Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- clinical neurology
- conversion disorder
- functional neurological disorder
- Guillain-Barre syndrome
The scientific community rapidly responded to the COVID-19 pandemic by developing novel SARS-CoV-2 vaccines (table 1). As of early June 2021, an estimated 2 billion doses have been administered worldwide.1 Neurological adverse events following immunisation (AEFI), such as cerebral venous sinus thrombosis and demyelinating episodes, have been reported. In some countries, these have led to the temporary halting of both vaccine trials and roll-out programmes. In the absence of clear evidence of causal associations between the vaccine and adverse events, or the rarity of the AEFIs themselves, programmes have thus far been restarted, although sometimes with modifications to recommendations.2
Transient influenza-like symptoms such as headache, myalgia and fatigue have been reported in up to 5% of SARS-CoV-2 vaccine recipients in clinical trials,3 4 although these symptoms often indicate an appropriate immune response to vaccination.5 More severe potential adverse effects in the open-label phase of vaccine roll-outs are being collected through national surveillance systems. In the USA, roughly 372 adverse events have been reported per million doses, which is a lower rate than expected based on the clinical trials.6
In the UK, adverse events are reported via the Coronavirus Yellow Card reporting website. As of early June 2021, approximately 250 000 Yellow Cards have been submitted, equating to around three to seven Yellow Cards per 1000 doses.7 For comparison, the 2010 Pandemrix vaccination had a rate of around 0.6 Yellow Card submissions per 1000 doses.8 Increased pharmacovigilance surrounding SARS-CoV-2 vaccinations may lead to surveillance bias,9 safety alerts may lead to notoriety bias10 and recall bias may also occur. Nevertheless, with large numbers of people being vaccinated, often with a two-dose schedule, rare complications of vaccination may be seen more frequently, and judicious assessment of causality is recommended in each case.
Widespread vaccine take-up is necessary to establish the required herd immunity to stop or slow the pandemic and to prevent the emergence of vaccine-resistant strains,11 as well as to decrease the neurological and neuropsychiatric complications that are associated with COVID-19 itself.12 13 As vaccine uptake is directly influenced by public confidence in the safety of the vaccine, it is critical that those reporting potential neurological and/or psychiatric complications apply stringent methodological approaches to assigning a causal association between the two, as has been undertaken for COVID-19 itself.14
Here we present a summary of what is currently known about the neurological and neuropsychiatric adverse effects of the currently available SARS-CoV-2 vaccines. We review evidence from previous vaccination campaigns and discuss the complexities. Finally, we propose system for AEFI causality assessments.
Neurological adverse events in previous vaccination campaigns
Previous programmes have raised the possibility of adverse neuropsychiatric events to vaccination. Autoimmune encephalitis following influenza and Japanese encephalitis vaccination have been documented,15 as has the likely autoimmune phenomenon of childhood-onset narcolepsy following both influenza H1N1 infection16 and the associated Pandemrix vaccination,17 which in 2010 was thought to lead to a small increase in narcolepsy rates in several countries.18 As well as the active immunogenic substance, vaccine excipients (eg, sodium taurodeoxycholate and thimerosal) and the manufacturing process have been potentially implicated in some vaccine reactions, such as febrile convulsions in children administered the inactivated influenza vaccination in 2010.19 Neurological events attributed to vaccines have also previously led to their cessation from use. A 1976 H1N1 influenza vaccine was withdrawn from use due to a small increase of Guillain-Barré syndrome (GBS) cases (leading to around an additional one case of GBS for every 100 000 vaccines).20 Similarly, an intranasal influenza vaccine was also withdrawn in Switzerland in 2001 as it was estimated to have contributed to an extra 13 cases of Bell’s palsy per 10 000 vaccinations.21
It is also recognised that there have been acute neurological responses in previous vaccination campaigns which, although precipitated by the vaccination procedure, are not directly related to the vaccine constituents. For example, human papillomavirus vaccination (HPV) vaccinations in Brazil have precipitated functional (non-epileptic) seizures or attacks (a subtype of functional neurological disorder (FND)),22 which have also been reported following the H1N1 vaccination in Taiwan23 and South Korea.24 The WHO Global Advisory Committee on Vaccine Safety and Immunisations (GACVS) working group of the WHO recognise such ‘immunisation stress related responses’ (ISRRs) as a disqualifier in their tool for assessing causality of an adverse event following immunisation, which include (but are not limited to) acute stress responses, vasovagal reactions and dissociative symptoms.25 Many of these ISRRs can be viewed as FND triggered by vaccines. Indeed, FND is commonly precipitated by, and can emerge from, a range of initial ‘organic’ symptoms that can be due to physical injury or illness. In these cases, vaccination due to the commonly encountered local and systemic reactions is sufficient to precipitate FND.26
COVID-19 vaccine and neurological manifestations
At least 16 SARS-CoV-2 vaccines have been approved for use by at least one national regulatory authority27 (table 1), and as of early June 2021, there are around a further 400 vaccines in development, with 102 in the clinical phase.27 Adverse events in the clinical trials—captured using the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use Medical Dictionary for Regulatory Activities, under system organ classifications such as central nervous system disorders and psychiatric disorders28–indicated limited reports of serious neurological adverse events.3 4 While phase III trials are powered to detect common adverse reactions, association with rare but significant events, such as neurological complications, may not be detected until phase IV or open-label studies. Commonly cited exemplars of this include cases of GBS in association with the 1976/77 US swine influenza vaccination programme29 and Bell’s palsy in association with inactivated intranasal influenza vaccine,21 as discussed previously.
Consistent with this, only isolated cases of neurological events were reported in clinical trials of major SARS-CoV-2 vaccines. Three cases of transverse myelitis were reported (2/12 174 in the vaccine group and 1/11 879 in the control arm) in phase III trial of ChAdOx1 vaccine, which led to a temporary halting of the study. Following evaluation, the cases were deemed unlikely to be related to the vaccine and the study continued.30 Medical litigation cases have also been raised, including an anonymous trial volunteer in India who sued the manufacturers for ‘acute encephalopathy’ occurring 10 days after ChAdOx1 vaccine administration. In this case, the national medical authorities concluded there was no causal link.31
Few cases of Bell’s palsy were reported in the BNT162b2 (Pfizer-BioNTech) trial (four in the vaccine arm and none in the placebo arm; each arm had >20 000 participants)3 and the mRNA-1273 (Moderna) trial (three in the vaccine arm and one in the placebo arm; each arm had >15 000 participants).4 Despite uneven distribution in the vaccine versus placebo arms, such numbers were in keeping with the expected rates based on the baseline prevalence of Bell’s palsy (20-40/100 000 annually).32 Aside from this, the BNT162b2 and mRNA-1237 trials reported no other notable patterns between treatment groups for other neurological or neuroinflammatory events.3 4
Numerical imbalances were also reported for some adverse neurological events in the phase III study of the Ad26.COV2.S (Johnson & Johnson/Janssen) of over 43 000 individuals (of whom >21 000 received the vaccine).33 In the vaccination group, there were four events of seizures (vs one in placebo group) and six events of tinnitus (vs 0). One case of GBS was also seen in the vaccine arm. There were also 11 thromboembolic events in the vaccine arm (compared with three in the placebo group), including one case of cerebral venous sinus thrombosis with cerebral haemorrhage. This case was associated with thrombocytopaenia, disseminated intravascular coagulation and later was also found to have developed antibodies to platelet factor 4. The trail was paused and subsequently restarted following an external review as causality could not be determined at the time.34 Causality has not been confirmed for other events, as underlying medical conditions may have predisposed the affected individuals. Detailed postapproval surveillance of these events was advised.33
The phase III trial of the Sputnik vaccine did not feature any significant neurological adverse events in over 16 000 participants in the vaccine group.35 Full peer-review publications of some other vaccine phase III trials, with an estimate of neurological adverse events, are currently awaited.
That rare, but significant, adverse events may only be detected following large-scale, open-label monitoring is exemplified by the emergence of vaccine-induced thrombosis with thrombocytopaenia (VITT). VITT resembles heparin-induced autoimmune thrombocytopaenia, manifesting with low platelet count and extensive thromboses. VITT has been reported to arise due to the development of antibodies to platelet factor 4, typically 4–28 days after the first dose of an adenovirus vector based vaccine.36 The most common neurological manifestation of VITT is cerebral venous sinus thrombosis (CVST), but arterial stroke is also a recognised complication.37
Concerns about the association of VITT with SARS-CoV-2 vaccination were first raised in March 2021,38 and cases reported in the literature followed shortly afterwards.39 40 The unique clinical features of the syndrome and the consistent temporal relationship with the vaccine led to the confirmation of its causative role. In response to this, many countries, including Germany, USA, Norway, Canada, the UK and others, suspended or amended the target age ranges of the implicated vaccines: ChAdOx1 and Ad26.CoV2.S.41 42 As of 2 June 2021, there have been 372 cases of VITT reported to the UK Medicines and Healthcare products Regulatory Agency (MHRA), of which 135 had CVST.7 This translates to an overall incidence of 14.2 cases per every million doses administered. An epidemiological study of Danish and Norwegian adults reported a higher incidence of CVST post-ChAdOx1 vaccine with an excess of 2.5 events for every 100 000 vaccines administered.43
Neurological adverse events are tracked by national regulatory bodies worldwide. This includes the UK MHRA’s Yellow Card system as well as the US Centers for Disease Control and Prevention’s Vaccine Adverse Event Reporting System. Both agencies run rapid cycle analysis as well as comparison of observed versus expected rates to allow early safety signal detection. Several hundred Yellow Card reports for Bell’s palsy, seizures, GBS and transverse myelitis have been submitted following almost 65 million doses of COVID-19 vaccinations administered in the UK.7 So far, these numbers do not exceed the natural expected case rates, although reports continue to be closely monitored. Similarly, there are only individual case reports of neurological adverse events following COVID-19 vaccines in the literature; these include GBS,44 Bell’s palsy,45 acute disseminated encephalomyelitis (ADEM),46 transverse myelitis,47 delirium48 and seizures.49 The authors of these articles warn against attributing causality based solely on the temporal association.
Several videos of neurological and neuropsychiatric postvaccination symptoms and signs have been shared on social media, many of which have been widely shared and in some cases picked up by news channels.50 In some of these widely circulated videos, from countries such as the USA, objective clinical features suggestive of FND have been identified, although clinical evaluation would be required to confirm this.51 Furthermore, 64 anxiety-related adverse events (such as syncope, sweating or dizziness) were reported in clusters following Ad26.CoV2.S vaccination in the USA.52 All were classified as non-serious, but the authors highlighted that the importance of recognising such events to provide reassurance to the affected individuals and to other members of public attending vaccination centres.
Considerations for causality assessments
There are several tools for assessing causality for postvaccination sequelae, including the WHO’s GACVS causality assessment tool.25 In table 2, we correlate the GACVS criteria with the well-known Bradford Hill criteria for causality14 and discuss the implications for establishing causality in neurological adverse events associated with COVID-19 vaccination, including FND.
Before proceeding to any causality assessment, the adverse event diagnosis must be first validated. This can be done using the Brighton Collaboration guidance, which provides definitions and levels of diagnostic certainty for neurological conditions such as GBS and Miller-Fisher syndrome, transverse myelitis, Bell’s palsy, encephalitis and ADEM as well as generalised convulsive seizures.53 Nevertheless, there are two principal limitations to using well-established case definitions. First, some recognised but atypical variants will not meet the diagnostic criteria, precluding causality assessment (an example of this is facial diplegia with paraesthesia variant of GBS).54 In these cases, clinicians may agree on working definitions for such condition to include them in the surveillance sensitivity analysis.55 Second, no case definitions are available for new and emergent adverse events, such as VITT. In the case of VITT, the consistent clinical presentation, specific laboratory features (such as thrombocytopaenia and positive platelet factor 4 antibodies), as well as the objective radiological findings allowed the scientific community to rapidly establish diagnostic criteria.37 40 Case definitions are being refined in larger cohort studies to reflect clinical variants not meeting the original definitions (eg, up to 5% of patients with VITT may have normal platelet count at presentation).56 Establishing case definitions is, however, a particular challenge for adverse events with variable presentations and without specific diagnostic tests, such as FND.
Once the clinical case definition is ascertained, one can proceed to the causality assessment. Using questions summarised in table 2, the WHO algorithm for AEFI groups events into five categories: consistent with causal association, indeterminate, inconsistent with causal association, unclassifiable (insufficient information) or unsuitable for causality assessment (not meeting clinical case definitions). Events are classified as inconsistent with causal association if there is a strong suspicion of alternative causes and as consistent with causal association if there is no alternative aetiology and previous evidence suggests association with the vaccination.25 A consequence of this is that for any new vaccine, all adverse events will be initially classified as ‘indeterminate’, until further evidence becomes available. This significantly limits the use of these criteria for assessment of adverse events following novel vaccines, such as large-scale programmes for SARS-CoV-2.
Instead, we propose classifying cases as: probable, possible and unlikely, considering the temporal relationship, individual risk factors and the likelihood of an alternative aetiology (table 3). This approach may support clinicians in assessing individual patients where the evidence from the literature is sparse, but it should not replace the statutory reporting to regulatory bodies.
High-quality randomised controlled trials may often be one of the strongest means by which to assess causality in AEFIs; however, in situations such as the ongoing SARS-CoV-2 vaccination campaign, undertaking such trials postlicensing may not be feasible or ethical.57 Instead, alternative methods of establishing causality, such as triangulation, may be used. Triangulation, a process akin to the Bradford Hill criterion of consistency, involves the integration of results from different epidemiological methods (eg, RCT and observational cohort studies) with different inherent biases. If two or more methods point towards a potentially causal event (particularly if the association is maintained among differing populations), this strengthens conviction in causal association.58 In addition to this, although a mechanism (the Bradford Hill criterion of ‘biological plausibility’) is often preferred in causality assessments, it may not be immediately apparent, and causality should not be refuted simply on the absence of biological mechanisms, particularly should the other evidence for causality be strong.
Although associations between SARS-CoV-2 vaccines and neurological adverse events may arise, these tools reiterate that caution must be exercised before causality is determined. Alternative aetiologies or recurrence of previous neurological disease may be responsible for the observed events, as was the case in some of the neurological AE associated with the ChAdOx1 vaccine.30 Furthermore, the baseline rate of a condition may account for its occurrence in vaccinated population.59
Additional attributes related to the current SARS-CoV-2 vaccination programmes that may need to be considered when assessing causality include (but are not limited to): (A) the two-dose nature of some vaccine programmes, (B) the possibility of some patients, for example, due to vaccination supply changes, receiving two separate brands of vaccine in a single dose schedule and (C) the individual’s prior exposure to SARS-CoV-2 (ie, antigen formation and/or development of neurological complications of COVID-19 itself). These factors will also remain important if in future a seasonal SARS-CoV-2 ‘booster’ vaccination is required.
While causality is under intense study, a practical issue for clinicians is advising on SARS-CoV-2 vaccine uptake in patients with a history of disabling or life-threatening GBS, acute disseminated encephalomyelitis or transverse myelitis, which are thought to have been linked to previous vaccination. Bayesian inference may suggest updating the probability of neurological complication based on an individual’s prior history. However, as there are no confirmed signals that these conditions occur with SARS-CoV-2 vaccines, cautious advice must be given. Other factors that may help shared decision making include the population prevalence of COVID-19, the individual’s risk of mortality from COVID-19 and patient preference.
In the current rapid worldwide vaccination campaign, healthcare regulatory authorities have tended to either halt vaccination campaigns or recommend their use only in those with risk factors for COVID-19 in response to potential adverse events.38 41 42 The latter appears to be a sensible course of action while causality is swiftly reviewed, given the high rates of known complications of COVID-19 itself.
The potential for neurological AEFIs (including FND) to SARS-CoV-2 vaccines raises important medicolegal questions. An effective and fair compensation scheme for AEFIs is integral to maintaining public trust in vaccination campaigns.60 In the UK, for example, vaccine recipients (including of a SARS-CoV-2 vaccine) are entitled to claim compensation for ‘severe disability’ resulting from a vaccination.61 Other methods of seeking compensation, including indemnification in clinical trials of vaccines and direct legal action against vaccine manufacturers, may also be applicable in the SARS-CoV-2 vaccination campaign. We are not aware of any precedent in this area regarding severe and disabling FND resulting from vaccination, but the fact that the vaccine constituents are not implicated, and significant or full recovery always remains possible, complicates an already complex field, in which 65% of historical claimants to the UK scheme have been unable to definitively prove causality.60
FND reactions to vaccination
As we have seen, an expected response to the vaccination programme, particularly in younger people, may be acute FND. Patients with FND may present with a range of neurological symptoms including, but not limited to, seizures, sensory abnormalities, gait or balance disturbance and weakness. Symptoms arise as a disorder of neurological function as opposed to structural lesions, and often mimic patients’ expectations of illness models rather than symptoms seen in organic neurological disorders. These factors can lead to misinterpretations of such symptoms as being under the patient’s control and fuel both stigma and rejection of the diagnosis.26 In addition to this, healthcare workers are often unaware of, or lack confidence eliciting, the ‘positive’ clinical features that can reliably distinguish it from other neurological disorders.26
Our understanding of FND has developed dramatically in the last decade, and current models of the disorder challenge stress or psychological processes as the sole aetiological factor. Stress is not always present or relevant, and its primacy in older ‘psychogenic’ or ‘conversion’ models of FND has been replaced by more nuanced biopsychosocial models that, while acknowledging the important role of stress, do not assume its relevance for all. These models focus on cognitive and neurobiological processes underpinning symptom formation and persistence. Alongside the reduced primacy of psychological factors has been the acknowledgement of the importance of positive neurological signs such as Hoover’s sign, which has a specificity of at least 95% in functional leg weakness.26 It is recognised that a compassionate, thorough and open explanation of the disorder improves understanding and acceptance of the diagnosis and is a critical first step to successful treatment.26
In many cases, FND is precipitated by physical disorders, sometimes relatively ‘minor’ injuries or accidents. In the case of vaccination, it is plausible that physiological reactions (eg, vasovagal symptoms or influenza-like symptoms), as well as unilateral pain from the injection site, may trigger and/or evolve into functional symptoms.62 Indeed, FND has been reported in response to previous vaccination campaigns, and case reports of FND precipitated by SARS-CoV-2 vaccination have been published.63 Despite this, clinicians must continue to be just as judicious in attributing causality of SARS-CoV-2 vaccines to FNDs as with any other neurological adverse event.
The capacity for functional symptoms to develop in response to vaccination may be increased due to wider factors of the global SARS-CoV-2 vaccination campaign. Patients with FND have been shown to have heightened suggestibility,64 which is currently understood in a Bayesian framework to reflect a tendency to form overprecise ‘priors’.65 Attentional focus on potential neurological side effects, particularly in the context of extensive media interest, may result in somatic hypervigilance for specific symptoms in some individuals, which in turn is likely to contribute to an increased risk of FND.51 Conceptually, this is supported by the improvement of cases of FND in response to previous vaccination campaigns once media coverage of the events ceased.66 Furthermore, the capacity of social media to facilitate the dissemination of information, as well as misinformation and disinformation, may contribute to the worldwide spread of functional reactions to vaccines.66
The influence of media coverage and public perception of vaccine reactions on susceptibility to FND offers a potential opportunity for healthcare professionals to reduce the incidence of expected vaccine-related FND. The better informed potential vaccine recipients and the wider public are about the potential for FND in response to SARS-CoV-2 vaccines, the better the outcomes are likely to be for those who develop FND in response to vaccination.67 Additionally, clinicians and wider health authorities are implored to strike a balance between thorough investigation and reassurance in patients in whom a functional reaction is identified. This is likely to prove challenging in some situations, particularly given the role of social media; however, sensitive and informed handling of situations is likely to lead to better outcomes for patients and the wider vaccination campaign.
Occasionally, functional symptoms can ‘spread’ between individuals as ‘mass psychogenic (functional) illnesses’ after vaccination. These clusters of functional illnesses occur most often within cohesive social groups such as schools or workplaces and can arise in many settings in response to stimuli that are perceived as noxious. There were cases of functional disorders following the H1N1 vaccination in Taiwan23 and South Korea.24 In both countries, these predominantly affected schoolchildren, and in some cases in Taiwan, symptoms of functional dizziness and weakness spread in clusters, all of which resolved without medical intervention. This phenomenon has been echoed in multiple other countries and in response to different vaccines.68 In situations in which functional illnesses have arisen contemporaneously with vaccination programmes, misunderstanding of causality and association has led to significant increases in vaccine hesitancy; in Colombia, following a large-scale outbreak of functional symptoms, HPV vaccine course uptake decreased from 88% to just 5%.69
Inappropriate recognition and management of FND may be disruptive to vaccination programmes. To this end, vaccine providers should have training on recognising functional disorders that should be clearly distinguished from other AEs, such as allergic reactions to the vaccine. Although symptoms of FND are disabling and in some cases can be chronic, patients can be reassured that FND does not equate to any structural damage to the CNS. In the absence of this clarity, vaccine-associated FND has the capacity to contribute significantly to vaccine hesitancy.70
Neurological and neuropsychiatric adverse events have been reported in clinical trials of various SARS-CoV-2 vaccines and, more often, in open-label monitoring. In many cases, no definitive evidence has yet supported causality. Despite this, there are recognised rare adverse events that have been causally linked to SARS-CoV-2 vaccines, for example, VITT. The necessary ongoing surveillance work is in progress; however, the current advice that the benefit of the vaccination outweighs the risk appears to be accurate from a neurological standpoint. Heightened reporting of adverse events, for example, via the Yellow Card system, may be in part due to increased pharmacovigilance.
In order to establish causality, clinical case definitions must be established, for example, via the Brighton collaboration guidelines for conditions recognised to be associated with vaccination and proactive clinician-led definitions in emergent conditions (such as VITT). In assessing causality, tools such as the WHO GACVS or Bradford Hill criteria may be used; however, we additionally propose criteria that classify associated neurological or neuropsychiatric events into probable, possible and unlikely cases, considering the temporal relationship, individual risk factors and the likelihood of an alternative aetiology. In such cases as the urgent SARS-CoV-2 vaccination campaign in which ongoing RCTs may be unfeasible and/or unethical, epidemiological methods of causality assessment such as triangulation may be used.
Regardless of current surveillance of neurological adverse events, it is increasingly likely that a significant minority of vaccine recipients will develop both organic and/or functional reactions to vaccine administration, which have already been noted in previous vaccination campaigns, as well as on videos and in case series during the COVID-19 pandemic. FND reactions are real and distressing, although require different management strategies, and lack of public understanding of FND reactions to vaccination may significantly increase the risk of vaccine hesitancy. Particular features of the SARS-CoV-2 vaccination campaign, including the role of social media in spreading (mis)information on vaccination side effects, as well as intense media interest, may mean individuals are further predisposed to FND reactions. Despite this, these avenues also offer clinicians opportunity to better inform potential vaccine recipients and in doing so potentially reduce the incidence of acute FND reactions.
It is important that those working in the clinical neurosciences are abreast of the evidence for and against potential associations, including background rates and the potential for FND manifestations to both care for our patients and to not hamper ongoing vaccination efforts. At the individual level, clinicians are encouraged to have collaborative discussions with patients on the potential for neurological and neuropsychiatric complications of SARS-CoV-2 vaccinations.
Patient consent for publication
MB and AT are joint first authors.
BDM and TRN are joint senior authors.
Twitter @mattbutlerpsych, @Tim_R_Nicholson
Contributors MB and AT produced the first draft. MB, AT, GKW, ME, RHT, IG, SP, TS, TAP, BDM and TRN significantly contributed to the first draft. BS contributed additional content post peer review. MB, AT, GKW, ME, RHT, IG, SP, BS, TS and TAP all reviewed final manuscript. BDM and TRN provided supervisory input and approved final manuscript.
Funding MB is an National Institute for Health Research (NIHR) Academic Clinical Fellow (ACF-2019-17-008). TAP is supported by a NIHR Clinical Lectureship. TS is supported by the NIHR Health Protection Research Unit in Emerging and Zoonotic Infections (Grant Nos. IS-HPU-1112-10117 and NIHR200907), NIHR Programme Grant for Applied Research (No. RP-PG-0108-10,048), NIHR Global Health Research Group on Brain Infections (No. 17/63/110) and the European Union's Horizon 2020 research and innovation program ZikaPLAN (Preparedness Latin America Network), grant agreement No. 734584. BDM is supported to conduct COVID-19 neuroscience research by the UKRI/MRC (MR/V03605X/1); for additional neurological inflammation research due to viral infection, BDM is also supported by grants from the MRC/UKRI (MR/V007181//1), MRC (MR/T028750/1) and Wellcome (ISSF201902/3).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.