A prospective study of risk factors associated with seroprevalence of SARS-CoV-2 antibodies in healthcare workers at a large UK teaching hospital

Objectives To describe the risk factors for SARS-CoV-2 infection in UK healthcare workers (HCWs). Methods We conducted a prospective sero-epidemiological study of HCWs at a major UK teaching hospital using a SARS-CoV-2 immunoassay. Risk factors for seropositivity were analysed using multivariate logistic regression. Results 410/5,698 (7·2%) staff tested positive for SARS-CoV-2 antibodies. Seroprevalence was higher in those working in designated COVID-19 areas compared with other areas (9·47% versus 6·16%) Healthcare assistants (aOR 2·06 [95%CI 1·14-3·71]; p=0·016) and domestic and portering staff (aOR 3·45 [95% CI 1·07-11·42]; p=0·039) had significantly higher seroprevalence than other staff groups after adjusting for age, sex, ethnicity and COVID-19 working location. Staff working in acute medicine and medical sub-specialities were also at higher risk (aOR 2·07 [95% CI 1·31-3·25]; p<0·002). Staff from Black, Asian and minority ethnic (BAME) backgrounds had an aOR of 1·65 (95% CI 1·32 – 2·07; p<0·001) compared to white staff; this increased risk was independent of COVID-19 area working. The only symptoms significantly associated with seropositivity in a multivariable model were loss of sense of taste or smell, fever, and myalgia; 31% of staff testing positive reported no prior symptoms. Conclusions Risk of SARS-CoV-2 infection amongst HCWs is highly heterogeneous and influenced by COVID-19 working location, role, age and ethnicity. Increased risk amongst BAME staff cannot be accounted for solely by occupational factors.


Background
With > 580 million cases and > 6 million deaths reported to date globally, the ongoing COVID-19 pandemic continues to impact daily life ( 1 ). A nationwide lockdown in the UK on 23 rd March 2020 succeeded in slowing infection rates during the "first wave" ( 2 ); however, subsequent waves and the emergence of novel dominant variants have continued to place unprecedented pressure on the NHS ( 3 , 4 ) and drive infections globally (5)(6)(7). The logistics of managing patients with COVID-19 presented unique challenges to hospitals and NHS trusts across the UK; evidence and practices evolved rapidly as experience was gained. Healthcare workers (HCWs) are at a higher risk of SARS-CoV-2 infection than the general population ( 8 , 9 ), and subsequent evidence has emerged for risk factors associated with SARS-CoV-2 infection in front-line HCWs (10)(11)(12)(13).
Protecting HCWs by identifying risk factors for SARS-CoV-2 infection will continue to be paramount 3 as the UK accepts coronavirus as a common endemic disease. Controlling transmission within a hospital setting, as well as from hospitals back into the community, was a key element in controlling the pandemic ( 14 , 15 ). However, defining HCW specific risk-factors remains a challenge. Additionally, higher rates of symptomatic SARS-CoV-2 infection, hospitalisation and death have been observed amongst patients from ethnic minority populations in the UK ( 16 ) and worldwide ( 17 , 18 ); the reasons for this disparity are unclear. Reported infections in HCW suggests higher mortality in HCWs from minority backgrounds ( 19 ); however, it is not yet clear to what extent workplace exposures influence infection. Here, we present the results of a large sero-epidemiological study of SARS-CoV-2 seropositivity in staff at a teaching hospital in the East of England undertaken during the first wave of the COVID-19 pandemic.

Setting
Cambridge University Hospitals NHS Foundation Trust (CUH) is a tertiary referral centre and teaching hospital with 1,0 0 0 beds and 11,545 staff serving a population of 580,0 0 0 people in the East of England. The facility is equipped with a 20-bed ICU, a 23-bed neurosciences and trauma ICU, and an Emergency Department that receives ∼14,0 0 0 attendees a month. Between March and June 2020, CUH treated 525 patients with PCR-confirmed COVID-19 ( Figure 1 ). The peak of COVID-19 admissions occurred in late March and early April 2020, with comparatively few COVID-19 admissions from June 2020 onwards. The definition of COVID-19 working for the purpose of risk stratification included clinical areas designated as either "Red" (patients with PCR-confirmed SARS-CoV-2 infection) or "Amber" (patients for whom there is a high clinical suspicion of COVID-19).
As of September 2020, the East of England reported 27,516 laboratory-confirmed cases of SARS-CoV-2 infection ( 20 ), with a corresponding population rate of 441 ·2 per 10 0,0 0 0 people as of September 2020. This rate was substantially less than the worst affected regions of North West England (772 ·9/10 0,0 0 0) and Yorkshire and the Humber (693 ·2/10 0,0 0 0) ( 20 ). According to the 2011 England and Wales census ( 21 )  An asymptomatic staff screening programme using SARS-CoV-2 PCR testing was established in April 2020 ( 23 ). A staff screening programme for SARS-CoV-2 serological testing was initiated on the 10 th of June 2020. All staff members were invited by email to participate in the serological screening programme and asked to selfrefer for a clinic appointment. Written informed consent was obtained from all participants enrolled into this study. As part of this process all participants were invited to join the NIHR BioResource -COVID-19 Research Cohort (IRAS 220277). At enrolment, participants completed a questionnaire asking about demographic characteristics, healthcare role, ethnicity, previous symptoms consistent with COVID-19 and previous results of SARS-CoV-2 PCR testing. A total of 7 ·8 ml of blood was collected, including one serum sample and one whole blood sample. The serum sample was assayed for total SARS-CoV-2 antibody; both residual serum and whole blood were stored for future analyses.

Laboratory assays
Serological testing for antibodies directed against SARS-CoV-2 was performed using the Centaur XP SARS-Cov-2 Total Antibody assay (Siemens Healthcare Limited, Surrey, UK). This method is a fully automated high throughput enzyme linked chemiluminescent bridging immunoassay which targets the S1RBD antigen of SARS-CoV-2 and can detect all Ig subclasses (IgG, IgM, and IgA). The quantity of SARS-CoV2 antibodies correlates directly with relative light units (RLU), which is converted to Index Values with a measuring interval of 0.05 -> 10 index, where values below 1 are reported as nonreactive and those ≥1.0 are reported as reactive, as validated by the manufacturer by clinical correlation. The method was independently validated by Public Health England and has a reported sensitivity and specificity of 98.1% (95% CI 96.6 -99.1) and 99.9% (95% CI 99.4 -100) respectively. Samples were processed in the Biochemistry laboratory at CUH following the SOP as stated by the manufacturer in their Instruction for Use (IFU) after a local verification using guidance from The Royal College of Pathologists ( 24 ).
As previously described, the RT-PCR assay used at CUH designates a cycle threshold (Ct) of ≤36 to correspond to a positive result ( 23 ).

Statistical analysis
Seroprevalence is reported as a percentage ([proportion with antibodies/number tested] x 100). Logistic regression was used for univariable and multivariable analyses of seroprevalence comparisons. The Wilcoxon rank-sum test was used for comparison of median Ct values. Data were analysed using Stata v14.2 (StataCorp, College Station, Texas).

Baseline information
The CUH staff serology screening clinic was operational between 10 th June and the 7th of August 2020. A total of 8,376 (73%) staff attended the clinic for SARS-CoV-2 serology; 5,697/8,376 (68%) of these consented to be enrolled in the study ( Figure 2 ). 1,700/5,967 (28 ·5%) of study participants reported that they had worked in a designated COVID-19 area within the CUH structure during the peak of the epidemic between February and June 2020. The median age of participants was 38 years (range 17-83 years) and 22 ·7% (1,293/5,697) were male ( Table 1 ). A total of 22 staff

Department
Staff working specifically in the ICUs had a seroprevalence of 6 ·33% (10/158), and staff working specifically in the Emergency Department had a seroprevalence of 9 ·1% (9/99). However, neither of these staff groups had significantly different seropositivity using univariate analysis ( p > 0 ·1 in both groups) compared to non-ICU and non-Emergency Department staff respectively.

Seroconversion after positive SARS-CoV-2 PCR
From 5,991 enrolled participants, 2,825 (47%) reported having had a SARS-CoV-2 PCR test between February 2020 and the time of blood sampling, primarily through the CUH HCW testing programme. Of these, 51 (2 ·05%) tested PCR positive for SARS-CoV-2 RNA, 47 had detectable SARS-CoV-2 antibodies, and four had no detectable SARS-CoV-2 antibodies. All serological samples in these cases were taken > 21 days after positive PCR tests. The median SARS-CoV-2 PCR Ct value in those who seroconverted was 30 (IQR 24 -34), in comparison to 36 (IQR 35 ·5 -37) in those who did not seroconvert ( p = 0 ·006). The four staff who had previously tested SARS-CoV-2 PCR positive and were antibody negative all reported having symptoms consistent with COVID-19 infection at the time of PCR testing, although none reported the loss of taste or smell. Nine (18%) of the staff previously testing SARS-CoV-2 PCR positive, and who were antibody positive, were asymptomatic at the time of PCR testing.

Discussion
In this comprehensive assessment of factors associated with seropositivity for SARS-CoV-2 antibodies in HCWs at a large UK tertiary referral centre we were able to identify key at-risk occupational groups. Specifically, staff working in areas where patients with confirmed SARS-CoV-2 infection are cared for, those employed as HCA or domestic and portering staff, those of younger age, and those working in acute medicine or a medical subspeciality were more likely to have SARS-CoV-2 antibodies. A reduced risk of SARS-CoV-2 seropositivity was associated with White ethnicity, being employed in an administrative role, and belonging to an older age group.
We found that the seroprevalence of SARS-CoV-2 antibodies in staff working in non-COVID facing areas was slightly higher (6 ·16%) than in the general population in the East of England (5 ·0%) ( 25 ) and comparable to the national prevalence (6 ·0%) ( 25 ). This is in keeping with previous retrospective serological HCW studies reporting relatively low seroprevalences in Germany (1 ·6%) ( 26 ), Wuhan (3 ·8%) ( 27 ) and Belgium (7 ·6%) ( 28 ). Amongst Asian staff working at CUH the seroprevalence was also comparable to East of Table 3 Unadjusted odds ratio (OR) and adjusted odds ratio (aOR) of SARS-CoV-2 seropositivity by reported symptoms England data (10 ·5% vs 10 ·1%, respectively), as was the seroprevalence amongst Black staff at CUH compared to regional data (18% vs 15%, respectively) ( 25 ). Overall, we observed significantly higher seroprevalence in all BAME staff compared to White staff, and to a greater extent in Black and Asian staff specifically. These differences have been observed nationally and are not unique to HCWs. The finding that the increased risk associated with BAME staff was not influenced by COVID-19 area working (and independent of job role), as well as the ethnic differences in symptomatic seroconversion rates, demonstrates that the increased prevalence of antibodies in BAME HCWs cannot be accounted for by purely occupational factors.
In staff who were previously SARS-CoV-2 PCR positive, 92% (47/51) had detectable antibodies. There was a significant difference in Ct values between those who did and did not seroconvert. A potential explanation for this difference is that higher viral loads may be required to generate a sustained antibody response ( 29 ). Alternatively, a false positive SARS-CoV-2 PCR result or the detection of viral nucleic acid without infectious virus would also explain a lack of seroconversion.
Consistent with previous studies, we demonstrate that whilst reporting prior symptoms consistent with COVID-19 increased the chances of seropositivity, differentiating previous COVID-19 infection from other common respiratory tract infections based on symptoms alone is unreliable ( 10 ). specifically, the only symptoms that significantly predicted seropositivity on a multivariable logistic regression model were the loss of sense of taste or smell, myalgia and fever. Prior reporting of cough or shortness of breath were not good predictors of the presence of SARS-CoV-2 antibodies in a multivariable model. These data also reiterate previous findings that asymptomatic SARS-COV-2 infection amongst healthcare workers is common, with 31% of seropositive staff having never reported consistent symptoms, and 18% of PCR positive staff never having reported consistent symptoms. Our data highlight the importance of the contribution of the asymptomatically infected population to the spread of the disease ( 30 , 31 ). Consequently, asymptomatic screening of staff in healthcare settings became a major component of routine disease surveillance ( 23 , 32 ), and is likely to be of benefit in future waves associated with novel variants and in future pandemics.
The development and widespread rollout of mRNA vaccines since this study was conducted has resulted in lower re-infection rates and symptomatic disease in HCWs in the UK, and elsewhere ( 33 , 34 ). However, the understanding of HCW infection risk remains critical in the protection of HCWs to novel variants (and vaccine escape), as well as those unable to receive vaccination.
We acknowledge several limitations to our study. Variables such as ethnicity, COVID-working location and job role were selfreported; however, we have no reason to think these variables were party to recall bias and it is unlikely to impact on the results to any large degree. The proportion of staff reporting being of Black ethnicity was relatively small, although the proportion of BAME staff is consistent with the wider NHS workforce, and our conclusions are therefore broadly generalisable. The terminology and designation of COVID-facing clinical areas was an evolving factor throughout the course of the epidemic and is likely to be variable between hospital trusts and regions, as will the re-distribution of workforces and workflows through hospitals. Additionally, there will have been heterogeneity in infection rates and admission pressures between different regions and between different hospitals within the same regions that may influence HCW exposure to infection differently. Consequently, this variation between practices may impact the specific risk factors assessed in this study to varying extents between different healthcare trusts. We were also unable to assess the use of PPE and adherence to PPE protocols in this study design. The selected assay may have reduced sensitivity in individuals who generated robust antibody responses to other SARS-COV-2 antigens or those producing low affinity antibodies during early disease. Similar considerations apply to other commercial assays ( 35 ), and a subsequent comparison demonstrated assay equivalence with the selected platform having higher accuracy ( 36 ), and an independently reported sensitivity and specificity of 98 ·1% (95% CI 96 ·6 -99 ·1) and 99 ·1% (95% CI 99 ·4 -100) respectively ( 36 ). We also note that symptom data were recorded retrospectively and may have been subject to recall bias.

Conclusions
Our study confirms prior findings and provides new information on the risk factors for SARS-COV-2 infection and antibody response in HCWs. We found that the occupational exposure to SARS-COV-2 is heterogenous across job roles, hospital department, and ethnicity. It is clear that HCWs who remain on the frontline of the COVID-19 pandemic require more protection from occupational exposure with accurate stratification of risk factors to develop mitigation strategies despite effective vaccines. The association with ethnic group is concerning and a deeper understanding of the societal and/or genetic factors predisposing the BAME population to SARS-COV-2 infection and seroconversion is needed.

Ethical approval
Ethical approval for this study was granted by the East of England -Cambridge Central Research Ethics Committee (IRAS ID: 220277).