SARS-CoV-2 Reinfections in Health-Care Workers, 1 March 2020–31 January 2023

Objective: To study SARS-CoV-2 reinfections in health-care workers (HCWs) of the University Health Agency Giuliano-Isontina (ASUGI), covering the provinces of Trieste and Gorizia (northeastern Italy) routinely screened for SARS-CoV-2 via nasopharyngeal swab. Design: Cohort study of HCWs (N = 8205) followed since the start of the pandemic (1 March 2020) through 31 January 2023. The risk of reinfection during the Omicron transmission period (after 30 November 2021) among HCWs previously infected by SARS-CoV-2 was estimated based on days since last dose of COVID-19 vaccine received, adjusting for age, sex, job task, workplace, number of doses of COVID-19 vaccines and number of swab tests performed. In the crude as well as adjusted incidence rate analysis, reinfections occurring 15+ days after a first dose of COVID-19 vaccine or 8+ days following a second or more dose were counted. Results: In a highly vaccinated population, during the entire study period (1 March 2020–31 January 2023) 5253 HCWs incurred at least one SARS-CoV-2 infection, 4262 HCWs were infected only once, and 1091 were reinfected. Reinfections almost entirely (99.1% = 1071/1091) occurred after 30 November 2021, peaking in July 2022 (N = 161). Six hundred eighty-three reinfections followed a pre-Omicron primary event against 408 reinfections following an Omicron event. Reinfections during the Omicron transmission period occurred a mean of 400 ± 220 days after primary SARS-CoV-2 infection; 512 ± 205 days following a pre-Omicron primary event, as opposed to 218 ± 74 days after an Omicron primary infection. Thirty-four hospitalizations were observed, all before the Omicron wave, following 18 (0.4%) primary SARS-CoV-2 infections and 16 (1.5%) reinfections. By excluding events occurring <15 days after a first dose or <8 days after a further dose of COVID-19 vaccine, 605 reinfections followed a pre-Omicron primary event (raw incidence = 1.4 × 1000 person-days) against 404 after a primary Omicron infection (raw incidence = 0.3 × 1000 person-days). Apart from nurse aids (slightly enhanced biological risk) and academic HCWs (remarkably lower risk with pre-Omicron primary events), the effect of occupation in terms of job task and workplace was marginal. Furthermore, whilst the risk of reinfection was lower in males and HCWs < 60 years old following a pre-Omicron primary infection, HCWs aged 30–50 were more likely to be infected after an Omicron primary event. Regardless of timeline of primary SARS-CoV-2 event, the risk of reinfection decreased with higher number of doses of COVID-19 vaccines, being lowest after the second booster. In particular, VE was 16% for one dose, 51% for two doses, 76% for the booster and 92% for the second booster with a pre-Omicron primary SARS-CoV-2 event. The latter figures increased to 72%, 59%, 74% and 93%, respectively, with Omicron primary infections. Conclusions: SARS-CoV-2 reinfections were frequent during the Omicron transmission period, though featured by mild or no symptoms. Whilst the impact of occupation on biological risk was relatively marginal, COVID-19 vaccination had the strongest protective effect against reinfection, with a 93% VE by second booster following an Omicron primary infection.

vaccination uptake combined with systematic screening for SARS-CoV-2 make HCWs an optimal target to assess the infection risk in relation to COVID-19 vaccination status.

Aims
In view of the above, we conducted a longitudinal study to estimate the risk of reinfections among HCWs of the University Health Agency Giuliano-Isontina (ASUGI), Northeastern Italy, from the start of the pandemic (10 March 2020) through 31 January 2023. To the best of our knowledge, no data on reinfection and vaccine effectiveness (VE) against Omicron were available till the end of January 2023.

Ethical Aspects
This study was approved by the Italian Medicine Agency (AIFA), the Ethics Committee of the Italian National Institute of Infectious Diseases (INMI) "Lazzaro Spallanzani" and the regional ethics committee (CEUR) of Friuli Venezia Giulia Region (FVG) (Reg N.188/2022).
In compliance with Italian legislation on privacy law, informed consent from study participants was waived since patients' data routinely collected for health care reasons were managed anonymously within the framework of an approve study protocol. The present cohort of HCWs contributed to the ORCHESTRA database, a European Union (EU) project funded by Horizon 2020. This study followed the Strengthening of Observational Studies in Epidemiology (STROBE) reporting guidelines.

Study Population
This study investigated incidence of SARS-CoV-2 reinfections from 1 March 2020 through 31 January 2023 among HCWs of ASUGI (N = 8205), including 4 hospitals and other public health services within the provinces of Trieste and Gorizia (FVG Region).

Data Collection
The cohort included HCWs employed by ASUGI on a permanent contract. Data included demographic characteristics (sex, age, occupation, and health care department), primary SARS-CoV-2 infection, reinfection, and hospitalization.
Since 15 April 2020, HCWs of ASUGI were tested by Polymerase Chain Reaction (Rt-PCR) with nasopharyngeal swabs weekly or monthly, depending on the level of their occupational biological risk [3]. HCWs were also swab-tested in cases of symptoms consistent with COVID-19 or for contact tracing [1,3].

Study Endpoint
The study investigated the incidence of SARS-CoV-2 reinfections from 1 March 2020 to 31 January 2023. Reinfection was defined as an infection in the same individual confirmed via rt-PCR at least 90 days after the primary event [26].
HCWs testing positive, with asymptomatic or mild/moderate COVID-19, had to observe home isolation for at least 5 days and were allowed to return to work only after a negative swab test confirmed via Rt-PCR. In compliance with guidelines from the USA Centers for Disease Control and Prevention (CDC), HCWs were exempted from screening tests against SARS-CoV-2 for three months following COVID-19 diagnosis in order to avoid possible false positive results [1].

Statistical Analysis
Continuous variables (age, number of swab tests performed, and time between primary SARS-CoV-2 infection and reinfection) were all non-normally distributed. Although we have presented both the respective mean ± standard deviation (SD) as well as the median (interquartile range, IQR), only the Wilcoxon test was used in comparison. Categorical variables were expressed as number and percentage of HCWs and compared by chi-square tests.
HCWs were followed up from the time of their last vaccine dose through eventual SARS-CoV-2 reinfection, thereby calculating person-days at risk among individuals previously infected. Follow up time for unvaccinated HCWs started on 1 December 2021 instead. Crude incidence rates of reinfections were calculated by explanatory factor and timeline of primary SARS-CoV-2 infection (before 1 December 2021 versus after 30 November 2021). HCWs re-infected before the first COVID-19 vaccine dose (N = 5), all following a pre-Omicron SARS-CoV-2 event, were included among those unvaccinated.
A multivariable Cox proportional hazard regression model was then fitted to investigate risk factors for SARS-CoV-2 reinfection from 1 December 2021 onward. Moreover, the latter analysis was also broken down by timeline of primary SARS-CoV-2 infection (before 1 December 2021 versus after 30 November 2021). Results were expressed as adjusted hazard ratio (aHR) with 95% confidence interval (95% CI). VE against SARS-CoV-2 reinfections was estimated by number of doses of COVID-19 vaccines received (VE = 1-aHR).
Cases with missing values were excluded and complete case analysis was performed. Statistical analyses were performed using STATA 16.0 (StataCorp LLC, College Station, TX, USA). Table 1 reports the characteristics of the study population (N = 8205) by total number of SARS-CoV-2 infections (only one vs. two). Over the entire study period, 4262 (51.9%) HCWs were infected only once, versus 1,091 (13.3%) reinfected. As can been seen from Table 1, 818 (10%) HCWs were never immunized with a COVID-19 vaccine, 79 (1.0%) received only one dose, 530 (6.5%) two doses, 5988 (73.0%) three doses, 773 (9.4%) four, and 17 (0.2%) five. Moreover, reinfections were less frequent in HCWs older than 60, yet more prevalent in HCWs who were employed in community services and/or unvaccinated (17.2% re-infected unvaccinated vs 5.2% infected once and unvaccinated). The vast majority of HCWs were vaccinated with Comirnaty (Pfizer BioNTech) and with three doses (73.0%). The mean number of swab tests performed during the omicron transmission period was higher in HCWs re-infected (16.2 ± 8.0 vs. 14.4 ± 8.5, respectively; p < 0.001; median 15 (IQR = 11; 20) versus 13 (IQR = 8; 19). Thirty-four hospitalizations were observed, all before the Omicron wave, following 18 (0.4%) cases of primary SARS-CoV-2 infections versus 16 (1.5%) re-infections (p < 0.001) ( Table 1). Figure 1 reports the frequency distribution of primary SARS-CoV-2 infections over time. As can be seen, the surge of primary infections in autumn 2020 (October through December 20) was followed by a sharp decline in January 2021 and a drop from February 2021 onward, in coincidence with the start of the COVID-19 vaccination campaign. Subsequently, in a highly vaccinated population such as HCWS of ASUGI, primary SARS-CoV-2 infections dramatically re-surged in November 2021, peaking in January 2022, declining thereafter, but maintaining levels of infections considerably higher than pre-Omicron waves and peaking again in July 2022.    Figure 3 shows the frequency distribution of SARS-CoV-2 reinfections from 1 December onward, following a pre-Omicron primary event (before 1 December 2021). It can be noted that these reinfections were mainly responsible for the first peak in January 2022, progressively declining thereafter.  Figure 4 shows the frequency distribution of reinfections from 1 December onward, following an Omicron primary event (after 30 November 2021). As can be seen, the frequency of these reinfections exhibited an opposite trend compared to those following a pre-Omicron primary event, progressively increasing over time, much contributing to the second peak of reinfections observed in July 2022.   Figure 5 report COVID-19 vaccine uptake over time by calendar month since 1 December 2021 through 31 January 2023. Only the 4th and 5th dose were a bivalent vaccine. Whilst the first booster (3rd dose) was still the Wuhan-Hu-1 vaccine, the second and third booster (4th and 5th dose respectively) were bivalent COVID-19 vaccines. Ninety-onepoint-two-percent (91.2%) of HCWs were immunized with Comirnaty (Pfizer BioNTech) at third dose. Under Italian law, unvaccinated HCWs were suspended from work or re-reassigned to job tasks not involving patient contact until the end of December 2022. As can be seen from Table 2, at the beginning of the Omicron transmission period, 48.9% (4015/8205) of HCWs were immunized with the booster, 33.8% (2773/8205) with two doses, 5.2% (=428/8205) with one dose, and 12.1% (989/8205) were unvaccinated. By the end of the study (31 January 2023), 9.4% (=773/8205) were immunized with the second booster, 73.0% (=5988/8205) just with the first booster, 6.5% (=530/8205) only with two doses, 1.0% (=79/8205) just with one dose, and 10.0% (=818/8205) were unvaccinated.   Table 3 reports days from primary SARS-CoV-2 infection until reinfection by sex and COVID-19 vaccination status. The group of unvaccinated subjects (N = 193) included 188 HCWs never immunized with COVID-19 vaccines and 5 individuals re-infected before the first dose. By excluding events occurring <15 days before the first dose or <8 days before a second or more dose of COVID-19 vaccines, 605 reinfections followed a pre-Omicron primary event. As can be seen, reinfections occurred after a median of 356 (IQR: 202; 567) days after primary infection in the entire cohort of HCWs, earlier in females (median = 425 days; IQR: 272; 643) than males (median = 328; IQR: 193; 536) days. Moreover, reinfections followed a median of 508 (IQR: 379; 664) days after a pre-Omicron primary event against 200 (IQR: 161; 282.5) days after an Omicron primary event, with no difference by sex. Finally, SARS-CoV-2 reinfections occurred considerably earlier among unvaccinated subjects (median = 185 days; IQR: 154; 300) since primary SARS-CoV-2 event) compared to HCWs immunized by one (median = 484.5 days; IQR: 406; 602), two (median = 361; 95% CI: 175; 574.5), three (median = 416; 95% CI: 258; 623.5) or four (median = 575; 95% CI: 306.5; 767) doses of COVID-19 vaccines. Table 4 shows the crude incidence rates of reinfections by explanatory factors and timeline of primary SARS-CoV-2 infection (before 1 December 2021 vs. after 30 November 2021). The total number of reinfections decreased from 1091 to 1009, since events occurring <15 days since the first vaccine dose or <8 days since the second or higher dose were not considered in the analysis.

Main Findings
By 1 December 2021, 12.1% of HCWs of ASUGI were unvaccinated, 5.2% were immunized with one dose, 33.8% with two doses and 48.9% with three doses. By the end of the study (31 January 2023), the rate of unvaccinated HCWs was 10%, 1.0% were immunized with one dose, 6.5% with two doses, 73.0% with three doses, 9.4% with four doses, and 0.2% with five.
The raw incidence of re-infections was 0.5 × 1000 P-d in the entire cohort, increasing to 1.4 × 1000 P-d if primary SARS-CoV-2 infection occurred before the Omicron transmission period, diminishing to 0.3 × 1000 P-d afterwards.
Apart from nurse aids (slightly enhanced biological risk) and academic HCWs (remarkably lower risk after a pre-Omicron primary event), the effect of occupation in terms of job task and workplace was marginal. Furthermore, whilst the risk of reinfection was lower in males and HCWs <60 years old following a pre-Omicron primary infection, HCWs aged 30-50 years were more likely to be infected after an Omicron primary event.
Regardless of timeline of primary SARS-CoV-2 event, the risk of reinfection decreased with higher number of doses of COVID-19 vaccines, being lowest for the second booster. In the entire cohort, VE was 16%, 51%, 76%, and 92% for one, two, three, and four doses of COVID-19 vaccine, respectively.
Moreover, in case of pre-Omicron primary SARS-CoV-2 event VE was 40% for one dose, 69% for two doses, 62% for the first booster, and 89% for the second booster. The latter figures increased to 72%, 59%, 84%, and 93%, respectively after an Omicron primary infection.

Interpretation of Findings
The higher communicability combined with a lower virulence of Omicron compared with previous VOC has been well established due to its ability to evade neutralizing antibody responses [16,27,28]. Likewise, SARS-CoV-2 reinfections were associated only with mild-moderate disease in the present study.
The raw incidence of reinfections was found to be much higher following a pre-Omicron primary event (1.4 × 1000) in the present study, diminishing afterwards (0.3 × 1000).
COVID-19 vaccination status was the main protective factor against reinfection during the Omicron transmission period in the present study, with a risk progressively decreasing with higher number of doses of COVID-19 vaccine, especially if primary SARS-CoV-2 infection occurred after 30 November 2021.
With pre-Omicron primary events, VE against SARS-CoV-2 reinfection increased from 40% for one dose up to 62% for the first booster and 89% with the second booster. If the primary SARS-CoV-2 event occurred after 30 November 2021, VE increased to 72% for one dose, 84% for the first booster and 93% for the second booster.
In a population base study conducted in Mexico from 3 March 2020 until 13 August 2022, 231,202 cases of SARS-CoV-2 reinfections were recorded, mostly occurring during the Omicron transmission period (89.8% = 207,623/231,202) and in unvaccinated individuals (41.6%). The respective case fatality rate was 0.22%, and vaccination protected against reinfection in the latter study, regardless of the temporal order of the immunizing event [29]. In another study from Mexico City conducted between 1 March 2020 and 28 February 2022, only 73 (5.6%) SARS-CoV-2 reinfections were noted out of 1388 total COVID-19 cases, including 97.3% individuals immunized with 2 doses of COVID-19 vaccine [30]. The overall rate of reinfection in the latter study was 0.231 per 1000 during the Omicron wave, a figure very close to that of the present study (0.3 × 1000). By contrast, higher rates of SARS-CoV-2 reinfections were noted among 11,474 HCWs from New Delhi (India) during the Omicron wave, with an incidence of 4.56 (95% CI: 4.29; 4.85) × 1000 p-d [31].
A systematic review and meta-analysis of 65 studies from 19 countries published before 31 Sep 2022 estimated the risk of SARS-CoV-2 reinfection against primary infection [32]. The high level of protection against reinfections observed with more virulent pre-Omicron VOC (Wuhan, Alpha, and Delta) decreased significantly under Omicron, although protection against severe COVID-19 remained consistently high. Despite waning over time, protection conferred by primary SARS-CoV-2 event against reinfection was at least as high as or even higher than a complete COVID-19 vaccination cycle (2 doses) of mRNA vaccines [32].
Protection against reinfection in previously SARS-CoV-2-infected individuals decreases with time since last immunizing event, regardless of whether any vaccine dose was received before or after infection [19]. However, protection from hybrid immunity acquired by vaccination combined with natural infection was reportedly higher than infection alone or vaccine-only [12,21,[33][34][35][36], especially with Omicron but also for previous VOC [19]. In a nationwide study conducted in Israel in August and September 2021, when Delta VOC was predominant, protection against reinfection by previous SARS-CoV-2 infection was higher than that provided by a second dose of vaccination among previously uninfected individuals following the same timeline since last immunizing event. However, a single dose of COVID-19 vaccine after infection reinforced protection against secondary infection [19].
A study was conducted on HCWs in Quebec (Canada) between 27 March and 4 June 2022 to examine the risk of reinfection in relation to vaccination status. Of 37,732 presumed BA.2 cases recorded, 2,521 (6.7%) were reinfections after pre-Omicron primary SARS-CoV-2 events, versus 659 (1.7%) reinfections following primary infection occurring during Omicron [36]. Pre-Omicron primary SARS-CoV-2 events were associated with a 38% reduction of BA.2 infection, further increasing to 56%, 69%, and 70% among HCWs immunized with one, two, or three doses of COVID-19 vaccines respectively. Higher protection (72%) against BA.2 infection was found with primary SARS-CoV-2 infection during the Omicron transmission period, further increasing to 96% in HCWs also immunized with 2 doses of COVID-19 vaccines. However, the booster dose did not further enhance protection (96%), questioning the additional benefit of further boosters doses against future SARS-CoV-2 variants in case of hybrid humoral immunity [36].
Differently from previous investigations in the same cohort [3,15], the biological risk was only marginally influenced by occupation task in the present study. The risk of reinfection was in fact slightly higher only for nurse aids in the entire cohort and remarkably lower for academic clinical tasks in case of primary SARS-CoV-2 infection occurred before Omicron. Likewise, the biological risk was not influenced by health care premises in the present study, suggesting extra-occupational spread of asymptomatic or mild-moderate SARS-CoV-2 infections in a highly vaccinated population of HCWs during the Omicron transmission period. The progressive lifting of mandatory health protection measures in the community allowed the spread of the SARS-CoV-2 outside health care premises, where face masks and social distancing were not systematically enforced any longer.
Lower risk of reinfection in males in the present study may be explained by their lower prevalence in patient-facing tasks; in fact, 78% of 2908 nurses and 79.3% of 1233 nurse aids were females. However, this estimate is in contrast with a previous investigations in the same cohort, reporting higher risk of SARS-CoV-2 infection and severe disease in males [3,15]. Likewise, lower risk of reinfection for HCWs older than 60 years, particular after pre-Omicron primary events, may be explained by the lower proportion of HCWs > 60 years assigned to patient-facing tasks [3,15].

Strengths and Weaknesses
This study investigated a large cohort (N = 8205) of highly vaccinated HCWs subject to systematic screening for SARS-CoV-2, from the start of the pandemic until 31 January 2023. This approach allowed the detection of asymptomatic HCWs and those developing symptoms after swab testing, allowing subsequent isolation to contain the spread of COVID-19 infection in health care premises.
However, our study has some limitations. First of all, although this cohort was systematically screened and followed up throughout the pandemic, without serology information it is impossible to exclude the possibility that some HCWs categorized as singly infected may have had undetected prior COVID-19 infection instead.
Moreover, we did not have information on individuals' risk factors for COVID-19, especially co-morbidities, which can influence the incidence rate of SARS-CoV-2 infection. Another limitation is the lack of data on COVID-19 symptoms, since only hospitalization was recorded.
Finally, this was a dynamic cohort, with a few HCWs who may have retired and some other may have been employed during the study period. However, this should have had only a limited impact on the calculation of the number of unvaccinated HCWs (by subtraction from the total number of HCWs, N=8205), since date of COVID-19 vaccination and date of SARS-CoV-2 infection were still available for retirees and newly employed HCWs from the Regional Data Registry of FVG, regardless of whether the HCW was in service during the study period.

Conclusions
In a highly vaccinated population of 8205 HCWs, 1091 cases of reinfection were notified since the start of the pandemic (1 March 2020) through 31 January 2023, almost entirely (99.1%) occurring after 30 November 2021 and peaking in July 2022. Six-hundred eighty-three reinfections occurred following a pre-Omicron primary event (raw incidence = 1.4 × 1000 P-d), versus 408 reinfections after an Omicron primary infection (raw incidence = 0.3 × 1000 P-d).
The risk of reinfection was only marginally influenced by occupation. Regardless of timeline of primary SARS-CoV-2 event, the risk of reinfection decreased with higher number of doses of COVID-19 vaccines, being lowest for the second booster. In particular, in the entire cohort VE was 16%, 51%, 76%, and 92% for one, two, three and four doses of COVID1-9 vaccine, respectively. In case of pre-Omicron primary SARS-CoV-2 infection, VE was 40% for one dose, 69% for two doses, 62% for the first booster, and 89% for the second booster. The latter figures increased to 72%, 59%, 84%, and 93%, respectively, in case of reinfections following an Omicron primary SARS-CoV-2 events.
Humoral immunity following natural SARS-CoV-2 infection should be taken into account by COVID-19 vaccination policies for HCWs and the general population.

Funding:
The study was partially funded by the EU project ORCHESTRA https://orchestra-cohort. eu. The ORCHESTRA project has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement No 101016167. The views expressed in this paper are the sole responsibility of the author, and the Commission is not responsible for any use that may be made of the information it contains.

Institutional Review Board Statement:
The study was conducted in accordance with the guidelines of the Declaration of Helsinki and approved by the regional regional ethics committee (CEUR) of Friuli-Venezia Giulia Region (Reg N.188/2022).
Informed Consent Statement: Patient consent was waived since, according to Italian privacy law (Legislative Decree 101/2018, D.Lgs 101/2018), patients' data routinely collected by the Italian National Health Service can be used for scientific purposes within the frame of approved studies/protocols, provided that sensitive information is anonymized.

Data Availability Statement:
The data generated and analyzed during the current study are not publicly available, since they were purposively collected by the authors for the present study, but they are available from the corresponding author on reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.