Experience of COVID-19 Vaccination of Healthcare Workers in a Hospital Setting

Introduction Several COVID-19 vaccines against SAR-CoV-2 have demonstrated high ecacy in clinical trials. This is the rst report describing their use in a healthcare setting. Methods We conduct a single centre observational study assessing vaccine uptake and apparent ecacy of the Pzer BioNTech vaccine among healthcare workers (HCW). Results Overall uptake was 60.8%, however we saw statistically signicant differences in uptake between age groups, ethnic origins, and job roles. In the 42 days after vaccination, 45 new cases of COVID were identied, of which 4 (8.9%) occurred in HCWs who were beyond 10 days of vaccination. Kaplan Meier curves for vaccinated and unvaccinated groups were congruent until day 14 and continued to diverge up to 42 days. Cox regression analysis showed a 79.0% (95%CI 21 – 95; p=0.02) risk reduction for COVID infection in vaccinated HCWs. Conclusions Initial vaccination rates among HCW were generally good, although uptake was lower in certain groups and efforts should focus on increasing uptake in these groups. The Pzer BioNTech vaccine is effective from 14 days post-vaccination in a frontline clinical setting and protection continues beyond 21 days post 1st dose without a 2nd dose, being given.

Introduction COVID-19 has had a signi cant impact on populations worldwide. Healthcare workers (HCW) have been disproportionately affected and therefore been designated a high priority group for vaccination by the Joint Committee on Vaccination and Immunization (JCVI) [1]. Vaccines have been shown to be highly effective in clinical trials with several now approved for clinical use [2] [3] [4]. However, vaccine uptake and apparent e cacy in an NHS healthcare setting has not yet been described. Here we describe early experience of vaccinating HCW at a single center in London.

Methods
We assessed uptake of COVID-19 vaccinations among HCWs at our tertiary hospital. At the time, we were treating COVID-19 patients in three wards and in an expanded ITU. All staff were offered a rst dose of the P zer BioNTech vaccine at our onsite vaccination center over an 8-day period up to 14th January 2021 [5]. Vaccines were administered by trained peer-vaccinators in accordance with the National Protocol or by patient speci c direction [6]. Our online booking system was accessible to all RNOH HCWs and was used to manage vaccination appointments. At the time of vaccination, a minimum dataset was uploaded to the National Immunization and Vaccination System (NIVS). All staff were observed for a minimum of 15 minutes post-vaccination in accordance with the MHRA temporary authorisation for P zer BioNTech vaccine and were asked to report any serious adverse events to the vaccination team or occupational health.
Human resource records containing baseline characteristics `including age, gender, ethnicity, and job role for all RNOH HCWs were collected. Contractor companies provided the same dataset for HCWs working onsite but not directly employed by the RNOH. We identi ed all HCWs who received the vaccine using data submitted to NIVs. Staff were asked to provide details of vaccination if they received it elsewhere.
All cases of laboratory con rmed COVID were reported to Occupational Health. New cases included symptomatic infection and asymptomatic cases identi ed through mandatory fortnightly screening of all HCWs working onsite. All HCWs were routinely asked questions relating to potential exposures and contacts, and more recently whether they had received a vaccine. New cases identi ed after the start of vaccinations were categorised as either: occurring within 10 days of vaccination, more than 10 day after vaccination or in an unvaccinated individual. A 10-day cut-off was chosen because it matches the criteria recommended by PHE for enhanced COVID surveillance [7]. We calculated a rolling 7-day average of new cases for our hospital for comparison with reported regional London data [8].

Statistical analysis
We summarised data using descriptive statistics. Categorical data on baseline characteristics were compared using Pearson's Chi-Square test.
We applied survival analysis to investigate the impact of COVID-19 vaccination on new COVID cases. The outcome variable was time to infection, constructed as the time between vaccination and symptom onset or time of rst positive test if asymptomatic (failure). A start date of 5th January 2021, coinciding with the start of the RNOH vaccination programme was used for unvaccinated HCWs. All HCWs not infected on 16th February 2021 were censored. The Kaplan-Meier method was used to plot cumulative hazards for vaccinated and unvaccinated groups. In the phase 2/3 safety and e cacy study of the P zer BioNTech vaccine, it was demonstrated that COVID-19 vaccination would unlikely have an impact on COVID infections until Day 14 post-vaccination [3]. Statistically this means that the hazards are unlikely to be proportional. There may be expected to be no effect in the rst 13 days, followed by an effect after this.
To allow for the likely non-proportional hazards, two sets of analyses were performed. The rst set compared the groups from Day 0 to Day 13. A second set of analyses compared the groups with the start point being Day 14, up until the end of the follow-up period. As it is unlikely that patients would have more than one COVID infection in such a short space of time, staff with a COVID-19 infection with the rst 13 days, in both groups, were omitted from the analysis of the second period. Due to the survival nature of the outcome, the analyses were performed using Cox regression for both time periods, two analyses were performed. Initially a simple 'unadjusted' comparison between vaccinated and unvaccinated groups was made. Subsequently, the groups were compared adjusting for demographic details found to vary signi cantly between groups.

Results
We vaccinated 1,373 (60.8%) out of a total 2,257 HCWs working at RNOH. The split by age, sex, ethnicity, and staff group are shown in Table 1. Uptake was higher for male (65.0%) than females (58.9%) (p = 0.006). Statistically signi cant differences were seen between ethnic groups with Whites and Asians more likely to be vaccinated but only 27.0% of Black / Afro-Caribbean and 43.0% of mixed-race staff being vaccinated (p = 0.001). Differences were also seen between staff groups with only approximately half of nursing (52.9%) and clinical support (49.2%) staff, and even fewer portering, domestic and catering staff (30.5%) were vaccinated (p = 0.001).
The number of new cases identi ed at the RNOH since October 2020 are shown in Fig. 2. There were 45 new cases identi ed in the 42 days following the start of vaccinations. Of these, 25 (55.6%) were unvaccinated, 16 were within 10 days of vaccination and 4 (8.9%) were beyond 10 days of vaccination. The curves appear to be congruent from day 0, then starts to diverge at day 14, then continues to diverge up to day 42. Separate Cox regression analyses performed for the two stages of the follow-up period, for vaccinated and unvaccinated groups are compared in Table 2. There was no signi cant difference between groups for the rst 13 days of follow-up. This was the case in both the unadjusted and adjusted analysis. No serious adverse events were reported in our staff and we submitted no yellow cards to the MHRA. (+) Expressed as hazard of COVID in Vaccinated group relative to Unvaccinated group.

Discussion
Overall vaccine uptake was 60.8% in our hospital. Whilst this is encouraging it is unclear what proportion need to be vaccinated to confer herd immunity. Several estimates have put the proportion between 70-80% across a population [9]. It is likely that the current rate will provide some protection against nosocomial spread, but hospital populations are dynamic with patients, often with relatives or carers, attending frequently. Therefore, it remains uncertain whether there will be an impact on the recommended level of infection prevention & control precautions.
We saw differences in uptake between groups. There was little difference between the sexes, but staff were more likely to be vaccinated with increasing age. This may be important if there are areas where staff are typically younger, for example general wards.
We saw lower uptake in nursing and clinical support staff, but the lowest rates were seen in portering, domestic and catering staff, all of whom are potentially more at risk of coming into close contact with patients with COVID or their environment.
It was striking that uptake is lower amongst Black / Afro-Caribbean staff and those of mixed heritage. This is concerning as these groups have been disproportionately adversely affected by COVID and potentially remain at risk. Additionally, these groups are over-represented amongst the staff groups above. Furthermore, London has a higher proportion of staff from these groups when compared with the rest of the UK which may have an impact on a health service level [8].
It is clear that, to improve uptake, more work is needed to understand the reasons why it is lower in certain groups. It would be important to know whether this represents an issue with access or whether there is true hesitancy in receiving the vaccine and if so, what are the reasons why staff may not wish to be vaccinated. Some work has already begun around this nationally [10]. We are currently conducting a survey locally to explore reasons. We are also conducting small focus groups and other engagement sessions over the coming weeks.
We did not see any serious adverse events and we did not routinely collect data on minor side effects. Consequently, we do not have enough data from our cohort to make reliable conclusions around safety, however, early indications are that the P zer-BioNTech vaccine appears to be safe.
Our 7-day rolling average of cases mirrored the London 7 day rolling average. This is not surprising as our hospital falls within the London area and suggests that a substantial proportion of cases amongst our staff re ect what is happening in the local community rather than nosocomial spread within our institution. Our rate appeared to rise, peak, and start to fall slightly before the London rate, and we do not have an explanation for this. We saw relatively fewer cases after Christmas and during the rst 2 weeks of January and we postulate that this was for 2 main reasons. Firstly, the onsite testing centre closed for a period over Christmas & New Year. Secondly, we suspect there was an effect due to more staff than usual being on annual leave.
Once the vaccination campaign began a clear difference in the groups testing positive for COVID began to emerge. Cases initially occurred in the unvaccinated and recently vaccinated groups, however, the groups diverged by 14 days with almost all subsequent cases occurring in the unvaccinated group. This appears to be similar to what was reported in the initial e cacy study and most likely re ects the development of an effective immune response in the vaccinated group [3]. It is reassuring that the survival analysis beyond 14 days, showed an 80% lower risk of COVID infection in vaccinated HCWs. Although this is a statistically signi cant difference in risk, there is a large amount of uncertainty around our estimate, with the 'true' risk reduction being anywhere from 21-95%. Larger studies are required to verify our ndings.
Importantly, our small study shows that the protective effect of this vaccine continues beyond 21 days in an hospital setting following the rst dose. This is reassuring given that the National Strategy is to delay administration of the 2nd dose of the vaccine to 90 days [11].

Conclusion
Initial vaccination rates among HCW were generally good although uptake was lower in certain groups and efforts should focus on increasing uptake in these groups.
The P zer BioNTech vaccine is effective from 14 days post-vaccination in a frontline clinical setting and protection continues beyond 21 days post 1st dose without a 2nd dose, being given.

Declarations
Con ict of interest the authors declare that there are no con icts of interest. The review was signed off locally as service evaluation.
No funding was received to produce this work.