COVID-19 vaccine effectiveness for healthcare workers in Georgia during the Omicron wave, January to June 2022

COVID-19 has caused high rates of illness and mortality in health care workers (HCWs) [1]. HCWs are at high risk of occupational exposure, and ensuring the safety and health of HCWs is critical for maintaining functioning healthcare systems in pandemic and interpandemic periods.

COVID-19 vaccines are an important tool to reduce morbidity and mortality from SARS-CoV-2 infections [2,3]. Reducing SARS-CoV-2 infections among HCWs can reduce work absences and also reduce the possibility of onward transmission to vulnerable patients [4,5]. While COVID-19 vaccine effectiveness (VE) has been widely studied in high-income countries (HICs), few studies have evaluated COVID-19 booster dose VE in middle-income countries (MICs) in Europe, where populations are different, and a variety of COVID-19 vaccines have been used in resource-limited settings.

Georgia is an upper-MIC in the South Caucasus with a population of 3.7 million people [6]. Four two-dose COVID-19 vaccines are currently approved for use among adults in the Republic of Georgia: Oxford/AstraZeneca AZD1222, Pfizer/BioNTech BNT162b2, Sinopharm (Beijing) BBIBP-CorV, and Sinovac (CoronaVac) [7]. HCWs were a priority group for vaccination after vaccines became available on 15 March 2021 [8]. As of May 2023, in Georgia, 34.6% of the population had received at least one dose, 31.6% had received a PVS, and 6.7% had received a booster dose [9].

In April 2021, we initiated a prospective cohort study of COVID-19 VE in HCWs in Georgia. VE estimates against the Delta variant from this study have been previously published [10], but VE estimates against the Omicron variant have not been previously published. Georgia was heavily affected by the Omicron wave; daily new cases peaked in early February, 2022 [11]. Here we evaluated COVID VE during the early Omicron period using data from the same cohort (January 1 – June 1, 2022).

This cohort study has been described previously [10,12,13]. Briefly, we enrolled HCWs from six hospitals in Georgia. We collected information on socio-demographics, underlying conditions, hospital role, self-assessed health status, prior SARS-CoV-2 infection, and COVID-19 history. Every participant provided a blood sample for serology testing at the time of enrolment, and quarterly during the course of the study. Serology samples were tested for anti-nucleocapsid antibody (Anti-N) and anti-spike antibody (Anti-S), as previously described [10].

Throughout the study timeline, participants completed a weekly symptom questionnaire, administered by study personnel. Participants who reported any COVID-19 symptoms included in the Georgia Ministry of Health (MoH) suspected COVID-19 definition (fever, cough, general weakness, fatigue, headache, muscle aches, sore throat, runny nose, shortness of breath, lack of appetite, nausea, vomiting, diarrhea, altered mental status, loss of taste, or loss of smell) provided a respiratory specimen, which was tested for SARS-CoV-2 by PCR or RAT. During the entire study period, as part of the national COVID-19 response, HCWs at all six hospitals were recommended to be tested every week for SARS-CoV-2 by RT-PCR or RAT, but in practice testing was done inconsistently.

Participants who tested positive for SARS-CoV-2 by RT-PCR or RAT were administered a follow-up questionnaire about their clinical course 30 days after the date of their positive test. Data on testing and vaccination were confirmed using national databases.

On January 1, 2022, there were 1592 HCWs enrolled in the study; 1253 were included in this analysis (Fig 2). Of those included in the analysis, the median age was 41 (Interquartile range: 30–53), 1058 (84%) were female, and 316 (25%) reported having at least one chronic condition. Most HCWs were nurses or midwives (39%) and medical doctors (21%). The majority of HCWs (64%) had evidence of prior SARS-CoV-2 infection at the beginning of the analysis period (Table 1).

At the beginning of the study period, 855 (68%) had received PVS, 141 (11%) had received a first booster dose, and 248 (20%) remained unvaccinated. Most boosters were BNT162b2 vaccine (92%) and BBIBP-CorV vaccine (8%), and most PVS were BNT162b2 vaccine (68%) and BBIBP-CorV vaccine (24%) (Table 1).

During the analysis period, there were a total of 372 PCR or RAT-confirmed SARS-CoV-2 symptomatic infections; 63 symptomatic infections among unvaccinated participants, 264 symptomatic infections among PVS, and 45 symptomatic infections among those who received a first booster dose (Fig 3). Overall, 267 (68%) of infected participants completed the clinical course at 30 days questionnaire. In total, 126 (47%) of infected participants sought medical care, and 8 (3%) were hospitalized.

Absolute booster dose VE against symptomatic SARS-CoV-2 infection was 40% (95% CI -56–77) in the 7–29 days following receipt of the first booster dose, -9% (95% CI -104–42) for 30–59 days following receipt of the first booster dose, and -46% (95% CI -156–17) for ≥ 60 days following receipt of the first booster dose (Table 2, Fig 4).

Relative VE of a booster dose compared to PVS against symptomatic infection was 58% (95% CI 1–82) for 7 days – 29 days from first booster dose receipt, 21% (95% CI -33–54) for 30 days – 59 days from first booster dose receipt, and -9% (95%CI -82–34) for ≥ 60 days from first booster dose receipt (Table 2, Fig 4).

Relative VE for a first booster dose of BNT162b2 vaccine compared to only a PVS (≥ 150d) of BNT162b2 vaccine against symptomatic infection was 77% (95%CI 3–94) for 7 days – 29 days from the first booster dose receipt, 26% (95% CI -47–63) for 30 days – 59 days from the first booster dose receipt, and -16% (95%CI -103–34) for ≥ 60 days from the first booster dose receipt (Table 2, Fig 4). VE for other booster vaccines could not be calculated due to small sample size.

Absolute VE of a first booster dose against medically attended infection was 45% (95% CI -121–86) for 7 days – 29 days from the first booster dose receipt, 29% (95% CI -71–71) for 30 days – 59 days from the first booster dose receipt, and 21% (95% CI -104–70) for ≥ 60 days from booster dose receipt VE (Table 2, Fig 4).

Relative VE for a booster dose compared to PVS against medically attended infection was 45% (95% CI -121–87) for 7d – 29d from booster dose receipt, 33% (95% CI -67–73), for 30d – 59 days from the first booster dose receipt, and 27% (95%CI -90–72). for ≥ 60d from the first booster dose receipt (Table 2, Fig 4).

During Omicron circulation in early 2022 in Georgia, we found moderate COVID-19 absolute VE (≥40%) of the first booster dose against both symptomatic and medically attended infection that rapidly waned after 29 days in a cohort of HCWs with a high rate of previous infection. Relative VE against symptomatic infection was slightly higher (58%) for a first booster dose compared to PVS, but relative VE also quickly waned after 29 days since vaccination (Table 2, Fig 4).

Our finding of moderate VE against symptomatic infection that rapidly waned during Omicron is similar to findings from other studies showing moderate VE using ancestral monovalent vaccines against Omicron that decreased over time. A systematic review of VE against symptomatic infection during the Omicron period found pooled estimates of VE after any PVS and any booster dose waned over time. VE after 1 month from PVS was 53%, after 6 months was 14%, and decreased to 9% after 9 months since PVS. VE after 1 month since receiving any booster dose was 60%, after 6 months VE was 23%, and after 9 months VE was 13% [18]. A multi-country study in Europe of COVID-19 VE against symptomatic infection during Omicron (Dec 2021 – June 2022) also found significant waning in protection. All-product PVS VE was 60% < 90 days since vaccination and 29% ≥ 180 after vaccination. All-product first booster VE was 56% < 90 days since vaccination and 3% ≥ 180 after vaccination [19].

A study of COVID-19 VE against symptomatic infection during Omicron from England (November 2021 to January 2022) showed that booster dose BNT162b2 VE against symptomatic infection was 67% following 1 week of booster dose receipt and then receded to 46% after 10 weeks [17]. Another study in England of COVID-19 VE against symptomatic infection during BA.1 and BA.2 (January – March, 2022) found VE of a first booster dose against symptomatic infection with BA.1 or BA.2 1 week after vaccination was 71% and 74%, respectively. VE against the same outcome >15 weeks after vaccination for BA.1 and BA.2 was 37% and 44%, respectively [20]. VE of PVS against symptomatic infection from <2 weeks since vaccination for BA.1 and BA.2 was 63% and 64%, respectively. VE of PVS against symptomatic infection from 25 + weeks since vaccination for BA.1 and BA.2 was 15% and 28%, respectively [20]. Another study of COVID-19 VE against symptomatic infection during Omicron from Canada (December 2021) showed that VE of PVS against symptomatic infection was 36% from 7–59 days since vaccination and 15% from 120–179 days since vaccination [21].

The above studies indicate overall a slower decline in VE compared to our study, which found a rapid decline in VE. Our HCW setting differs from these studies, all of which are in the general population. HCW are likely to have higher exposure levels to SARS-CoV-2 than the general population [15]. Within our study population, the epidemic curve was extremely steep (Fig 3); 95% of HCWs infected during the study period had a symptomatic SARS-CoV-2 infection between 1 January and end of February. This comprises 28% of the total study population. This force of infection, and the context of COVID-19 vaccine being at least a partially leaky vaccine [22,23], meaning that protection may be affected by a number of exposures, may explain at least in part the more rapid decline in VE observed in our study.

The higher point estimates for relative VE compared to absolute VE suggest that the risk of symptomatic infection among individuals who completed their primary series more than 150 days ago was higher than the risk of symptomatic infection among the unvaccinated. We would expect that the risk of symptomatic infection among those vaccinated with primary series more than 150 days ago would be the same or if anything slightly lower than the unvaccinated, in case of residual protection. This difference may reflect differences between unvaccinated HCWs and those vaccinated more than 150 days ago in terms of virus exposure or variability in seroprotection related to previous infection. However, as confidence intervals overlap between estimates, there is uncertainty in this difference, and it may be due to random error.

While we found that moderate VE against symptomatic infection rapidly waned, we were not able to evaluate VE against more severe endpoints than medically attended COVID-19. Studies have found that the COVID-19 vaccine continues to be effective in preventing severe disease. A multi-country study in Europe during a period of Omicron-predominant circulation (December 2021 – July 2022) found that COVID-19 VE against hospitalization was 43% for PVS and 59% for a first booster dose for ≥ 150 days between the last PVS and symptom onset or last PVS and booster dose with 85% VE if illness onset was 14–59 days after booster receipt [24]. A study in the Netherlands from October – December 2023 looking at VE in those aged ≥60 during found VE against hospitalization was 71% [25].

Our study had several strengths. We used PCR-confirmed infection and serology to determine participants with prior infections. We used both PCR testing and rapid antigen tests to identify infections more completely among study participants. We were able to adjust for different confounders in our VE analysis. We were also able to evaluate VE against the more severe endpoint of medically attended infections.

Our study also had limitations. Our study was underpowered to detect the true effect in some sub-group analyses, particularly those with a low number of events. Therefore, the results of sub-group analyses should be interpreted cautiously. Our study was not powered to assess VE against more severe outcomes such as hospitalization, ICU admission, or death. Our sample size was low for estimating certain time periods of VE and medically attended VE, which resulted in wide CIs. We did not account for variability in routine testing of asymptomatic participants. However, we focused our analysis on symptomatic infection, an outcome which is less likely to be affected by the variability in routine asymptomatic testing. Because we could not use serology tests to determine previous infection status among the 22% of participants who had received inactivated vaccines, we likely underestimated the number of previous infections in this population. However, because we did not conduct a stratified analysis based on previous infection, this underestimation should not have biased our VE estimates. We did not have the resources to do micro-neutralization tests which would have given us more accurate results. We may have missed previous infections not captured by PCR testing or serology testing. While we could not find published data on the sensitivity of the ECLIA serology test against Omicron subvariants, the pooled sensitivity of the test against the original strain for IgG and IgM were 0.93 and 0.85, respectively, compared to rRT-PCR [26]. We were unable to stratify VE by prior infection status or estimate vaccine products other than BNT162b2 due to the small sample size. The high rates of prior infections among participants likely led to an underlying level of immunity not seen in uninfected populations. While the immunity likely blunted the magnitude of VE, it reflects the real world setting of Georgia at the time of the study. HCWs in our study enrolled voluntarily, which may have caused selection bias. Participants may have been healthier and more likely to be compliant with the study criteria compared to HCWs in the hospital. However, we do not know if this would have an overall impact on VE. While the hospitals included in our study are large hospitals in the Georgian capital, we do not have demographic and comorbidity data for all Georgian HCWs, and therefore we are unable to determine the extent to which our study population is representative of all HCWs in the country. However, we managed to enroll 40% of HCWs in the 6 study hospitals [10]. Enrolling a large proportion of HCWs helps generalize our findings to all HCWs in Georgia.

Overall, we found that COVID-19 booster doses were moderately effective at preventing symptomatic infection in hospital based HCWs in Georgia but VE rapidly waned. Our overall VE results should be interpreted cautiously as we used both inactivated and mRNA vaccines in our study. This study was conducted in the context of high previous infection and a high force of infection. These results suggest that timing COVID-19 vaccination campaigns prior to periods of anticipated high SARS-CoV-2 incidence, if practical, could be important to reduce COVID-19 infections. The moderate but waning VE in this study against symptomatic and medically attended infection combined with the more durable COVID-19 VE against severe outcomes described in other studies, underscores the need for broader efforts in Georgia to increase vaccine coverage among priority groups such as HCWs, and to encourage the adoption of guidelines that encourage high-risk individuals to receive annual COVID-19 revaccination, as recommended by WHO [27].