Title: Honors and costs of being heroes: risk factors and epidemiology of COVID-19 infection among health care workers of a severely hit COVID-19-referral Hospital in Italy

Purpose: During the COVID-19 pandemic, the health care workers (HCWs) at the frontline have been largely exposed to infected patients, running an high risk of being infected by the SARS-CoV-2 virus. This study investigates the epidemiological, clinical and lifestyles characteristics that might play roles in the susceptibility of HCWs to COVID-19 in a hit Italian hospital. Methods:Demographic, lifestyle, work-related and comorbidities data of 1447 HCWs which underwent a nasopharyngeal swab for SARS-CoV-2 were retrospectively collected. For the 164 HCWs positive for SARS-CoV-2, data about safety in the workplace, symptoms and clinical course of COVID-19 were also collected. Cumulative incidence of SARS-CoV-2 infection was estimated. Risk factors for SARS-CoV-2 infection were assessed using a multivariable Poisson regression. Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 27 August 2020 doi:10.20944/preprints202008.0588.v1 © 2020 by the author(s). Distributed under a Creative Commons CC BY license. Results: The cumulative incidence of SARS-CoV-2 infection among the screened HCWs was 11.33 (9.7213.21). Working in a COVID-19 ward, being a former smoker (vs being a person who never smoked) and BMI were positively associated with SARS-CoV-2 infection, whereas being a current smoker was negatively associated with this variable. Conclusions: Assuming an equal accessibility and proper use of PPE of all the HCWs of our Hospital, the great and more prolonged contact with COVID-19 patients remains the crucial risk factor for SARS-CoV2. Therefore, increased and particular care needs to be focused specifically on the most exposed HCWs groups, which should be safeguarded. Furthermore, in order to limit the risk of asymptomatic spread of SARS-CoV-2 infection, the HCWs mild symptoms of COVID-19 should be considered when evaluating the potential benefits of universal staff testing.

of the end of April 2020, about 12000 between doctors and nurses were infected by COVID-19, and 228 doctors and 26 nurses had died [7,8].
However, data on HCWs real-life infection risk and clinical characteristics are currently scarce.
This study tries to fill at least part of this gap by reporting the infection risk of HCWs of a severely hit COVID-19-referral Hospital in Italy. We aim to investigate the epidemiological, clinical and lifestyles characteristics that might play roles in the susceptibility of HCWs to COVID-19 in the most critical period of the outbreak in Northern Italy.

Study design, population and data collection
This is an observational retrospective cohort study carried out in the IRCCS Policlinico San Matteo of Pavia, Lombardy. From February 22, 2020, this Hospital was appointed as a COVID-19 referral center and its 4632 HCWs have been involved in this unprecedented health emergency [4,5]. Two sources of data were queried and are described below.
Data has been collected by medical direction in partnership with occupational medicine, which deals with the health surveillance of workers, and Nursing Direction. Before analysis, all data have been made anonymous.
Firstly, records of the 1447 HCWs which underwent a nasopharyngeal swab for SARS-CoV-2 RNA detention from February 22 to May 8, 2020 were collected. These HCWs were the employees which were tested, in accordance to the Local Guidelines at the beginning of the pandemic, because they had symptoms suggestive of COVID-19 or because they had high-risk contacts with SARS-CoV-2 positive individuals. For each of these HCWs, we collected the demographic (age and gender) and occupational (length of service, job roles) characteristics using administrative databases, and we retrospectively collected the potential risk factors for COVID-19 infection such as lifestyle variables (BMI, smoking habit, alcohol consumption), work-related characteristics (work in COVID-19 wards, work in contact with CPAP helmets) and comorbidities (hypertension).
Secondly, we selected the subsample of HCWs who were found positive for SARS-CoV-2. For these HCWs, we also collected data on safety at the workplace and community as correct use of personal protective equipment (PPE) and workstation safety. Sick days, exposure to confirmed case, number of positive nasopharyngeal swabs and time length of negativization were also collected. In addition, we collected clinical data from electronic medical records referring to symptoms reported at the time of the nasopharyngeal swab: fever, cough, shortness of breath, sore throat, conjunctivitis, gastrointestinal (GI) symptoms, asthenia, ageusia, anosmia, headache or neurological symptoms.

HCWs protection protocols
During the pandemic the Infections Committee issued a protocol aimed to design HCWs protection strategy which is structured as follows: with regards to safety at the workplace, training courses on the correct use of PPE were mandatory for the involved personnel. The Hospital stocked Class 2/3 Filtering Face-Piece respirator, surgical masks, liquid-repellent gowns certified for biological risk, hair cap, overshoses, googles/faceshield for all the involved HCWs according to the WHO guidelines [9]. A continuous supply of the aforementioned PPE was then guaranteed to all HCWs during the study period.
All these data were retrospectively collected on june 2020 by trained medical management doctors.
The study has been approved by the Ethics Committee of the IRCCS Policlinico San Matteo and has followed the principles of the Declaration of Helsinki.

Statistical analysis
The characteristics of the HCWs were described using medians and interquartile ranges (IQR) for the quantitative variables and absolute/relative frequency values for the qualitative ones. Comparison between HCWs who were infected and HCWs who were not infected by COVID-19 were performed using non-parametric Mann-Whitney U test for quantitative variables and Chi-square or Fisher exact test for categorical variables. COVID-19 cumulative incidence and 95% confidence intervals were estimated considering the whole population of San Matteo health care workers as well as considering only the workers who underwent at least a nasopharyngeal swab for SARS-CoV-2 RNA detention from

Results
Of the 4632 HCWs working in our Hospital on February 22, 2020, 66,6% were females and their median age was 45.4 years (IQR: 32.2-54.2). With regards to job role, 1525 (32.9%) were physicians, 1321 (28.5%) were nurses, 860 (18.6%) were health care assistants while the remaining 926 (20.6%) were administrative staff members. The median length of service was 11 years (IQR: 2-22). As of May 8 2020, the infection was eventually confirmed by at least one nasopharyngeal swab test in 164 HCWs out of the 1447 screened.

COVID-19 incidence and risk factors in the HCWs
The COVID-19 cumulative incidence in the San Matteo Hospital in the study period was 3.54 (95%CI: 3.04-4.13) per 100 HCWs, while the cumulative incidence among the screened HCWs can be estimated as 11.33 (95%CI: 9.72-13.21) per 100 HCWs.
The epidemiological, lifestyles characteristics and risk factors for SARS-CoV-2 infection among our Hospital screened HCWs are shown in table 1. Comparison between 164 HCWs who were infected and 1283 HCWs who were not infected showed similar median age, length of service in the hospital and a similar distribution of job roles, while males were slightly more prevalent among infected HCWs (36.6% vs 29.6%, p=0.068). The working environment was significantly associated to COVID-19 infection Eventually, 7 (4.3%) infected and 150 (11.3%) not infected HCWs worked in the Hospital without any contact with patients at all. Contact with CPAP helmets (which were present in some of the COVID-19 wards) was also significantly associated to COVID-19 infection (48.4% of the infected vs 16% of the notinfected, p<0.0001). Median BMI was higher (24.4 vs 23.4, p=0.01) among infected vs not infected workers while hypertension was similarly distributed (9.8% vs 8.3%, p=0.538). More than 60% of the infected and not infected HCWs never smoked, but there was a higher proportion of former smokers (20.1% vs 11.0%) and a lower proportion of current smokers (11.1% vs 24.3%) among infected w.r.t not infected (p<0.0001); alcol consumption was similarly distributed, conversely (p=0. 34).   Furthermore, being a current smoker, compared with being a person who had never smoked, was associated to a reduced rate of infection (IRR: 0.43, 95%CI: 0.23-0.80).

Characteristics of HCWs who were infected by SARS-Cov-19
There were no differences between the epidemiological and immunological profile or symptoms frequency of infected HCWs working in a COVID-19 ward w.r.t those working in other clinical wards, as well as there were no differences comparing infected males and females HCWs (p>0.05 for all features). Therefore, Table 3 describes specific features and clinical characteristics of the whole sample of COVID-19 infected HCWs.
On average positive HCWs were absent from work for 18 days and the mean time to negativation of swab was 16 days. The SARS-CoV-2 transmission through person-to-person contact and, consequently, also among HCWs, might result in a harmful shortage of medical staff, which fuels the concern of a collapse of the healthcare systems in the most hit regions worldwide.
The role of HCWs as "heroes" and their best practice in a dramatic, pandemic context have been often enhanced by widespread media reports of fatigue and burnout [14,15].
However, data on risk factors and clinical or lifestyle characteristics of infected HCWs are widely lacking and surveillance for new reports on this subject is ongoing [16].
Since the non-aligned timing of COVID-19 pandemic worldwide, we believe that sharing our experience as a hard and early hit country has important implications for ensuring the protection of essential workers from the infection risks.
Among the screened HCWs of our Hospital, we found a cumulative incidence of SARS-CoV-2 infection of 11.33% (95%CI: 9.72-13.21). The disease was found to be mild in most cases, requiring hospitalization in only 5% of cases and nobody died. In our research, the relative incidence rate of SARS-CoV-2 was higher between HCWs working in COVID-19-dedicated wards than HCWs working in other wards.
Among HCWs, occupational exposure to the virus is certainly of great concern. Specifically, due to the possibility of transmission by droplets, manoeuvres like intubation, non-invasive ventilation and manipulation of oxygen masks or continuous positive arterial pressure (CPAP) helmets might be considered as potentially risky. Consequently, Intensive care (IC) and first aid personnel has often been regarded as the most exposed. In a similar manner, in the view of a large amount of SARS-CoV-2 hospitalized infected patients in the infectious and respiratory diseases units, working in the aforementioned facilities has been assumed to be risky too. Accordingly, the prevalence of infection among the medical staff members in centres receiving COVID-19 patients has been reported as much higher than that of centres not receiving COVID-19 patients [31]. In the same way, physicians, nurses, and health technicians with direct contact with COVID-19 patients have been more likely to be infected than those without, like clerical workers [18]. However, several studies have not identified a statistically significant difference in the proportion of infected HCWs from hospital units firstly involved in close contact with COVID-19 patients compared with intermediate or low-risk units [22,32,33] At a first glance, we might be surprised that all these potentially hardest hit groups of HCWs have not been the most infected by the virus at the very end.
One might however observe that this finding is not that surprising, since PPE and other general protective measures have been initially unavailable in clinical departments other than first-line infectious and respiratory diseases units. Accordingly, non-first-line HCWs might have been exposed to a heightened risk of infection.
Notably, in our Hospital, the Infections Committee issued a protocol to all HCWs regarding the management of suspected or ascertained COVID-19 cases on January 31, 2020, before the first Italian case was confirmed. This helped estimating in advance the real urgency for PPE for all the staff, not only for the most exposed wards. Furthermore, and perhaps most importantly, our Hospital immediately trained all the HCWs on the appropriate use of PPE. Over 1.200 HCWs attended several courses before and during the outbreak.
Consequently, we may safely agree that, assuming the equal accessibility and proper use of PPE of all the HCWs of our Hospital, the great and more prolonged contact with COVID-19 patients remained a crucial risk factor for SARS-CoV-2 infection among HCWs. This finding is indeed confirmed by the literature [37,38].
The majority of the available studies have attributed a mild or moderate disease severity to the COVID-19 infected HCWs [16,21,[39][40][41]. Moreover, although possibly under-reported, the case-fatality rate varies between countries and time of disease outbreak but remains quite low [42] [43].
Despite the well and sadly known high case fatality rate in our country [44], and the warning of a world-wide echo in the early pandemic phases [45], the sheer consistency of our results should come as no surprise to us.
Hence, severe illness and death predominantly occurs in elderly patients with underlying medical comorbidities and, conversely, affected HCWs are usually younger people with less predisposing conditions. Furthermore, early symptoms are more easily noticed by HCWs themselves and treatment urgently started.
Since several conditions have been associated with severe illness and mortality in the community [46], HCWs risk factors and medical comorbidities have not been uniformly reported in literature [15,42,47].
We have considered HCWs lifestyle and medical history of our Hospital staff. In our experience, being an active smoker is associated to a reduced rate of SARS-CoV-2 infection. Though, the impact of current smoking on COVID-19 is controversial [48,49]. Our results are in line with those by Lippi et al [50] but we believe that this is an intriguing diatribe. Whether cessation of smoking indeed improves pulmonary function, it is as well known that this benefit is decreased by the cumulative injury of smoke to the lungs. Therefore, it is strongly associated with the smoking period of time. This figure, which indicates a complicated relationship between smoking history and the severity of COVID-19, often lacks in literature and further research is warranted.
We are aware of several limits of this study. Firstly, there are some missing data on all the tested HCWs. Specifically, clinical data are available only for the positive HCWs sample. Furthermore, due to the retrospective nature of this study, recall missing data results tricky.
Secondly, we have tested the HCWs only in presence of typical symptoms or unprotected contacts with COVID-19 patients. Indeed, in the eye of the storm, in our hardly hit Hospital at the beginning of the pandemic, nasal swabs for SARS-CoV-2 RNA detention in HCWs have been justified by epidemiological and clinical criteria. With the gift of hindsight, we currently know that this has been a weakness.
However, the decisions of those days have been determined by the emergency circumstances, definitely unrivalled in the history of medical care, which have overwhelmed the health care system on all fronts and have not certainly ceased.
In the event of a regrettably increasingly plausible second wave of contagions, which seems we are now aware of the several, potential benefits of universal staff testing [23]. On the one hand, it would boost working staff depletion by identifying only the symptomatic HCWs who really have COVID-19 and avoiding the substantial proportion of faulty self-isolation of the others. On the other, it would undoubtedly limit the risk of asymptomatic spread of SARS-CoV-2 [17,18,32,51]. Asymptomatic HCWs might become, indeed, a significant risk factor for patients, colleagues, and the community.

Authors contributions
All authors provided critical feedback and helped shape the research, analysis and manuscript. VN, AM,