Infection Prevention Strategies in COVID-19 Units Highly Protective While Primary Risks to Healthcare Professionals Come From Coworkers and the Community

Background: Early evaluations of healthcare professional (HCP) COVID-19 risk occurred during insucient personal protective equipment and disproportionate testing, contributing to perceptions of high patient-care related HCP risk. We evaluated HCP COVID-19 seropositivity after accounting for community factors and coworker outbreaks. Methods: Prior to universal masking, we conducted a single-center retrospective cohort plus cross-sectional study. All HCP 1) seen by Occupational Health for COVID-like symptoms (regardless of test result) or assigned to 2) dedicated COVID-19 units, 3) units with a COVID-19 HCP outbreak, or 4) control units from 01/01/2020-04/15/2020 were offered serologic testing by an FDA-authorized assay plus a research assay against 67 respiratory viruses, including 11 SARS-CoV-2 antigens. Multivariable models assessed the association of demographics, job role, comorbidities, care of a COVID-19 patient, and geocoded socioeconomic status with positive serology. Results: Of 654 participants, 87 (13.3%) were seropositive; among these 60.8% (N=52) had never cared for a COVID-19 patient. Being male (OR 1.79, CI 1.05-3.04, p=0.03), working in a HCP-outbreak unit (OR 2.21, CI 1.28-3.81, p<0.01), living in a community with low owner-occupied housing (OR=1.63, CI=1.00-2.64, p=0.05), and ethnically Latino (OR 2.10, CI 1.12-3.96, p=0.02) were positively-associated with COVID-19 seropositivity, while working in dedicated COVID-19 units was negatively-associated (OR 0.53, CI=0.30-0.94, p=0.03). The research assay identied 25 additional seropositive individuals (78 [12%] vs. 53 [8%], p<0.01). Conclusions: Prior to universal masking, HCP COVID-19 risk was dominated by workplace and community exposures while working in a dedicated COVID-19 unit was protective, suggesting that infection prevention protocols prevent patient-to-HCP transmission. Blood samples from participating HCPs were assessed using two platforms, a novel, high- sensitivity Fingerstick Coronavirus Antigen Microarray (COVAM) measuring IgG and IgM antibodies against 67 respiratory viruses, including 11 SARS-CoV-2 antigens, and a Food and Drug Administration Emergency Use Authorized serology assay (FDA-EUA). 14–17 The primary FDA-EUA assay was the Diazyme SARS-CoV-2 IgG detecting antibodies against SARS-CoV-2 nucleocapsid (N) and spike (S) proteins. 17 To address potential variability in performance characteristics, results were reexed for conrmation by an alternate FDA-EUA assay (Beckman Access SARS-CoV-2 IgG assay against S protein or Abbott Architect SARS-CoV-2 IgG assay against the N protein) when Diazyme results were non-reactive in PCR-conrmed or clinically suspected prior SARS-CoV-2 infection. 17–20 COVID-19 RT-PCR testing was performed using either DiaSorin Molecular Simplexa, m2000 RealTime, or Xpert Xpress SARS-CoV-2. 21–23 PCR results from other facilities were obtained from OH records.


Introduction
Exposure to transmissible diseases is a known occupational hazard for healthcare professionals (HCPs), which warrants robust infection prevention protocols. Early reports from China demonstrated HCP COVID-19 infection acquisition rates as high as 44% and subsequent large-scale studies in Spring of 2020 in the UK and U.S. showing a 12-fold higher COVID-19 test-positivity rate among HCPs compared to the community, leading the international healthcare community to adopt robust measures to protect against occupational exposure to COVID-19. [1][2][3][4][5] Evidence is beginning to emerge that COVID-19 seroprevalence among HCPs mirrors the communities in which they live. 6-9 Nevertheless the perception of risk among HCPs during patient care remains high and calls into question the effectiveness of infection prevention strategies deployed to protect them. 10,11 Many currently available studies assess incidence or prevalence of COVID-19 among randomly sampled HCPs and infer occupational risk by de ning exposure broadly as having worked in healthcare or in a COVID patient care unit without addressing ongoing clusters (outbreaks) related to coworkers working while ill or community exposures. 2,4,5,7−9 HCP COVID-19 prevalence may also be exaggerated by testing bias from occupational health screening and ready access to tests. 2 Finally, HCPs are essential workers that require in-person activity, increasing the overall number of both community and work-related interactions within a given day. Epidemiologic studies are needed that assess the added risk borne by HCPs due to COVID-19 patient care compared to in-person coworkers exposures within the healthcare infrastructure and the communities where they live.
In this study, we evaluated whether HCP roles, documented COVID-19 patient care, HCP-outbreak associated coworker exposures, and geocoded community characteristics were associated with the likelihood of polymerase chain reaction (PCR) and serology con rmed COVID-19, enriching for those presenting to Occupational Health (OH) with symptoms, those working in COVID-19-designated care units, those on units with COVID-19 HCP outbreaks, and control units without HCP outbreaks that were non-COVID-19 designated units.

Study Design and Population
We conducted a retrospective cohort study combined with a cross-sectional seroprevalence survey of HCP at an academic medical center in Orange County, California seen by Occupational Health for COVID-19-like symptoms (regardless of COVID-19 PCR result) between January 1, 2020-April 15, 2020. To enrich for COVID-19 cases, all HCP assigned to designated COVID-19 units (3), hospital units that experienced a COVID-19 HCP outbreak (3), and matched control units not designated for COVID-19 care and without an active HCP outbreak (3) during the study period were included. HCPs included those involved in direct patient care (e.g. doctors, nurses, nursing assistants, physical/speech/respiratory therapists), and those assigned to the unit for non-patient care duties (e.g. environmental services, pharmacy, dietary, social work, case management).
Eligible HCPs were invited to obtain free serologic testing between May 1-June 30, 2020. This study was conducted under hospital operations jointly with approval for the use of a research serologic platform from the University of California Irvine IRB.

Data Collection
Demographic data and HCP job title/role were obtained from human resources records. Care of an infectious COVID-19 patient or patient room entry was identi ed using electronic health record (EHR) records from start of the study period through 2 days before COVID-19 serology blood collection. HCPs not identi able through EHR records (e.g., environmental services) were con rmed by managers to have assigned duty locations during the relevant timeframe. OH records provided assessment dates for COVID-19 symptoms and PCR results. Epidemiology and Infection Prevention Program records provided details about outbreak investigations and response activities, which included interviews of COVID-19-positive HCP for exposure assessment and evaluations for any linkages between cases by time, location, and activities. In addition, HCP zip codes were geocoded to obtain census-based socioeconomic status (SES) variables focusing on income and housing characteristics (Table 1). 12 Eligible HCP who participated in serologic testing also completed a REDCAP survey requesting demographic information, residential zip code, comorbid conditions, height/weight, COVID-19 patient-care (including aerosol generating procedures (AGP)), and history of COVID-19 symptoms (including type/time of onset).
Community COVID prevalence was obtained through the Orange County Health Care Agency. 13 Laboratory Testing Blood samples from participating HCPs were assessed using two platforms, a novel, high-sensitivity Fingerstick Coronavirus Antigen Microarray (COVAM) measuring IgG and IgM antibodies against 67 respiratory viruses, including 11 SARS-CoV-2 antigens, and a Food and Drug Administration Emergency Use Authorized serology assay (FDA-EUA). [14][15][16][17] The primary FDA-EUA assay was the Diazyme SARS-CoV-2 IgG detecting antibodies against SARS-CoV-2 nucleocapsid (N) and spike (S) proteins. 17 To address potential variability in performance characteristics, results were re exed for con rmation by an alternate FDA-EUA assay (Beckman Access SARS-CoV-2 IgG assay against S protein or Abbott Architect SARS-CoV-2 IgG assay against the N protein) when Diazyme results were non-reactive in PCR-con rmed or clinically suspected prior SARS-CoV-2 infection. 17 income below the median household income, where 25% or more live in a house with 1.5 or more occupants per room, and where 50% households live in structures containing > = 5 units. 24,25 We assessed the associated between Latino HCPs and living in areas with higher percentages of Latino residents (above median, 35%) to assess whether ethnicity can be considered a community-level exposure risk. Percent owneroccupied housing was evaluated using the median cutpoint (58%) of the dataset since the owner occupancy of Orange County, California is higher than the national average and highly variable due to a large wealth gradient across the county.
Separately, we evaluated symptom association with seropositivity, creating symptom groups using correlation matrices. Symptom groupings were then assessed for an association with seropositivity using multivariable logistic regression. Finally, performance of each serologic assay was compared.

Results
A total of 1,320 HCPs were invited, including 476 seen by OH, 494 assigned to a designated COVID-19 unit, 388 assigned to a unit that experienced a COVID-19 HCP-outbreak, and 378 assigned to a matched control unit (non-COVID unit and without HCP-outbreak  were associated with COVID-19 (Table 3). proportion of seropositive HCPs who reported any COVID-19 symptoms, speci cally fever, chills, myalgias, fatigue, anosmia, cough, and shortness of breath, while COVAM had a higher proportion of seropositive HCPs who were asymptomatic, had only non-febrile illness, or symptoms > 75 days prior to testing. When restricting to the 41 PCR-con rmed participants (Table 5), all were seropositive by the COVAM assay while 38 (92.7%) were seropositive by the FDA-EUA assay; among the two additional patients detected by COVAM, both had symptoms more than 60 days before blood draw.  During the study period, which occurred before implementation of universal masking and before mandatory N95 use, there were three units with HCP-outbreaks involving a total of 18 HCPs; of these, 8 (44.4%) were exposed to an ill coworker, 6 (33.3%) had no known exposure source, and 3 (16.7%) had a community exposure source. Only 1 (5.6%) HCP infection was plausibly related to patient exposure due to breach of personal protective equipment (PPE). The rst outbreak began with an HCP who had traveled to an area with widespread COVID-19 and had never cared for a COVID-19 patient; this HCP developed muscle and joint pains within the incubation period while at work followed by a sore throat the next day, prompting symptom report and testing. While symptomatic, the HCP interacted directly with two other HCP who subsequently developed COVID-19 within 4-5 days. Their interactions involved hand-off of a nursing cell phone and sharing lunch in a breakroom.
The second outbreak began with a HCP who had no clear source for COVID-19 at work or in the community. The HCP developed symptoms while working and likely infected three other coworkers who worked the same shift. These HCPs also developed symptoms while working and resulted in a cascade of four additional COVID-19 infections in HCPs who worked during the same shift and/or shared spaces (e.g., breakroom, nursing station, skills class). One physician who did not regularly work on the unit spent less than 1 hour reviewing charts at the nursing station and developed COVID-19 without having entered a COVID-19 patient room.
In the third outbreak, a potluck led to 6 HCPs developing COVID-19. Preceding the potluck, a patient tested positive for COVID-19 after being unrecognized while admitted. This patient underwent emergent resuscitation and intubation before COVID-19 diagnosis, but none of the code blue providers developed COVID-19. Five of the six HCPs had not provided care for a COVID-19 patient in the weeks prior to developing symptoms. The last HCP was an administrative staff that assisted with obtaining supplies during the code blue and did not have direct patient contact; based on exposure history, this HCP most likely acquired illness from symptomatic coworkers in the nursing station.

Discussion
Appropriate attribution of true HCP exposure risk must be contextualized by the exposure sources faced in community settings and non-patient care work activities, while simultaneously accounting for the infection prevention strategies in place within healthcare settings to mitigate patient exposures. The top predictors of COVID-19 seropositivity were working in an HCP-outbreak unit, Latino ethnicity, and living in zip code with lower owner-occupied housing, re ecting the important contribution of coworkers and community exposures above and beyond documented COVID-19 patient care. In fact, we found that working in a COVID-19 unit (with contact, eye, and droplet-based mask precautions) was protective against COVID-19, after accounting for the above community and work-related factors, including HCP role and documented care of a COVID-19 patient, suggesting that infection prevention protocols and practices are highly effective in preventing patient-to-HCP transmission.
The majority of HCP workhours are spent performing indirect patient care tasks, such as charting, rounding, or discussing and coordinating patient care. Direct patient-facing care constitutes 20-40% of HCP time; in contact precautions rooms, this time decreases by 18%. [26][27][28] In the setting of a pandemic, heightened awareness of personal risk during COVID-19 patient care results in high compliance with PPE and hand hygiene, further decreasing the likelihood of direct patient care exposures. [29][30][31][32] HCPs spend comparatively more time in communal settings, often in con ned and shared spaces such as nursing stations, physician workrooms, breakrooms, and conference rooms. 33,34 Further exacerbating coworker exposure risk is the propensity to work while actively ill, an unintended consequence of strong work ethics that lead to working long hours despite physical discomfort or sickness. 35 Our nding that 60% of HCPs involved in COVID-19 outbreaks never cared for a COVID-19 patient and that HCP infections appeared to propagate between coworkers highlights the importance of robust daily symptom screening, enforcement of working well policies, and strict compliance with universal masking and social distancing in communal spaces. Current regulatory agencies emphasize high standards for hand hygiene, PPE compliance, and environmental cleaning practices for patient safety; translating these standards to shared HCP spaces and activities is imperative, particularly during a pandemic. 36,37 This includes increasing the number and strategic placement of hand hygiene stations, environmental cleaning products for high touch items in workstations and breakrooms, work ow and structural modi cations to minimize crowding, and robust systems-level enforcement of HCP compliance with protocols.
COVID-19 cases in our county were higher among Latino and densely populated communities, mimicking national trends showing disproportionately greater burden in socioeconomically disadvantaged areas. 13,24 COVID-19 exposures are more likely in communities with more household crowding, low-wage essential workers, and reliance on public transportation, increasing the likelihood of encountering crowded conditions. 24,38 This has two important implications for hospital pandemic response. First, infection prevention strategies to reduce HCP risk should also address community level risks within the workforce, targeting outreach to HCPs from high risk communities, providing culturally and linguistically appropriate education on how to minimize exposure risks both within healthcare and community settings. Since HCPs in socioeconomically disadvantaged communities live with household members who also carry heightened risk for acquiring infection, assuring HCPs have an understanding of home-based household infection prevention protocols may afford additional protection. Second, essential workplaces must partner with public health to educate, support contact tracing and quarantining strategies, and facilitate timely testing and management in high risk communities where their workers reside.
HCP perception of risk is integral to promoting behaviors that reduce exposure risk. Fear and concern about acquiring COVID-19 results in high adherence to hand hygiene and PPE during patient interactions. 30 In contrast, HCPs perceive coworkers or community exposures as less risky, which can increase transmission opportunities in common spaces and breakrooms. The earliest reports of high COVID-19 risk among HCPs occurred in the setting of inadequate PPE and nascent infection prevention protocols. 1,3,30 Subsequently, national and state regulatory bodies, nursing unions, and the media assumed inadequate PPE was the primary exposure risk in healthcare settings, missing important contributions from coworker or community risk. This led to increasingly intensive direct-care PPE requirements without equal attention to the contribution of coworker and community exposures. Notably, the outbreaks in this study occurred before N95 requirements for patient care and we found that droplet-based PPE successfully prevented COVID transmission from patients-to-HCP, while coworker-to-coworker exposures were the sources of unit HCP outbreaks.
This study allowed comparison between the FDA-EUA assays and a novel microarray assay capable of differentiating 67 separate respiratory virus antigen, including in uenza and four common cold coronaviruses. 15 Though we found high concordance between the two assays, the microarray assay identi ed additional individuals who had asymptomatic or non-febrile infection, or illness beyond 75 days prior to testing. The microarray assay could help differentiate between COVID-19 and other respiratory viruses, which could be particularly helpful during the cold and u season.
Our ndings are limited by a single institutional experience; however infection prevention programs across the nation report similar experiences with coworker-associated COVID-transmissions. Second, our county has an overall higher per capita income compared to other counties although the wealth gradient across the county is notable and able to identify SES-based risk factors. Third, participation was voluntary which could introduce sampling bias, although participation across the invited cohorts were similar.
HCP COVID-19 exposure risks must be evaluated and interpreted within the full context of workplace and community exposure sources. When accounting for the protections in place for direct patient care activities, workplace and community exposures appear to dominate COVID-19 risk. Additional investments are needed to account for familiar behaviors and shared meals among co-workers and improve infection prevention strategies to reduce transmission. In addition, extensive investment is needed to target pandemic response efforts to lower socioeconomic status communities where essential workers live. Healthcare systems should consider opportunities to partner with public health to leverage COVID-prevention expertise to these areas.

Declarations
(1) Ethics approval and consent to participate: This study was conducted under hospital operations jointly with approval for the use of a research serologic platform from the University of California Irvine IRB. (2)