Effectiveness of the Comirnaty (BNT162b2, BioNTech/Pfizer) vaccine in preventing SARS-CoV-2 infection among healthcare workers, Treviso province, Veneto region, Italy, 27 December 2020 to 24 March 2021

Data on effectiveness of the BioNTech/Pfizer COVID-19 vaccine in real-world settings are limited. In a study of 6,423 healthcare workers in Treviso Province, Italy, we estimated that, within the time intervals of 14–21 days from the first and at least 7 days from the second dose, vaccine effectiveness in preventing SARS-CoV-2 infection was 84% (95% confidence interval (CI): 40–96) and 95% (95% CI: 62–99), respectively. These results could support the ongoing vaccination campaigns by providing evidence for targeted communication.

By 24 March 2021, the coronavirus disease (COVID-19) pandemic has caused over 3.4 million cases and 105,000 deaths in Italy [1]. Although non-pharmaceutical interventions implemented in Italy were effective in reducing the impact of the first and second wave [2,3], there is urgency, now with the availability of approved vaccines, to accelerate the COVID-19 vaccination campaigns.
The first stage of the vaccination campaign in Italy started on 27 December 2020, which initially targeted healthcare workers (HCW) and residents in long-term care facilities. The Comirnaty, (BNT162b2, BioNTech/ Pfizer, Mainz, Germany/New York, United States) vaccine was used because it was the only vaccine approved by the Italian Medicines Agency at that date [4]. Recommended administration was two doses 21 days apart.
Although efficacy of the Comirnaty vaccine has been proven in clinical trials [5], there is a need to evaluate its effectiveness in real-world settings. Based on surveillance data, this study aimed to estimate the effectiveness of the Comirnaty vaccine in preventing severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in frontline HCW employed at the local health unit that serves the entire province of Treviso in the Veneto region (LHU-TV).

Vaccination coverage and characteristics of healthcare workers included in the study
We conducted a retrospective cohort study of 9,878 HCW employed at the LHU-TV. From the local COVID-19 surveillance database, we retrieved information on demographic and professional characteristics, recorded dates of vaccine administration (all HCW were vaccinated with the Comirnaty vaccine) and the recorded date of SARS-CoV-2 infection, based on a positive antigenic test (SARS-CoV-2 Ag Test, LumiraDx, Alloa, United Kingdom (UK); sensitivity = 97.6% and specificity = 96.6% according to the manufacturer's indications) confirmed by RT-PCR on the same day.
A total of 6,423 HCW were included in the analysis, after exclusion of 1,285 (13.0%) HCW infected with SARS-CoV-2 before the vaccination campaign, and 2,170 (22.0%) HCW working outside hospitals and district outpatient centres or who were support and administrative staff. The mean age of the included HCW was 47.1 years (standard deviation (SD): 10.8 years), most of them female (n = 4,986 (56.5%)) ( Table 1). A total of 3,630 HCW were nurses (56.5%), 1,469 were medical doctors (22.9%), and 1,324 were social HCW (20.6%) ( Table 1). All the included HCW were screened approximately every 8 days and at any other time if presenting symptoms consistent with COVID-19.
The percentage of unvaccinated HCW was higher in women than in men (17.9% vs 13.5%), and in those aged 30-39 years (23.0%) compared with other age groups ( Table 1). The highest percentage of complete vaccination with both doses was highest in medical doctors (85.7%) and HCW working in hospitals (82.1%).

Effectiveness of the Comirnaty vaccine
We also conducted a time-to-event analysis using the number of days elapsed from vaccine administration to measure the length of follow-up. We estimated the effectiveness of one and two dose administration of the Comirnaty vaccine to prevent SARS-CoV-2 infection at different time intervals using a multivariable Cox proportional hazard model, including sex, age group, professional category, work context, and starting week of exposure as covariates. The adjusted hazard ratios
a The follow-up analysis was started on day 7 from the start of the vaccination campaign when the number of vaccinated HCW was sufficiently high to allow robust estimates [10].  (HR) were used to calculate vaccine effectiveness (VE) as ((1 − HR) × 100).
In the time interval of 14-21 days after the administration of the first dose, VE in preventing all (both asymptomatic and symptomatic) and only symptomatic SARS-CoV-2 infections was estimated at 84% (95% CI: 40-96) and 83% (95% CI: 15-97), respectively ( Table  2). In the time interval of at least 7 days after the administration of the second dose, VE increased to 95% (95% CI: 62-99) and 94% (95% CI: 51-99) in the two groups. The analysis showing the Kaplan-Meier failure curve by vaccination status according to time since vaccination or start of exposure for unvaccinated HCW is presented in Supplementary Figure S2.

Trend of COVID-19-associated hospital admissions in the study area, immunisation rate, and number of SARS-CoV-2 infections among healthcare workers
Finally, we analysed the trend of COVID-19-associated hospital admissions in the study area, together with the trend of immunisation rate and of the number of SARS-CoV-2 infections among HCW included in the study. We found that from mid-February 2021, when the potential long-term immunisation rate among HCW was ca 70%, the number of newly diagnosed cases in this group remained stable despite a higher risk of exposure due to the rapid increase of COVID-19 hospital admissions ( Figure 2). a The level of restrictions implemented in the Veneto region according to risk scenarios ranged from high (red band) to medium (orange band) and low (yellow band) [11].

Ethical statement
The dissemination of COVID-19 surveillance data was authorised by the Italian Presidency of the Council of Ministers on 27 February 2020 (Ordinance no. 640).

Discussion
Our analysis suggests that the Comirnaty vaccine had a high effectiveness in preventing SARS-CoV-2 infection in HCW during the time intervals after administration where protection may be expected [6]. Data on trends of COVID-19-associated hospital admissions in the study area, immunisation rate, and number of SARS-CoV-2 infections among HCW support this finding.
Moreover, also Italian national data suggest that the vaccination campaign in HCW was successful. A recent report has shown how, from mid-January 2021, COVID-19 incidence among HCW started to decrease rapidly, while it increased in the general population where vaccination coverage was still low [7].
To our knowledge, the few studies evaluating the Comirnaty vaccine effectiveness in preventing SARS-CoV-2 infection among HCW that included a control group were conducted in the UK and Israel [8,9]. We found that VE was 85% (95% CI: −35 to 98) after 21 days from the first dose administration compared with 72% (95% CI: 58-86) in the UK [8], although our estimate suffers from lack of precision because of the reduced time of follow-up in this interval (the large majority of subjects received the second dose of vaccine within few days from the scheduled date on day 21 from the first dose). When comparing VE 7 days after administration of the second dose, we found a higher VE of 95% (95% CI: 62-99) compared with 86% in the UK (95% CI: 76-97). Our estimates of VE in preventing symptomatic infections during the time intervals 1-14 days and 15-28 days from administration of the first dose were 40% (95% CI: 9-60) and 86% (95% CI: 33-97), respectively. This closely reflects the estimated VE of 47% (first dose, 95% CI: 17-66) and 85% (second dose, 95% CI: 71-92) estimated in Israel [9]. However, despite the general agreement, these comparisons should be interpreted with caution as they may be biased by several factors, such as differences in case definition and surveillance procedures.
Overall, 17% of the eligible HCW were not yet vaccinated almost 3 months after the start of the vaccination campaign, probably because of refusal. In accordance with findings from the UK, female individuals and HCW under 40 years of age had a lower tendency to be vaccinated, while medical doctors were the professional category showing the highest coverage [8].
This study has several limitations. It included only HCW and cannot be assumed as representative of the general population. We had no information about possible occurrence of adverse effects after vaccine administration, although there was no evidence of severe complications (no post-vaccination hospital admissions in vaccinated HCW). Unfortunately, the date of testing was not recorded in case of a negative result, and we were therefore unable to assess adherence to routine testing. However, the analyses evaluating VE in preventing all infections or only symptomatic infections did not show great differences. Given the probability that testing was performed in a timely manner in the event of symptoms, we feel confident that asymptomatic cases were also detected early. However, a residual differential bias could remain. For example, vaccinated people may have less rigorously adhered to testing, based on the belief they were protected, thus leading to an overestimate of VE. It is also possible that we missed mild or asymptomatic infections undetectable through the first-line antigenic test. It was not possible to accurately estimate VE in time intervals where the number of person-days of follow-up were much reduced (i.e. after 21 days from the administration of the first dose, before receiving the second dose). Finally, information about the SARS-CoV-2 variants linked to infections was not available and it was therefore not possible to evaluate VE according to this variable.

Conclusions
In a real-world setting in northern Italy, during time intervals after vaccine administration where protection may be expected, we found a high effectiveness of the Comirnaty vaccine. This result could help to promote the ongoing vaccination campaign in the general population and among the still unvaccinated HCW by reinforcing communication based on evidence.