‘Is group immunity realistic?’ and six other questions on epidemiology and coronavirus

20 April 2021

The Academy regularly organises webinars about coronavirus from different perspectives. The webinar ‘Epidemiology of infectious diseases and the COVID-19 pandemic’ took place on 11 March. Following the webinar, the audience had the opportunity to ask the speakers questions. Below you will find a selection of questions asked and answers given.

Question 1

How can you distinguish the fraction of the population that already has natural immunity from the R0 in a fully susceptible population? You could see that in hindsight when natural immunity is reached, but are there ways to model this early in e.g. contact matrices or clustering models?

Assessing the acquisition of natural immunity is complicated at the start of the epidemic. It can most readily be determined later on in the epidemic/pandemic if certain age groups are underrepresented among the cases. This under-representation can be a consequence of passively transferred immunity from mothers to infants or the acquisition of immunity in older populations who may have experienced (genetically similar) infections before. Acquired immunity cannot be reliably incorporated in models without first observing such signals in the data. 

Question 2

Azra Ghani showed that herd immunity is not easy to achieve even with vaccination. Is herd immunity for a respiratory virus still a realistic end point or should vaccine efforts rather focus on risk groups and the elderly like is done for respiratory infections like influenza with yearly boosters adapted to the latest variants?

Herd immunity that results in the complete elimination of the virus may be difficult to achieve. Prioritizing risk groups for severe disease is the most sensible prioritization strategy to prevent mortality and reduce the burden on clinical care. It is important to notice that also outside risk groups, there is (severe and sometimes long-lasting) disease. A dichotomization into risk and none-risk groups is therefore complicated. Moreover, higher vaccine coverage in the general population, provided the vaccine reduces transmission (which, recent data show, is the case), will also help reduce the force of infection and thus the likelihood that people will get infected. This is also achieved without achieving complete herd immunity. 

Question 3

Would a model as offered by Grote Griep Meting if continued, have helped us to correct for over- or underestimation? https://www.dutchbuttonworks.com/tag/griepmeting

These initiatives from society should be lauded (e.g. participate in the COVID Radar of LUMC). However, the collected information is to some extent determined by (varying) levels of attendance, which makes their use for modelling challenging. Their main value is in early warning of sudden events (outbreaks) or detecting worrying increasing trends at an early stage, so that timely action can be taken.

Question 4

How do you test the quality of the model in the complicated environment?

For some models, the best test-case has been to see if disease dynamics in the natural situation follow that of models if interventions are introduced. The challenge with this approach for COVID-19 is that interventions are composite interventions (many at the same time) and adherence to interventions/coverage is difficult to quantify. This makes it very difficult to validate models prospectively. Of course the models are based on biological and medical principles and findings, including findings from other diseases. Models are therefore not a shot in the dark but the best possible approximation of the situation ‘in silico’.

Question 5

Is it not intrinsically debatable to use the number of positive tests for computing R_t ?

Yes it is. Different testing strategies or variation in the degree of contact tracing efforts will influence the R_t trends. This can to some extent be corrected for, but never perfectly. Hospital data therefore provide better information, but unfortunately after a longer delay. It is best to look at both.

Question 6

There are indications of SARS-CoV-2 cases in Europe from early 2020 and even (early) December 2019, based on an retrospective analysis of samples from patients admitted with unexplained pneumonia, without epidemiological links to Wuhan. How confident are we that the very first (index) case actually arose in Wuhan and that the virus was not primarily introduced into humans recently elsewhere in (or even outside) China, with Wuhan market simply representing the first superspreader event?

It is not fully impossible, but just very unlikely. A virus with these characteristics would probably have spread rapidly in that other region as well, certainly given the initial absence of control measures.

Question 7

Is the way the spike proteins in the corona of the virus work in some way comparable to spermatozoids linking to an ova and making combining of the DNA/RNA inside the ovula possible?

Not exactly. The spike protein binds a specific receptor (ACE2) and that leads to a fusion reaction between the viral membrane and the host cell membrane. If you want to compare this to a cellular process, you should look into syncytins, and how they are involved in placenta formation. This process finds its origin in the fusion proteins of endogenous retroviruses (i.e. retroviruses that have become a part of the genome in the evolutionary history of humans and other mammals).