Primary Series and Booster Coronavirus Disease 2019 Vaccine Effectiveness in a Cohort of Healthcare Workers in Albania During a BA.1 and BA.2 Variant Period, January–May 2022

Abstract Background Healthcare workers (HCWs) have experienced high rates of coronavirus disease 2019 (COVID-19) morbidity and mortality. We estimated COVID-19 2-dose primary series and monovalent booster vaccine effectiveness (VE) against symptomatic severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Omicron (BA.1 and BA.2) infection among HCWs in 3 Albanian hospitals during January–May 2022. Methods Study participants completed weekly symptom questionnaires, underwent polymerase chain reaction (PCR) testing when symptomatic, and provided quarterly blood samples for serology. We estimated VE using Cox regression models (1 – hazard ratio), with vaccination status as the time-varying exposure and unvaccinated HCWs as the reference group, adjusting for potential confounders: age, sex, prior SARS-CoV-2 infection (detected by PCR, rapid antigen test, or serology), and household size. Results At the start of the analysis period, 76% of 1462 HCWs had received a primary series, 10% had received a booster dose, and 9% were unvaccinated; 1307 (89%) HCWs had evidence of prior infection. Overall, 86% of primary series and 98% of booster doses received were BNT162b2. The median time interval from the second dose and the booster dose to the start of the analysis period was 289 (interquartile range [IQR], 210–292) days and 30 (IQR, 22–46) days, respectively. VE against symptomatic PCR-confirmed infection was 34% (95% confidence interval [CI], −36% to 68%) for the primary series and 88% (95% CI, 39%–98%) for the booster. Conclusions Among Albanian HCWs, most of whom had been previously infected, COVID-19 booster dose offered improved VE during a period of Omicron BA.1 and BA.2 circulation. Our findings support promoting booster dose uptake among Albanian HCWs, which, as of January 2023, was only 20%. Clinical Trials Registration. NCT04811391.

during periods of Omicron circulation [5].Differences in population health and demographics, differences in health systems, and time of access to COVID-19 vaccines between highincome countries and middle-income countries may variably impact vaccine performance, underscoring the need for VE studies in middle-income countries to guide policy [6].Furthermore, local data demonstrating positive VE estimates can be helpful in promoting vaccine uptake in countries where vaccine acceptance is low.
In early 2022, the spread of SARS-CoV-2 Omicron, a variant with higher transmissibility compared to previous variants [7], led to elevated pressure on healthcare systems across Europe, and many infected HCWs missed work [8].To date, studies of COVID-19 VE against Omicron, mostly from high-income countries [9], have shown sustained moderate to high primary series VE against severe disease (76% within 6 months after the last dose) but much lower VE against milder symptomatic infection (35% within 6 months after last dose) [10]; however, VE estimates against mild infection increased to 62% [11] and 71% [10] following booster doses.
In Albania, an upper-middle-income country of 2.8 million inhabitants in Southeast Europe, COVID-19 vaccination with BNT162b2 (Comirnaty, Pfizer-BioNTech) started on 11 January 2021, and HCWs were a high-priority group for vaccination.Additionally, ChAdOx1-S (Vaxzevria, AstraZeneca) and CoronaVac (Sinovac Life Sciences) were introduced in mid-March 2021.On 15 October 2021, the Albanian National Technical Advisory Group for Immunizations recommended a booster vaccine for all HCWs who had completed their primary series at least 6 months prior.As of 31 January 2023, while most Albanian HCWs (83%) had received primary series vaccine, only 20% had received a booster (third) dose [4].

Study Design
We conducted an interim analysis of an ongoing prospective cohort study that started in February 2021.Our objective was to estimate VE of COVID-19 primary vaccine series and booster dose against symptomatic and asymptomatic SARS-CoV-2 infection among HCWs at 3 hospitals in Albania, from January through May 2022.Study methods and early primary series VE results have been previously published [12,14].

Data Collection
At enrollment (February-May 2021), participants completed a questionnaire that included questions about demographics, health status, hospital role, and prior COVID-19 vaccination.During the study, participants completed weekly symptom questionnaires through telephone interviews; participants who reported having any symptom included in the Albanian Ministry of Health COVID-19 case definition [12] provided a nasopharyngeal specimen that was tested for SARS-CoV-2 by reverse-transcription polymerase chain reaction (PCR) at the Institute of Public Health Laboratory in Tirana, Albania.
Study staff cross-checked the Albanian National Surveillance of Infectious Diseases electronic information system to identify any PCR or rapid antigen tests (RATs) performed in laboratories and hospitals outside of the study, and entered these data into the study database.Self-tested RAT results were not recorded.Participants who tested positive for SARS-CoV-2 by PCR or RAT were administered a follow-up questionnaire about their course of illness 30 days after their positive test.Study staff verified participants' COVID-19 vaccination status through the national integrated immunization information system and the family care physicians' web-based medical data system (E-vizita).All study data were entered securely and stored in REDCap [15].

Laboratory Procedures
We collected blood specimens from participants at enrollment and then every 3 months throughout the study.Specimens were tested for total anti-nucleocapsid antibody using the Platelia SARS-CoV-2 Total Antibody Assay (Bio-Rad Laboratories, Hercules, California).Cutoff values were determined according to instructions from the package insert.
PCR-positive specimens were sent to the Institute of Virology-Charité (Berlin, Germany), where a representative subset of specimens based on dates and location and cycle threshold value <24 was selected to undergo whole genome sequencing (WGS) [16].We also inspected Albanian molecular surveillance data from the Global Initiative on Sharing All Influenza Data (GISAID) [16].

Vaccine Effectiveness Analysis
We measured primary series and booster dose VE compared to unvaccinated participants.Additionally, we measured the relative VE (rVE) of booster dose, comparing HCWs who received primary series and booster dose vaccination with HCWs who received primary series alone.For both VE and rVE, the primary analytic outcome was PCR-confirmed symptomatic SARS-CoV-2 infection.For the booster dose VE analyses, HCWs started contributing person-time only when they were eligible to receive the booster dose.We also measured VE against 2 secondary outcomes: an outcome of any PCR-or RAT-confirmed symptomatic SARS-CoV-2 infection, and a combined outcome of symptomatic and asymptomatic SARS-CoV-2 infection, confirmed by 1 of 3 tests: PCR, RAT, and/or seroconversion.Because inactivated vaccines can produce anti-nucleocapsid antibodies, we excluded participants who received CoronaVac from the analysis that included seroconversion as an endpoint.We defined a symptomatic SARS-CoV-2 infection as symptom onset between 14 days before and 4 days after the collection date of a SARS-CoV-2-positive swab.We defined seroconversion as a positive anti-nucleocapsid antibody test preceded by a negative anti-nucleocapsid antibody test 3 months earlier.
For participants who seroconverted without having a positive PCR test or RAT, we estimated the date of infection using 2 approaches: For participants who reported symptoms on their weekly questionnaire during the period of seroconversion, the inferred date of infection was the date of symptom onset.For those without reported symptoms, we estimated the date of infection as the midpoint between the last negative serological test and the date 3 weeks before the subsequent positive serological test [17].
We defined prior infection as a history of a positive PCR test, RAT, or anti-nucleocapsid serology test at the start of the analysis period.For each analysis described above, we also conducted stratified analyses to evaluate VE among participants who did and did not have evidence of previous SARS-CoV-2 infection at any point before the study start.We also assessed the combined protective effect of vaccination and prior SARS-CoV-2 infection using 4 levels of exposures: (1) primary series and no prior infection (reference group), (2) primary series and prior infection, (3) booster vaccination without prior infection, and (4) booster vaccination and prior infection.Finally, we assessed the impact of time since vaccination on VE of primary series vaccination compared to unvaccinated.

Statistical Model
We used Cox proportional hazards models to estimate VE as (1 -adjusted hazard ratio) × 100; therefore, all reported VE estimates are adjusted.Vaccination was a time-varying exposure as vaccination status could have changed over time.Thus, the same participant could contribute person-time to >1 exposure category.Calendar time was used as the underlying time in the Cox regression.We used the Schoenfeld residual test to check the proportional hazard assumption.All analyses were performed using R software [18].
We calculated unadjusted and adjusted hazard ratios and estimated VE, including hospital as a fixed effect to account for the multisite design.We considered the following a priori covariates to be added in the multivariable regression model: previous infection, age, sex, household size, and any chronic condition.We included the aforementioned confounders that changed the VE estimate by >5%.Person-time contribution started either on 1 January 2022 or from the start of time at risk for those with prior SARS-CoV-2 infection (90 days after positive test or inferred infection date); it ended at the first occurrence of any of the following events: (1) the day of the SARS-CoV-2 infection, (2) the day of the last weekly questionnaire before complete loss to follow-up (28 days of not completing weekly symptom questionnaire), (3) the day of withdrawal from the study, or (4) the censor date for the analysis period (31 May 2022).Participants who received a second or third vaccine dose were excluded for 14 days, after which they were considered to be fully immunized with their respective dose and added to the corresponding exposure category.Person-time of participants vaccinated with only 1 dose was excluded from the analysis.For the secondary analysis that included seroconversion as an outcome, person-time also ended on the day of the last serological result or receipt of an inactivated vaccine.We also performed 2 sensitivity analyses (Supplementary Material).

Outcomes
The completion rate of all weekly symptom questionnaires that were administered was 89% during the follow-up period.
Overall, 167 participants reported symptomatic episodes during the study period; 2 participants reported 2 symptomatic episodes, ≥30 days apart.All participants who reported symptoms had a SARS-CoV-2 PCR or RAT performed; 104 were positive (94 by PCR and 10 by RAT), and 63 participants were negative by PCR or RAT.Among participants included in the analysis, 86 (45%) had symptomatic SARS-CoV-2 infections and 107 had asymptomatic infections (55%) (Figure 1B).Of the 86 symptomatic infections, 76 were detected by PCR and 10 by RAT.Of the symptomatic infections, 12 were among unvaccinated participants, 70 were in participants who had received primary vaccine series, and 4 were in participants who had received a booster dose (Figure 1A).Most of the 107 asymptomatic infections (101 [94%]) were detected through seroconversion (Figure 1B).Combining symptomatic and asymptomatic infections, there was a total of 19 infections among unvaccinated participants, 156 infections in participants with primary vaccine series, and 18 infections in participants who received a booster dose.Almost all symptomatic infections (85/86 [99%]) occurred in January and February, a period when Omicron BA.1 and BA.2 were circulating in Albania (Supplementary Figure 1B).Overall, 17 (22%) samples underwent genomic sequencing, of which 16 (94%) were of BA.1 or BA.2 Omicron sublineage.BA.1 and BA.2 sublineages accounted for >96% of sequenced samples from Albania reported to GISAID in January and February 2022 [16] (Supplementary Figure 1B).

DISCUSSION
Using data from an ongoing cohort study of HCWs in Albania [12,14], we found that COVID-19 booster dose VE against symptomatic SARS-CoV-2 infection during a period of Omicron BA.1 and BA.2 predominance was high, at 88%.Conversely, primary vaccine series VE was lower (34%) with low precision.In our cohort, only 17% of HCWs had received a booster dose at the end of the follow-up period, and as of mid-January 2023 only 20% of all HCWs in Albania had received a COVID-19 booster dose [4].Our high booster VE findings, which to our knowledge reflect the first published booster dose VE results in Southeast Europe, provide evidence to support increased uptake of booster doses among HCWs in Albania.
The high booster dose VE against symptomatic infection, reflecting mainly BNT162b2 booster doses given within 1 month before the start of follow-up, is slightly higher than findings from studies in the United Kingdom [11] and Qatar [19], which found recently administered BNT162b2 booster VE of 67% and 52%, respectively, against symptomatic Omicron BA.1 and BA.2 infection, with CIs that overlap with those from our study.We also found that rVE was 57% for booster dose compared to HCWs who had received the primary vaccine series a median time of 10 months prior to follow-up.The 95% CIs around rVE estimate were wide, with lower bounds that crossed zero.The limited precision of our VE estimates may reflect the relatively small number of events in this analysis.Studies of booster dose rVE against symptomatic Omicron infection in the United   States [20] and Qatar [21] found low to moderate VE, but rVE was much higher against severe outcomes such as hospitalization and death.
In our study, we found that primary vaccine series provided low VE (34%) against symptomatic Omicron infection, although precision was low due to small numbers.This estimate differs from the higher primary series VE estimate we previously found against symptomatic infection during a Delta-predominant period (67%) in the same cohort [14].In our study, median time since receiving the second dose was almost 10 months; therefore, in addition to the increased immune escape of Omicron, waning of vaccine immunity likely contributed to low primary series VE.Waning VE against symptomatic Omicron infection has been shown in other studies [9].In the United Kingdom [11], BNT162b2 VE dropped from 66% in the first month after completion of the primary series to 9% 6 months after.In our study, we could not observe significant waning immunity following primary vaccine series, likely due to the low power secondary to the low number of events.
We found that VE was highest among individuals who were both boosted and had prior SARS-CoV-2 infection.These findings are consistent with many published studies that have demonstrated the benefits of hybrid immunity [22], and support the Albanian Ministry of Health and Social Protection recommendations that individuals should receive a primary vaccine series regardless of prior infection history and that HCWs and other priority populations should receive a booster even if they were previously infected.This message is particularly important in the context of populations with high rates of previous infection; a study in the United Kingdom reported lower vaccine acceptance among individuals who were previously infected and thought the vaccine would no longer be useful [23].
This study has several strengths.First, regular serological testing of HCWs allowed us to detect seroconversions, infer asymptomatic infections, and estimate VE against any SARS-CoV-2 infection.Serology testing also increased our sensitivity in detecting prior infection.Additional strengths of the study are its prospective cohort design and the very high retention rate of the participants after 1 year of follow-up.The completion rate of the weekly symptom questionnaire remained high during the follow-up period.Finally, all participants with symptoms were tested for SARS-CoV-2.The high sampling rate reflects 2 main factors: (1) There was a mandatory requirement in Albania that all symptomatic healthcare workers be tested for SARS-CoV-2; and (2) study coordinators were also hospital employees, so they knew the participants well and could follow the participants closely.
Our study has several limitations.First, our findings may not be perfectly generalizable to the overall population due to differences in SARS-CoV-2 exposure, behavior, and demographics.In addition, due to the observational, nonrandomized study nature, there were differences in age, profession, comorbidities, and prior infection status among those unvaccinated, those with primary series only, and those who received a booster vaccine.Due to the high coverage of the primary vaccine series in our cohort, only a small number of unvaccinated HCWs were available for a comparison group, and this group may have been different from vaccinated participants with respect to their exposures to SARS-CoV-2 and disease risk [24].To overcome this limitation, we also calculated relative VE of booster dose compared to primary vaccine series.In addition, our study was not powered to estimate VE against severe but less common outcomes like hospitalization or death.Other studies have shown high, durable primary series VE against severe outcomes during Omicron for BNT162b2 [9,10].
Although we were able to detect prior infections through  multiple testing platforms, we were not able to analyze time since previous infection.We could not confirm through WGS that all positive participants were infected with BA.1 or BA.2 SARS-CoV-2 Omicron sublineage.There were limited molecular surveillance data from Albania from the first 2 weeks of January 2022.Finally, because of the low number of events and the relatively low VE, some of our VE estimates had wide CIs and overlapped zero.
In our cohort study of Albanian HCWs, we found that a monovalent COVID-19 booster dose offered better protection against symptomatic SARS-CoV-2 infection compared to primary vaccine series during a period of BA.1/BA.2Omicron circulation.These findings underscore the importance of the Albanian Ministry of Health and Social Protection's recommendation that all HCWs receive a booster dose 6 months following receipt of their primary vaccine series.Our results, which reflect the first locally generated booster VE data, could be communicated to Albanian HCWs to decrease vaccine complacency and increase booster uptake.Our findings could be further generalized and used to promote booster dose uptake among HCWs in neighboring countries in Southeast Europe and in the general population, where booster uptake remains low.

Table 2 . Continued
Abbreviations: CI, confidence interval; HCWs, healthcare workers; HR, hazard ratio; PCR, polymerase chain reaction; RAT, rapid antigen test; VE, vaccine effectiveness.a Adjusted for hospital site, time since previous infection, age, sex, and household size.b Participants eligible for booster dose.

Table 3 . Relative Vaccine Effectiveness of Booster Dose Compared to Primary Vaccine Series Against Any Severe Acute Respiratory Syndrome Coronavirus 2 Infection During a Period of Omicron (BA.1 and BA.2) Predominance, Stratified by Previous Coronavirus Disease 2019 Infection, Albania, 1 January-31 May 2022
: CI, confidence interval; HCWs, healthcare workers; HR, hazard ratio; PCR, polymerase chain reaction; RAT, rapid antigen test; VE, vaccine effectiveness.
Abbreviationsa Adjusted for hospital site, age, and sex.