Based on data from the first wave of Covid-19, epidemic modellers in Zurich estimate Switzerland would have suffered 22,000 deaths without the restrictions on daily life introduced in March 2020.
The modelling, done by Fadoua Balabdaoui and Dirk Mohr at the Swiss Federal Institute of Technology in Zurich (ETHZ), estimates nearly 75% of Switzerland’s population would be infected by now – see chart a on page 8.
One of the difficulties decision makers face when responding to pandemics is the prevention paradox, something first formally described in 1981 by the epidemiologist Geoffrey Rose. Success almost guarantees some degree of criticism. If measures to reduce viral transmission are successful some will claim that everything was fine and there was no need to act. Essentially, there is no do-nothing version of reality to prove the critics wrong.
However, Balabdaoui’s and Mohr’s model presents a do-nothing scenario that attempts to show what would have happened without the measures. It also attempts to quantify the impact of the various measures that were deployed to slow the spread.
Reproduction numbers (R0) measure viral spread. An R0 of 1 means one infected person infects one other. Estimates of the R0 for SARS-CoV-2 range from 2.2 – 6.6, but much of the research underpinning these estimates has not been peer reviewed.
This model estimates the reproduction number of SARS-CoV-2 in Switzerland decreased from 5.0 to 2.4 due to social distancing and then further from 2.4 to 0.7 due to the stay home policy and the decline in human contact at other locations. If these measures were applied strictly until the end of 2020, the model predicts a total death toll in Switzerland of 2,300. Under a second wave scenario it predicts around 7,000 deaths. With none of the restrictions introduced in March it predicts the total death toll would have been around 22,000.
Any model that attempts to model something as complex as the spread of a virus will produce uncertain results. The extent to which the structure of models and the data entered into them reflects reality is uncertain and unavoidable. This model is particularly sensitive to assumptions on the reproduction number, social contact patterns, hospitalisation numbers and age group infection rates. Any results should be interpreted with caution.
For example, under a scenario with no anti-spread measures the model assumes a relatively low infection rate among those over 65, a high risk group. It assumes that roughly 80% of those 65 and under are infected but only 50% of those over 65 are. This assumption means those over 65 are 38% less likely to be infected than those under 65.
These assumptions are based on infection rates during Switzerland’s first wave. The model authors state that older age-groups in Switzerland appear to be protected by the contact patterns among age groups in Switzerland.
Infection percentages for this older group from a large antibody study in Spain contrast with the infection rates used in the model. In the Spanish study those over 65 were more heavily infected than those 65 and under. Those in the older group were 27% more likely to be infected than those under 65 rather than less likely.
If a Swiss do-nothing scenario was modelled based on infection data from the Spanish study, which reflect different group contact patterns from a different country and culture, the hypothetical number of deaths would be higher. Given that 95% of Covid-19 related deaths in Switzerland were among those over 65, a do-nothing scenario based on the Spanish infection rates might produce a hypothetical death estimate closer to 40,000.
Is such a scenario plausible? Possibly. Possibly not. It partly depends on why infections among the older age segment of Switzerland’s population were lower and whether it would be maintained in a high-spread high-infection scenario.
In any case, given the inherent uncertainty of epidemic modelling these modelling results will never be a substitute for what would have actually happened if no anti-spread measures had been introduced. Too much is unknown and reality is too complex.
Those proposing or deciding on anti-spread measures can probably never avoid criticism.
Research paper (in English)