High SARS-CoV-2 Prevalence among Healthcare Workers in Cochabamba, Bolivia

Healthcare workers (HCWs) are at increased risk of SARS-CoV-2 infection. The aim of the study was to estimate the SARS-CoV-2 seroprevalence among HCWs in Cochabamba, Bolivia and to determine the potential risk factors. In January 2021, a cross-sectional SARS-CoV-2 seroprevalence study was conducted in 783 volunteer clinical and non-clinical HCWs in tertiary care facilities. It was based on IgG detection using ELISA, chemiluminiscence, and seroneutralisation tests from dried blood spots. Analysis revealed a high seroprevalence (43.4%) of SARS-CoV-2 IgG antibodies. The combination of anosmia and ageusia (OR: 68.11; 95%-CI 24.83–186.80) was predictive of seropositivity. Belonging to the cleaning staff (OR: 1.94; 95%-CI 1.09–3.45), having more than two children in the same house (OR: 1.74; 95%-CI 1.12–2.71), and having been in contact with a close relative with COVID-19 (OR: 3.53; 95%-CI 2.24–5.58) were identified as risk factors for seropositivity in a multivariate analysis. A total of 47.5% of participants had received medication for COVID-19 treatment or prevention, and only ~50% of symptomatic subjects accessed PCR or antigenic testing. This study confirms a massive SARS-CoV-2 attack rate among HCWs in Cochabamba by the end of January 2021. The main risk factors identified are having a low-skilled job, living with children, and having been in contact with an infected relative in the household.


Introduction
Since December 2019, the COVID-19 pandemic has represented a major crisis worldwide. Particularly in developing countries, the health system is on the verge of collapse, and the shortage of oxygen supply and healthcare workers (HCWs) [1] makes management particularly challenging. In addition, transmission is exacerbated by the shortage of personal protective equipment and diagnostic tests.
In Bolivia, the first confirmed case of coronavirus disease 2019 (COVID-19) was reported on 10 March 2020 [2], followed by a nationwide lockdown that began on 22 March and ended on 10 May [3]. In total, national authorities reported 160,124 confirmed positive cases, with 9165 deaths by the end of 2020 [4]. To date, three epidemic waves have been reported in Bolivia: the first epidemic peak occurred in July and August 2020, the second peak occurred in January and early February 2021, and a third peak occurred in May and June 2021 [5].Vaccination campaigns for COVID-19 began on 29 January 2021, with doses reserved for HCWs [6].
Cochabamba is the fourth largest city in Bolivia, with~826,316 inhabitants in 2020 [7]. It is located in a valley in the Andes mountain range. Cochabamba has been one of the most affected cities in Bolivia during the COVID-19 pandemic, with frequent cases among HCWs in June 2020, leading to the closure of some hospitals.
In this context, we conducted a seroprevalence study among HCWs in Cochabamba, Bolivia to estimate exposure and the risk factors associated with SARS-CoV-2 infection.

Study Design and Participants
In January 2021, during the second epidemic wave and before the start of vaccination campaigns, 783 clinical and non-clinical HCWs from different healthcare facilities in Cochabamba, Bolivia  All participants were included, irrespective of whether they had had COVID-19 symptoms or not. All consenting subjects were asked to answer a questionnaire and to provide a fingerstick capillary blood sample on blood cards.
The questionnaire allowed for collection of information concerning demographics, occupation, hospital department, working time since the beginning of the pandemic (including lockdown), household size and composition, blood type, tobacco smoking, types of face mask used (surgical, FFP2/KN95, cloth), contacts with confirmed COVID-19 cases (patients, family, friends, and/or colleagues), symptoms since the beginning of the pandemic (ageusia, anosmia, or at least two of the followings: fever, cough, tiredness, rhinitis, sore throat, headache, conjunctivitis, or diarrhea), comorbidities, previous COVID-19 diagnosis (PCR, antigen, and serology tests), hospitalization, treatment, and chlorine-dioxide ingestion.

Ethical Approval
The ethics committee of the Dr. Mario Ortíz Suárez hospital in Bolivia (No. 09/2020) approved this study. HCWs agreed to participate in the study by providing written informed consent prior to enrollment. Blood samples, personal data, and biological results were irreversibly anonymized.

Serological Testing
Sera were recovered from dried blood spots (DBS) using a standardized quantitative elution protocol. Briefly, four 4.7 mm discs were punched and stored in 0.5 mL 2D FluidX 96-format tubes (Brooks life sciences, Chelmsford, MA, USA) [8]. Elution was performed with 380 µL of PBS, shaken at 1050 rpm for one hour and left for 18 h at room temperature before analysis. The sample eluates were tested with three different SARS-CoV-2 serological tests: (i) a semi-quantitative chemiluminescent immunoassay (CLIA) detecting IgG against the receptor binding domain (RBD) (Access SARS-CoV-2 IgG II, Beckman, CA, USA), run on the DXI instrument; (ii) a commercial semi-quantitative ELISA detecting IgG against the S1 domain of the spike protein (Anti-SARS-CoV-2, EUROIMMUN, Lübeck, Germany; sensitivity 87%, specificity 97.5% [8]), run on the EUROLabWorkstation instrument; (iii) a highly specific in-house virus-neutralization test (VNT) cytopathic effect (CPE)-based, 100 TCID50 of the SARS-CoV-2 BavPat1 strain, with serum dilutions from 1:20 to 1:160, as previously described [9]. The two commercial assays were performed according to the manufacturer's instruction. However, for the EUROIMMUN S1 ELISA assay, the cut-off ratio to define equivocal results was decreased to 0.7 (instead of 0.8). CPE-based VNT was performed for all samples with either a positive or equivocal ELISA result or a positive CLIA result.

Diagnostic Interpretation
Samples were considered positive when at least two tests were positive among the S1 ELISA (Ratio ≥ 1.1), the CLIA RBD assay (titre > 10 AU/mL), and the VNT test (titre ≥ 40).
Continuous variables are presented as means with standard deviations (SD), and categorical variables as numbers and percentages with odds ratio (OR) and 95% confidence interval (CI) when relevant. Variables were compared using the Pearson's chi-square test to identify those associated with the presence or absence of IgG antibodies against SARS-CoV-2. Variables used to evaluate risk factors for seropositivity were age, sex, hospital, occupation, clinical or non-clinical occupational group, working during lockdown, household and composition, blood type, tobacco smoking, face-mask types used, and COVID-19 contact. Variables with p-values < 0.05 were considered statistically significant (Bonferroni-adjusted p < 0.05) and were included in multivariate logistic-regression analysis to determine whether each variable was an independent factor for seropositivity. Analyses were performed using IBM-SPSS Statistics v 24.0.0.0 (Chicago, IL, USA) and GraphPad Prism 7.00 (San Diego, CA, USA) software.

Characteristics of the Study Population
The mean age of participants was 39.29 (SD 12.3) years, and participants were predominantly female (579/783, 73.9%). The O blood type was the most common, followed by the A blood type (76.7% and 16.7%, respectively, in 708 participants, who provided the information). An underlying disease was reported by 23 Most of the study population (656/769, 85.3%) worked during the lockdown, and 24/763 (3.1%) were hospitalized due to COVID-19. Details of demographics, occupational exposure, symptoms, and tobacco smoking are presented in Tables 1, S1 and S2. Of note, prior to sample collection, among the health care facilities involved in the study, ten deaths were reported, and six HCWs did not return to work permanently due to COVID-19 sequelae. Missing, n (%) 14 8 1 Ageusia, anosmia, or at least two of the followings: fever, cough, tiredness, rhinitis, sore throat, headache, conjunctivitis, or diarrhea. Abbreviations: n: number; NA: Not applicable; SD: Standard Deviation. Significant values are in bold (after Bonferroni correction for variables for more than 2 categories).

Diagnostic Tests
Three hundred nineteen out of 340 (93.8%) seropositive participants had positive results for all three assays (ELISA, CLIA, and VNT). For ELISA and CLIA tests, the sensitivity was 98.2% and 97.0%, the specificity was 97.3% and 99.0%, the positive predictive value was 97.0% and 98.8%, and the negative predictive value was 98.4% and 96.5%, respectively, with almost perfect agreement between both tests (Cohen's k: 0.91). In VNT, only positive or equivocal ELISA and CLIA results were tested, showing a VNT sensitivity and specificity of 100%.

Diagnostic Tests
Three hundred nineteen out of 340 (93.8%) seropositive participants had positive results for all three assays (ELISA, CLIA, and VNT). For ELISA and CLIA tests, the sensitivity was 98.2% and 97.0%, the specificity was 97.3% and 99.0%, the positive predictive value was 97.0% and 98.8%, and the negative predictive value was 98.4% and 96.5%, respectively, with almost perfect agreement between both tests (Cohen's k: 0.91). In VNT, only positive or equivocal ELISA and CLIA results were tested, showing a VNT sensitivity and specificity of 100%. Among seropositive participants, 22.9% (76/332) did not report any of the symptoms mentioned in the questionnaire. Anosmia (OR: 30.42; 95% CI 16.09-57.50) and the combination of anosmia and ageusia (OR: 68.11; 95% CI 24.83-186.80) were highly predictive of SARS-CoV-2 seropositivity (Figure 1). Of participants who reported symptoms or were seropositive for IgG,~50% had a history of previous PCR and/or antigen testing. Participants who reported an underlying disease were 4.14 times more likely to be hospitalized (OR: 4.14; 95% CI 1.82-9.42).

Diagnostic Tests
Three hundred nineteen out of 340 (93.8%) seropositive participants had positive results for all three assays (ELISA, CLIA, and VNT). For ELISA and CLIA tests, the sensitivity was 98.2% and 97.0%, the specificity was 97.3% and 99.0%, the positive predictive value was 97.0% and 98.8%, and the negative predictive value was 98.4% and 96.5%, respectively, with almost perfect agreement between both tests (Cohen's k: 0.91). In VNT, only positive or equivocal ELISA and CLIA results were tested, showing a VNT sensitivity and specificity of 100%.
Consistent with previous reports in the international literature, anosmia and the combination of anosmia and ageusia were the most specific predictive symptoms associated with evidence for previous SARS-CoV-2 seropositivity [23,24]. A total of 22.9% of seropositive participants did not mention any symptoms [23,25]. The number of asymptomatic cases and the lack of diagnosis in almost 50% of symptomatic HCWs could have contributed to a substantial number of secondary infections and to the high seroprevalence rates observed.
In univariate analysis, seropositivity did not differ by sex, health facility, age, or the type of protective mask used. In multivariate analysis, whilst clinical HCWs were more exposed to COVID-19 patients than non-clinical HCWs (77.4% vs. 46.4%), cleaning workers were associated with the highest seroprevalence (57.9%), as previously described [26,27]. This excess risk in low-income occupational groups may be related to their frequent use of public transportation, less adherence to social distancing and the use of face masks, living in overcrowded housing (50% lived with more than five persons in our study), and limited access to diagnostic tests. Being in contact with a close relative with COVID-19 and living with more than two children were also significantly associated with seropositivity. As previously reported [8], children may play an important role in household transmission, reinforced by the frequent occurrence of asymptomatic infections [28].
Active tobacco smokers exhibited lower IgG seroprevalence compared with nonsmokers [8,[29][30][31]. This should not be taken as a false message to promote smoking to avoid infection [32]. On the one hand, tobacco smoking is independently associated with a clear excess risk of severe disease (e.g., cancer, respiratory and cardiovascular diseases), and, on the other hand, it is possible that tobacco smokers infected with SARS-CoV-2 are prone to develop severe disease [33]. Previous studies identified a lower risk of SARS-CoV-2 infection among people with blood type O [9,[34][35][36][37]. We did not find this trend in our study, but the overwhelming proportion of blood group O in Bolivia [38] (76.7% in our study) makes this type of assessment difficult and suggests that larger studies are needed to precisely assess this association.
In Bolivia, the spread of COVID-19 was accompanied by great concern among the general population, misinformation, and common self-medication [39]. The antiparasitic ivermectin and the antibiotic azithromycin, alone or combined with the anti-malarial hydroxychloroquine, grabbed attention [40,41] and were widely used as treatment or prophylaxis for SARS-CoV-2 infection after having shown some antiviral activity in vitro [42,43] and sometimes recommended by the health authorities [44]. We found that ivermectin (33.8%) and azithromycin (26.9%) were frequent used (also occurred among HCWs) amid the fear of contracting the disease (28.7% in asymptomatic participants) or to treat it (64.3% in symptomatic participants). In addition, chlorine dioxide, a derivate of bleach, which is potentially toxic and can cause serious and potentially life-threating side effects, falsely marketed as "Miracle Mineral Solution" for prevention and treatment of COVID-19 [45], has been commonly used by the Bolivian population (11.3% in our study) and even promoted by health authorities [1].
Our study has some limitations. It was carried out among volunteer healthcare workers, which may have resulted in a bias towards those who have already been infected with SARS-CoV-2, either presumed or confirmed. Therefore, the prevalence figures provided should be considered as order-of-magnitude indications. On the other hand, the strength of this research program lies in the consistency of the information collected (with reference to other international studies and to what is known about the dynamics of the epidemic in Bolivia and the behavior of the population) and in the possibility of collecting seroprevalence data prior to the implementation of vaccination campaigns.

Conclusions
In conclusion, this study shows high seroprevalence rates among HCWs in Cochabamba, Bolivia, suggesting that the first wave of COVID-19 had a considerable impact on this occupational group. Better information and protective measures, continuous awareness and sensibilization, and a greater access to diagnostic tests are needed to reduce the risk of infection.