Friday Report: Issue 28
By Matt Fletcher, Dan Ryan and John Roberts
COVID-19 Actuaries Response Group – Learn. Share. Educate. Influence.
COVID-19 is still one of the hottest topics for scientific papers and articles. The COVID‑19 Actuaries Response Group will provide you with a regular Friday update with a curated list of the key papers and articles that we’ve looked at recently.
Clinical and Medical News
Vaccine breaking news
This week saw encouraging announcements from 3 of the leading vaccine producers.
BioNTech / Pfizer
After interim results last week from BioNTech and Pfizer, the BNT162b2 phase 3 trial reached its final endpoint with 170 confirmed cases of COVID-19, 8 in the vaccine group and 162 in the placebo group. Attention is particularly focused on effectiveness in the older population. The study protocol identified that 41% of the 41,135 trial participants on the 2-dose regime across 6 countries were between 56-85, and that vaccine efficacy was >94% for those over age 65. Global production is expected to be 50 million vaccine doses in 2020 up to 1.3 billion doses by the end of 2021, with 2 doses needed for each person. Of these, the UK has secured 10 million this year with a further 30 million next year.
Earlier in the week, Moderna released interim results on their mRNA-1273 phase 3 trial with 95 confirmed cases. 90 of these cases occurred in the placebo group, and 5 in the vaccinated group, leading to an efficacy rate of 94.5%. All of the 11 cases of severe COVID-19 occurred in the placebo group. The UK secured 5 million doses this week for delivery from Spring 2021.
As the trials are not periodically swabbing, all cases are likely to have followed symptomatic infection of the participant or a close contact. The evidence so far is therefore focused on the ability of vaccines to prevent symptomatic infection and severe COVID-19, rather than transmission.
AstraZeneca / University of Oxford
Finally, whilst interim and final results for the AstraZeneca/University of Oxford phase 3 trial are not expected for weeks, results published in the Lancet on Thursday on the earlier phase 2 trial indicated that 208 out of 209 on the boost vaccination regimen showed neutralising titres of antibody by 14 days after vaccination.
The early announcements from BioNTech/Pfizer and Moderna reflect the rapid increases in the number of confirmed cases in the US, tripling to over 150,000 per day in the last month. At the same time, the US arm of the Astrazeneca/University of Oxford trial was on hold from September 6 to October 23 pending review by the FDA following an earlier case of transverse myelitis.
COVID-19 attacking brain astrocytes
Many studies have confirmed that COVID-19 is a systemic disease affecting many organs. A further pre-print study last month found evidence of structural and functional changes in the brains of some COVID-19 patients, even those with mild symptoms. The prime target appears to be astrocytes, which provide support to both neurones and those cells that form the protective blood-brain barrier. The structural changes are correlated with increased symptoms of anxiety and impaired cognition. A study in Nature last week identified the crucial role that astrocytes play in co-ordinating brain activity, and urged more attention and research on these “white-gloved conductors” rather than the synapses between neurones.
News on Tocilizumab from the REMAP-CAP trial
Initial findings from the REMAP-CAP trial, led by Imperial College London and ICNARC in the UK, on the use of tocilizumab (an immunosuppressant drug, used mainly for treatment of rheumatoid arthritis), suggest that those receiving the treatment were 1.87 times more likely to have better outcomes in terms of survival and not needing organ support than those whose immune system was not being modulated. Whilst this is very positive news, the trial is not expected to complete for some months and as yet there is no data as to how treatment with tocilizumab compares to other immune system modulators.
Personalised risk prediction from Medicare
A pre-print study of 534,023 US Medicare patients with a diagnosis of COVID-19 has developed a personalised risk prediction calculator to identify optimal candidates for early vaccine and therapeutic allocation (www.predictcovidrisk.com) using a random forest model. The study identified that those co-morbidities with the highest risk included sickle cell disease (OR 1.73), chronic kidney disease (OR 1.32), leukaemias and lymphomas (OR 1.22), heart failure (OR 1.19) and diabetes (OR 1.18).
Benefits of universal mask use
The IHME COVID-19 Forecasting Team has published a pre-print study that attempts to estimate the global benefits of mask use, currently assumed to be close to 60%. It concludes that the universal use of masks could prevent over 730,000 deaths over the period from October 2020 to March 2021, including 55,000 deaths in the UK. The latter represents one-third of the predicted number of deaths by the group for this period.
COVID-19 at the cellular level
COVID-19 is a war waged at the cellular level, and a pre-print study has provided an estimate of the number of SARS-CoV-2 virions in circulation. Every infected person is expected to be carrying 1-100 billion virions at the peak of infection with the greatest concentrations in the lungs and tonsils. At current rates of infection this would mean a total mass of between 100g and 1kg of virions – a sobering thought in this David vs Goliath battle that continues to rage.
Imperial College Reports
Report 35: How can we keep schools and universities open? Differentiating closures by economic sector to optimize social and economic activity while containing SARS-CoV-2 transmission (link)
This paper aims to identify the control strategies which maximise economic production given the constraints of 1) allowing schools and colleges to remain open, and 2) ensuring emergency hospital capacity is not exceeded.
To do this, an integrated economic-epidemiological model was developed, accommodating risks of infection within and between different sectors of the economy.
The paper concludes that a differentiated sectoral closure strategy over six months could lead to a GDP gain of between £163Bn and £205Bn, compared to a blanket lockdown of all non-essential services. For example, they consider that such sectors as accommodation, food services, and sports would need to be closed down at various points over autumn and winter 2020/21.
They also note that increased hospital capacity could significantly increase GDP by allowing a more open economy; this would presumably imply more hospitalisations and deaths, though these aspects are not considered in the paper.
Report 36: Modelling ICU capacity under different epidemiological scenarios of the COVID-19 pandemic in three western European countries (link)
This paper estimates spare capacity of four key ICU resources (beds, nurses, doctors, ventilators) in France, Germany and Italy over the winter period between late October 2020 and March 2021 using a dual-demand (COVID / non-COVID) patient model.
Suppression strategies of varying effectiveness are considered, triggered based on the number of COVID-19 patients surpassing ICU capacity thresholds.
Perhaps unsurprisingly, unmitigated scenarios would be expected to lead to very high patient numbers (exceeding those from the first peak) and induce capacity deficits – although even these scenarios are not expected to simultaneously produce a deficit in all resources over the projection horizon.
The paper notes that a combination of strategies (reducing the number of non-COVID patients, supply-side interventions and lockdowns) are likely to be required to ensure that ICU capacity is not breached, and that as many patients as possible (COVID and non-COVID) can be treated. The trade-offs in each potential strategy are noted, and the paper recommends “careful, continuous decision-making” will be needed by those making policy, to ensure relatively favourable outcomes across winter 2020/21.
Center for the Mathematical Modelling of Infectious Disease (CMMID) – London school of Hygiene and Tropical Medicine
Assessment of tiered restrictions and a second lockdown on COVID-19 deaths and hospitalisations in England: a modelling study (link)
This study has not yet been peer reviewed. It looks at the impact of the tiered restrictions and options for national lockdown in England, varying in terms of stringency, timing and length.
Alongside an unmitigated outbreak and tiered restrictions, the study considers lockdown measures similar to Northern Ireland and Wales, with schools open and closed. Outcomes are measured in terms of hospital pressure and cumulative deaths.
The authors conclude that an unmitigated COVID-19 epidemic could lead to over 30,000 deaths in England in the period to March 2021, and that lockdowns consistently outperform less stringent restrictions. They find that the policy adopted in England, with a 4 week national lockdown, is a reasonable compromise – either a lockdown greater than 4 weeks or school closure would be expected to reduce the total number of expected deaths somewhat, but would not be expected to reduce pressure on hospital services significantly.
Demographic and Occupational Influences in Hospital Staff
A report (link) published in the Oxford Journal of Public Health investigates the relationship between demographic factors in seropositivity of hospital staff, with a focus on ethnicity. With an 11% positivity rate across the sample of nearly 11,000 staff, it confirms that ethnic minority staff were more likely to have been infected with COVID early in the first wave than their white counterparts: the results were 12% and 21% for South Asian and Black staff respectively, compared with 9% for White staff.
What is less clear however is how much of this difference is due to other factors, in particular deprivation. The analysis notes that more senior and professional roles (such as doctors) are less likely to have been infected, and suggests that a combination of working conditions (less patient contact), staff social interaction (less sharing of breakroom space) and external factors (more affluent living conditions) are relevant.
ONS Infectivity Update
The latest survey published today (link) covers the first week after the second lockdown was introduced in England and shows overall infection levels level at around 1-80 of the community population. New infections have fallen slightly. Having bounced around 50,000 per day for three weeks, they are now put at 39,000. These random samples are taken throughout the week, so we would not have expected a large fall to be showing yet – next week should be much more revealing as to the effect the lockdown is having.
There are significant regional variations though, with northern areas showing clear falls, whereas there is a much more mixed position in the south. The stringency of tiering restrictions immediately prior to the lockdown may well be influencing these results.
Also encouraging are signs of a reduction in infectivity at older ages. In contrast, school age children have shown increases of late. It may be that the half term dampened infection levels here, and we are seeing a bounce back with the return after half term. Schools of course remain open during the second lockdown.
Elsewhere, both Wales (shown below) and Northern Ireland are recording clear falls, as a result of the stringent restrictions that both devolved administrations put in place in the second half of October. The ONS describes Scotland as levelling off – figures suggest that there may be a slight fall, but it is much less clear than the trends seen in Wales and Northern Ireland.
Figure: Welsh Infectivity Levels
The SAGE estimate of R has fallen slightly this week, from a range of (1.0 -1.2) to (1.0 – 1.1) for the UK. The range for England is also (1.0 – 1.1). There are regional variations and most notably the NW estimate is now below 1 at (0.8 – 1.0), consistent with the ONS survey discussed above.
By their nature these estimates are backward looking, as they rely on evidence such as tests and admissions data, which follow the actual infections by some days. Accordingly we shouldn’t be too concerned yet that these numbers haven’t fallen below 1 more consistently following the lockdown on 5 November. Hopefully next week will see a further reduction as the effect becomes apparent.
The SAGE estimate continues to be consistent with our own view based on hospital admissions. This is now showing a steady fall back towards 1 following a brief surge last week which commentators are widely attributing to a burst in activity immediately prior to the lockdown. We update this view on Twitter every Tue/Thu/Sat in the early evening.
En passant …
There appears to be increased interest in the ancient game of chess during the pandemic. The increase has also been linked to the success of the new Netflix series “The Queen’s Gambit”, which traces the ascent of an orphaned girl to the heights of the chess world.
Chess site lichess.org reports a significant increase in the maximum number of concurrent players since the release of the series; this follows a jump of almost 40% in the number of monthly games played between February and April this year.
It’s worth noting that one of our authors (Matt) is a keen chess player – if you have a (free) account, you can skirmish with him on lichess. His handle is “wmiltti” – let us know how it goes!
20 November 2020