Specific exposure of ICU staff to SARS-CoV-2 seropositivity: a wide seroprevalence study in a French city-center hospital

Background Most hospital organizations have had to face the burden of managing the ongoing COVID-19 outbreak. One of the challenges in overcoming the influx of COVID-19 patients is controlling patient-to-staff transmission. Measuring the specific extent of ICU caregiver exposure to the virus and identifying the associated risk factors are, therefore, critical issues. We prospectively studied SARS-CoV-2 seroprevalence in the staff of a hospital in Lyon, France, several weeks after a first epidemic wave. Risk factors for the presence of SARS-CoV-2 antibodies were identified using a questionnaire survey. Results The overall seroprevalence was 9% (87/971 subjects). Greater exposure was associated with higher seroprevalence, with a rate of 3.2% [95% CI 1.1–5.2%] among non-healthcare staff, 11.3% [8.9–13.7%] among all healthcare staff, and 16.3% [12.3–20.2%] among healthcare staff in COVID-19 units. The seroprevalence was dramatically lower (3.7% [1.0–6.7%]) in the COVID-19 ICU. Risk factors for seropositivity were contact with a COVID-19-confirmed household (odds ratio (OR), 3.7 [1.8–7.4]), working in a COVID-19 unit (OR, 3.5 [2.2–5.7], and contact with a confirmed COVID-19 coworker (OR, 1.9 [1.2–3.1]). Conversely, working in the COVID-19-ICU was negatively associated with seropositivity (OR, 0.33 [0.15–0.73]). Conclusions In this hospital, SARS-CoV-2 seroprevalence was higher among staff than in the general population. Seropositivity rates were particularly high for staff in contact with COVID-19 patients, especially those in the emergency department and in the COVID-19 unit, but were much lower in ICU staff. Clinical trial registration NCT04422977 Supplementary Information The online version contains supplementary material available at 10.1186/s13613-021-00868-8.


Background
Since its outbreak in January 2019 in Wuhan, China, the coronavirus disease 2019 (COVID-19) pandemic has burdened healthcare systems worldwide [1][2][3]. As suspected and recently identified, one of the main difficulties in managing the outbreak has been the specific exposure of hospital staff [4][5][6][7]. Evidence and an understanding of the risk factors for staff infection remain lacking, however. Hospitals concentrate many infected patients and may be significant spreaders of the disease. Intra-hospital clusters among staff may emerge and induce secondary nosocomial infections. The effectiveness of personal protective measures and equipment for caregivers has been well demonstrated, but its use remains imperfect for several reasons: insufficient training, errors of use, and caregivers fatigue [8,9]. The spread of COVID-19 has been facilitated by the frequency of mildly symptomatic and Open Access *Correspondence: evivier@ch-stjoseph-stluc-lyon.fr 1 Medecine Intensive Reanimation, Centre Hospitalier Saint Joseph Saint Luc, 20 Quai Claude Bernard, 69007 Lyon, France Full list of author information is available at the end of the article asymptomatic forms, and by the length of the incubation period of the virus [10]. Contact tracing and screening and the tracking of intra-hospital epidemic outbreaks are, therefore, very difficult [11,12]. The unusual and rapid influx of patients has required hospitals to reorganize, with the accelerated training of healthcare teams further increasing the risk of an imperfect implementation of protective measures.
We report a large prospective study of severe acute respiratory syndrome coronavirus 2 (SARS-COV-2) seroprevalence among the staff of a city-center hospital in Lyon, France, where there was an influx of COVID-19 patients in March-April 2020. The aims of the study were to determine the overall SARS-COV-2 seroprevalence rate in the hospital staff, and to identify and quantify risk factors for symptomatic and asymptomatic staff infection. The main research questions were (i) whether seroprevalence was higher in healthcare personnel than in non-healthcare staff, (ii) whether working in units managing COVID-19 patients increased the risk of infection and (iii) whether a specific exposure was associated with working in intensive care unit.

Hospital description and epidemic background
Saint Joseph Saint Luc Hospital is a tertiary city-center hospital located in Lyon, in the Auvergne-Rhône-Alpes region in France, where the first wave of the COVID-19 pandemic occurred in March and April 2020. The hospital was strongly involved in tackling the epidemic, mainly through its emergency department and intensive care unit (Fig. 1). The first case was detected on February 26th, and 250 patients with confirmed COVID-19 were eventually hospitalized until June, including 11% in the intensive care unit. The rapid influx of a large number of COVID 19 patients relative to the hospital's capacity (320 beds) meant that an equally rapid reorganization of the hospital's treatment units was required. Elective surgery was stopped and specific COVID units were opened to screen patients with suspected COVID-19 (and await CT and/or RT-PCR results) and treat patients with confirmed COVID-19. The capacity of the intensive care unit was increased by 40%. Work meetings were limited and when possible, staff were placed on leave or worked remotely. Family visits were banned. The staff restaurant remained open, but strict social distancing measures were enforced. Each confirmed case among staff was investigated to limit further intra-hospital infections by contact tracing and screening of symptomatic suspects.

Study design
After the first European wave of the epidemic had passed, SARS-CoV-2 antibody testing and a dedicated survey questionnaire were offered to all persons who had worked at the Saint Joseph Saint Luc hospital in Lyon, Auvergne-Rhône-Alpes, France, from March 1 to April 30 2020. This included all clinical and nonclinical employees, the staff of the subcontracted cleaning company, medical interns and nursing students. Testing was delayed by 6 weeks after the end of the first wave to maximize the sensitivity of the antibody test. Enrollment was free, anonymous and voluntary. The study was registered at ClinicalTrials.gov (NCT04422977) and was approved by the local ethics committee (Comité de Protection des Personnes SUD-EST II, Bron, France). Written informed consent was obtained from all participants. The inclusion period ran from June 8 to June 30, 2020, 1 month after the end of the lockdown in France.

Survey
Participants were asked to complete a dedicated questionnaire (see Questionnaire in Additional file 1). The following parameters were recorded: age, gender, presence of symptoms during the outbreak, and prior COVID-19 RT-PCR results. To determine risk factors for COVID-19 infection, participants were asked to state their occupation, the main locations of their work in the hospital, whether they were in direct contact with patients (healthcare workers) or not (non-healthcare workers) and whether they worked day or night shifts. They were also asked to indicate whether they had been in contact with a RT-PCR confirmed COVID-19 household, colleague or patient. Finally, participants were asked about their use of personal protective equipment, social distancing, where they worked in the hospital, and any mandated work time reduction or telework.

Serological analysis
The samples were collected in lithium heparin gel separator tubes, centrifuged and kept at − 20 °C until analysis. All tests were carried out after calibrating the analyzer. Before performing the test, the samples were thawed and re-centrifuged. The SARS-Cov-2 serology tests were performed using electrochemiluminescence immunoassays (Elecsys Anti-SARS-CoV-2, C6000, E601 analyzer Roche Diagnostics). In these tests, a recombinant nucleoprotein is used to detect total anti-SARS-CoV2-2 immunoglobulins against nucleopcapsid antigen (IgA, IgM and IgG), a technique whose sensitivity and specificity 14 days after a positive SARS-Cov-2 RT-PCR test have been reported as > 99% [13][14][15]. Participants were considered seropositive for SARS-CoV-2 if their test results were above the manufacture-specified threshold.

Statistical analysis
As the population seroprevalence of SARS-CoV-2 in Auvergne-Rhône-Alpes was unknown at the time of designing the study, the necessary sample size was calculated considering hypothetical seropositivity rates of 15% in unexposed patients and 30% in exposed patients. The estimated sample size required to detect a statistically significant difference between exposure and non-exposure at an α level of 0.05 and a power (1 − β) of 0.9 was 100 participants per group. The expected participation rate was 70% of staff, i.e., 900 persons.
Overall and group-specific seroprevalences were calculated as the ratios of subjects tested with positive serology tests divided by the total number of subjects in each group, expressed as absolute and relative frequencies. Ninety-five percent confidence intervals were estimated using the asymptotic approximation. Age was expressed as median [IQP]. In univariate analysis, proportions were compared using χ 2 tests or Fisher exact tests (depending on sample size) and ages were compared using Mann-Whitney U tests. Reported symptoms and exposure factors associated with positive serology tests were then successively assessed by means of multivariate logistic regression analyses. The maximal model included all exposures associated at P < 0.15 with a positive serology test. The final model included all variables significantly associated (P < 0.05, two tailed) with a positive serology test. All statistical

Symptoms
Forty-two percent of subjects (410/971) reported at least one prior symptom but this was not a highly sensitive ([76/87] 87%) or specific ([550/884] 62%) sign. Of the 87 members of staff with positive SARS-CoV-2 antibody tests, 11 (12.6%) reported no prior symptoms. Overall, the positive predictive value of declaring one or more prior symptoms was very low (76/410 [18%]). However, anosmia (OR = 55.3) and fever (OR = 3.5) were strongly associated with subsequent seropositivity (Additional file 4: Table S3). Fifty of the 87 subjects with positive SARS-CoV-2 antibody tests had previously had a RT-PCR test for SARS-CoV-2, which was positive in 41/50 cases (84%). One subject who had had a positive RT-PCR test had a negative antibody test.
Reporting a personal contact with a confirmed COVID-19 household, a confirmed COVID-19 coworker

Discussion
This cross-sectional seroprevalence study confirms that hospital workers may have a higher risk of SARS-CoV-2 infection. The risk seems to be particularly high for staff in direct contact with COVID-19 patients, the groups with the highest seropositivity rates in this study being emergency department and COVID-19 unit workers. In contrast, ICU staff had relatively lower seropositivity rates.

Global seroprevalence
The overall SARS-CoV-2 seroprevalence rate among hospital staff (9%) was twice the estimated community  seroprevalence in the hospital catchment area during the screening period (3-5%) [16,17]. The prevalence of infections among hospital staff across Europe was characterized by a great diversity during the first phase, ranging from 2 to 30% according to countries and area [5,[18][19][20]. This high variability is related not only to local prevalence of community infection but also to specific hospital measures to contain the infection. This result is significant, because only 41/87 subjects had previously been diagnosed by RT-PCR. While RT-PCR remains essential for contact tracing, this technique is limited by its low sensitivity and the fact that the virus may no longer be present at the time of screening [21]. The positive predictive value of a symptom declaration for positive serology was very weak in our study, this uncommon result could be explained by technical (weak sensibility of the electrochemiluminescence test) or methodological biases (suggestive questionnaire, nonspecific symptoms, long study period, sero-reversion before the survey). Furthermore, the rate of asymptomatic seropositive workers was only 13%, which contrasts with other series (20-30%) [20,22]. This low rate could also be explained by overdeclaration of symptoms, low sensitivity of the test, possible sero-reversion and a global low seroprevalence in our cohort.

Exposure risk factors
In keeping with the results of a recent hospital-wide survey in Belgium, we found that the primary risk factor for SARS-CoV-2 seropositivity was contact with a confirmed COVID-19 household [19]. However, this factor must be interpreted with caution, as the delay between contact and the seroprevalence study makes it impossible to determine whether the contact with confirmed COVID-19 was indeed the initial source of the infection or was secondarily infected by the member of staff. Our study also highlights the risk of intra-hospital spread, contact with a COVID-19-positive coworker being associated with a twofold increase in the risk of SARS-CoV-2 seropositivity. Working in units where COVID-19 patients were treated was a strong risk factor for subsequent SARS-CoV-2 seropositivity. This may be due in part to the shortage of personal protective equipment at the start of the epidemic, as reported in other series [20,23]. Another special difficulty was to apply social distancing measures in a hospital environment, especially during the staff meals. Indeed, eating in a break room of a COVID care unit was strongly associated with a higher seroprevalence, in univariate analysis. Saint Joseph Saint Luc is a small single-block city-center hospital with small and tightly packed treatment areas. By contrast, having lunch in the staff restaurant seems to have been a protective factor. The restaurant is more spacious and ventilated and the social distancing seems to have been better defined and respected there.
A similar seroprevalence study conducted in a Spanish hospital facing a higher local epidemic wave within the same period reveals some concordant conclusions: higher infection rate among caregivers vs. non-caregivers [20]. But differently from us, the authors showed the lack of PPE and a previous contact with COVID-19 patients were independent factors that were associated with SARS-CoV-2 infection. The seroprevalence was the highest among porters.
We found that staff treating COVID-19 patients in the ICU had a much lower than average SARS-CoV-2 seropositivity rate. This result is in line with a similar previous study [22]. It is of particular clinical significance as it concerns both regular ICU staff and (about 90) additional crisis support personnel. This relatively low seroprevalence may stem from the particular focus on intensive care at the start of the epidemic, with less shortage in protective personal equipment and high standards of infection control. The closed circuit ventilation of intubated patients may also have limited the spread of the virus and the risk of aerosolization with non-invasive techniques seems to be limited [24,25]. The staff/ patient ratio is higher in intensive care than in standard care units with lower patient turnover, which may have reduced individual risks of exposure [26]. Furthermore, patients are often admitted to intensive care 5-8 days after first presenting symptoms and may, therefore, have a lower viral load during their ICU stay [27].
Interestingly, ICU workers had no less household contact but significantly less COVID-19 coworker contact than others hospital workers (5.1% vs. 5.7%, P = 0.76 and 20.9% vs 30.1%, P = 0.01). These data highlight the responsibility of staff clusters in the intra-hospital spread of SARS-Cov-2.

Strengths and limitations
The strengths of the study are the large number of staff included throughout the hospital (including both healthcare and non-healthcare workers), the short data collection time and the optimal delay between the peak of the epidemic and the survey. Recent data show that seroconversion can be delayed by up to 3 weeks after infection with SARS-CoV-2, especially in cases of mildly symptomatic disease [28]. This could explain the discrepancies between the results of our study and those of Steensels et al. [19]. The limitations of the study are first, its single center design, which means that any generalization of the results should be made with caution. Second, since some staff members chose not to participate in the study, selection bias cannot be excluded. Third, to respect patient confidentiality, the questionnaires were self-administered and anonymous, leading to possible errors or misunderstandings. Fourth, we used an automated electrochemiluminescence immunoassay to detect total anti-SARS-CoV-2 antibodies and not a differential detection of IgM and igG with ELISA (enzyme linked immunosorbent assay); however, recent data suggest a good agreement between the two methods [14]. Fifth, given the three-month delay between the first case of COVID-19 in a healthcare worker and the survey, some degree of sero-reversion may have contributed to underestimating global seroprevalence. Recent data suggest that antibody responses could decrease rapidly (especially in moderate forms) [29,30].

Conclusion
In this study, hospital staff treating COVID-19 patients were at a high risk of intra-hospital infection. This suggests that prevention measures should be improved and infection clusters tracked carefully in COVID-19 units.
Our results indicate that the risk of exposure seems to be unevenly distributed throughout hospitals, seropositivity rates being higher than average in the emergency department and (standard care) COVID-19 unit where patient turnover was high, and this is where prevention efforts should be focused in managing the epidemic. The relatively low seropositivity rate observed in COVID-19 intensive care staff suggests that intra-hospital patient-to-staff transmission is limited in this area.