The UK is an outlier by allowing children to remain unvaccinated at a time when lifting of restrictions will increase covid-19 infection rates
We have heard much about how vaccination is breaking—or weakening—the link between SARS-CoV-2 and the clinical manifestations of covid-19. We consider the nature of this link from the perspective of quantitative modelling—and what it means for risks following exposure to the virus. In brief, it suggests we should shift our focus from mortality to morbidity, particularly in children.
The modelling question is dynamic causal modelling (DCM) of viral transmission, which combines conventional epidemiological models with agent-based behavioural modelling to quantify how viral spread affects behaviour (e.g., social distancing and self-isolation)—and how behaviour affects viral transmission (e.g., through fluctuating contact rates and transmissibility).
For the past six months, DCM estimates and long-term forecasts have been released on a weekly basis: everything has been unfolding largely as expected, until the past few weeks. Hitherto, DCM could explain the prevalence of infection and subsequent hospitalisations (and fatalities) with ease. However, it was unable to explain the recent rises in notification rates (new daily tests) and symptoms (as assessed by the KCL-Zoe model). In short, under a model that best explains the first and subsequent waves, there appears to be an unexplainable excess of notifications and symptoms, which are about twice the number one would anticipate, given current vaccination levels and surveys of prevalence. So, what is going on?
To date, DCM modelled vaccination in a coarse-grained fashion by precluding infection. Technically, this is known as sterilising immunity. However, there are several links in the chain from infection to fatality. The DCM features two states of infection (exposed and infected), in which one cannot, and can, transmit the virus, respectively. Similarly, there are two levels of clinical symptomatology (mild and systemic), that do not, and do, require hospitalisation, respectively. Crucially, this kind of model allows for asymptomatic infection and the possibility of having symptoms without being infectious.
This construction begs the question, where does vaccination enter the game? Vaccination could preclude (i) infection, (ii) transmission, (iii) serious (systemic) illness when symptomatic or (iv) death when seriously ill. When these links are included in the DCM, one sees evidence for an effect of vaccination at all four points in the chain. As one might anticipate, this more expressive model explains the recent increases in notification and death rates—as a response to unlocking and increased transmission risk due to the Delta variant—that does not translate into systemic illness and or fatalities. Quantitatively, current efficacy estimates—that best explain a wide-range of data—are as follows: [1]
Efficacy of preventing infection: 13.3% (CI 7.4 to 18.9)
Efficacy of preventing transmission: 84.8% (CI 82.0 to 87.2)
Efficacy of preventing serious illness when symptomatic: 61.1% (CI 58.3 to 63.6)
Efficacy of preventing death when seriously ill: 93.2% (CI 91.6 to 94.5)
The remarkable aspect of these estimates is that vaccination has a very small effect (7.4 to 18.9%) on the risk of becoming infected and subsequently symptomatic—despite providing substantial protection against severe illness and death. In terms of personal risk, this means it may be wise to consider yourself at the same risk of contracting coronavirus when doubly vaccinated as prior to vaccination—even though you are less likely to be hospitalised or die.
At the population level, this suggests that the profile of symptoms could shift away from the symptoms associated with systemic illness and towards those symptoms associated with a predominantly mucosal infection. Indeed, there is anecdotal evidence to suggest that this is the case.
While this is good news in terms of mortality, there remain three reasons for concern. First, our estimate of the efficacy of preventing serious illness when symptomatic is only 61%. This suggests that the NHS could still face a surge in hospitalisations when restrictions are lifted entirely on 19 July, given that less than 60% of the UK population will be fully vaccinated at that time. Second, vaccination in the presence of high community transmission increases the opportunity for new variants to emerge, which may have different profiles of symptoms, transmissibility, virulence, and vaccine escape.
Third, the most vulnerable populations that are unprotected by vaccination are the poorest groups, and all children under the age of 16. The US Centre for Disease Control (CDC) estimates over 22 million children aged 5-17 have been infected, with 155 deaths among children aged 0-4 (1 in 29,000 cases) and 316 deaths in the 5-17 years age group (1 in 68,000 cases). [2] In terms of childhood morbidity, in the UK, multisystem inflammatory syndrome in children occurs with a frequency of around 0.05% (1 in 2000). [3] In the US, 4018 cases have been reported to CDC with many others not counted. Perhaps most worrying is that the latest UK estimates for long covid in children aged 12-16 who experience prolonged symptoms (for at least 12 months) are 0.12% (0.06-0.17) or 1 in 830, with possible but unknown effects on developing brain structure based on recent adult studies. [4,5] Despite the low risk of myocarditis in vaccinated children, all of whom have recovered, the USA, European Union, Israel and other countries judge the benefit-risk ratio to strongly favour vaccination of older children. The UK is an outlier by allowing children to remain unvaccinated at a time when lifting of restrictions will increase infection rates.
Karl J. Friston, Scientific Director: Wellcome Centre for Human Neuroimaging. Professor, Queen Square Institute of Neurology, University College London. Honorary Consultant, The National Hospital for Neurology and Neurosurgery
Anthony Costello, UCL Institute for Global Health, University College London. Professor of Global Health and Sustainable Development.
Competing interests: Karl Friston and Anthony Costello declare they have no conflicts of interest or competing interests. Interestingly, one of us is currently suffering from covid-19, despite being doubly vaccinated.
References:
1. Specifically, 34 kinds of data made available by the Office of National Statistics and Public Health England. The data fits can be found here. The estimates of efficacy can be read as the estimates that best explain the progression of the epidemic is reflected in these data. In other words, the population is behaving ‘as if’ these efficacies were in play. For example, the efficacy of preventing infection is the effective efficacy that may reflect people ‘letting their guard down’.
2. Howard, J. COVID-19 and balancing the risks: The vaccine or the virus. Science Based Medicine,
https://sciencebasedmedicine.org/covid-19-and-balancing-the-risks-the-vaccine-or-the-virus/]
3. Flood J, Shingleton J, Bennett E, et al. Paediatric multisystem inflammatory syndrome temporally associated with SARS-CoV-2 (PIMS-TS): Prospective, national surveillance, United Kingdom and Ireland, 2020 Lancet Reg Health Eur. 2021 Apr;3:100075. doi: 10.1016/j.lanepe.2021.100075. Epub 2021 Mar 22.
4. Prevalence of ongoing symptoms following coronavirus (COVID-19) infection in the UK
Office for National Statistics, July 1 2021, Table 6. https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/datasets/alldatarelatingtoprevalenceofongoingsymptomsfollowingcoronaviruscovid19infectionintheuk
5. Douaud G, Lee S, Alfaro-Almagro F, et al. Brain imaging before and after COVID-19 in UK Biobank. MedRxiv (pre-print) doi: https://www.medrxiv.org/content/10.1101/2021.06.11.21258690v2
6. Howard, J. COVID-19 and balancing the risks: The vaccine or the virus. Science Based Medicine,
https://sciencebasedmedicine.org/covid-19-and-balancing-the-risks-the-vaccine-or-the-virus/]