Note: This is an automated translation (using DeepL) of the original German article.
This article can be found, with an interactive version of the data visualization, on the news side of dwh GmbH.
The major challenge in assessing the COVID-19 situation is that much necessary information and data is not available. On the one hand, this is due to organizational reasons (among others, data are not consistently collected and merged): this concerns, for example, the linkage of data on the severity of diseases or previous illnesses to vaccination status. Second, it is due to the nature of the disease itself. For example, asymptomatic disease is only recorded if the affected person is tested (by chance) at the right time. The same is true for immunity; vaccination protects very reliably, but just not 100% of vaccinated persons. In addition, people who have been vaccinated or have recovered also lose their immunity after a certain time.
For a well-founded assessment of the situation and, above all, to be able to estimate the further dynamics, it is important to understand how many people are currently still infectious and how many are not. Together with other data, such as measures taken, data on the virus, existing test programs, but also the current weather situation and mobility, a forecast can be made. For this purpose, all available information and data are combined. With the help of the simulation model, in which all (almost) 9 million Austrians are represented as “digital doubles”, the actual situation can then be calculated realistically. As far as immunization is concerned, it becomes apparent that there are major differences between the data (recovered and vaccinated persons) and those who are actually immunized.
The diagram illustrates how these data are related using the status as of 8/20/2021. On the left side are vaccination progress and recoveries based on official data. So how many people in Austria are currently …
|modeling without dark rate based on real data 8/20/2021||modeling including dark rate based on real data 8/20/2021||
modeling of effective immunization in the population
|Calibration approximates real data* as closely as possible and takes the view of the health care system, i.e. only identified cases are represented.||In the same model, the view is taken that the modeled overall dynamics are fully known. People have at least one contact with virus and/or vaccination in the model.||Due to lack of vaccine efficacy, partial loss of immunity over time and other effects modeled “real immunization” with full protection|
Real data 8/20/2021 incl. delay of vaccination efficacy based on this modell: partially vaccinated 2.2%, fully vaccinated 53.9%, not vaccinated, not recovered 36.6%, recovered & fully vaccinated 2.1%, recovered & partially vaccinated 1.2%, recovered 4.1%, BMSGPK.
It can be seen that a good third (36.7%) of the population is neither vaccinated nor recovered. The vaccinated people are divided, for example, into partially vaccinated, fully vaccinated and recovered, and partially or fully vaccinated. Note that the time lag of vaccination is already included here, i.e. the figures are not directly comparable with the direct vaccination figures.
In the middle axis, the unconfirmed cases of the disease (dark figure) are additionally taken into account. This results in an encouraging increase in the number of people who have already developed immunity through contact with the virus and/or vaccination.
This intersection of the recorded numbers (vaccinations & recoveries) with the dark figure based on different antibody studies and published data is possible due to the model calculation on the virtual population. The “dark rate groups” are divided into people who had an unconfirmed illness and are either fully vaccinated, partially vaccinated or non-vaccinated. For example, recovered & vaccinated have particularly good protection - this group is at more than 11% with the model assumptions. Such statements - even though they have many uncertainties - help to fathom how stable immunization already is in the population. Thus, the proportion of the group not vaccinated and not recovered (red) drops to below 30%. In concrete terms, therefore, more than 70% have already achieved some form of immunity by August.
How many people are fully immunized is shown on the far right. This does not mean how many people are doubly immunized, but it is the number of people who are really protected against infection in the virtual world. To do this, the loss of immunity must be taken into account, as well as limitations on vaccine effectiveness. In the computer model, the epidemic progression in the virtual population is simulated with the data from the vaccination program, studies (including vaccination protection see e.g. http://www.dexhelpp.at/site/assets/files/2311/impfmodell_3.png or loss of immunity) or other modeling (including loss of immunity https://www.tuwien.at/tu-wien/aktuelles/news/news/covid-19-modellrechnungen-genesene-bleiben-lange-geschuetzt).
These data show the modeled actual state for the effective immunity of the Austrian population (right axis of the graph). Taking into account the loss of immunity, the non-100% effectiveness of vaccination and other effects, the proportion of individuals with full immunity is between 50% and 55%. Especially in the groups partially vaccinated, unconfirmed recovered and fully vaccinated there are many people who have no or only reduced immunity or have already lost it again - at least partially. Explanations for this are that the vaccination or recovery was already (too) long ago, that a partial vaccination only protects inadequately, and that the vaccination protection does not reach 100%.
This “retrospective” modeling thus allows us to classify where we stand. 50-55% percent are effectively immunized according to the current model status and are no longer infectious (in the simplified model world). They currently represent a human “barrier” to the simulated viruses. Even more than 70% have already had contact with a simulated virus variant or have been vaccinated. They are at least partially protected.
Based on these results, “prospective” scenarios can now be calculated together with modeled vaccination strategies. It can be seen quite stably that we have already achieved quite a bit and that already lower measures than a year ago are effective in flattening the curve. This is plausible because, positively, vaccination effects and other measures such as testing add up. In a recent estimate, we assume that as of Sept. 24, 2021, another 800,000 initial vaccinations would be sufficient in any case to keep the spread of SARS-CoV-2 in check solely through immunity (apart from sensible hygiene measures, which are also sensible for other infectious diseases). This number could even be reduced depending on how many more people are added to those who have recovered. It should be noted that especially children under 12, as long as vaccination is not possible in this group, should be further protected and tested. And: to further improve the scenarios, more antibody studies and data are important.
An update on the monthly model-based immunity estimate will follow in early October; you can find the most up to date information here. Read more about convalescents and vaccination in Mayr’s magazine (ORF) at https://tv.orf.at/program/orf2/mayrsmagaz146.html.