Prevalence and Risk Factors Associated With Sars-cov-2 Infection in Healthcare Personnel of the Spanish Central Military Hospital. Results of a Case-control Study

Background. Spain is the country with the highest number of health care workers affected by COVID-19 disease in the world. The aim of this study was to describe the epidemiology of SARS-COV-2 infection in HCW at the Gómez Ulla Military Hospital. Methods. A case-control study was conducted during the rst outbreak of COVID 19 at GUMH. The study was extended to the total number of HCW in the hospital who met the inclusion criteria. HCW of the GUMH were civilian and military personnel that included Nursing and Medicine students of Spanish Military Academy, Medicine specialty residents and Nursing residents of Medical Surgical Specialty in Operations (EMQO). An ad hoc questionnaire of 33 items was made up. The questionnaire was sent by e-mail to the entire population of study. With this questionnaire personnel were classied into sick (cases) or healthy (controls). Results and Conclusions. 150 professionals answered the questionnaire. Cases were dened as those who tested positive in the diagnostic tests (n=28, 20.7 %) and controls were those who tested negative (n=107, 79.3 %). Therefore, the prevalence of SARS-CoV-2 in the GUMH was 20.7 %. Of the total number of cases, 64.3 % were men (p<0.05), with a mean age of 47.1 (SD 13.3), with a mean BMI of 25.3 (SD 3.8), 48.2 % were overweighted. Of the total cases, 59.3 % had A blood group type and 69.2 % were Rh positive. 50 % were physicians, 32.1 % were nurses and 17.9 % were auxiliary nurses (p<0.05). Cases and controls with vitamin D deciency and who took supplements had a lower risk of suffering COVID-19, with signicant differences. Fever, cough and diarrhea were found in at least 50% of the sample with signicant differences.


Introduction
The pandemic caused by SARS-Cov-2 (COVID- 19) rst identi ed in Wuhan, China, remains as a global health problem. Following the experience during the 2002 outbreak of severe acute respiratory syndrome (SARS), an important factor in controlling transmission is the protection of healthcare workers (HCW), as 1725 cases among HCW were infected with SARS (1). In the ongoing epidemic, infection among HCW has been increasing since the rst 15 affected cases were reported in Wuhan (2). It is necessary to know the risk factors and nosocomial transmission to protect the health of staff.
Spain is the country with the highest number of HCW affected by COVID-19 disease in the world. A report by the European Centre for Disease Control and Prevention (ECDC) highlights that 20% of COVID-19 in Spain affects this group, while in Italy this percentage is 10%, and in the United States and China 3% and 3.8%, respectively (3). Up to May 21st, 2020, a total of 250 287 cases of COVID-19 were reported to the National Epidemiological Surveillance Network (RENAVE) in Spain. 40921 cases correspond to HCW (4).
It is necessary to adopt primary preventive measures to reduce morbidity and mortality among HCW. To establish such measures, it is important to know which factors are linked to the different health determinants of the most exposed armed forces HCW. To know these causes will contribute to the design, development, and implementation of public health measures to respond effectively to the ongoing SARS-CoV-2 epidemic.
The aim of this study was to describe the epidemiology of SARS-COV-2 infection in HCW at the Gómez Ulla Military Hospital (GUMH, Madrid, Spain). Secondary aims were to describe the prevalence of COVID-19 infection in healthcare personnel at baseline, to analyze the probability of exposure to different risk factors in cases versus controls, to detail the signs and symptoms presented by cases, and to assess which factors in uenced the onset of signs and symptoms.

Design, setting and participants
A case-control study was conducted during the rst outbreak of COVID 19 at GUMH. The total study population were 1171 health professionals working in that hospital. No sampling was selected, as the study was extended to the total number of HCW in the hospital who met the inclusion criteria. HCW of the GUMH were civilian and military personnel that included Nursing and Medicine students of Spanish Military Academy, Medicine specialty residents and Nursing residents of Medical Surgical Specialty in Operations (EMQO).
All HCW who were not physicians, nurses, Medicine or Nursing residents, or Nursing assistants, were excluded from the study.

Study variables
The collected variables collected were age, sex, height, weight; body Mass Index (BMI) was calculated as weight (kg)/height 2 (m 2 ); blood type (ABO and Rh); type of healthcare professional; physician or nurse specialty if they had; civilian or military personnel; years of experience, (coded as < 10 years, 10-20 years and > 20 years); eld of work (hospital emergency, emergency areas enabled for this purpose, intensive care, hospitalization, other); to receive seasonal in uenza vaccine in 2019-2020 (yes, no); to suffer high blood pressure (HBP), diabetes, heart disease, Chronic Obstructive Pulmonary Disease (COPD), asthma, malignant tumors or other chronical disease; tobacco consumption (yes, no, ex-smoker); alcohol consumption (never drank, less than once a week, drink at least once a week); measurement of stress and distress in professionals through the Perceived Stress Scale (PSS), in its validated Spanish version, reduced 4-item version. In addition, the following variables were collected in relation to COVID-19 infection: currently working (yes, no); COVID-19 signs and symptoms currently present (yes, no); to have been tested for COVID-19 (yes, no); type of COVID-19 test (PCR, rapid antigen test, ELISA); test result (positive, negative); to be daily exposed to COVID-19 patients (yes, little exposure, no exposure); average time of exposure to COVID patients, per shift (< 5 hours, 5-8 hours, > 5 hours); work shifts performed (morning shift, afternoon shift, night shift, 24-hour shifts, 12-hour shifts, back-up); number of hours of daily work (8, 10,12,14,16,24); days off weekly in the period of COVID-19 care (0, 1, 2, more than 2); extent to which you have received adequate training in handling personal protective equipment (PPE) (a lot, a lot, a little, not at all); PPE removal training (always, often, occasionally, rarely, hardly ever, never); to keep the removal steps safely (always, often, occasionally, hardly ever, never); whenever left a more contaminated room or area, to remove the outermost layer of PPE (always, often, occasionally, rarely, almost never, never); amount of protective material (a lot, quite enough, little, none); adequate material (yes, no); to re-use the material (yes, no); if re-used material, to indicate which (gloves, mask, impervious gown, tights, hat, protective screen, goggles).
An ad hoc questionnaire of 33 items was made up. The questionnaire was sent by e-mail to the entire population of study. With this questionnaire personnel were classi ed into sick (cases) or healthy (controls).

Ethical aspects
All individuals received an initial section in the questionnaire in which they were informed in detail of the details, objectives, and other aspects of the study. Likewise

Statistical analysis
All risk factors were assessed and the probability of having been exposed or not to these factors was calculated versus the controls.
Variables were described with absolute and relative frequencies. To assess which quantitative variables followed a normal distribution, the Kolmogorov-Smirnof test was performed. To assess whether there was a statistically signi cant association (p < 0.05) between the case-control status and of the study variables, Pearson Chi-Squared Test (qualitative variables) and Student's t test (quantitative variables) were used. To assess the independent association of the factors studied with the disease, odds ratios (OR) with their 95% con dence intervals were calculated using logistic regression models. The models were adjusted for the confounding variables that were considered appropriate, and the existence of interaction between them was evaluated. Statistical processing of the data was carried out using SPSS (Statistical Package for Social Sciences) V25.0 for Windows.

Results
Of the 1171 health professionals of the initial population, 703 (62.0 %) were women, had a mean age of 48.7 years, 577 (49.3 %) were nurses, 365 (31.2 %) physicians and 229 (19.5 %) nursing assistants. 150 professionals answered the questionnaire. 56.2 % were women, with a mean age of 46.7 years (SD 13.4). 47.6 % were nurses, 29.6 % were physicians and 27.3 % were auxiliary nurses. 63.2 % had 10 years of experience and 66.4 % were military personnel. 36.7 % worked in intensive care, 35.9 % in hospitalization service and 32.0 % in the emergency service. 62.5% were aged 50 years or over. Mean BMI was 24.9 kg/m 2 (SD 3.9) and 61.7% were normal weighted. 47.7 % had A blood group and 77.3 % were Rh positive. The results presented in this study refer to the personnel who answered to the questionnaire (Table 1). * Chi-cuadrado, ** T-Student, p-value < 0.05, M -arithmetic mean, DT-standard deviation, NS/NC-don't know/no answer, nabsolute frequency, BMI-body mass index.
Of the 150 professionals who answered the questionnaire, 15 did not undergo any diagnostic test, so they could not be classi ed as cases or controls. Cases were de ned as those who tested positive in the diagnostic tests (n = 28, 20.7 %) and controls were those who tested negative (n = 107, 79.3 %). Therefore, the prevalence of SARS-CoV-2 in the GUMH was 20. week. In addition, the mean stress among the cases was 2.86 (2.1) (p < 0.05) and 10.7 % reported feeling a medium level of stress (Table 1).
Concerning the factors of direct exposure to COVID-19, which could have in uenced transmission, 82.1 % of the cases reported having been exposed to COVID-19 patients, 64.3 % were exposed for at least ve hours or more, 77.8 % of the cases worked during the morning shift (p < 0.05), with an average daily working hours of 8.6 (SD 1.0) and a mean number of days free of work of 1.25 (SD 0.7). The percentage of cases that reported having done a lot of training for handling PPE was 3.7 % compared to 32.1 % of cases that reported not having done any training. There were no cases among those who reported having always been supervised in the removal of PPE versus 57.7% of the cases who reported having never been supervised. The 25.0 % of the cases reported having safely followed the steps for PPE removal compared to 33.3 % who did not, nding a statistically signi cant association (p = 0.009). 52.2 % of the cases did not always remove the outermost layer when leaving contaminated areas, while 69.8 % of the controls always did so. The percentage of cases and controls that reported that there was little protective material was 81.5 % in cases and 77.6 % controls. 60.7 % of the cases and 68.3 % of the controls rated the material they had as adequate, and 78.6 % of the cases and 90.5 % of the controls reported that they had re-used some element of the protective material sometimes (Table 2).  (Table 3). Differences were also found in relation to fever and sex, being higher in men (88.2 %) than in women (11.8 %), in those who presented chronic comorbidities (58.8 %) versus those who did not (41.2 %), as well as in those who reported not having vitamin D de ciency (57.1 %) versus those who did (42.9 %). A statistically signi cant association was found between sore throat and those who did not receive the u vaccine (54.5 %) vs. those who did (45.5 %) (p < 0.05). Differences were also found between cases with pneumonia and overweight (100.0 %), 0 blood group (50.0 %), and those who did not perform physical activity (75.0 %). Cough was associated with those cases with 10 years or more of experience (61.5 %). On the other hand those cases that presented diarrhea were signi cantly associated with not taking vitamin D supplements (57.1 %)(p < 0.05) and cases with respiratory distress were more among those with 0 blood group (75.0 %)(p < 0.05). In addition, women presented other respiratory signs in a higher percentage (75 %) than men (25 %), with a signi cant association (Table 3).

Discussion
The questionnaire was distributed to 1171 individuals of a total of 1260. 89 professionals were discarded because they did not meet the inclusion criteria. The response rate was 12.8%, which is lower than in other studies, where the response rate ranged from 86.7% (only 83 individuals were sent the questionnaire) [5] to 73.8% [6]. It should be taken into account that of the 150 professionals, 15 did not undergo any diagnostic test and therefore could not be classi ed as a case or control, so there were 135 classi ed participants. A sample design with random sampling and strati cation by sex and type of health professional should have included 107 women (72 were included) and 66 men (60 were included), 84 nurses (compared to 61 in our study), 53 physicians (compared to 38 in our study) and 34 nursing assistants (35 were included). The COVID 19 prevalence obtained in this Hospital was 20.7%. In other studies the seroprevalence ranged from 0.75% (95% CI 0.0-8.13)[6], to 16.4 % in a study done in Spain [7] or 18.5% in a study done in Ontario [8]. A total of 54.5 % of the healthcare personnel who answered to the questionnaire were women, similar gures were found in the controls (59.6%), but the majority of cases were men (64.3 CI 95% 56.1-72.5), and a statistically signi cant association was found. Other studies have found different results. In two of them carried out in Spain on health personnel (2020), 23.4 % and 23.8% of the cases, respectively, were men [4] and another study in Girona showed a percentage of women of 90.1% [7]. Another study in Wuhan had 50.0% of women [5], another in Canada had 81.7% [8], another in Buenos Aires had a percentage of women of 71.8% [6] and another in Germany had a percentage was 73.0%. This difference may have been because many medical o cers from the Military Health Corps were commissioned, in reserve status (personnel who cease to be on active duty when they reach 61 years of age). All the personnel were male and with a high age range, which may have increased the number of cases and explained the increase in male cases (41.7 % vs. 28.6 %). The mean BMI (24.9 kg/m 2 , SD 3.9) of the total and controls was within the normal weight range, while the mean BMI of the cases was elevated (25.3 kg/m 2 , SD 3.8) and was in the overweight range. One study used data from the UK Biobank (n = 285 817) to show that being overweight increased the risk of COVID-19 by 44.0 % (relative risk [RR] = 1.44; 95 % CI, 1.08-1.92; p = 0.010) [10]. Two studies showed that the odds of having COVID-19 were increased by 30 % (OR = 1.30; 95 % CI, 1.09-1.54; p = 0.003) [11] and 38 % (OR = 1.38; p < 0.001) [10] respectively, among overweight/obese individuals. The mechanisms responsible for the increased prevalence of COVID-19 in overweight/obese individuals are unknown. Knowledge of other viral infections, such as in uenza, provide insight into how excess weight increases the risk of COVID-19 severity. Understanding why this pathology increases the risk of severe COVID-19 is critical to ensure appropriate preventive and interventional therapies against this novel coronavirus.
The percentage of cases with A + blood type was higher than the other groups. The difference was not statiscally signi cant, but clinically relevant. Similar data were found in a study done in China where group A was associated with a higher risk of COVID-19 infection compared to AB and O types [12]. In another study authors suggest that while A type may play a role in increasing susceptibility to COVID-19 infection, O type may be somewhat protective. However, when infected, blood type does not appear to in uence clinical outcome [13]. Another study done in Changsha First Hospital, China, the proportion of patients with A blood type in the COVID-19 group was signi cantly higher than that in the control group (36.9 % vs.27.5 %, p = 0.006) and concluded that patients with A blood type had a higher risk of SARS-CoV-2 infection, while O type was associated with a lower risk, indicating that certain ABO blood groups correlated with susceptibility to SARS-CoV-2 [14].
Concerning the professional group physicians were overrepresented in the cases with respect to the controls, with a statistically signi cant difference, as was found in another study conducted in a Wuhan Hospital [5]. Different results were found in other studies where the professional groups that presented higher susceptibility were nurses and assistants who spent more time in contact with patients where the values ranged from 20.2-52%[8] [7]. This fact could be due to the performance of bronchoscopies, intubations, or site examinations where exposure levels were higher compared to the care activities performed by nurses and auxiliaries, except for the suctioning activities of ICU staff [15]. Cases with 10 or more years of experience and membership in the military administration were proportional to the total and controls.
The percentage of cases who received the in uenza vaccine was very low compared to those who did not, but in the same proportion between cases, controls and totals, and no difference was observed. Several studies show that in uenza vaccination does not increase the risk of COVID-19 infection [16]. This study provides trust against speculation that in uenza vaccine increases the risk of COVID-19 infection. In addition, other studies claim that in uenza vaccination may be associated with lower severity and lower mortality from COVID-19 [17][18].
Concerning vitamin D, those with vitamin D de ciency who also took supplements had a lower risk of COVID-19, in the total and both in the cases and controls, with a signi cant association. The factor that produces positive effects, reducing transmission and improving prognosis, is vitamin D supplementation and not the de cit itself. Several studies reported an association between CRP positivity and vitamin D de ciency. The exact e cacy of vitamin D supplementation for the prevention or as adjunctive treatment of COVID-19 remains to be determined, but several ongoing randomized controlled trials (RCTs) are actively investigating these potential bene ts [19] [20].
Low stress was associated to COVID-19 (low stress level in 89.3 % cases vs. medium stress level 10.7 % cases). These results were not consistent with another study that reported that the presence of psychosocial risks had important consequences for the quality of care and increased the probability of errors [21].
Likewise, the number of cases among personnel who reported being more exposed was higher than the rest (82.1 % vs. 0.0 %), and no differences were found, since the number of exposed controls was also very high with respect to those not exposed. This fact could suggest that exposure is not as important as nosocomial or community transmission. The same results were not found in a study done in Spain, where greater direct exposure led to a higher number of infections [7]. It would be advisable to maintain a safe distance between professionals, hand hygiene and the use of personal protective equipment. The number of cases among personnel who worked during the morning shift and who had fewer days off than the controls was higher than in other shifts or in the case of the controls, with a statistically signi cant difference. The morning shift is the one with the greatest work overload and the highest number of health professionals. These two factors, nosocomial transmission and overload, are two important factors to be taken into account for prevention [24]. It would be advisable to increase the number of healthcare personnel during this shift to reduce overload; other studies have obtained similar data [25].
Continuing with the epidemiological risk antecedents, the cases had fewer days off (1.25 SD 0.7) than the controls (1.44 SD 0.7). In relation to protective equipment, there were hardly any cases among those who reported adequate training in the use of the equipment.
Concerning signs and symptoms, fever, cough and diarrhea were found in at least 50% of the sample with signi cant differences, and the proportion of cases that also showed other symptoms such as headache or sore throat was not negligible [26] [27]. In a study conducted in Spain [4], fever and cough were also the most representative signs. Fever was identi ed more frequently and with signi cant differences, in men with chronic diseases and without vitamin D de ciency [4]. Sore throat was identi ed among those who had not been vaccinated against in uenza. In a speci c analysis, it was observed that health workers with pneumonia were signi cantly overweight and did not practice any type of sporting activity compared to those who did not have pneumonia. On the other hand, cough was signi cantly more frequent in healthcare personnel with more than 10 years of experience.
The most important limitation of this study refers to the sample. This difference could have been due to the short time available to the health personnel or to exhaustion after the practice of health care. Likewise, if a sample size calculation had been made with an error percentage of 5 % and a con dence level of 95 %, the number of individuals should have been 173 as opposed to the 150 that made up our study.

Conclusions
The prevalence of SARS-CoV-2 in the HCDGU HCWS was 20.7%.
64.3 % of the cases were men (p < 0.05), with a mean age of 47.1 (SD 13.3), the mean BMI of the cases was elevated to the overweight range. In the distribution by professional group in cases with respect to controls and total study participants, physicians were overrepresented, with statistically signi cant differences. Cases and controls with vitamin D de ciency and who took supplements had a lower risk of suffering COVID-19, with signi cant differences. The number of cases among morning shift personnel and with fewer days off than controls was higher than in other shifts, with a statistically signi cant association.
Fever, cough and diarrhea were found in at least 50% of the sample with signi cant differences, in addition to headache or sore throat.

Declarations
Authors' contributions MJAB designed this study and obtained funding. CFM, LEBO and LHP analyzed and collected the data. MJAB, MPV, JW and NPF wrote the manuscript. NPF, CFM, MPV and LHE provided critical revisions that are important for the intellectual content. All authors read and approved the nal manuscript.