SARS-CoV-2 Seroprevalence in Germany

A Population-Based Sequential Study in Seven Regions

Background: Until now, information on the spread of SARS-CoV-2 infections in Germany has been based mainly on data from the public health offices. It may be assumed that these data do not include many cases of asymptomatic and mild infection.

Methods: We determined seroprevalence over the course of the pandemic in a sequential, multilocal seroprevalence study (MuSPAD). Study participants were recruited at random in seven administrative districts (Kreise) in Germany from July 2020 onward; each participant was tested at two different times 3–5 months apart. Test findings on blood samples were used to determine the missed-case rate of reported infections, the infection fatality rate (IFR), and the association between seropositivity and demographic, socio-economic, and health-related factors, as well as to evaluate the self-reported results of PCR and antigenic tests. The registration number of this study is DRKS00022335.

Results: Among non-vaccinated persons, the seroprevalence from July to December 2020 was 1.3–2.8% and rose between February and May 2021 to 4.1–13.1%. In July 2021, 35% of tested persons in Chemnitz were not vaccinated, and the seroprevalence among these persons was 32.4% (07/2021). The surveillance detection ratio (SDR), i.e., the ratio between the true number of infections estimated from seroprevalence and the actual number or reported infections, varied among the districts included in the study from 2.2 to 5.1 up to December 2020 and from 1.3 to 2.9 up to June 2021, and subsequently declined. The IFR was in the range of 0.8% to 2.4% in all regions except Magdeburg, where a value of 0.3% was calculated for November 2020. A lower educational level was associated with a higher seropositivity rate, smoking with a lower seropositivity rate. On average, 1 person was infected for every 8.5 persons in quarantine.

Conclusion: Seroprevalence was low after the first wave of the pandemic but rose markedly during the second and third waves. The missed-case rate trended downward over the course of the pandemic.

Eighteen months into the pandemic, information about the extent of SARS-CoV-2 infections in Germany is still based largely on the number of COVID-19 cases reported to health authorities. The first wave in spring 2020 was characterized by low age-standardized case fatality estimates and low excess mortality (1). The second and third wave from October 2020 to April 2021 had significant excess mortality (2). Only a small number of the many seroepidemiological studies published in Germany have assessed the general population (3); most are based on a selective sample and focus on hotspots or specific non-representative population groups (4, 5, 6, 7), and thus do not permit generalization.

Because the current estimates of SARS-CoV-2 infection activity in Germany may not include unnoticed asymptomatic or mild infections, they are not reliable (8, 9). Population-based studies measuring IgG antibodies can:

• Shed light on the number of persons with prior SARS-CoV-2 exposure independent of clinical manifestations
• Determine the infection fatality rate (IFR)
• Help to judge the effectiveness of population-based interventions
• Guide prevention and vaccination strategies

To address the lack of data on SARS-CoV-2 spread and permit better comparison with other countries, in July 2020 we established the Multilocal and Serial Prevalence Study of Antibodies against SARS-2 Coronavirus in Germany (MuSPAD) .

Methods

Design

We followed the WHO protocol (10) for SARS-CoV-2 seroprevalence studies and the STROBE statement as guideline for observational studies (11). MuSPAD is a population-based seroepidemiological study incorporating the WHO recommendations regarding study design and consisting of successive cross-sectional studies with longitudinal components (eSupplement: MuSPAD sampling periods, eSupplement Figure 1). Each follow-up comprises a new cross-sectional study. The data collection period was from July 2020 to August 2021. Residents’ registration offices drew random samples according to the age and sex (12) distribution of the respective rural district (RD) or urban district (MD) (12). Randomly sampled residents were invited by mail (Figure, eSupplement: Data acquisition).

Figure

Timeline

We set up study centers at each location (eSupplement Figure 2). Data and blood collection followed standardized operating procedures (SOPs). Invited persons booked appointments at the centers or—if symptomatic, immobile, or frail—with mobile teams. Individuals with self-reported SARS-CoV-2 infection were not excluded.

The data were acquired using the Prospective Monitoring and Management app (PIA) (13) (eSupplement: Data acquisition).

Sample preparation and analysis

Blood sampling was performed using barcoded serum-gel Monovettes. After centrifugation, samples were stored at 4–8 °C until analysis. Spike S1-specific IgGs were measured using SARS-CoV-2 IgG ELISA (ELISA, enzyme-linked immunosorbent assay) from Euroimmun.

Data analysis

We derived crude and—in the interests of generalizability—weighted prevalence with 95% confidence intervals (14) for each district. Weighting was calculated by determining the proportion of participants and the number of residents in each age and sex group in the general population and dividing by the proportion in the sample. We provide estimates of seropositivity with uncertainties accounting for test performance (sensitivity 88.3%, specificity 99.2%) as in (4) by applying Bayesian hierarchical models (15) (eSupplement: Test performance).

We analyzed the influence of comorbidities, housing situation, and work-related factors in multivariable logistic regression analysis (variables determined beforehand). Expected infections were calculated for each site as a product of seroprevalence and inhabitants. We computed surveillance detection ratios (SDR), dividing the number of persons expected to be infected according to the calculated seroprevalence estimates by the number of persons notified as cases (up to 14 days before sampling [development of detectable antibodies takes about 2 weeks]). We calculated the number of persons who had to go into quarantine in order to ensure that one infected person was quarantined (nnq) (eSupplement: Calculation of estimates). The statistics of the Robert Koch Institute (RKI) were the source of the data for reported cases and deaths. Data up to the beginning of the sampling period at each study center started were considered. We present the IFR, calculated as the reported deaths at the time of sampling divided by the expected number of persons infected according to the seroprevalence at each site (eSupplement: Calculation of estimates). Analyses were conducted using STATA (version 14 and 16) and R version 4.0.2.

Results

Participants’ characteristics

The eFigure shows the number of persons invited, the response rate, and the number of participants included in the data set.

eFigure

Weighted seroprevalence in unvaccinated study participants

We recruited 18 638 participants in the first sampling periods in Reutlingen, Freiburg, Aachen, Osnabrück, Magdeburg, Chemnitz, and Vorpommern-Greifswald (July 2020 to August 2021) and 18 210 participants in a second sampling period (some of them follow-up participants) in the same regions, except for Vorpommern-Greifswald (October 2020 to August 2021) (eFigure), all aged 18–99 years. The response rates ranged from 15.3% to 43.9%. For the analysis presented here, we evaluated the data of 25 712 participants who had attended for the first time and had not yet been vaccinated (eFigure). Proportions of self-reported chronic conditions varied across regions and sampling periods. For example, the prevalence of diabetes ranged from 4% in Freiburg (July 2020) to 10.1% in Magdeburg (November 2020) (Table 1, eSupplement-Table 1). The proportion of daily smokers was 9.9% in Freiburg (November 2020) and 23.8% in Greifswald. Seventy-seven percent of the participants shared a household with other persons, including children in 23.5%, and 17.6% lived alone. The proportion of individuals with higher education ranged from 38% (Chemnitz) to 70.2% (Freiburg). The proportion of participants who experienced no changes in their working life varied between 61.9% in Magdeburg (November 2020) and 21% in Aachen (February 2021) and Freiburg (November 2020) (eSupplement-Table 1).

Table 1

Participants’ characteristics

Symptoms, exposure, quarantine, and testing

Fifty percent of all participants reported no symptoms after February 2020. Nine percent had experienced fever, 22.2% cough, 22.3% had complained of fatigue, and 4.6% had suffered anosmia (Table 2, eSupplement-Table 2).

Table 2

Self-reported COVID-19 symptoms, quarantine, contact with confirmed cases, and tests performed

Eleven percent reported contact with confirmed cases, and 15.4% had been in quarantine (Table 2). Household members of 25.1% of the participants and in total 28.2% of the participants themselves (n = 7256; eSupplement-Table 2) had been tested for SARS-CoV-2 (self-reported PCR or rapid antigen tests). Three hundred thirteen participants (eSupplement-Table 3) had tested positive (Euroimmun). The proportion of household members ever tested for SARS-CoV-2 varied across regions and sampling times from 7.7% in Reutlingen (July 2020) to 60.0% in Greifswald (May 2021).

Table 3

Multivariable analysis of MuSPAD participant characteristics influencing SARS-CoV-2 seropositivity

Seroprevalence estimates

In eSupplement-Table 4 we report unweighted seroprevalences. Weighted according to age- and sex-specific distribution in the general population, seroprevalences were 2.6% in Reutlingen (July 2020), 1.5% in Freiburg (August 2020), 2.3% in Aachen (September 2020), 1.3% in Osnabrück and 2.8% in Reutlingen (October 2020), and 2.4% in both Freiburg and Magdeburg (November 2020). After the vaccination campaign had started, weighted seroprevalence estimates in those not vaccinated were 5.4% in Aachen (February 2021), 13.1% in Chemnitz and 4.1% in Osnabrück (March 2021), 6.0% in Magdeburg (April 2021), and 11.6% in Greifswald (May 2021). The last district sampled was Chemnitz (second sample, July 2021); the weighted seroprevalence among the unvaccinated was 32.4% (Figure, eSupplement-Table 4).

Weighted seroprevalence estimates adjusted for test performance (eSupplement-Table 5) were 2.0% in Reutlingen (July 2020), 1.2% in Freiburg (August 2020), 2.0% in Aachen (September 2020), 1.1% in Osnabrück and 2.1% in Reutlingen (October 2020), 2.0% in both Freiburg and Magdeburg (November 2020), 5.2% in Aachen (February 2021), 14.3% in Chemnitz and 3.7% in Osnabrück (March 2021), 9.2% in Magdeburg (April 2021), 13.6% in Greifswald (May 2021), and 37.4% in Chemnitz (July 2021).

Seroprevalence was age dependent and was highest in the age group > 79 or > 59 years (eSupplement-Table 4).

Surveillance sensitivity

We express surveillance sensitivity as the ratio of the infected cases detected by serological analysis to the reported cases. The ratios (calculated on the basis of weighted seroprevalences) were 4.1 (Reutlingen, July 2020), 3.0 (Freiburg, August 2020), 4.9 (Aachen, September 2020), 2.6 (Osnabrück) and 3.5 (Reutlingen; both October 2020), 5.1 (Magdeburg) and 2.2 (Freiburg; both November 2020), 1.8 (Aachen, February 2021), 2.9 (Chemnitz) and 1.3 (Osnabrück; both March 2021), 2.8 (Magdeburg, April 2021), and 2.9 (Greifswald, May 2021) (Figure, eSupplement-Table 4). In Chemnitz the SDR was 4.5 in the unvaccinated, with a high proportion of self-reported known SARS-CoV-2 infections (17.3%) (July 2021) (Figure, eSupplement-Tables 2 and 4).

Infection fatality estimates

The IFR over all age groups were 1.3% [95% confidence interval] [1.0; 1.7] (Reutlingen, July 2020), 2.4% [1.8; 3.3] (Freiburg, August 2020), 1.0% [0.7; 1.3] (Aachen, September 2020), 1.6% [1.2; 2.1] (Osnabrück) and 1.2% [0.9; 1.6] (Reutlingen; both October 2020), 0.3% [0.2; 0.4] (Magdeburg) and 1.7% [1.1; 2.0] (Freiburg; both November 2020), 1.5 [1.3; 1.8] (Aachen, February 2021), 1.4% [1.3; 1.6] (Chemnitz) and 1.9% [1.5; 2.5] (Osnabrück; both March 2021), 1.2% [0.8; 2.0] (Magdeburg, April 2021), 1.2% [1.0; 1.4] (Greifswald, May 2021), and 0.8% [0.7; 1.0] (Chemnitz, July 2021). The IFR were highest in the oldest age group (>79) (eSupplement-Table 4).

Exposure to SARS-CoV-2 and risk of seropositivity

Among the 608 persons who reported having had a previous positive SARS-CoV-2 test result, between 81% (Freiburg/Greifswald) and 86% (Chemnitz) were seropositive in our analysis.

Fourteen percent of those ever quarantined were seropositive, in contrast to 2.1% of those who had not been quarantined. This yields an nnq of 8.5 [8.1; 8.8] persons who have to be quarantined in order to include one infected person.

In a logistic regression, participants who reported two typical COVID-19 symptoms during recent months were more likely to be seropositive (adjusted odds ratio [aOR] 3.7 [3.2; 4.3]) (eSupplement-Table 6).

Multivariable analysis

A weighted logistic regression accounting for clustering by study region revealed evidence of a higher risk of seropositivity in participants with a general secondary school certificate than in those with a university entrance level certificate (aOR 1.7 [1.2; 2.4]. The risk of a seropositive result was lower in daily smokers (aOR 0.4 [0.3; 0.7]) (Table 3).

Discussion

The findings of this study regarding seroprevalence and IFR during the course of the pandemic are in line with other available population-based reports from Germany (16). Comparable studies in Germany indicate low seroprevalences of between 0.4 and 1.4% with an under-reporting factor of 2–8 and IFR of 0.5–1.5% in studies up to November 2020 (4, 17, 18). Estimates in our study were similar among the study regions at similar times, especially up to November 2020. This is in contrast to earlier hotspot studies conducted in smaller communities (17).

Among participants sampled in Freiburg, Reutlingen, Aachen and Osnabrück up to October 2020, the majority of reported cases resulted from infections during the first wave, indicating seroprevalences below 3% and under-reporting (SDR 3–5) during this period. In study locations sampled either for the first or the second time during and after the second and third wave up to May 2021 (Reutlingen, Freiburg, Aachen, Chemnitz, Osnabrück, Vorpommern-Greifswald) an additional 2–10% of the population were infected. The comparably high seroprevalence (>30%) among the unvaccinated in Chemnitz in July 2021, with vaccination coverage > 65% in those recruited, is partly due to self-selection of participants with known infection but may also be an indicator of higher frequency of acquired infection in the unvaccinated.

Under-reporting of cases during the early months of the pandemic was higher in young adults and in 60- to 79-year-olds and decreased substantially, with SDRs between 1.3 and 2.9, after December 2020. The SDR in Chemnitz in July was probably overestimated due to the high number of unvaccinated persons with known confirmed SARS-CoV-2 infection. The tendency towards lower under-reporting during the latter phase of the pandemic might be attributable to the fact that the eligibility criteria, strategy, and capacity for PCR testing in Germany changed over time, with increases in capacity particularly after April 2020 and after December 2020 (19, 20).

In most regions the IFR at our study sites ranged from 1% to 2.4%, the difference being attributable mainly to the age structure of the infected in these locations. This is also reflected in the age-specific IFR of the study locations, which is in line with previous estimates (16). The low estimate for the overall IFR in Magdeburg (November 2020) is largely explained by the seroprevalence, which reflects the early increase in cases during the second wave, and by our cut-off point for reported deaths, which was too early to show the actual IFR. If we had instead chosen a cut-off for reported deaths at the end of December, the IFR would have been in line with the other results and above 1% (18).

Our findings suggest that a low level of education is one of the few factors that increase the risk of seropositivity. Lower educational level has been shown to raise barriers to isolation at home (21), probably because it is associated with lower socioeconomic status and certain kinds of work cannot be done at home (22).

As shown by previous studies (23), seroprevalence was lower in smokers than in non-smokers. For those infected, however, smoking is a risk factor for more severe disease (24). It is possible that this result is due to selection bias, as the proportion of smokers in our study is lower than in the German population as a whole. This means that among the smokers we recruited there may be more health-conscious individuals with a lower risk of infection with SARS-CoV-2.

The nnq of 8.5 shows high contact tracing efficiency. This is in line with estimates of secondary infection rates of 7.2% in those who do not live in the same household vs. 13.0% in those who do (16). Our nnq is low compared with the figure found in population-wide contact tracing. There is little evidence on how this compares with other countries, underlining the importance of efficient contact tracing during the pandemic (25, 26, 27).

The strengths of our study include its large sample size, the multilocal approach, and the sampling strategy. Its limitations include the sequential sampling; we did not measure the seroprevalence at all locations simultaneously. Measurement of under-reporting after 12/2020 may be overestimated, as we did not account for self-selection of participants with knowledge or suspicion of a SARS-CoV-2 infection. The proportion of seropositive participants who also reported a positive PCR test indicates a higher waning of antibodies in regions where the second sampling was performed after a larger interval. Another limitation is that our results can only be generalized to adults who read German, as all study materials were provided solely in German and we only recruited adults.

We have presented results from a large SARS-CoV-2 seroprevalence study in Germany which indicates low seroprevalence from July to December 2020 and substantial additional infections from October 2020 to August 2021. Under-reporting of infections decreased after December 2020. The IFR was > 1% in most regions and was especially high in the elderly. Given the heterogeneity in age- and region-specific under-reporting estimates, we recommend that infrastructures be created to enable real-time estimation of population-based infection activity and that models incorporate regional age-specific under-reporting estimates to predict infection dynamics.

Ethics
Ethical approval was given on 21 June 2020 by the ethics committee of Hanover Medical School (9086_BO_S_2020).

Acknowledgments

We sincerely thank the administrative districts for supporting our study and for the fruitful cooperation. We are grateful to our colleagues Anne Ulrike Marzian, Angelika Rath, Christina Suckel, and Nicole Grupe for entering data and dealing with participant’s questions and concerns; to Astrid Hans for administrative support; and to Neha Warikoo for providing software assistance. We appreciate the work of our colleagues in the Plauen and Osnabrück laboratories. We thank Kevin Grigorian and the Johanniter-Unfall-Hilfe, and Tim Balz and BOS112 for collaborating with us. We appreciate the excellent technical assistance of IPSOS in arranging appointments. Our deepest debt of gratitude is to all the participants for donating their blood and their time.

Funding

This work was supported by The Helmholtz Association, the European Union‘s Horizon 2020 research and innovation program [grant number 101003480] and by intramural funds of the HZI. The serohub platform for sharing data from SARS-CoV-2 seroprevalence studies (www.serohub.net) was supported by the German Center for Infection Research (DZIF) and the Network for University Medicine (NUM).

Data sharing statement

The data used for this study can be made available in anonymized form to other academic researchers. The following data variables will be shared with the applicant: study site information, recruitment status, essay information, biosample type, demographic information, self-administered diagnostic histories, and seroprevalence test results. For more details, please contact [e-mail]. Academic institutions can apply for the data via [e-mail]. The serohub is the virtual seroprevalence research environment that stores case-based data from the MuSPAD study and other (inter-)national studies. The application process includes scrutiny of the intended research question and research method. After approval, a link to share the data will be send to the applicant. The shared data meet international data standards (further details can be found in csv format at https://www.covid19dataportal.org/support-data-sharing-covid19.

Conflict of interest statement
The authors declare that no conflict of interest exists.

Manuscript received on 10 May 2021, revised version accepted on
8 October 2021

Corresponding author
Daniela Gornyk
Helmholtz-Zentrum für Infektionsforschung
Abteilung für Epidemiologie
Inhoffenstr. 7
38124 Braunschweig, Germany

Box

Collaborators

Cite this as:
Gornyk D, Harries M, Glöckner S, Strengert M, Kerrinnes T, Heise JK, Maaß H, Ortmann J, Kessel B, Kemmling Y, Lange B, Krause G, on behalf of the MuSPAD Team: SARS-CoV-2 seroprevalence in Germany—a population-based sequential study in seven regions. Dtsch Arztebl Int 2021; 118: 824–31. DOI: 10.3238/arztebl.m2021.0364

►Supplementary material

eReferences, eFigure, eSupplement:
www.aerzteblatt-international.de/m2021.0364

References

1. Vanella P, Basellini U, Lange B: Assessing excess mortality in times of pandemics based on principal component analysis of weekly mortality data—the case of COVID-19. Genus 2021; 77: 16

2. Statistisches Bundesamt: Number of deaths and excess mortality www.destatis.de/EN/Themes/Cross-Section/Corona/Society/population_death.html2021 (last accessed on 30 July 2021).

3. Pritsch M, Radon K, Bakuli A, et al.: Prevalence and risk factors of infection in the representative COVID-19 cohort Munich. Int J Environ Res Public Health 2021; 18: 3572

4. Santos-Hovener C, Neuhauser HK, Rosario AS, et al.: Serology- and PCR-based cumulative incidence of SARS-CoV-2 infection in adults in a successfully contained early hotspot (CoMoLo study), Germany, May to June 2020. Euro Surveill 2020; 25: 2001752

5. Streeck H, Schulte B, Kümmerer BM, et al.: Infection fatality rate of SARS-CoV2 in a super-spreading event in Germany. Nat Commun 2020; 11: 1–12

6. Fischer B, Knabbe C, Vollmer T: SARS-CoV-2 IgG seroprevalence in blood donors located in three different federal states, Germany, March to June 2020. Euro Surveill 2020; 25: 2001285

7. Aziz NA, Corman VM, Echterhoff AK, et al.: Seroprevalence and correlates of SARS-CoV-2 neutralizing antibodies from a population-based study in Bonn, Germany. Nat Commun 2021; 12: 1–10

8. Pan X, Chen D, Xia Y, et al.: Asymptomatic cases in a family cluster with SARS-CoV-2 infection. Lancet Infect Dis 2020; 20: 410–1

9. Rothe C, Schunk M, Sothmann P, et al.: Transmission of 2019-nCoV infection from an asymptomatic contact in Germany. NEJM 2020; 382: 970–1

10. World Health Organization: Population-based age-stratified seroepidemiological investigation protocol for coronavirus 2019 (COVID-19) infection. World Health Organization 2020.

11. von Elm E, Altman DG, Egger M, et al.: The Strengthening the Reporting of Observational studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol 2008; 61: 344–9

12. Winkler V, Leitzmann M, Obi N, et al.: Response in individuals with and without foreign background and application to the National Cohort in Germany: which factors have an effect? Int J Public Health 2014; 59: 555–63

13. Helmholtz-Zentrum für Infektionsforschung: Prospective assessment of Incident health events-App. www.info-pia.de/ (last accessed on 14 April 2021).

14. Statistische Ämter des Bundes und der Länder: Bevölkerung nach Geschlecht und Altersjahren (79) – Stichtag 31.12. – (ab 2011) regionale Tiefe: Kreise und krfr. Städte. www.regionalstatistik.de/genesis/online (last accessed on 8 March 2021).

15. Gelman A, Carpenter B: Bayesian analysis of tests with unknown specificity and sensitivity. J R Stat Soc Series C 2020; 69: 1269–83

16. Wymant C, Ferretti L, Tsallis D, et al.: The epidemiological impact of the NHS COVID-19 App. Nature 2021; 594: 408–12

17. Hippich M, Holthaus L, Assfalg R, et al.: A public health antibody screening indicates a 6-fold higher SARS-CoV-2 exposure rate than reported cases in children. Med (N Y) 2021; 2: 149–63.e4

18. Poethko-Müller C, Prütz F, Buttmann-Schweiger N, Fiebig J, Sarganas G, et al.: German and international studies on SARS-CoV-2 seroprevalence. J Health Monit 2020; 5(S4): 2–15.

19. Levin AT, Hanage WP, Owusu-Boaitey N, Cochran KB, Walsh SP, Meyerowitz-Katz G: Assessing the age specificity of infection fatality rates for COVID-19: systematic review, meta-analysis, and public policy implications. Eur J Epidemiol 2020; 35: 1123–38

20. Robert Koch-Institut: SARS-CoV-2 Auswertungen und Daten. www.ars.rki.de/Content/COVID19/Reports.aspx 2021(last accessed on 30 July 2021).

21. Sy KTL, Martinez ME, Rader B, White LF: Socioeconomic disparities in subway use and COVID-19 outcomes in New York City. Am J Epidemiol 2020; 190: 1234–42

22. Bezerra ACV, Silva CEMD, Soares FRG, Silva JAMD: Factors associated with people‘s behavior in social isolation during the COVID-19 pandemic. Cien Saude Colet 2020; 25: 2411–21

23. Farsalinos K, Barbouni A, Poulas K, Polosa R, Caponnetto P, Niaura R: Current smoking, former smoking, and adverse outcome among hospitalized COVID-19 patients: a systematic review and meta-analysis. TACD 2020; 11: 2040622320935765

24. Reddy RK, Charles WN, Sklavounos A, Dutt A, Seed PT, Khajuria A: The effect of smoking on COVID 19 severity: a systematic review and meta analysis. J Med Virol 2021; 93: 1045–56

25. Quilty BJ, Clifford S, Hellewell J, et al.: Quarantine and testing strategies in contact tracing for SARS-CoV-2: a modelling study. Lancet Public Health 2021; 6: e175–e83

26. Bradshaw WJ, Alley EC, Huggins JH, Lloyd AL, Esvelt KM: Bidirectional contact tracing could dramatically improve COVID-19 control. Nat Commun 2021; 12: 1–9

27. Kretzschmar ME, Rozhnova G, van Boven M: Isolation and contact tracing can tip the scale to containment of COVID-19 in populations with social distancing. Front Phys 2021; 8: 677

e1. EUROIMMUN Medizinische Labordiagnostika AG: Anti-SARS-CoV-2 ELISA (IgG). www.coronavirus-diagnostics.com/documents/Indications/Infections/Coronavirus/EI_2606_D_UK_A.pdf (last accessed on 17 December 2020).

e2. Patel EU, Bloch EM, Clarke W, et al.: Comparative performance of five commercially available serologic assays to detect antibodies to SARS-CoV-2 and identify individuals with high neutralizing titers. J Clin Microbiol 2021; 59

e3. Stan Development Team: RStan: the R interface to Stan. R package version 2.21.2. http://mc-stan.org/ 2020 (last accessed on 22 October 2021).

e4. Meredith M, Kruschke J: HDInterval: highest (posterior) density intervals. R package version 0.2.2. https://CRAN.R-project.org/package=HDInterval 2020 (last accessed on 22 October 2021).