for Pfizer’s vaccine; other manufacturers’ data cannot be requested until the vaccines are approved, which the Moderna and Johnson & Johnson vaccines are not. Industry, which holds the raw data, is not legally required to honour requests for access from independent researchers.

Like the FDA, and unlike its Canadian and European counterparts, the UK’s regulator—the Medicines and Healthcare Products Regulatory Agency—does not proactively release clinical trial documents, and it has also stopped posting information released in response to freedom of information requests on its website.26

Transparency and trust

As well as access to the underlying data, transparent decision making is essential. Regulators and public health bodies could release details27 such as why vaccine trials were not designed to test efficacy against infection and spread of SARS-CoV-2.28 Had regulators insisted on this outcome, countries would have learnt sooner about the effect of vaccines on transmission and been able to plan accordingly.29

Big pharma is the least trusted industry.30 At least three of the many companies making covid-19 vaccines have past criminal and civil settlements costing them billions of dollars.31 One pleaded guilty to fraud.31 Other companies have no pre-covid track record. Now the covid pandemic has minted many new pharma billionaires, and vaccine manufacturers have reported tens of billions in revenue.32

The BMJ supports vaccination policies based on sound evidence. As the global vaccine rollout continues, it cannot be justifiable or in the best interests of patients and the public that we are left to just trust “in the system,” with the distant hope that the underlying data may become available for independent scrutiny at some point in the future. The same applies to treatments for covid-19. Transparency is the key to building trust and an important route to answering people’s legitimate questions about the efficacy and safety of vaccines and treatments and the clinical and public health policies established for their use.

Twelve years ago we called for the immediate release of raw data from clinical trials.1 We reiterate that call now. Data must be available when trial results are announced, published, or used to justify regulatory decisions. There is no place for wholesale exemptions from good practice during a pandemic. The public has paid for covid-19 vaccines through vast public funding of research, and it is the public that takes on the balance of benefits and harms that accompany vaccination. The public, therefore, has a right and entitlement to those data, as well as to the interrogation of those data by experts.

Pharmaceutical companies are reaping vast profits without adequate independent scrutiny of their scientific claims.33 The purpose of regulators is not to dance to the tune of rich global corporations and enrich them further; it is to protect the health of their populations. We need complete data transparency for all studies, we need it in the public interest, and we need it now.

Footnotes

• Competing interests: We have read and understood BMJ policy on declaration of interests and declare that The BMJ is a co-founder of the AllTrials campaign. PD was one of the Cochrane reviewers studying influenza antivirals beginning in 2009, who campaigned for access to data. He also helped organise the Coalition Advocating for Adequately Licensed Medicines (CAALM), which formally petitioned the FDA to refrain from fully approving any covid-19 vaccine this year (docket FDA-2021-P-0786). PD is also a member of Public Health and Medical Professionals for Transparency, which has sued the FDA to obtain the Pfizer covid-19 vaccine data. The views and opinions do not necessarily reflect the official policy or position of the University of Maryland.

• Provenance and peer review: Commissioned; externally peer reviewed.

It seemed a truth universally acknowledged that the human population had no pre-existing immunity to SARS-CoV-2, but is that actually the case? Peter Doshi explores the emerging research on immunological responses

Even in local areas that have experienced some of the greatest rises in excess deaths during the covid-19 pandemic, serological surveys since the peak indicate that at most only around a fifth of people have antibodies to SARS-CoV-2: 23% in New York, 18% in London, 11% in Madrid.[1][2][3] Among the general population the numbers are substantially lower, with many national surveys reporting in single digits.

With public health responses around the world predicated on the assumption that the virus entered the human population with no pre-existing immunity before the pandemic,[4] serosurvey data are leading many to conclude that the virus has, as Mike Ryan, WHO’s head of emergencies, put it, “a long way to burn.”

Yet a stream of studies that have documented SARS-CoV-2 reactive T cells in people without exposure to the virus are raising questions about just how new the pandemic virus really is, with many implications.

Not so novel coronavirus?

At least six studies have reported T cell reactivity against SARS-CoV-2 in 20% to 50% of people with no known exposure to the virus.[5][6][7][8][9][10]

In a study of donor blood specimens obtained in the US between 2015 and 2018, 50% displayed various forms of T cell reactivity to SARS-CoV-2.[5][11] A similar study that used specimens from the Netherlands reported T cell reactivity in two of 10 people who had not been exposed to the virus.[7]

In Germany reactive T cells were detected in a third of SARS-CoV-2 seronegative healthy donors (23 of 68). In Singapore a team analysed specimens taken from people with no contact or personal history of SARS or covid-19; 12 of 26 specimens taken before July 2019 showed reactivity to SARS-CoV-2, as did seven of 11 from people who were seronegative against the virus.[8] Reactivity was also discovered in the UK and Sweden.[6][9][10]

Though these studies are small and do not yet provide precise estimates of pre-existing immunological responses to SARS-CoV-2, they are hard to dismiss, with several being published in Cell and Nature. Alessandro Sette, an immunologist from La Jolla Institute for Immunology in California and an author of several of the studies (box 1), told The BMJ, “At this point there are a number of studies that are seeing this reactivity in different continents, different labs. As a scientist you know that is a hallmark of something that has a very strong footing.”

Swine flu déjà vu

In late 2009, months after the World Health Organization declared the H1N1 “swine flu” virus to be a global pandemic, Alessandro Sette was part of a team working to explain why the so called “novel” virus did not seem to be causing more severe infections than seasonal flu.[12]

Their answer was pre-existing immunological responses in the adult population: B cells and, in particular, T cells, which “are known to blunt disease severity.”[12] Other studies came to the same conclusion: people with pre-existing reactive T cells had less severe H1N1 disease.[13][14] In addition, a study carried out during the 2009 outbreak by the US Centers for Disease Control and Prevention reported that 33% of people over 60 years old had cross reactive antibodies to the 2009 H1N1 virus, leading the CDC to conclude that “some degree of pre-existing immunity” to the new H1N1 strains existed, especially among adults over age 60.[15]

The data forced a change in views at WHO and CDC, from an assumption before 2009 that most people “will have no immunity to the pandemic virus”[16] to one that acknowledged that “the vulnerability of a population to a pandemic virus is related in part to the level of pre-existing immunity to the virus.”[17] But by 2020 it seems that lesson had been forgotten.

Researchers are also confident that they have made solid inroads into ascertaining the origins of the immune responses. “Our hypothesis, of course, was that it’s so called ‘common cold’ coronaviruses, because they’re closely related,” said Daniela Weiskopf, senior author of a paper in Science that confirmed this hypothesis.[18] “We have really shown that this is a true immune memory and it is derived in part from common cold viruses.” Separately, researchers in Singapore came to similar conclusions about the role of common cold coronaviruses but noted that some of the T cell reactivity may also come from other unknown coronaviruses, even of animal origin.[8]

Taken together, this growing body of research documenting pre-existing immunological responses to SARS-CoV-2 may force pandemic planners to revisit some of their foundational assumptions about how to measure population susceptibility and monitor the extent of epidemic spread.

Population immunity: underestimated?

Seroprevalence surveys measuring antibodies have been the preferred method for gauging the proportion of people in a given population who have been infected by SARS-CoV-2 (and have some degree of immunity to it), with estimates of herd immunity thresholds providing a sense of where we are in this pandemic. Whether we overcome it through naturally derived immunity or vaccination, the sense is that it won’t be over until we reach a level of herd immunity.

The fact that only a minority of people, even in the hardest hit areas, display antibodies against SARS-CoV-2 has led most planners to assume the pandemic is far from over. In New York City, where just over a fifth of people surveyed had antibodies, the health department concluded that “as this remains below herd immunity thresholds, monitoring, testing, and contact tracing remain essential public health strategies.”[19] “Whatever that number is, we’re nowhere near close to it,” said WHO’s Ryan in late July, referring to the herd immunity threshold (box 2).

Calculating the herd immunity threshold

In theory, outbreaks of contagious disease follow a certain trajectory. In a population that lacks immunity new infections grow rapidly. At some point an inflection in this growth should occur, and the incidence will begin to fall.

The 1970s gave rise to a theory that defined this inflection point as the herd immunity threshold (HIT) and offered a straightforward formula for estimating its size: HIT=1−1/R0 (where R0 is the disease’s basic reproduction number, or the average number of secondary cases generated by an infectious individual among susceptible people). This simple calculation has guided—and continues to guide—many vaccination campaigns, often used to define target levels of vaccination.[20]

The formula rests on two assumptions: that, in a given population, immunity is distributed evenly and members mix at random. While vaccines may be deliverable in a near random fashion, from the earliest days questions were raised about the random mixing assumption. Apart from certain small closed populations such as “orphanages, boarding schools, or companies of military recruits,” Fox and colleagues wrote in 1971,[21] truly random mixing is the exception, not the rule. “We could hardly assume even a small town to be a single homogeneously mixing unit. Each individual is normally in close contact with only a small number of individuals, perhaps of the order of 10-50.”

Nearly 50 years later, Gabriela Gomes, an infectious disease modeller at the University of Strathclyde, is reviving concerns that the theory’s basic assumptions do not hold. Not only do people not mix randomly, infections (and subsequent immunity) do not happen randomly either, her team says. “More susceptible and more connected individuals have a higher propensity to be infected and thus are likely to become immune earlier. Due to this selective immunization by natural infection, heterogeneous populations require less infections to cross their herd immunity threshold,” they wrote.[22] While most experts have taken the R0 for SARS-CoV-2 (generally estimated to be between 2 and 3) and concluded that at least 50% of people need to be immune before herd immunity is reached, Gomes and colleagues calculate the threshold at 10% to 20%.[22][23]

Ulrich Keil, professor emeritus of epidemiology from the University of Münster in Germany, says the notion of randomly distributed immunity is a “very naive assumption” that ignores the large disparities in health in populations and “also ignores completely that social conditions might be more important than the virus itself.” He added, “Tuberculosis here is the best example. We all know that the immune system is very much dependent on the living conditions of a person, and this depends very much on education and social conditions.”

Another group led by Sunetra Gupta at the University of Oxford has arrived at similar conclusions of lower herd immunity thresholds by considering the issue of pre-existing immunity in the population. When a population has people with pre-existing immunity, as the T cell studies may be indicating is the case, the herd immunity threshold based on an R0 of 2.5 can be reduced from 60% of a population getting infected right down to 10%, depending on the quantity and distribution of pre-existing immunity among people, Gupta’s group calculated.[24]

But memory T cells are known for their ability to affect the clinical severity and susceptibility to future infection,[25] and the T cell studies documenting pre-existing reactivity to SARS-CoV-2 in 20-50% of people suggest that antibodies are not the full story.

“Maybe we were a little naive to take measurements such as serology testing to look at how many people were infected with the virus,” the Karolinska Institute immunologist Marcus Buggert told The BMJ. “Maybe there is more immunity out there.”

The research offers a powerful reminder that very little in immunology is cut and dried. Physiological responses may have fewer sharp distinctions than in the popular imagination: exposure does not necessarily lead to infection, infection does not necessarily lead to disease, and disease does not necessarily produce detectable antibodies. And within the body, the roles of various immune system components are complex and interconnected. B cells produce antibodies, but B cells are regulated by T cells, and while T cells and antibodies both respond to viruses in the body, T cells do so on infected cells, whereas antibodies help prevent cells from being infected.

An unexpected twist of the curve

Buggert’s home country has been at the forefront of the herd immunity debate, with Sweden’s light touch strategy against the virus resulting in much scrutiny and scepticism.[26] The epidemic in Sweden does seem to be declining, Buggert said in August. “We have much fewer cases right now. We have around 50 people hospitalised with covid-19 in a city of two million people.” At the peak of the epidemic there were thousands of cases. Something must have happened, said Buggert, particularly considering that social distancing was “always poorly followed, and it’s only become worse.”

Understanding this “something” is a core question for Sunetra Gupta, an Oxford University epidemiologist who developed a way to calculate herd immunity thresholds that incorporates a variable for pre-existing innate resistance and cross protection.[24] Her group argues that herd immunity thresholds “may be greatly reduced if a fraction of the population is unable to transmit the virus.”

“The conventional wisdom is that lockdown occurred as the epidemic curve was rising,” Gupta explained. “So once you remove lockdown that curve should continue to rise.” But that is not happening in places like New York, London, and Stockholm. The question is why.

“If it were the case that in London the disease hadn’t disseminated too widely, and only 15% have experienced the virus [as serology tests indicate] . . . under those circumstances, if you lift lockdown, you should see an immediate and commensurate increase in cases, as we have observed in many other settings,” Gupta told The BMJ, “But that hasn’t happened. That is just a fact. The question is why.”

Possible answers are many, she says. One is that social distancing is in place, and people are keeping the spread down. Another possibility is that a lot of people are immune because of T cell responses or something else. “Whatever it is,” Gupta added, “if there is a significant fraction of the population that is not permissive to the infection, then that all makes sense, given how infectious SARS-CoV-2 is.”

Buggert’s study in Sweden seems to support this position. Investigating close family members of patients with confirmed covid-19, he found T cell responses in those who were seronegative or asymptomatic.[10] While around 60% of family members produced antibodies, 90% had T cell responses. (Other studies have reported similar results.[27]) “So many people got infected and didn’t create antibodies,” concludes Buggert.

Deeper discussion

T cell studies have received scant media attention, in contrast to research on antibodies, which seem to dominate the news (probably, says Buggert, because antibodies are easier, faster, and cheaper to study than T cells). Two recent studies reported that naturally acquired antibodies to SARS-CoV-2 begin to wane after just 2-3 months, fuelling speculation in the lay press about repeat infections.[28][29][30]

But T cell studies allow for a substantially different, more optimistic, interpretation. In the Singapore study, for example, SARS-CoV-1 reactive T cells were found in SARS patients 17 years after infection. “Our findings also raise the possibility that long lasting T cells generated after infection with related viruses may be able to protect against, or modify the pathology caused by, infection with SARS-CoV-2,”[8] the investigators wrote.

T cell studies may also help shed light on other mysteries of covid-19, such as why children have been surprisingly spared the brunt of the pandemic, why it affects people differently, and the high rate of asymptomatic infections in children and young adults.

The immunologists I spoke to agreed that T cells could be a key factor that explains why places like New York, London, and Stockholm seem to have experienced a wave of infections and no subsequent resurgence. This would be because protective levels of immunity, not measurable through serology alone but instead the result of a combination of pre-existing and newly formed immune responses, could now exist in the population, preventing an epidemic rise in new infections.

But they were all quick to note that this is speculation. Formally, the clinical implications of the pre-existing T cell reactivity remain an open question. “People say you don’t have proof, and they’re right,” says Buggert, adding that the historical blood donor specimens in his study were all anonymised, precluding longitudinal follow-up.

There is the notion that perhaps T cell responses are detrimental and predispose to more severe disease. “I don’t see that as a likely possibility,” Sette said, while emphasising that we still need to acknowledge the possibility. “It’s also possible that this absolutely makes no difference. The cross reactivity is too small or weak to affect the virus. The other outcome is that this does make a difference, that it makes you respond better.”

Weiskopf added, “Right now, I think everything is a possibility; we just don’t know. The reason we’re optimistic is we have seen with other viruses where [the T cell response] actually helps you.” One example is swine flu, where research has shown that people with pre-existing reactive T cells had clinically milder disease.[12][13][14]

Weiskopf and Sette maintain that compelling evidence could come through a properly designed prospective study that follows a cohort of people who were enrolled before exposure to SARS-CoV-2, comparing the clinical course of those with and without pre-existing T cell responses.

Understanding the protective value of pre-existing SARS-CoV-2 T cell reactivity “is identical to the situation on vaccines,” said Antonio Bertoletti, professor of infectious disease at Duke-NUS Medical School in Singapore. “Through vaccination we aim to stimulate antibodies and T cell production, and we hope that such induction of immunity will protect … but we need a phase III clinical study to really demonstrate the effect.”

German investigators came to the same conclusion, arguing that their T cell findings represented a “decisive rationale to initiate worldwide prospective studies” mapping pre-existing reactivity to clinical outcomes.[31] Other groups have called for the same thing.[6]

“At the start of the pandemic, a key mantra was that we needed the game changer of antibody data to understand who had been infected and how many were protected,” two immunologists from Imperial College London wrote in a mid-July commentary in Science Immunology. “As we have learned more about this challenging infection, it is time to admit that we really need the T cell data too.”[32]

Theoretically, the placebo arm of a covid-19 vaccine trial could provide a straightforward way to carry out such a study, by comparing the clinical outcomes of people with versus those without pre-existing T cell reactivity to SARS-CoV-2. A review by The BMJ of all primary and secondary outcome measures being studied in the two large ongoing, placebo controlled phase III trials, however, suggests that no such analysis is being done.[33][34]

Could pre-existing immunity be more protective than future vaccines? Without studying the question, we won’t know.

Acknowledgments

I thank Juan-Andres Leon and Angela Spelsberg for comments on a draft of this article.

Footnotes

• Competing interests: I am a colleague of Ulrich Keil, quoted in this article. A generic statement of competing interests may be found at https://www.bmj.com/about-bmj/editorial-staff/peter-doshi

• Provenance and peer review: Commissioned; externally peer reviewed.

This article is made freely available for use in accordance with BMJ's website terms and conditions for the duration of the covid-19 pandemic or until otherwise determined by BMJ. You may use, download and print the article for any lawful, non-commercial purpose (including text and data mining) provided that all copyright notices and trade marks are retained.

https://bmj.com/coronavirus/usage

View Abstract