this week MOONSHOT Q&A page 338 • ROLE OF GPs page 340 • COVID JARGON page 343

£100bn Moonshot test scheme revealed

The government has drawn up plans to carry lockdown before a vaccine, something the at a press out 10 million covid-19 tests a day by early country cannot aff ord,” it says. conference last week after next year as part of a £100bn expansion of Martin McKee, professor of European was leaked to The BMJ its national testing programme, documents at the London School of leaked to The BMJ show. Hygiene and Tropical Medicine, said, “The Funding for Operation Moonshot will plan disregards the enormous problems almost match what is spent on the NHS in with the existing testing and tracing England each year (£130bn) to create a programmes.” He added, “It focuses on only mass testing regime “to support economic one part of the problem, testing, and says activity and a return to normal life.” The nothing about what will happen to those current capacity is 350 000 tests a day. found positive, a particular concern given The leaked documents reveal a heavy the low proportion of those who isolate. reliance on the private sector to achieve the “What parliamentary scrutiny will there goal and give details of “letters of comfort” be of a programme that would cost almost LATEST ONLINE that have been signed with companies to as much as the annual budget for the NHS?” reach three million tests a day by December. , professor of global public Covid-19: Doctors Critics are concerned that the programme health at Edinburgh University, said, “’m working outside ignores the current problems with testing concerned about the reliance on the private their expertise are and appears to have been devised with little sector. There is a case for giving the extra unlikely to face input from scientists or local health experts. billions to the NHS and asking it to deliver.” GMC charges Under the plan, testing will be rolled A Department of Health and Social Reports from out to workplaces, entertainment venues, Care spokesman said, “We are increasing the BMA’s GP surgeries, pharmacies, and schools to capacity to 500 000 tests a day by the end of virtual annual representative improve access. Digital immunity passports October, and the ability to get rapid, on-the- meeting will also be launched to allow people who spot results will signifi cantly increase our test negative to return to work and travel. ability to stop the spread, and for our NHS funds medicinal The memo says that implementing mass economy to recover.” cannabis for fi rst testing is a “top priority” for Boris Johnson. Gareth Iacobucci, Rebecca Coombes, The BMJ time since the law “This is described by the prime minister as Cite this as: BMJ 2020;370:m3520 changed in 2018 our only hope for avoiding a second national  NEWS BRIEFING p 338 the bmj | 19 September 2020 337 THE MOONSHOT DOCUMENTS

Operation Moonshot Memos prompt questions over cost, evidence, and the private sector’s role

The £100bn plan to carry out 10 million covid-19 tests a day by early next year as part of an expanded national testing programme raises many concerns. Gareth Iacobucci examines the leaked documents

What is Operation Moonshot? it can walk. Chris Ham, former chief Martin McKee, professor of It refers to the government’s plan executive of the King’s Fund, said, “I European public health at the to deliver a mass population testing wish the government were much more London School of Hygiene and programme for covid-19 by early focused on getting our current system Tropical Medicine, said, “This is 2021, with the aim to test the whole working eff ectively, particularly given yet another standalone initiative, UK population each week. the increases in positive cases, because developed without any apparent A confi dential briefi ng memo sent we are at a very critical moment now.” involvement of those on the ground I wish the to Scotland’s fi rst minister and cabinet or acknowledgement of existing secretaries, seen by The BMJ, reveals How did they arrive at £100bn? structures. Worse, it envisages a government plans to grow the UK’s testing capacity The documents do not explain major role for Deloitte, a company were much from the current 350 000 tests a day to how the £100bn cost was calculated. that has presided over many of these more focused up to 10 million by early 2021, costing Ministers have refused to be drawn problems.” on getting “over £100bn to deliver.” since The BMJ fi rst published the Since the start of the pandemic our current The plans are dependent on a huge details on 9 September but have not Deloitte has played a prominent role system working upscaling in diagnostic capacity denied that it was used in internal in building and managing logistics at effectively and the use of as yet unvalidated correspondence. To put the number drive-through testing facilities and in Chris Ham technologies (p 340). They envisage a in context, £100bn is almost a year’s introducing home testing. But it faced prominent role for the private sector in budget for the NHS in England planning and delivery. (£130bn) and more than double the “STRATEGIC PARTNERS” UK’s annual defence budget (£41bn). Supply What is the desired outcome? The documents say that the • GSK The government’s aim is to government’s Scientifi c Advisory • Smith & Nephew “utilise the full range of testing Group for Emergencies (SAGE) is • Thermofi sher approaches and technologies to help modelling the potential eff ect of a mass • QuantuMDx reduce the R rate, keep the economy testing programme on the R number, • Optigene open and enable a return to normal while the Treasury is doing the same Digital life.” The Whitehall memo makes it for the economy. • X-Lab Systems clear that the prime minister views Nigel Edwards, chief executive of the • Ordnance Survey mass testing as the UK’s “only hope for Nuffi eld Trust, said he was stunned by • NHS Login avoiding a second national lockdown the fi gure being quoted, which he said • Gov.Verify before a vaccine, something the could fund 150 new NHS hospitals. “It • Equifax/Experian This is yet country cannot aff ord.” seems an astonishingly large amount Logistics and warehousing another Many suspect that the desire for of money to be talked about without • Boots and Sainsbury’s standalone positive headlines is also driving this a proper assessment of the costs and • DHL initiative, particular agenda. benefi ts,” he said. • Kuehne+Nagel developed • G4S without any What’s new about the strategy? Who is leading the programme? • Workforce apparent The current advice is that only A presentation prepared by the people with covid symptoms should consultancy fi rm Boston Consulting • Universities involvement have a test. Moonshot moves way Group, seen by The BMJ, reveals a • Society of Microbiologists of those on beyond this by planning to test the heavy reliance on the accountancy and • British Society of Immunology the ground whole population “where testing consultancy fi rm Deloitte to manage Laboratories Martin McKee enables economic or other vital activity and oversee the plans, alongside other • AstraZeneca and builds societal confi dence.” fi rms such as PA Consulting. Channels and distribution With the current test and trace The presentation is packed with • Engineering & Logistics Staff system struggling to operate eff ectively, jargon heavy fl ow diagrams and Corps (British Army) and people still being asked to travel metaphorical references such as • CILT (Chartered Institute of Logistics and Transport) hundreds of kilometres to get a “Moonshot headquarters” and • CIM (Chartered Institute of standard polymerase chain reaction “Mission HQ.” The lack of clinical and Marketing) (PCR) swab test, there are concerns system-wide input has drawn criticism • Job Centres that the government is running before from experts.

338 19 September 2020 | the bmj pace sounded “deeply implausible.” this backfi ring.” He said, “Even if you He said, “If half of the cost is on testing have a test which is 99% specifi c, so and the other half is on staff (as in only 1% of people get a false positive the NHS), we could be talking about result, if you then test 60 million employing 1.6 million people.” people we will be classifying a group the size of the population of Sheffi eld What are other countries doing? as wrongly having covid.” If these China and South Korea initially people and their close contacts had to deployed testing on a wide scale to isolate, this, Deeks noted, would create try to reduce the spread of covid-19, “substantial economic harm and MALCOLM WILLETT MALCOLM and the US and India are now testing massive need for further testing.” criticism in April after reports that widely. Denmark, with a population of GPs were left without access to many six million, announced on 18 May that What does SAGE say? thousands of their patients’ covid-19 it would off er mass testing, including In a consensus statement test results carried out at its centres. to people without symptoms. France dated 31 August the advisers struck has also tried population testing. But a cautious note. It said, “Careful How will capacity be scaled up? none of these countries has rolled out consideration should be given The documents show the plan is testing on the same whole population to ensure that any mass testing to eff ectively build a new diagnostics scale as the UK is proposing. programme provides additional industry to hugely increase capacity. Devi Sridhar, professor and chair of benefi t over investing equivalent They indicate that “letters of comfort global public health at the University resources into improving the speed are in place with companies to reach of Edinburgh, said a commitment to and coverage of NHS Test and Trace for 3 million tests per day by December.” wider testing can cause problems if symptomatic cases.” Potential partners listed include capacity becomes stretched. “We don’t It added that “mass testing can drug companies such as GSK on the want to make the same mistakes as only lead to decreased transmission if supply side and AstraZeneca for France, where there is broader testing individuals with a positive test rapidly increasing laboratory capacity. Boots, being off ered and it is being used by undertake eff ective isolation.” Sainsbury’s, DHL, Kuehne+Nagel, lots of ‘worried well,’ leading to queues G4S, and Serco are earmarked for and delays in results,” she said. Is the plan achievable? logistics and warehousing (box, left). Experts are highly sceptical. Some experts have expressed alarm at Is there evidence for mass testing? Edwards said, “My strong suspicion is the sums of public money that could Experts fear Moonshot has had this will end up like the airport on the be channelled into private fi rms. little involvement from scientists, island or the garden bridge [promised , a former clinicians, or experts in screening. Jon by Johnson when he was London director of maternal and child health Deeks, professor of biostatistics at the mayor but not delivered]. Quite a lot at the World Health Organization and leader of eff ort will be expended, but I will and professor at University College of the Cochrane Collaboration’s covid- be quite surprised if it survives a more London, described the plans as 19 test evaluation activities, said, rigorous look at the economics.” “waste/corruption on a cosmic scale.” “The document lacks insight into how McKee was equally sceptical, screening works, particularly the need saying the plans bore the hallmark of How will it be staffed? to balance the harms you can create a government “whose ambition far The documents state These plans through false positives against the exceeds its ability to deliver.” that the government will work show waste on benefi ts from true positives.” Gareth Iacobucci, The BMJ with universities, the Society of a cosmic scale The plans rely on as yet unvalidated Cite this as: BMJ 2020;370:m3580 Microbiologists, and the British Society Anthony Costello tests, and Deeks warned of a “risk of  PERSONAL VIEW, p 358 of Immunology to grow the diagnostic workforce. There are also plans to off er Moonshot’s “mission analysis” from the leaked documents covid-19 tests at general practices and Mission Objective Test setting Geography pharmacies to boost access, as part Mass test general population to contain spread Pilot LAMP test in Greater Train stations; additional of what the government calls a “huge and build societal confidence (see p 340) settings to be agreed new operational infrastructure.” Test asymptomatic NHS employees regularly to Deliver a pilot for asymptomatic Hospitals; laboratories , ensure NHS can operate effectively testing in NHS and other (Lighthouse, NHS) Basingstoke, The documents say, “This will settings; build body of evidence Manchester include a new warehousing and an Test target institutions (such as schools and a) Lateral flow testing Schools or universities (TBC) UK expanded logistics network, and a new universities) regularly using a risk based model to b) Cohort pooling Schools (TBC) UK enable safe return to normal life workforce with the appropriate skills c) ePCR TBC UK and expertise to deliver, administer Develop an agile and comprehensive testing Develop use of mobile and Closed institutions; open UK capability, beyond current approach, to identify quick turnaround technologies institutions; community and process our new testing and contain outbreaks gatherings and local community technology in every corner of the UK.” Enable private sector organisations to deliver TBC Employers; point of entry to UK But there is little detail, and Edwards testing safely and effectively, including to facilitate venues said that recruitment on this scale and economic activity or protect their workforces

the bmj | 19 September 2020 339 THE MOONSHOT DOCUMENTS

sequencing. However, there is not much publicly available evidence for the use of these tests on SARS-CoV-2. What is accessible is mainly from non- peer reviewed preprints of research carried out by the manufacturers. For example, in the case of RT-PCR a preprint paper from the company DnaNudge claimed that its point-of-care test, which involves nasopharyngeal swabs, had an average sensitivity of 94.4% (95% confidence interval 86% to 98%) and an overall specificity of 100%. It concluded that the test was “specific and rapid” and could detect SARS- CLIVE BRUNSKILL/GETTYIMAGES CoV-2 “without laboratory handling or sample pre-processing.” Operation Moonshot Plan relies A study assessing LamPORE (also available as a preprint), carried out on technology that does not exist by its manufacturer, Oxford Nanopore Technologies, claimed that the test was “rapid, sensitive, and highly elivering mass testing way we should be tackling something There is not scalable.” It reported that results could on the scale and level of when people are dying right now: much publicly be obtained from 12 samples in about ambition set by prime thinking about things we have not available an hour, when starting with extracted minister Boris Johnson got. We should be thinking about the evidence for RNA, and from 96 samples in less than D will probably require things we have got and we know work. the use of two hours. It also said the test correctly “testing technology that currently Backing a horse that hasn’t yet been identified 79 of 80 positive samples. does not exist,” say leaked documents born is a really bad strategy.” these tests on However, Deeks pointed to revealed by The BMJ. The Moonshot documents list several SARS-CoV-2 inaccuracies in the documents The Operation Moonshot plans, tests being considered for rollout concerning the tests. He said, “I which could see the government across the UK, despite some having have spoken to the person doing the spend over £100bn to ensure lower sensitivity than polymerase chain DnaNudge evaluation, and he said 10 million covid-19 tests a day, show reaction (PCR) tests. These would be they have not done it on saliva—they it’s likely that new testing technology used “for screening/enabling purposes, have only done it on swab [samples]. would need to be developed, with PCR used to confirm positive So why has that list got saliva on it?” validated, procured, and made results or in situations where accuracy is Also, Deeks said that, despite operational within months to meet the needed for highest risk individuals,” say LamPORE being listed as taking 90 early 2021 deadline. the documents. minutes, it actually needs 6.5 hours to Jon Deeks, professor of biostatistics run. “The 90 minutes is the middle bit. at the University of Birmingham and What do we know about the tests? That does not include plating up the leader of the Cochrane Collaboration’s The listed tests are reverse samples, doing the DNA extraction, covid-19 test evaluation activities, has transcriptase (RT) PCR, Endpoint PCR, and things like that. So from sample to described the plan as a “nice dream.” LamPORE, Direct LAMP, lateral flow result it’s about 6.5 hours.” He told The BMJ , “This is not the antigen tests, and whole genome He added, “No publicly available data for a lot of the tests [are available] Potential technologies for expanding testing as listed in Moonshot documents . . . It’s not good science. If these were Technology Time from Potential Key characteristics drugs, the government would have had test to result sample to register these studies on the Clinical qrt-PCR 2-24 hours Swab High sensitivity (>90%); high cost (£40 per test) Saliva Trials Register, with the protocol, and Endpoint PCR <24 hours Swab Low cost (£2-5 per test); high sensitivity (>90%); to publish the results.” Saliva high volume LamPORE 90 minutes Swab High sensitivity; moderate cost; near-patient options Potential harms Saliva Deeks’s main concern was that the Direct LAMP 20-60 minutes Swab Higher cost (£10-20 per test); rapid results; “whole of this programme has been Saliva lower sensitivity (80-100%) Lateral flow antigen tests 10-30 minutes Saliva Higher cost (£5-30 per test); rapid results; built without thinking about the harms (Swab) lower sensitivity (80-100%) it could do.” Whole genome sequencing 2-24 hours Swab Higher cost (£20 per test) He said, “The mathematical laws Saliva as to what happens when you start

340 19 September 2020 | the bmj WHAT IS HAPPENING IN GREATER GP surgeries could be used to MANCHESTER? improve access to covid-19 tests The Moonshot documents emphasise that trialling new models for delivering General practices and be better spent on NHS CLIVE BRUNSKILL/GETTYIMAGES testing in local areas will be “critical.” This includes a large pharmacies could be public health and primary trial in Salford, , where the prevalence of used to make covid-19 care services.” covid-19 is currently higher than in most of the UK. testing more available However, Jane The Salford City Council website says that the trial involves to the public, according Wilcock, a GP in Salford, weekly testing of the saliva of people who do not have to the leaked Operation pointed out that not covid-19 to “identify any positive cases early and allow Moonshot documents . all GP premises could those who know they are coronavirus free to go about their normal lives.” They state that a safely off er testing. “We We see patients “huge new operational have been enormously The test being used is the LAMP test, and though the who find it hard infrastructure” would be successful at not passing council has said it will be rolled out in phases across the to get a test area, it gives no dates for when this will happen. needed to process results on covid-19 to patients in and to “develop novel the community, and we because they A statement on the website says, “It is proposed that methods of allowing don’t want this to change. don’t have a car there will be an initial two-week period to test the ‘proof Jackie Applebee of concept.’ This would enable us to test end-to-end the people to access testing.” But it is essential there reliability of the different elements of what is a complex T his would include the is local testing available practical process. The ‘proof of concept’ testing will take use of “familiar locations” where it is possible to have place with smaller groups of our community and enable such as GP surgeries and a separate pyrexia room,” us to ensure each of these elements is working well before pharmacies, as well as she said. then rolling out a Salford-wide programme.” other local venues. “Also, when you’re “A new workforce not well it is not always screening mass populations are not with the appropriate possible to travel a long in the right direction in terms of the skills and expertise to way for a test, so having harms that will be done. The key thing deliver, administer and this available locally is specificity, and the documents process our new testing would be benefi cial.” When you’re do not mention specificity or false technology in every Primary care networks, not well it is not positives, which is how you get harm. corner of the UK” would which Oxford GP Joe always possible If you start using tests in 60 million also be needed, said the McManners admitted to travel a long people, even if they are 99% specific Backing a horse documents. had fallen off the radar way for a test you will end up giving false positives to Several GPs told The during the pandemic, Jane Wilcock hundreds of thousands of people.” that hasn’t yet BMJ they were keen to have been set up partly to Deeks warned that using less been born is be included in plans to integrate health systems reliable tests to enable people a really bad control the pandemic. and should, he said, be to attend events such as football strategy Jackie Applebee, a GP in well placed to enable GPs matches would be “dangerous,” as it Jon Deeks London, said local public to work together to decide could give people false confidence and health services, including which general practices see them ease protective measures general practices, were in a local area could act as such as . frustrated at being testing sites. However, he was positive about excluded from testing, Using these networks expanding testing, provided that it was tracing, and supporting would ensure all done correctly. “Some of these tests patients aff ected by covid. England’s localities had Primary care have potential to improve how well “In general practice we an accessible testing networks we can do test and trace. If they can don’t know which patients site, assuming staffi ng should be improve the capacity and speed, that have had covid-19 and funding issues were well placed to would be very useful. We just need to be because we are not aware resolved, said Azeem enable GPs to very clear as to where they are going to of test results, and we see Majeed, head of primary work together go and what their purpose is,” he said. patients who fi nd it hard care at Imperial College Joe McManners Elisabeth Mahase, The BMJ to get a test because they London. Improving access Cite this as: BMJ 2020;370:m3585 don’t have a car,” she said. to tests will be particularly “Local health bodies important in the winter, have felt excluded when, when other viruses are “IF YOU USE tests in 60 million in fact, local systems are circulating, said Majeed. best placed to get on top “A saliva test would be a people, even if they are 99% specific of the pandemic. The big step forward,” he said. you will end up giving false positive results to money going to Serco, Elisabeth Mahase , The BMJ hundreds of thousands of people” Jon Deeks Deloitte, and others would Cite this as: BMJ 2020;370:m3552 the bmj | 19 September 2020 341 SEVEN DAYS IN Scotland launches app, with England and Wales to follow Scotland has launched a contact tracing app for covid-19 that is based on the same Apple and Google toolkit used across Ireland since July. A day aft er the ’s announcement, the Department of Health and Social Care for England said an app will be available in England and Wales on 24 September. Its launch was delayed while NHSX, the NHS’s technology unit, tried to develop a system in which anonymised data could be held on an NHS database. The Protect Scotland app was downloaded nearly 600 000 times in the two days aft er its launch. First minister (left ) said, “The more people who download and use the app, the more eff ective it can be in helping to make connections that may otherwise have been missed. This will allow people to self-isolate quickly if they are exposed to the virus.” All the apps use Bluetooth to alert users if they have been in contact with someone who has tested positive and advises them to self-isolate. They do not store identity details or location but use encrypted codes between smartphones to tell users they have spent at least 15 minutes within 2 m of someone who has tested positive.

Jacqui Wise, London Cite this as: BMJ 2020;370:m3566

Covid-19 is being rationed in the US. training junior doctors while they Drug companies vow Doctors and politicians called are working at their hospitals not to rush vaccine on the Trump administration to to help tackle the NHS backlog The heads of AstraZeneca, use existing laws to increase caused by the pandemic. The BioNTech, GlaxoSmithKline, supplies of the drug by allowing NHS has been using operating Johnson & Johnson, Merck, other companies to make generic theatres and clinical facilities Moderna, Novavax, Pfizer, and versions of the drug and to permit in sector to Sanofi, which are all working imports of generic versions from tackle the backlog of elective towards a covid-19 vaccine, foreign manufacturers. procedures caused by covid-19. in Thorax . Workers from ethnic issued a joint statement Cliff Shearman, vice president of minority backgrounds were nearly promising to “only submit London hospital “must the Royal College of Surgeons of twice as likely to have already had [a vaccine] for approval or improve infection control” England, said, “It’s only right that the infection as white staff. emergency use authorisation Hillingdon Hospital NHS NHS funded treatment should after demonstrating safety Foundation Trust, which was hit help train the NHS workforce of Lucentis and efficacy through a phase 3 by a covid-19 outbreak in July the future.” Companies are fined £412m clinical study” and “to always that required 70 staff to self- for unfair marketing make the safety and well-being isolate, has been ordered to take Infection risk is lowest France’s Competition Authority of vaccinated individuals our stringent measures to control in intensive care staff fined the Swiss drug companies top priority.” The statement infection. An investigation found A study of 545 staff who worked at Novartis and Roche €444m followed a letter from seven drug that a nurse with covid-19 had University Hospitals Birmingham (£412m) for abusing their industry CEOs last week urging unwittingly infected 16 others NHS Foundation Trust at the dominant market position to that “political considerations during a training session on 30 height of the pandemic found steer eye doctors towards the wet should be put aside” in covid-19 June, described by one doctor as that cleaners (34%; 10/29) macular degeneration treatment drug and vaccine development a “super-spreading event.” The were most likely to test positive Lucentis (ranibizumab), which and that clinical data should be Care Quality Commission, which for antibodies to SARS-CoV-2, they sell jointly, at the expense publicly disclosed. carried out an unannounced followed by clinicians working in of the cancer drug Avastin inspection on 4-5 August, has acute medicine (33%; 10/30) (bevacizumab), which is often US “should end Gilead’s used its urgent enforcement or general internal medicine used off label in its place and monopoly” on remdesivir powers to place conditions (30%; 30/99). The lowest costs 30 times less. Novartis, The US government should end on the trust’s registration to seroprevalence was seen which was found to have the shortage of the antiviral protect patients and staff. among staff working in “unjustifiably exaggerated” drug remdesivir by eliminating intensive care (15%; 9/61), Avastin’s risks, will pay about Gilead’s monopoly, said a report Private hospitals emergency medicine 85% of the fine and plans to from Public Citizen, a consumer commit to (13%; 2/15), and appeal. Roche said that it was interest pressure group. training juniors general surgery studying its options. In 2014, Remdesivir, an unapproved Private healthcare (13%; 3/23), Italian authorities fined the two investigational drug used in providers in findings companies €180m over similar patients with severe covid-19, committed to reported allegations.

342 19 September 2020 | the bmj SIXTY SECONDS MEDICINE ON . . . JARGON

Public health Karsten, a Brazilian brewer, adaptedd IT’S EVERYWHERE Food and drink companies its logo during the covid crisis You’re telling me. Terms that used to be the “exploit pandemic” preserve of scientists are suddenly being Companies trading in alcohol, bandied around by politicians on television, tobacco, junk foods, gambling, brought up over a coff ee with friends, or infant milk formula, and fossil appearing in the family WhatsApp group. fuels are “leveraging” the to the R number and coronavirus crisis to boost contact tracing have all entered the general their brands, often to the lexicon since SARS-CoV-2 emerged. detriment of public health and BUT THEY MAKE ME SOUND CLEVER sustainability goals, showed Uhm. While it’s good that everyone is research from the NCD Alliance involved in the debate, and some terms can and the SPECTRUM consortium be helpful, problems arise when they aren’t of researchers. For example, in without a prescription because used correctly or properly explained. Brazil the brewer Karsten adapted cuts to services are making it MMR its logo to resemble a pair of harder for women to access ARE WE NOT ALL O N THE SAME PAGE? Not always, says public health consultant lungs with the slogan, “Good contraception, said the All UPTAKE Angela Raff le. False positive and negative beer is like air: you can’t live Party Parliamentary Group on top her list of words that should be banned without it,” while encouraging Sexual and Reproductive Health. Counts of vaccinations in from general use. “They mean diff erent three steps to survive: “Isolate, Postpartum contraception in all things to diff erent people. When talking with use sanitiser, and drink beer maternity settings should also children aged the general public we cannot be sure they for fun.” The research authors be funded, it said, and a national 12 to 18 months have the same shared understanding.” called for a tough response from digital contraception service in England were governments and regulators. should be developed to help CAN YOU PUT THAT IN CONTEXT? women get the contraceptives Exactly. Raff les says that when these terms almost 20% are used without any context it creates E-cigarettes are safer they need. Such a service confusion. For example, a false positive can but not without risks would “protect contraceptive lower in the first mean someone who tests positive for covid- Electronic cigarettes are provision in the event of another few weeks after 19, but what they have actually got are viral significantly less harmful than ‘lockdown,’ even out inequalities lockdown than in fragments from an infection they had a long tobacco but are not risk-free, in remote access , and streamline the same period time ago, meaning they’re not infectious or concluded the Committee on care pathways for women,” the in 2019, although a new case. However, to someone working Toxicity of Chemicals in Food, group’s report advised. they have now in a laboratory a false positive would mean Consumer Products, and the returned to near a test that is done incorrectly and therefore Environment. If smokers switched Climate crisis indicates the virus is present when it is not. completely to e-cigarettes harm “Include emission targets” normal levels would be cut, but some risks in economic recovery IT’S TESTING OUR PATIENTS would fall more than others, Government action to help (−3.3%) You can say that again. Pillar 1 and pillar 2 testing are more examples of confusing its report said. For example, the economy recover from the [Public Health terms. They were fi rst used to loosely a “considerable reduction pandemic should also aim England] separate out hospital based (pillar 1) and in risk of lung cancer would to achieve net zero carbon community based (pillar 2) testing. “It’s be anticipated,” but the risk emissions, said the Climate jargon. I want to know which are the tests reduction would be much less for Assembly UK, which comprises done diagnostically because other conditions such as asthma. 108 members of the public someone had symptoms, which commissioned by six House of are done as safety checks before Contraception Commons select committees. operations, which are done for Make some pills available The assembly would most like the employment purposes, and over counter, say MPs government to limit investment which are done because they are The progestogen-only pill (POP) in high carbon industries, rethink contacts of cases. When you are should be made available over and invest in infrastructure, muddling all those together it the counter in pharmacies support low carbon industries, does not give you the information you need,” make the most of opportunities argues Raff le. created by moving to net zero SO, FEW OF US REALLY KNOW WHAT emissions, and deal with covid- WE’RE TALKING ABOUT? 19 and the climate crisis together That’s right. But try telling that to politicians. where possible. Elisabeth Mahase , The BMJ Cite this as: BMJ 2020;370:m3564 Cite this as: BMJ 2020;370:m3567

the bmj | 19 September 2020 343 Edmond Adedeji Saad Al-Dubbaisi Krishan Gopal Arora Medhat Sobhy Atalla

Abdul Chowdhury Mohinder Dhatt Amged El’Hawrani Sadeq Elhowsh

Kamlesh Kumar Masson Karamat Ullah Mirza Poornima Nair Yusuf Patel

Jitendra Rathod Manjeet Singh Riyat Alfa Saadu Anton Sebastianpillai

Furqan Siddiqi Erwin Spannagl Adil El Tayar Peter Khin Tun

344 19 September 2020 | the bmj THE BIG PICTURE Remembering the UK doctors who have died from covid-19

Fayaz Ayache

Many healthcare workers have lost their lives to covid-19 in the line of duty. The BMA has been collecting the names of doctors in the UK who were reported to have died while working during the pandemic, and The BMJ has created a memorial page to honour their lives (bmj.com/covid-memorial). The list highlights the devastating toll on doctors from ethnic minority backgrounds, including many migrant Syed Zishan Haider Thaung Htaik Nasir Khan workers on whom the NHS depends. Fiona Godlee, The BMJ’s editor in chief, said, “The web page honours doctors who have lost their lives working for the good of others under the most diffi cult of circumstances in this covid- 19 pandemic. Each name represents an irreplaceable gap in a family and a workplace. “No one should have to risk their lives or health because of their work, and we honour those who have paid this ultimate sacrifi ce. In doing so we Rudresh Pathak Mamoona Rana Vishna Rasiah commit to all eff orts that will bring this pandemic to an end and that will ensure the safety and wellbeing of everyone working on the front line of healthcare.” Chaand Nagpaul, chair of the BMA council, said, “The death of a fellow doctor is always tragic, but to lose so many at the hands of the virus is devastating. “We off er our profound sorrow and heartfelt condolences to the families, friends, and colleagues of these committed clinicians who cared Abdorreza Sedghi Tariq Shafi Mohamed Sami Mahmood Shousha for patients in the most challenging of times, battling against this highly infectious and deadly virus. “They are the GPs and hospital doctors who treat us when we are sick, and they are our friends and colleagues, who dedicated their lives to the pursuit of helping people get better. “We owe them our gratitude, our respect, and a pledge that we will remember them.” Juliet Dobson, editor, bmj.com Craig Wakeham David Wood Habib Zaidi

the bmj | 19 September 2020 345 EDITORIAL Are scientists underestimating seroprevalence of SARS-CoV-2? Current antibody tests fail to identify people who had mild infections

esting for severe acute Tests have Other problems with test calibration recent seroprevalence survey of 1473 respiratory coronavirus been evaluated include the eff ect of demographic residents (79% of the local population) 2 (SARS-CoV-2), which mostly on factors such as age, sex, and ethnicity in Ischgl, Austria, using a combined causes covid-19, is people who on antibody responses and hence IgG and IgA approach found SARS- 7 Tcomplex and politically experienced assay results, and the eff ect of CoV-2 antibodies in 42.4% of those sensitive. Seroprevalence studies use severe covid-19 timing, since early testing before tested, far higher than rates in previous antibodies as markers of pathogen symptoms seroconversion may result in false population based surveys of other exposure to estimate the proportion of negative results. Preliminary reports infection hotspots.19 Similarly, IgA the population that has been infected. showing rapid decline in virus specifi c antibodies were detected in 11% of Considerable variation has been IgG levels suggest that testing too late 1862 people sampled from the general observed in the results of SARS-CoV-2 may also miss cases.8 population in Luxembourg, whereas seroprevalence studies.1 A recent Test performance is also infl uenced IgG antibodies were found in only survey in Spain suggested that a by the choice of antibody. Of the FDA 1.9%.20 small fraction of the population was authorised tests, most detect only IgG Finally, mucosal and bloodborne seropositive, despite the country being and IgM antibodies, the dominant immune responses may provide severely aff ected by the virus.2 components of the bloodborne complementary information crucial Seroepidemiological studies may antibody response. But IgA also has an for accurate assessment of viral underestimate the true seroprevalence important role in the immune response exposure in both individuals and of SARS-CoV-2 for several reasons. to respiratory tract infections and populations. In a cross sectional study Accuracy demands the use of an seems immunologically relevant in of UK healthcare workers, combined assay sensitive enough to reliably covid-19, particularly in asymptomatic IgG, IgA, and IgM testing for SARS- detect antibody responses to mild people.9 10 Cov-2 spike protein in saliva samples infection across diff erent post- identifi ed 15% of participants as exposure scenarios. The selection of Look for IgA positive despite a negative serum test target antigen is critical, with recent SARS-CoV-2 enters cells by interacting result.4 data showing that the trimeric spike with host proteins expressed in In conclusion, current glycoprotein off ers superior detection the respiratory tract, cornea, and seroprevalence studies may fail to to the nucleocapsid in people with gastrointestinal tract.11 IgA is the detect people who have had mild low level antibody responses.4 Of predominant immunoglobulin covid-19. Standardised approaches are 12 the 24 serological diagnostic tests Stephen Burgess, expressed at these mucosal surfaces, required so seroprevalence estimates that the FDA initially authorised for statistician, School and IgA responses with neutralising are comparable. Specifi c consideration emergency use, six consider only of Clinical Medicine, capability are described for several should be given to the selection of the nucleocapsid, including high University of viral pathogens.9 10 13 IgA antibodies the SARS-CoV-2 antigen in diagnostic throughput tests in widespread use. specifi c to SARS-CoV-2 have now assays, calibration of assay thresholds, The nature of the pandemic means sb452@medschl. been detected in various biological the breadth of the antibody response, that tests have been evaluated mostly cam.ac.uk specimens, including serum, saliva, and the role of mucosal antibody on people who experienced severe Mark J Ponsford, and breast milk.4 14 15 responses. Application of these covid-19 symptoms.5 Recent evidence clinical Serum IgA antibody responses may principles in future seroprevalence immunologist, describes a clear link between the be detectable earlier than IgG and surveys may off er more accurate Immunodeficiency 16 17 magnitude of serological responses Centre of Wales, IgM responses and can persist for insight into the population dynamics 4 6 and severity of illness. This implies University Hospital at least 38 days in hospital patients of covid-19 and help inform 18 that unless assay performance is also Wales, Cardiff recovering from covid-19. This is epidemiological modelling strategies assessed in mild and convalescent Dipender Gill, consistent with a recent Cochrane and public health policy. cases, the threshold for a positive clinician scientist, review, which found that IgA based Cite this as: BMJ 2020;370:m3364 result may be too high, resulting in serological testing had greater St George’s, Find the full version with references at 5 missed community cases. University of London sensitivity than other methods. A http://dx.doi.org/10.1136/bmj.m3364

346 19 September 2020 | the bmj EDITORIAL UK’s record on pandemic deaths Recent changes in defi nition can’t disguise the country’s poor international ranking

he grim daily count of Almost half and Wales than the 2015-19 average people received healthcare for life covid-19 deaths in the (44%) of all (fi gure).1 This welcome respite has a threatening conditions during UK is nearing single excess deaths darker side, however, as it suggests this period.7 9 10 Such deaths 1 digits. But evidence of occurred in many of the earlier deaths had been could mount over time as the the UK’s higher overall premature, for which ONS analyses NHS struggles to cope with the T care homes 7 death toll during the fi rst wave of provide further corroboration. backlog of deferred care alongside the pandemic relative to comparable winter pressures and a possible countries is unequivocal. Hidden covid-19 deaths resurgence of covid-19. Although England had the highest excess Early indications suggest that many other countries also face these all-cause mortality rate among 23 of the excess deaths not related to challenges, the UK is less well European countries in the fi rst fi ve covid-19 since March were not the equipped to deal with them, with months of 2020 compared with result of reduced care for serious non- overstretched health and social care 2015-19, followed by Spain and covid conditions. For example, the services that historically have been Scotland, with mortality being spread sharp surge in total and non-covid under-resourced and understaff ed throughout the country in contrast to deaths followed by a return to near compared with other high the more localised patterns in Europe.2 normal levels mirrors the trajectory of performing health systems.11 12 England also had the second (after covid-19 deaths,1 whereas the eff ect Among many brutal realities Spain) highest peak of excess all of reduced healthcare would have of the pandemic has been a cause mortality and the slowest fall to been more persistent. clear amplifi cation of existing normal levels—so the longest period of Furthermore, excess non-covid socioeconomic and ethnic excess deaths. deaths occurred predominantly inequalities, both in the UK 13 -16 and Recent changes to the defi nition of among frail older adults, many in beyond.17 a covid-19 death in England (from all care homes, and included a sharp The lessons of the fi rst wave deaths after a positive test to deaths rise in deaths from dementia and ill of covid-19 must inform policy within 28 days) have reduced the UK’s defi ned conditions, suggesting that decisions for tackling any resurgence. offi cial covid-19 death toll by 16%.1 3 many such deaths were related to While controlling the pandemic is But the change doesn’t alter the UK’s undiagnosed covid-19.6 7 8 clearly a priority, it’s also imperative poor ranking among European peers. Comparisons with 2015-19 also to take the long view as many of the Excess mortality rates based on suggest a substantial and continuing risk factors for dying from covid- death certifi cation data2 circumvent “displacement” of non-covid deaths 19—such as cardiovascular disease, diff erences in how covid-19 deaths are from hospitals to private homes diabetes, obesity, and deprivation— counted and also include deaths from and, earlier in the pandemic, to care are also leading contributors to the the wider eff ects of the pandemic. homes.1 7 This could have contributed lacklustre mortality improvements The overall death rate for England Veena S Raleigh, to a greater proportion of deaths and widening inequalities prevailing from 1 January to 31 July was the senior fellow, King’s occurring in other settings with lower in the UK before the pandemic. 6 highest since 2009. The year got off Fund, London rates of testing. Cite this as: BMJ 2020;370:m3348 to a good start, with a mild infl uenza v.raleigh@ Even so, lives have undoubtedly Find the full version with references at season and almost 5000 fewer kingsfund.org.uk been lost as substantially fewer http://dx.doi.org/10.1136/bmj.m3348 deaths in England and Wales up to early March 2020 than the 2015-19 average.1 6 7 But in the ensuing fi ve months, there were over 58 000 more deaths than the 2015-19 average, of which almost 52 000 (89%) were related to covid-19.1 Almost half (44%) of all excess deaths occurred in care homes. Total deaths returned to near normal levels some weeks ago, as they did in other European countries, then they fell below normal. In the eight weeks to 7 August there were about 1700 (2%) fewer all cause deaths in England Weekly deaths in England and Wales: week ending 13 March to week ending 21 August 2021 the bmj | 19 September 2020 347 The usual “essential” of full data transparency HEAD TO HEAD before prescription should become a “nice to yes have” in this urgent, fraught emergency Should Raymond M Johnson, associate professor of medicine Unlike conventional clinical trials, adaptive (infectious diseases), Yale School of Medicine, New cascading trials are not intended to answer a doctors Haven, Connecticut [email protected] prespecifi ed question to garner approval and publication; they are intended to rapidly optimise recommend In normal circumstances, insisting on full data medical treatment during a pandemic. transparency and limiting decision making to Adaptive clinical trials are not a new idea,7 but, covid-19 published data alone is rightly paramount. But formulated through existing health bureaucracies a pandemic is far from normal, and to insist on reliant on publication and peer review, they are treatments normal practice adds delay to interventions that likely ineff ective. Public health agencies are the could cost lives. A pandemic gives us little choice antithesis of nimble and adaptable. other than unpublished manuscripts (preprints) Academic medical centres can contribute to and vaccines to guide therapeutic decision making. They pandemic responses by scrutinising primary should be used, thoughtfully. published data and preprints to design and when full Of course, physicians would prefer to perform adaptive clinical trials. Therapeutics can prescribe treatments and vaccines that have evolve during pandemics, or we can use what we data are been thoroughly tested and scrutinised in peer have and hope to do better next time. reviewed, randomly controlled trials with full not publicly data transparency. But this isn’t always possible, RECOVERY trial when we don’t know how much data collection We have one real life example already. The is enough and we lack the understanding of a RECOVERY trial is an adaptive clinical trial that available? disease to know how to interpret the fi ndings. The argues for using unpublished data to shape usual “essential” of full data transparency before medical practice during a pandemic. The trial prescription should become a “nice to have” or structure incorporates several critical features an “as much as possible” in the urgent, fraught for research in such a situation. First, it identifi ed emergency circumstances we fi nd ourselves in. a limited number of treatment arms to allow defi nitive comparisons between them. Second, Preprint data and adaptive trials it intentionally focused on clinical outcomes, Beyond surge capacity, our medical systems rather than mechanistic investigations, to reach need prepositioned, randomised, adaptive clinical effi cacy endpoints. An independent data cascading trials to evaluate treatments. Quality monitoring committee analyses the interim data preprints can identify therapeutics, or inform to identify benefi t and harm early, to “adapt” the study arms, during adaptive clinical trials. This trial as it moves forward. applies especially to repurposed or off -label RECOVERY was designed to close and add drug use where prescribing them on the basis treatment arms over time. Unpublished data were of unpublished data has a lower threshold than released as a press release (though this is not adopting newer treatments or vaccines, because without its problems), and then as a preprint,11 the treatment and its eff ects and side eff ects are to showing a mortality benefi t for steroid treatment some extent known, and the infected patients are in covid-19 patients developing hypoxemia. acutely and specifi cally at risk for harm. Because RECOVERY is a high quality trial its Hypothetically, back in March, the fi rst unpublished data should be guiding decisions iteration of a covid-19 randomised trial could now. Because pandemics are temporally and have included standard of care versus geographically dynamic, limiting decision hydroxychloroquine (HQ)2 versus lopinavir/ making to full published data adds a delay that ritonavir,3 on the basis of published SARS- adversely aff ects the design of adaptive trials and, CoV-2 in vitro and MERS case-control data. potentially, pandemic outcomes. In April, when a 150 subject randomised The critical part in using unpublished data is control trial preprint was released showing content review and ensuring as much as possible no HQ virologic or clinical benefi t,4 that preprint data contain key information—the hypothetical investigators could have closed protocol, summary data tables, and the statistical the hydroxychloroquine arm and substituted analysis used. Qualifi ed parties should review steroids based on experience in China.5 When preprints, pharmacodynamics, and toxicities lopinavir/ritonavir was subsequently found to assess biologic plausibility and risk before ineff ective,6 the lopinavir/ritonavir arm could incorporating therapeutics into adaptive trials or have closed and an alternative arm opened, practice. Thus, unpublished results can deliver such as convalescent plasma. crucial interventions without sacrifi cing integrity.

348 19 September 2020 | the bmj This is about a chain of trust that stems from knowing that judgments have With knowledge of the no been scrutinised and challenged new coronavirus less than a year old, treatment Peter Doshi, associate professor, Department But the benefi ts of transparency go beyond of Pharmaceutical Health Services Research, a truer understanding of product safety and remains fraught with University of Maryland School of Pharmacy, effi cacy: earning public trust, for a start. uncertainty. Preprint Baltimore, Maryland [email protected] Jobbing doctors and patients alike reasonably data and adaptive clinical David Healy, professor, Department of Family expect any licensed covid-19 treatment or Medicine, McMaster University, Hamilton, Ontario trials are imperfect vaccine to work as advertised. This is about a chain of trust: only open data can allow but can guide active The trust we place in licensed medicines other researchers with the ability to analyse decision making in life-or- is a strong reason for insisting on full data it to do so, generating the trust that stems death situations, says transparency and reporting, even in a from knowing that judgments have been pandemic. Few would disagree with the scrutinised and challenged. Raymond M Johnson. importance of transparency, but even during Data transparency also creates the optimal But Peter Doshi and normal times it remains a challenge—so, why environment for products—there will be many David Healy argue demand it during a pandemic? The reason is covid-19 products, to be sure—to compete on that data transparency builds the foundation the strength of their evidence base, not on the that without complete for information we can trust. Data secrecy, by strength of promotion and buzz. transparency, products contrast, creates risks too large to take. should not be endorsed The fi rst critical risk is that of an No legitimate barriers as being based on science exaggerated estimate of a product’s benefi ts Finally, it must be recognised that there are when relying on scientifi c publications no legitimate barriers to data transparency alone, not the underlying data. When during the covid-19 pandemic. Companies the underlying clinical study reports for can have little basis for claiming commercial oseltamivir were fi nally made public they confi dentiality, as most products with any revealed that the data collection on lower prospect of market entry have already been respiratory tract complications relied on guaranteed massive profi ts through advance patients’ self-reporting, which makes sense government purchases. There should also be for some outcome measures, such as pain, but no concern about patient privacy: guarantees not pneumonia. The result was a complete to patients and trial participants regarding loss of confi dence in the quality of data the privacy of their data should be honored, collected for the key performance assumption and such patient level data can and should be underpinning global stockpiling.12 duly de-identifi ed. The second critical risk is underestimating Nor should data release cost us valuable a product’s side eff ects. A year after novel time. While it does take time to vaccines were manufactured and rolled out data for sharing, the core work involves on expedited timelines to tackle the threat of de-identifi cation, and the trial specifi c 2009 H1N1 swine fl u, post-marketing reports methods can be determined in advance while of narcolepsy emerged in some Pandemrix trials are ongoing, for easy release when data vaccine recipients. But it would take a further collection is complete. seven years—and a lawsuit—to unearth Before any treatment or vaccine is made internal pharmacovigilance reports by the widely available, study protocols should be manufacturer, which had suggested that in the public domain, along with statistical problems with the vaccine’s safety had been analysis plans, clinical study reports, patient produced in real time during the pandemic.13 level data, and copies of the correspondence Copious evidence already shows that with regulators and other key stakeholders. adverse event data collected in trials are Data transparency is not a “nice to under-reported in journal publications.14 have.” Claims made without access to the Moreover, serious adverse events may data—whether appearing in peer reviewed disappear if classifi ed under rubrics such publications or in preprints without peer as “intercurrent illness” or “new medical review—are not scientifi c claims. Products histories,” which do not require serious can be marketed without access to the data, adverse event reports—as has happened in but doctors and professional societies should vaccine and treatment trials.1516 publicly state that, without complete data Only publicly available full datasets will transparency, they will refuse to endorse allow for a thorough assessment of side covid-19 products as being based on science. eff ects. Cite this as: BMJ 2020;370:m3260 the bmj | 19 September 2020 349 COVID-19 Do many people have a pre-existing immunity to covid-19? It has been commonly accepted that the human population had no protective response to SARS-CoV-2 before it emerged last year, but is that actually the case? Peter Doshi reports

ven in local areas that specimens obtained in the US CoV-2, they are hard to dismiss, have experienced some between 2015 and 2018, 50% coming from diff erent laboratories in of the greatest rises in displayed various forms of T cell diff erent continents and with several excess deaths during reactivity to SARS-CoV-2. A similar being published in Cell and Science. E the covid-19 pandemic, study that used specimens from The Researchers are also confi dent serological surveys since the peak Netherlands reported T cell reactivity that they have made solid inroads indicate that at most only around in two of 10 people who had not into ascertaining the origins of the a fi fth of people have antibodies been exposed to the virus. immune responses. to SARS-CoV-2: 23% in New York, In Germany reactive T cells were “Our hypothesis, of course, was 18% in London, 11% in Madrid. detected in a third of SARS-CoV-2 that it’s so called common cold Among the general population the seronegative healthy donors (23 of coronaviruses, because they’re numbers are substantially lower, 68). In Singapore a team analysed closely related,” said Daniela with many national surveys reporting specimens taken from people with Weiskopf, senior author of a paper in single digits. no contact or personal history in Science that confi rmed this With public health responses of SARS or covid-19; 12 of 26 hypothesis. “We have really shown around the world predicated on the specimens taken before July 2019 that this is a true immune memory assumption that the virus entered showed reactivity to SARS-CoV-2, and it is derived in part from common the human population with no as did seven of 11 from people who cold viruses.” pre-existing immunity before the were seronegative against the virus. Taken together, this growing body pandemic, serosurvey data are Reactivity was also discovered in the of research may force pandemic leading many to conclude that the UK and Sweden. planners to revisit some of their virus has, as Mike Ryan, WHO’s head Though these studies are foundational assumptions about of emergencies, put it, “a long way small and do not yet provide how to measure population to burn.” precise estimates of pre-existing susceptibility and monitor the extent In a study of donor blood immunological responses to SARS- of epidemic spread.

Swine flu deja vu In late 2009, months after the World Health Organization declared the H1N1 “swine flu” virus to be a global pandemic, Alessandro Sette (left) 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 influenza. 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.” Other studies came to the same conclusion: those with pre-existing reactive T cells had suffered less severe H1N1 disease. In addition, a study carried out during the 2009 outbreak by the US CDC reported 33% of those over 60 years old had cross-reactive antibodies to the 2009 H1N1 virus, leading the CDC to conclude “some degree of preexisting immunity” to the new H1N1 strains existed, especially among adults over age 60. The data forced a change in views at WHO and CDC, from an assumption prior to 2009 that most people “will have no immunity to the pandemic virus” to one that acknowledged “the vulnerability of a population to a pandemic virus is related in part to the level of pre-existing immunity to the virus.” But by 2020, it seems that lesson had been forgotten.

350 19 September 2020 | the bmj 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 This is a true immune memory Many people got infected and by T cells, and while T cells and derived from common cold viruses didn’t create antibodies antibodies both respond to viruses Daniela Weiskopf Marcus Buggert in the body, T cells do so on infected cells, whereas antibodies help Population immunity: nowhere near close to it,” said prevent cells from being infected. underestimated? WHO’s Ryan in late July. But memory T cells are known Unexpected twist Seroprevalence surveys measuring for their ability to aff ect the clinical of the curve antibodies have been the preferred severity and susceptibility to future method for gauging the proportion infection, and the T cell studies Buggert’s home country has been at of people in a given population documenting pre-existing reactivity the forefront of the herd immunity who have been infected by SARS- to SARS-CoV-2 in 20-50% of people debate, with Sweden’s light touch CoV-2 (and have some degree of suggest that antibodies are not the strategy against the virus resulting immunity to it). full story. in much scrutiny and scepticism. That only a minority of people, “Maybe we were a little naive to The epidemic in Sweden does even in the hardest hit areas, take measurements such as serology seem to be declining, Buggert says. display antibodies against testing to look at how many people Something must have happened, SARS-CoV-2 has led most public were infected with the virus,” the he says, particularly considering health planners to assume the Karolinska Institute immunologist that social distancing was “always pandemic is far from over. In Marcus Buggert told The BMJ. “Maybe poorly followed, and it has only New York City, where just over there is more immunity out there.” become worse.” a fi fth of people surveyed had The research off ers a powerful Understanding this “something” antibodies, the health department reminder that very little in is a core question for Sunetra concluded that “as this remains immunology is cut and dried. Gupta, an Oxford University below herd immunity thresholds, Physiological responses may have epidemiologist who developed a monitoring, testing, and contact fewer sharp distinctions than in the way to calculate herd immunity tracing remain essential public popular imagination: exposure does thresholds that incorporates a health strategies.” Whatever not necessarily lead to infection, variable for pre-existing innate that threshold number is, “we’re infection does not necessarily lead resistance and cross protection. Her

Calculating the herd immunity threshold In theory, outbreaks of contagious population, immunity is distributed Gabriela Gomes, large disparities in health among disease follow a certain trajectory. evenly and members mix at random. University of populations and “also ignores In a population lacking immunity, While vaccines may be deliverable Strathclyde completely that social conditions new infections grow rapidly. At some in a near random fashion, from might be more important than the point, an inflection in this growth the earliest days, questions were virus itself.” He adds, “Tuberculosis should occur, and the incidence will raised about the random mixing infected and thus are likely to become here is the best example. We all know begin to fall. assumption. Apart from certain immune earlier. Due to this selective that the immune system is very much The 1970s gave rise to a theory small closed populations such as immunisation by natural infection, dependent on the living conditions of that defined this inflection point as “orphanages, boarding schools or heterogeneous populations require a person and this depends very much the herd immunity threshold (HIT), companies of military recruits,” truly less infections to cross their herd on education and social conditions.” and offered a simple formula for random mixing is the exception not immunity threshold,” they wrote. Researchers led by Sunetra Gupta estimating its size: HIT = 1 − (1/R0) the rule. While most experts have taken at University of Oxford have arrived (where R0 is the disease’s basic Nearly 50 years later, Gabriela the R0 for SARS-CoV-2 (generally at similar conclusions by considering reproduction number, or the average Gomes, an infectious disease estimated between 2 and 3) and the issue of pre-existing immunity. number of secondary cases generated modeller at University of Strathclyde, concluded that at least 50% of people When a population has people by an infectious individual among is reviving concerns that the theory’s need to be immune before HIT is with a pre-existing immunity, as susceptible people). This calculation basic assumptions do not hold. Her reached, Gomes and colleagues T cell research indicates may be the has guided—and continues to guide— team says that not only do people calculate the threshold at 10 to 20%. case, the herd immunity threshold many vaccination campaigns, often not mix randomly, infections (and Ulrich Keil, professor emeritus based on a R0 of 2.5 can be reduced used to define target vaccination subsequent immunity) do not happen of epidemiology from University from 60% of a population getting levels. randomly either. “More susceptible of Muenster says the notion of infected to 10%, depending on the The formula rests on two and more connected individuals randomly distributed immunity quantity and distribution of immunity, assumptions—that, in a given have a higher propensity to be is “a very naive” that ignores the according to their calculations. the bmj | 19 September 2020 351 group argues that herd immunity Sette, an immunologist from La thresholds “may be greatly reduced Jolla Institute for Immunology if a fraction of the population is in California, while emphasising unable to transmit the virus.” that we need to acknowledge the “The conventional wisdom is that possibility. “It’s also possible that lockdown occurred as the epidemic this absolutely makes no diff erence. curve was rising,” Gupta explains, The cross reactivity is too small or “So once you remove lockdown that weak to aff ect the virus. The other curve should continue to rise.” But outcome is that this does make that is not happening in places like a diff erence, that it makes you New York, London, and Stockholm. respond better.” The question is why. Weiskopf adds, “Right now, I Possible answers are many, Gupta think everything is a possibility; says. One is that social distancing is There is a significant fraction of the population we just don’t know. The reason in place, and people are keeping the that is not permissive to the infection we’re optimistic is we have seen spread down. Another possibility Sunetra Gupta with other viruses where [the T is that a lot of people are immune cell response] actually helps you.” because of T cell responses or cells generated after infection with One example is swine fl u, where something else. “Whatever it is, if related viruses may be able to protect research has shown that people there is a signifi cant fraction of the against, or modify the pathology with pre-existing reactive T cells had population that is not permissive to caused by, infection with SARS- clinically milder disease. the infection, then that all makes CoV-2,” the investigators wrote. sense, given how infectious SARS- T cell studies may also help shed Acquiring the evidence CoV-2 is.” light on other mysteries of covid-19, Buggert’s study in Sweden such as why children have been Weiskopf and Sette say that seems to support this position. surprisingly spared the brunt of the compelling evidence could come Investigating close family members pandemic, why it aff ects people through a properly designed of patients with confi rmed covid- diff erently, and the high rate of prospective study that followed a 19, he found T cell responses in asymptomatic infections in children cohort of people who were enrolled those who were seronegative or and young adults. before exposure to SARS-CoV-2, asymptomatic. While around The immunologists I spoke to comparing the clinical course of 60% of family members produced agreed that T cells could be a key those with and without pre-existing antibodies, 90% had T cell factor that explains why places like T cell responses. responses. (Other studies have New York, London, and Stockholm Understanding the protective value reported similar results.) “So many seem to have experienced a wave of pre-existing SARS-CoV-2 T cell people got infected and didn’t create of infections and no subsequent reactivity is “identical to the situation antibodies,” concludes Buggert. resurgence. This would be because on vaccines,” says Antonio Bertoletti, protective levels of immunity, professor of infectious disease Deeper discussion not measurable through serology at Duke-NUS Medical School in alone but instead the result of a Singapore. “Through vaccination we T cell studies have received scant combination of pre-existing and aim to stimulate antibodies and T cell media attention, in contrast to newly formed immune responses, production, and we hope that such research on antibodies, which seem could now exist in the population, induction of immunity will protect . . . to dominate the news (probably, preventing an epidemic rise in new but we need a phase III clinical study says Buggert, because they are infections. to really demonstrate the eff ect.” easier, faster, and cheaper to study But they were all quick to note “At the start of the pandemic, a than T cells). Two recent studies that this is speculation. Formally, key mantra was that we needed the reported that naturally acquired the clinical implications of the game changer of antibody data to antibodies to SARS-CoV-2 begin to pre-existing T cell reactivity remain understand who had been infected wane after just 2-3 months, fuelling an open question. “People say and how many were protected,” speculation in the lay press about you don’t have proof, and they’re two immunologists from Imperial repeat infections. right,” says Buggert, adding that the College London wrote in a mid-July But T cell studies allow for historical blood donor specimens commentary in Science Immunology. a substantially diff erent, more in his study were all anonymised, “As we have learned more about this optimistic, interpretation. In the precluding longitudinal follow-up. challenging infection, it is time to Singapore study, for example, There is the notion that perhaps admit that we really need the T cell SARS-CoV-1 reactive T cells were T cell responses are detrimental data too.” found in SARS patients 17 years after and predispose to more severe Peter Doshi , associate editor, The BMJ infection. “Our fi ndings also raise disease. “I don’t see that as a likely [email protected] the possibility that long lasting T possibility,” says Alessandro Cite this as: BMJ 2020;370:m3563

352 19 September 2020 | the bmj