Clinical and molecular fingerprint of SARS-CoV-2 among hospital employees in a period of Omicron BA.2 dominance

Abstract

In the spring of 2022, SARS-CoV-2 Omicron BA.2 peaked in Germany. The main burden was staff shortage. To achieve effective identification and management of infected persons as well as early reintegration of recovered persons, an infection-control outpatient clinic was established at the Bundeswehr Central Hospital Koblenz. This article reports a secondary data analysis of 663 people with 1,174 visits to the outpatient clinic. For asymptomatic contacts, no correlation was observed between PCR result and testing time or frequency. Although no significant symptoms were documented, a high correlation was found between a positive antigen self-test and positive PCR. For clearance, a median time until a negative test was obtained was 8-11 days. The PCR gold standard was compared with ECLIA antigen testing for all indications. The results of this study challenge the rationale for testing asymptomatic contacts. Solely symptom-driven diagnostics by PCR also do not seem to be effective. However, contact persons or symptomatic persons with a positive rapid antigen test should be tested further. Whether this testing is done by ECLIA or PCR does not seem to matter. Clearance testing after recovery prior to day 8 is also not appropriate.

Keywords: COVID-19; COVID-19 symptoms; ECLIA; Omicron BA.2; SARS-CoV-2; antigen testing; clearance testing; contact testing; hospital hygiene; testing strategy.

Table 1. Description of examined staff (whenever a person visited the outpatient clinic more than once, only the first visit was counted for demographic analysis)

Table 2. 2x2 tables for observed symptoms and positive antigen self-test results in relation to the PCR result for visits of suspicious cases (whenever a person visited the outpatient clinic more than once, all symptoms ever mentioned were counted for analysis. Significant p-values are written in bold; n=371. n=number of evaluable results excluding missing results)

Table 3. Suspicious cases with either symptoms and/or positive rapid antigen test, with or without contact person status (2x2 table for comparison of PCR and ECLIA with calculation of positive predictive value (PPV), negative predictive value (NPV), sensitivity (sens), specificity (spec) and prevalence (prev); square brackets give the 95% confidence intervals; n=371, 7 presentations had missing PCR results due to loss of sample.)

Table 4. Contact persons without symptoms or positive rapid antigen test (2x2 table for comparison of PCR and ECLIA with calculation of positive predictive value (PPV), negative predictive value (NPV), sensitivity (sens), specificity (spec) and prevalence (prev); 95% confidence intervals in square brackets; n=423. One presentation had a missing PCR result due to loss of sample.)

Table 5. Clearance (real data, where PCR with at least one gene <30 counts as positive, all genes >30 or 0 count as negative; 2x2 table for comparison of PCR and ECLIA with calculation of positive predictive value (PPV), negative predictive value (NPV), sensitivity (sens), specificity (spec) and prevalence (prev); 95% confidence intervals in square brackets; n=423. One presentation had a missing PCR result due to loss of sample)

Figure 1. ROC analysis for ECLIA values in the clearance group (the best cut off – identical in Liu, Youden or nearest point method [12]) is 0.9505 (red dot); real cut off is 1.000).

Figure 2. Estimated median time until successful clearance based on Kaplan-Meier curves. Red: estimator for 50% positive tests; blue: estimator for 50% negative tests; light red and light blue: 95% confidence intervals; green lines: span for the estimators in case of ECLIA (A) and PCR (B); contra-factual modelling of ECLIA cut-offs by nearest point to 0.1 (C); red lines: lower limit for the estimated median, which is identical (8 days) in all three variants.