Bacillus Calmette-Gu (cid:1) erin vaccine to reduce healthcare worker absenteeism in COVID-19 pandemic, a randomized controlled trial * Clinical Microbiology and Infection

controlled trial Trained immunity Objectives: The COVID-19 pandemic increases healthcare worker (HCW) absenteeism. The bacillus Calmette-Gu (cid:1) erin (BCG) vaccine may provide non-speci ﬁ c protection against respiratory infections through enhancement of trained immunity. We investigated the impact of BCG vaccination on HCW absenteeism during the COVID-19 pandemic. Methods: HCWs exposed to COVID-19 patients in nine Dutch hospitals were randomized to BCG vaccine or placebo in a 1:1 ratio, and followed for one year using a mobile phone application. The primary endpoint was the self-reported number of days of unplanned absenteeism for any reason. Secondary endpoints included documented COVID-19, acute respiratory symptoms or fever. This was an investigator-funded study, registered at ClinicalTrials.gov (NCT03987919). Results: In March/April 2020, 1511 HCWs were enrolled. The median duration of follow-up was 357 person-days (interquartile range [IQR], 351 to 361). Unplanned absenteeism for any reason was observed in 2.8% of planned working days in the BCG group and 2.7% in the placebo group (adjusted relative risk 0.94; 95% credible interval, 0.78 e 1.15). Cumulative incidences of documented COVID-19 were 14.2% in the BCG and 15.2% in the placebo group (adjusted hazard ratio (aHR) 0.94; 95% con ﬁ dence interval (CI), 0.72 e 1.24). First episodes of self-reported acute respiratory symptoms or in 490 and 443 (60.2%) participants, respectively (aHR: 1.13; 95% CI, 0.99 e 1.28). Thirty-one serious adverse events were reported (13 after BCG, 18 after placebo), none considered related to study medication. Conclusions: During the COVID-19 pandemic, BCG-vaccination of HCW exposed to COVID-19 patients did not reduce unplanned absenteeism nor documented COVID-19. Thijs ten Doesschate, Clin Microbiol Infect 2022;28:1278 © 2022 The Author(s). Published by Elsevier Ltd on behalf of European Society of Clinical Microbiology and Infectious Diseases. This is an open access article under the CC BY license (http://creativecommons. org/licenses/by/4.0/). randomized trial determine


Introduction
The COVID-19 pandemic, caused by the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), represents an important cause of morbidity in healthcare workers (HCWs) caring for COVID-19 patients. Moreover, the burden of COVID-19 patients puts enormous pressures on healthcare system capacities worldwide [1]. Early in the pandemic, when the effectiveness and timely availability of effective SARS-CoV-2-specific vaccines was not yet defined, preventing HCW absenteeism by other means was imperative to safeguard patient care. Non-specific vaccines were suggested as a means to reduce the incidence and severity of COVID-19.
Bacillus Calmette-Gu erin (BCG) vaccine, widely used to prevent tuberculosis, also provides non-specific effects against other infectious diseases [2]. These non-specific effects lead to a faster and more efficient innate immune response to heterologous pathogens, a process termed 'trained immunity' [3]. Both in vitro and in vivo beneficial effects have been observed for distinct viral pathogens, such as respiratory syncytial virus, influenza virus, and attenuated yellow fever vaccine virus [4].
We hypothesized that trained immunity induced by BCG vaccination may offer protection against COVID-19. We conducted a randomized trial to determine the efficacy of BCG vaccine to protect against all-cause unplanned absenteeism and absenteeism due to COVID-19 in HCWs during the COVID-19 epidemic in the Netherlands.

Trial design and oversight
BCG-CORONA was a parallel, double-blind, placebo-controlled randomized trial conducted in nine hospitals in the Netherlands. The study protocol was approved by the institutional review board of the University Medical Centre (UMC) Utrecht, registered at clinicaltrials.gov (Identifier: NCT04328441), and published [5]. All participants provided written informed consent. The trial was investigator-initiated and funded by the UMC Utrecht and Radboud UMC Nijmegen, the Netherlands.

Trial participants
Participants were adult (18 years) HCWs working in the participating hospitals or their affiliated ambulance services, with expected exposure to COVID-19 patients as part of their clinical duties. Four participating hospitals were university hospitals and five were non-university teaching hospitals. Primary exclusion criteria were known allergy to BCG, active or latent Mycobacterium tuberculosis infection, any other active infection, immunocompromised state, and current or planned pregnancy. The diagnosis of latent M. tuberculosis infection was judged by the principal investigator. A positive tuberculin skin test after previous BCG vaccination or a positive interferon gamma release assay test or tuberculin skin test after a fully treated latent or active M. tuberculosis infection were excluded from this definition. A full list of eligibility criteria is provided in Table S1.

Procedures
Participants were randomized to BCG or placebo (1:1) using a computer-generated dynamic randomization algorithm, stratified for participating hospital in random blocks of 2, 4, or 6 sequences. Participants received either 0.1 mL of the Danish strain 1331, Statens Serum Institut (SSI), Denmark, equivalent to 0.075 mg attenuated Mycobacterium bovis, or 0.1 mL of normal saline solution as an intradermal injection in the left upper arm.
Participants and study personnel conducting participant followup were blinded to treatment allocation. Study personnel preparing and administering the study vaccines were not blinded but could not influence treatment allocations or data collection. The trial statistician that conducted the interim and final analyses was not blinded throughout the trial but was not otherwise involved in trial conduct.
Participants used a mobile phone application (Research Follow App, Your Research BV, Huizen, The Netherlands) to complete daily questionnaires regarding respiratory and systemic infection symptoms, adverse events, absenteeism, and weekly questionnaires regarding SARS-CoV-2 exposures, testing and healthcare visits. Push notifications and emails were sent if participants did not complete the questionnaire for seven consecutive days. If someone did not respond to this, study personnel contacted them by telephone. After 6 months of follow-up, the daily questionnaire was integrated into a weekly questionnaire to improve user convenience.

Efficacy assessments
The primary endpoint was the number of days of unplanned absenteeism for any reason, reported as the average proportion of sick-days out of planned work days. Secondary endpoints are listed in Table S2.
As part of routine hospital infection prevention procedures, HCWs were urged to stay at home in case of symptoms (e.g. respiratory symptoms or fever), were offered nasopharyngeal polymerase chain reaction (PCR) testing for SARS-CoV-2 free-of-charge at the hospital where they worked, and were allowed to return to work when cleared by a hospital-affiliated occupational physician. HCWs also had access to PCR tests or rapid antigen test free-ofcharge at public health service test sites. In symptomatic HCWs, negative antigen tests were confirmed with a PCR test. From 1 December 2020 onwards, HCWs identified as a close contact of an index case were also tested. Documented COVID-19 required self-reporting of any respiratory tract symptom (cough, dyspnea, nasal cold, sore throat, or loss of smell or taste of any reported severity) or fever (body temperature 38.0 C) within 7 days before or up to 14 days after a positive SARS-CoV-2 PCR or rapid antigen test. From participants in the UMC Utrecht, we verified the self-reported COVID-19 with the COVID-19 PCR results in the hospital-affiliated laboratory. Unplanned absenteeism because of documented COVID-19 was counted from the first day that the participant reported symptoms of any severity prior to the test date to the first day at work after the test date. If the HCW had reported symptoms for more than 7 days prior to the test date, the start date was fixed at 7 days prior to the test date. Presence of fever required a self-measured body temperature of 38.0 C and respiratory symptoms required the presence of cough or dyspnea with a severity score of 2 (mild symptoms), or the presence of nasal congestion or sore throat with a severity score of 3 (moderate symptoms).

Safety assessments
Participants were instructed throughout to contact the study team in case of (serious) adverse events. Participants that reported a hospital admission in the mobile phone application were contacted by the research team. Known systemic and local injection site side effects of BCG vaccination were solicited during the first 7 days after injection, and local side effects again at the end of the trial.

Statistical analysis
The primary endpoint was analyzed as total counts (i.e. one total count per participant) using a Bayesian negative binomial regression and corrected for participant baseline characteristics. Secondary count endpoints (e.g. number of days reporting symptoms) were analyzed using maximum likelihood estimation (frequentist model) and are reported as relative risk (RR) with 95% confidence interval (CI), and secondary time-to-event endpoints (e.g. incidence of COVID-19) using Cox-proportional hazard models reporting hazard rations (HR) and 95% CIs, with adjustment for the same baseline participant characteristics as the primary analysis. COVID-19 related absenteeism was analyzed using the same method as the primary endpoint.
Interim analyses were performed as described for the primary outcome and were performed biweekly from weeks 4 to 26 and monthly from week 26 to study end (March 2021). Unblinded results of the interim analysis and incidence of (serious) adverse events were reported to the data safety monitoring board, once per month from week 4 to 26 and once per two months from week 26 to study end.
Data were analyzed using the R statistical package v3.6.3 [6]. The full statistical analysis plan and sample size calculation are provided in Table S3.

Participants
From 24 March 2020 to 23 April 2020, 1511 participants were randomized (Fig. S1). Follow-up ended on 27 March 2021, and the database was locked on 18 April 2021. The median duration of follow-up was 358 person-days (interquartile range [IQR] 351e361) in the BCG group and 355 (IQR 351e361) in the placebo group. The proportions of follow-up days missing were 2% (IQR 0e7) and 1% (IQR 0e9), respectively.
The BCG and placebo groups were well balanced ( Table 1,  Table S4eS5). During follow-up, 578 participants in the placebo group vs. 593 in the BCG group reported having been tested for COVID-19 for a total of 1648 and 1680 tests, respectively. The median number of days between testing and reporting was 5 (IQR 3ae8) in both intervention groups and not different between the first and second 6-month period. From January 2021 onwards, 439 participants reported to have received at least one dose of a COVID-19 vaccine, mostly the BioNtech/Pfizer vaccine, which was equally balanced between groups (Fig. S2 Fig. 2. Only three participants were hospitalized due to COVID-19, one in the BCG group and two in the placebo group.
In post-hoc analyses, results were consistent among subgroups of participants who had prior BCG-vaccination and/or a positive tuberculin skin reaction test, or in subgroups of participants aged below and above 50 years. Fig. S3 to S10 and Tables S7 to S12 provide the outcomes of the exploratory and post-hoc analyses results.
The cumulative incidence of documented COVID-19 and absenteeism rate were relatively low in HCWs affiliated to the intensive care unit (ICU) and high in HCWs affiliated to the internal medicine, pulmonology, and emergency ward (Table S13).
In the affiliated laboratories of the UMC Utrecht and the Noordwest Ziekenhuisgroep Alkmaar, covering 122 of 224 documented COVID-19 cases, 9 participants had positive COVID-19 PCR tests that were not reported in the mobile application and 3 participants had negative COVID-19 PCR tests, registered as positive in the mobile application. A sensitivity analysis of these laboratoryconfirmed data did not lead to different results (Table S14).

Safety
Solicited (within 7 days) and unsolicited adverse events were higher in the BCG group than in the placebo group, and consisted mainly of pain and local skin reactions at the injection site (Table 3). At the end of the study, 18.0% of the participants in the placebo group had reported any complaints at the injection site, compared to 90.6% of the participants receiving BCG.
In total, 31 serious adverse events occurred; 13 in the BCG and 18 in the placebo group, including one participant in the BCG group who died as a complication of chemotherapy. None of these were considered to be related to BCG vaccination. Tables S15 to S18 summarize all (serious) adverse events.

Discussion
In this multicenter double-blind placebo-controlled trial, BCG vaccination did not reduce unplanned absenteeism due to any cause in HCWs. In addition, BCG vaccination failed to reduce the incidence of documented COVID-19, or acute respiratory symptoms or fever. Our findings corroborate those from a similarly designed study in elderly in the Netherlands, also performed during the COVID-19 pandemic, described in an accompanying manuscript [5].
The trial started during the peak of the first epidemic wave in the Netherlands in March 2020 and was completed during the third epidemic wave in March 2021. The percentage of unplanned absenteeism ran parallel to the COVID-19 incidence. However, only a fraction of unplanned absenteeism (25.4%) was attributable to documented COVID-19. A similar fraction (26.3%) was a consequence of being quarantined or isolated while physically capable of working. The aim of the trial was to ensure the continuity of health care during the COVID-19 pandemic. Therefore, the primary endpoint was unplanned absenteeism by any cause which captured all potential beneficial effects of BCG vaccination, while secondary endpoints, such as documented COVID-19, were explanatory. BCG vaccination has been associated with lower child mortality in low-income countries, and with a reduced incidence of respiratory infections, mainly from presumed viral origin, in patients who had been discharged from an intensive care unit in in Greece [7e9]. There was considerable in vitro evidence that BCG augments the innate immune response in a non-specific manner [10]. There was a worldwide interest in BCG potentially offering partial protection against COVID-19. Some retrospective and ecological studies from countries with differing BCG vaccination coverages suggested a protective effect of BCG [11], whereas others did not [12]. These are, however, inherently limited by confounding and other biases. Randomized controlled trials were required to resolve the question whether BCG protects against COVID-19 [13]. Unfortunately, our trial showed that BCG vaccination does not alter the incidence of COVID-19 in HCWs in a high-income European country.    Our trial had several study limitations. One of the inclusion criteria was 'hospital personnel expected to provide care to patients with COVID-19', while among 30% of the included participants did not expect to work on a COVID-19 department. Even if this hospital personnel does not work on a COVID-19 department, they may still have provided care for COVID-19 patients, e.g. radiology personnel involved in Computer Tomography (CT) scanning. No structural differences existed between the incidence of documented COVID-19 and the absenteeism rate between COVID-19 departments and other departments. The choice for self-reporting of symptoms and COVID-19 test results by participants carried a risk of underdetection and/or misclassifications of endpoints. However, COVID-19 PCR test results from two hospitals corresponded well with self-reported results in the mobile application. Sending push notifications and email alerts, and the adjustment to the weekly instead of daily self-reporting of symptoms, provided a risk of recall bias. This risk is probably lower for absenteeism and COVID-19 test results. Blinding was only partially possible due to the local injection site reactions that occur after BCG vaccination in most individuals. On the other hand, this did not result in a noticeable differential loss to follow-up, uptake of COVID-19 testing, or uptake of COVID-19 specific vaccination.
Our trial population of fit adults aged 65 years experienced few serious COVID-19 infections that resulted in hospitalization, let alone ICU admission or death. To determine the effect of BCG on severe endpoints, inclusion of older and more vulnerable Fig. 2. Cumulative incidences of documented COVID-19 (above) and first episode self-reported acute respiratory symptoms or fever (below). COVID-19 is defined as PCR-based or rapid antigen based detection of severe acute respiratory syndrome coronavirus 2 in a respiratory sample in combination with respiratory tract symptoms (cough dyspnoea, nasal cold, sore throat, or loss of smell or taste of any reported severity) or fever (body temperature 38.0 C) in the period 7 days before and up to 14 days after the positive test. Selfreported acute respiratory symptoms are defined as the presence of cough or dyspnoea with a severity score of 2 or higher, or the presence of nasal cold or sore throat with a severity score of 3 or higher. BCG, bacillus Calmette-Gu erin.
participants is required. The ACTIVATE-2 randomized trial found a significantly reduced incidence of COVID-19 after BCG in frail elderly [14]. Parallel to the BCG-CORONA trial, we conducted two additional trials in elderly, the BCG-CORONA-ELDERLY trial [15], and the BCG-PRIME trial (NCT04537663). Several additional BCG trials with COVID-19 endpoints are ongoing around the world, including the BRACE study (NCT04327206), and a real-time individual data meta-analysis is being conducted to monitor the effect of BCG on severe endpoints (PROSPERO ID: CRD42021213069).
In conclusion, this trial found no effect of BCG vaccination on unplanned absenteeism or documented COVID-19 among HCWs during the SARS-CoV-2 pandemic in the Netherlands.

Transparency declaration
There were no conflicts of interest. No external funding was received.