In the present study on universal SARS-CoV-2 admission screening encompassing 44 weeks and more than 20,000 patients, around 3% of patients undergoing a SARS-CoV-2 PCR on hospital admission tested positive of which 12% were asymptomatic. The rate of asymptomatic patients with a positive SARS-CoV-2 test on admission paralleled, but slightly lagged the population incidence.
The proportion of asymptomatic patients with a positive SARS-CoV-2 test on admission (0.36%) is similar to previous findings in our hospital15 but higher compared to reports from other hospitals in Switzerland (0.15%)16 or abroad (0.005% and 0.16%)17,18. Most asymptomatic SARS-CoV-2 positive patients on admission did not develop COVID-19 symptoms during hospital stay; this finding resembles a recent report from another Swiss tertiary care centre where one out of six asymptomatic SARS-CoV-2-positive patients became symptomatic over the course of the hospitalization16.
In contrast to the studies of Stadler et al.16 and Jung et al.17, we identified a significant correlation of the population incidence with both, the proportion of individuals with a positive test on admission and the proportion of asymptomatic patients with a positive test on admission. In line with our findings, Krüger et al. reported variation in the proportion of individuals with a positive test on admission screening by population incidence18. The authors divided an overall observation period of 4 weeks in high (before peak incidence in Bavaria during observation period) and low (after peak incidence in Bavaria during observation period) incidence periods and compared the proportions. Taking the dynamics of the pandemic better into account, we addressed the correlation between cantonal incidence and results of admission screening on a weekly basis.
For the period from calendar week 22 in 2020 until to calendar week 8 in 2021, the cantonal dataset also included information on the proportion of positive SARS-CoV-2 tests performed in the canton of Zurich. In time series analysis plots, the dynamics in the proportion of positive SARS-CoV-2 tests on the cantonal level paralleled the cantonal incidence rate. This finding is similar to the results of Nesteruk19. In line with the findings for the cantonal incidence rate, significant associations of the proportion of positive SARS-CoV-2 tests performed in the canton and (a) the proportion of individuals with a positive SARS-CoV-2 test and (b) the proportion of asymptomatic SARS-CoV-2 infected individuals identified in universal admission screening were detected. The parallel increase in the incidence rate and the proportion of positive SARS-CoV-2 tests on the cantonal level reflects insufficient testing during periods with high SARS-CoV-2 activity.
Following the exponential increase in population incidence, a one-week lag time was observed until a parallel dynamic in the proportion positive of admission screening was detected, regardless of having symptoms or being asymptomatic. One hypothesis for this lag time could be that the initial phase of the exponential incidence growth in the population is driven by younger age groups. Among hospitalized patients in Switzerland, older age is common with 36.9% of patients ≥ 65y in 201920. Recent reports indicated an increase in the proportion of younger individuals among all SARS-CoV-2 infections21. Even if younger individuals are less likely to suffer from severe COVID-19 resulting in hospitalization, some might transmit SARS-CoV-2 to elderly individuals. US data indicated a preceding increase in the percentage of positive SARS-CoV-2 test results among younger adults prior to an increase in elderly individuals22. If an intermittent universal admission screening is considered, one strategy could be an introduction immediately after a rapid rise in population incidence. Future studies confirming our observation of a certain delay until an increase of SARS-CoV-2 in the community is associated with a higher proportion of a- or presymptomatic SARS-CoV-2 individuals at hospital admission, seem desirable, to support the proposed strategy.
Throughout the study period, mainly SARS-CoV-2 wild type and later alpha variant predominated in Switzerland23. Vaccination started at the end of 2020 focussing initially on vulnerable populations. At the end of the study period vaccination coverage was still low with 9.47 vaccinated individuals per 100,000 inhabitants24. The applicability of our findings for settings with predominance of other SARS-CoV-2 variants and much higher vaccination coverage remains unsure and future studies seem warranted.
The present study has several strengths. We evaluated data collected over a time span of 44 weeks encompassing more than 20,000 admitted patients. The vast majority of prior studies on this research question were characterized by a shorter duration15,16,17,18 or limited sample size15,16,18. The availability of corresponding longitudinal data on the cantonal level enabled a comparison of cantonal incidence with the proportion of individuals with a positive test on admission screening irrespective of symptoms of COVID-19. We were able to identify non-contagious individuals with prior COVID-19 and residual viral replication; this allowed us to focus the analysis on contagious individuals. In addition, medical chart review of individuals with a positive test on admission but without symptoms of COVID-19 enabled us to assess potential contagiousness during hospital stay and to distinguish between a- and presymptomatic infection.
The study also has several limitations. It was a single centre study and the classification of symptomatic or asymptomatic at admission was performed retrospectively. However, as our hospital participated in an antecedent study with prospective collection of COVID-19 symptoms at admission15, we have confidence that documentation in patient charts has improved following this intervention. In addition, the method we used for the correlation analysis between admission screening and cantonal incidence data assumes independence between data points, which is not the case. The degree of evidence resulting from this approach is likely overestimated. However, independently of the statistical significance of the associations, we found very high correlations (rho > 0.90) arguing that this effect is relevant. One specific limitation of the Poisson regression is that it does not allow for overdispersion. To address this limitation, we added negative binomial regression models that confirmed the findings. The use of saliva samples in the Department of Obstetrics may have resulted in some underestimation of infectiousness among SARS-CoV-2 positive individuals. A prior study with participation of our centre comparing SARS-CoV-2 detection in saliva and nasopharyngeal swabs supported saliva as a generally reliable specimen for SARS-CoV-2 detection, but average Ct values were lower in saliva samples compared to nasopharyngeal swabs25. No information on the frequency of symptoms of individuals tested in the Canton of Zurich was available, thus hindering an analysis of a correlation with results of universal admission screening.
Future studies taking secondary attack rates of a- and presymptomatic SARS-CoV-2 infected individuals for benefit assessment into account, seem desirable. Notably, a recent meta-analysis supported lower secondary attack rates from asymptomatic individuals compared to presymptomatic or symptomatic individuals26.