To the Editor—The safety of healthcare workers (HCWs) is a major challenge for healthcare systems. In the course of a severe acute respiratory coronavirus virus 2 (SARS-CoV-2) infection, immunoglobulin G (IgG) antibodies may be detected after a median of 14–24 days (interquartile range [IQR], 10–18) after onset of symptoms. ^{Reference Schmidt, Gruter and Boltzmann1}

In France, the coronavirus disease 2019 (COVID-19) pandemic reached a peak on April 7, 2020. HCWs had mobility and flexibility inside the Paris Center university hospital, where there was a cluster in the pandemic. We investigated the prevalence of IgG antibodies against SARS-CoV-2 among all HCWs in this hospital. We also sought to determine the correlation between RT-PCR test and serology and to compare our seroprevalence with that of other European countries.

From May 14, 2020, to June 17, 2020, all HCWs were asked by the occupational health department to participate in serologic screening. The Abbott-Architect test (Abbott Laboratories, Abbott Park, IL) was used to detect IgG anti-SARS-CoV-2. During blood sampling, clinical information was recorded using a standardized self-questionnaire on presented symptoms, comorbidities, and the reverse-transcriptase polymerase chain reaction (RT-PCR) test if one had been previously performed. Blood samples were collected >28 days after the first symptoms from those who were symptomatic.

The seroprevalence and 95% confidence interval were estimated using the Fisher exact method. The t test and the χ ^{Reference Korth, Wilde and Dolff2} test were performed to compare quantitative and qualitative variables, respectively. Simple and multivariate logistic regressions were performed to assess risk and symptoms associated with seroprevalence respectively. Statistical analyses were performed using SAS software (SAS Institute, Cary, NC). The local institutional review board approved this study. All subjects participated voluntarily under pseudonyms.

Of 5,021 workers present during the study period, 4,607 (91.8%) were included in the study. The mean age was 41.8 years (SD, 12.6), and 75% were female. Furthermore, 45% were paramedical staff members, 36% were physicians (including medical students), and 19% were in administrative and other professions.

Overall, the prevalence of IgG antibodies was 11.5% (95% confidence interval [CI], 10.6–12.4), and it was significantly higher (ie, 13%) for paramedical staff (P = .04). Age and gender did not differ significantly according to seroprevalence. Furthermore, 5 clinical symptoms were independently associated with positive serology: asthenia, fever, myalgia, ageusia, and anosmia, for which the highest odd ratio (OR) was observed (OR, 11.1; 95% CI, 7.4–16.6) (Table S1). Notably, although anosmia appeared to be the most specific factor, 64.3% of subjects with antibodies did not experience this symptom. The proportion of asymptomatic subjects with a positive serology was 21.4%. When considering comorbidities, positive serology was significantly associated with a lesser prevalence in smokers (OR, 0.41; 95% CI, 0.29–0.58) and a higher prevalence of diabetes (OR, 1.78; 95% CI, 1.04–3.03) (Table S1).

Comparison with European countries

In our literature review, we retained only studies with IgG antibody testing; we excluded those with IgA or IgM serologies. The 11.5% prevalence of IgG in our HCWs is similar to the reported prevalences in Belgium or the United Kingdom (Table 1). Different protective measures, date of blood screening, and/or population structure in each country could explain the variation in IgG serology from 1.6% reported by Korth et al ^{Reference Korth, Wilde and Dolff2} up to 14.5% reported by Bampoe et al. ^{Reference Bampoe, Lucas and Neall3} In our hospital, masks are compulsory, and protective equipment has been available since March 17.

Table 1. Comparison of Seroprevalence IgG in European Countries

Note. CI, confidence interval.

^a Present study.

Of the 233 HCWs participants with RT-PCR positive, 29 (12.4%) have no detectable antibodies. This result parallels that of Garcia-Basteiro et al, ^{Reference Garcia-Basteiro, Moncunill and Tortajada4} who also reported 15% of individuals with positive RT-PCR and negative serology. A recent study by Patel et al ^{Reference Patel, Thornsburg and Stubblefield5} showed the possibility of decreased antibodies over 60 days, which implies transiently detectable antibodies.

Our study has some limitations. During the lockdown period, some HCWs were isolated at home on a case-by-case basis for reasons of severe personal or familial comorbidities. RT-PCR swab tests were conducted at the time of suspected illness only in symptomatic or in individuals who had had contact with COVID-19 patients. Thus, 902 of 4,607 (19.6%) had this test at the time of onset of symptoms.

The detection of asymptomatic cases by RT-PCR is essential to isolating or avoiding quarantine of HCWs to prevent risk of contamination for vulnerable patients and to reduce the risk of interprofessional staff-to-staff transmission.

To limit virus transmission, we emphasize the necessity of large-scale screening for exposed HCWs, even those who do not present any symptoms. Further investigations are needed to explore negative serology in subjects with positive RT-PCR for understanding population immunity and the potential risks of reinfection and disease in HCWs.