Effectiveness of COVID-19 Vaccines in the General Population of an Italian Region before and during the Omicron Wave

We performed a cohort analysis of the entire population of Abruzzo, Italy, to evaluate the real-world effectiveness of SARS-CoV-2 vaccines against infection, COVID-19 hospitalization or death, over time and during the Omicron wave. All resident or domiciled subjects were included, and official vaccination, COVID-19, demographic, hospital and co-pay exemption datasets were extracted up to 18 February 2022. Multivariable analyses were adjusted for age, gender, hypertension, diabetes, major cardio- and cerebrovascular events, COPD, kidney diseases, and cancer. During the follow-up (average 244 days), 252,365 subjects received three vaccine doses (of BNT162b2, ChAdOx1 nCoV-19, mRNA-1273 or JNJ-78436735), 684,860 two doses, 29,401 one dose, and 313,068 no dose. Overall, 13.4% of the individuals were infected with SARS-CoV-2 (n = 170,761); 1.1% of them had severe COVID-19, and 0.6% died. Compared with the unvaccinated, those receiving two or three vaccine doses showed an 80% to 90% lower risk of COVID-19 hospitalization or death. Protection decreased during the Omicron wave and six months after the last dose, but it remained substantial. Lethal disease was uncommon during the Omicron wave and in the young population, even among the unvaccinated. Some of the current policies may need a re-evaluation in light of these findings. The results from the Omicron wave will inevitably require confirmation.

However, several questions remain unanswered. Firstly, the length of follow-up is inevitably short, and vaccine long-term safety and effectiveness is uncertain. Indeed, a growing body of literature suggests that vaccine immunity may decrease over time as a result, among other factors, of the emergence and spread of new virus variants [14][15][16][17][18]. Secondly, it is unclear whether current vaccines are still effective against the virus variant B.1.1.259, commonly named "Omicron" [19], as early investigations found a decrease of 23% in the efficacy against Omicron-related hospitalizations [20,21]. Third, as a response to the potential waning of vaccine-induced immunity and Omicron spread, many countries introduced a third, booster dose of vaccine, the effectiveness of which remains to be evaluated in large population studies [22,23].
In this cohort analysis of the entire population of an Italian region, we evaluated the risk of infection, COVID-19 hospitalization and death of the unvaccinated subjects versus a.
The subjects who received only one dose of BNT162b2, mRNA-1273 or ChAdOx1 nCoV-19 vaccines, were included in the group "1 dose"; b.
The subjects who received only one dose of JNJ-78436735 vaccine, or only two doses of BNT162b2, mRNA-1273 or ChAdOx1 nCoV-19 vaccines, were included in the group "2 doses"; c.
The subjects who received two doses of JNJ-78436735 vaccine, or three doses of BNT162b2, mRNA-1273 or ChAdOx1 nCoV-19 vaccines, were included in the group "3 doses".
For the analyses evaluating the effectiveness of the various vaccines vs. no vaccines, the subjects who received two or three different vaccines were included in a global category named "mixed vaccines".
The start of follow-up varied across vaccine categories, as all vaccine groups were separately compared versus unvaccinated subjects: a.
For the analyses comparing unvaccinated versus those receiving only one vaccine dose (set of analyses "1 dose"), the start of follow-up was set 14 days after the single vaccine dose (to account for the time required for seroconversion) [27] for vaccinated subjects and 14 days after the first administration of the first vaccine dose (16 January 2021) for the group "unvaccinated". b.
For the analyses comparing unvaccinated versus those receiving two vaccine doses only (set of analyses "2 doses"), the start of follow-up was set 14 days after the second vaccine dose for vaccinated subjects and 14 days after the first administration of the second vaccine dose (31 January 2021) for the group "unvaccinated". c.
For the analyses comparing unvaccinated versus those receiving three vaccine doses (set of analyses "3 doses"), the start of follow-up was set 14 days after the third vaccine dose for vaccinated subjects and 14 days after the start of the mass administration of the third vaccine dose (17 September 2021) for the group "unvaccinated". For the outcome "infection", the end of follow-up was set: − On 18 February 2022 for those who had no positive swabs during the follow-up; − On the day of the first positive swab for those who were not infected before the start of the follow-up ("uninfected" cohort).
For the outcome "COVID-19 hospitalization", the end of follow-up was set on the day of the hospital admission, or on 18 February 2022, for all the subjects who were not hospitalized with COVID-19 related symptoms. For the outcome "death", the end of follow-up was set on the day of the death with a positive swab, or on 18 February 2022 for all others. A schematic definition of the study groups, outcomes and follow-up is reported online in Table S1.

Data Collection
All the information about vaccines, laboratory tests, demographic, anamnestic and clinical residents' data are routinely entered into NHS official datasets, updated daily, and sent to the Italian Institute of Health [29]. We extracted all data from the official vaccination, COVID-19, demographic, hospital (Italian SDO) and co-pay exemption (Italian "Esenzioni Ticket" file) datasets of the Abruzzo region. Individual data were merged through encrypted fiscal code.

Statistical Analyses
The main analyses separately compared the risk of the primary infections, COVID-19 hospitalization and COVID-related death of unvaccinated subjects versus individuals who received one, two, or three vaccine doses. Each of these comparisons were stratified by age category (0-29y, 30-59y, 60 + y), vaccine type, and time since the last vaccine dose (≤ or >182 days, or six months, which was the duration of the "green pass" set by the Italian Government for subjects who recovered after a SARS-CoV-2 infection) [30]. The age ranges were identified to be consistent with the periodical reports of the Italian Institute of Health [31], and with the recommendations of the Italian Government [32], which identified subjects aged ≥ 60 years as priority targets for vaccination.
As the survival analyses could be biased due to the severe unbalance in the average follow-up duration between the unvaccinated individuals (367 days) and those who received two (181 days) or three (71 days) vaccine doses, and considering that more than 80% of the infections occurred during the restricted time span of the Omicron wave (the last 50 days of follow-up), we opted to use multiple logistic regression instead of the typical Cox proportional hazards analysis to evaluate the independent association between each outcome and vaccination [33]. All models were adjusted for age, gender, hypertension, diabetes, major cardio-and cerebrovascular events, COPD, kidney diseases, and cancer, all included a priori, and the results of such covariates were reported in the online Table S2. Standard post-estimation tests were used to assess the validity of final models, performing multicollinearity and influential observation analyses (using standardized residuals, change in Pearson and deviance chi-square), and testing for potential interactions between vaccination and other covariates [34]. The goodness of fit was checked using Hosmer-Lemeshow test, and predictive power was assessed through C-statistics (area under the receiving operator curve).
The analyses were repeated, including only the events that occurred during the Omicron wave (from 27 December 2021 to 18 February 2022, when the proportion of Omicron variant in the available positive swabs was higher than 50%) [35]. In these analyses, all the subjects with a positive SARS-CoV-2 swab before 27 December 2021 were classified as "infected before the start of follow-up" and excluded.
A two-sided p-value of <0.05 was considered to be significant for all analyses, which were carried out using Stata, version 13.1 (Stata Corp., College Station, TX, USA, 2014).

Results
The STROBE flowchart of the study participants is shown in Figure 1. From 2 January 2021 to 18 December 2021, a total of 313,068 unvaccinated (24.5%) and 966,626 vaccinated (who received 2,156,216 doses) residents or domiciled in the Abruzzo region, Italy were included in the analyses. The main demographic characteristics of the population and the proportion of selected comorbidities are reported in Table 1 cron wave (from 27 December 2021 to 18 February 2022, when the proportion of O variant in the available positive swabs was higher than 50%) [35]. In these analyse subjects with a positive SARS-CoV-2 swab before 27 December 2021 were class "infected before the start of follow-up" and excluded.
A two-sided p-value of <0.05 was considered to be significant for all analyse were carried out using Stata, version 13.1 (Stata Corp., College Station, TX, USA,

Results
The STROBE flowchart of the study participants is shown in Figure 1. From 2 2021 to 18 December 2021, a total of 313,068 unvaccinated (24.5%) and 966,626 va (who received 2,156,216 doses) residents or domiciled in the Abruzzo region, Ita included in the analyses. The main demographic characteristics of the population proportion of selected comorbidities are reported in Table 1. Overall, 29,401 sub ceived only one vaccine dose (2.3% of the population), 684,860 received two doses and 252,365 received three doses (19.7%).
During the follow-up, 13.4% of the population was infected with SARS-CoV-2 (n = 170,761): 19.1% of the unvaccinated, 13.8% of those receiving one dose, 13.1% of those receiving two doses, and 6.5% of those who received three doses (Table 2). During     Table 1 footnote " G ", for the analyses comparing the group "1 dose" versus "unvaccinated", the follow-up started 14 days after the day of the single vaccine dose or, for the unvaccinated, on 16 January 2021 (14 days after the start of the immunization campaign). For the category "2 doses", the follow-up started 14 days after the second dose or, for the unvaccinated, on 31 January 2021 (14 days after the start of the administration of the second doses). Therefore, the subjects who tested positive for SARS-CoV-2 between 16th January and 31st January 2021 were included in the comparison "dose 1 versus no vaccination"; however, they were excluded from the comparison "dose 2 versus no vaccination" as they were infected before the start of the follow-up of the second dose. Given that the main comparison was "two doses vs. none", the numbers in the categories "total" and "unvaccinated" are the numbers of the subjects that were included in the comparison "2 doses" and may not be equal to the sum of the three categories. E Includes only the outcomes that occurred from the start of the follow-up to 26 December 2021, when the proportion of Omicron variant in the available positive swabs was lower than 50%. The average follow-up of the "3 doses" group was very short (19 days) during the pre-Omicron period. F Includes only the outcomes that occurred from 27 December 2021 to 18 February 2022, when the proportion of Omicron variant in the available positive swabs was higher than 50%. The subjects who were infected before the predominance of Omicron were excluded from the sample. G Subjects who received two or three different vaccines. H Two separate analyses were conducted: the first only included events that occurred 182 days (6 months) from the last vaccine dose (start of follow-up); the second only included the events that occurred >6 months after the last dose of vaccine. The subjects with follow-up shorter than 6 months, as well as the subjects that had a positive swab within 6 months of follow-up, were excluded from the analysis of the events occurring after six months of follow-up.
The mean follow-up was 244 ± 99 days, but varied widely by vaccine group, being shortest (71 ± 18 days) for the analyses of the effectiveness of the third dose, which was widely administered from September, 2021.

Risk of Primary Infection
The highest incidence of infection was observed in the young population (0-29 years): 21.8%, as compared to 6.3% among the adults aged ≥ 60 years. In the multivariable analysis (Table 3), when adjusting for age, gender, and comorbidities, the risk of infection was significantly lower for vaccinated subjects. As compared to the unvaccinated, the individuals who received one, two, or three doses showed the following adjusted odds ratios (ORs), respectively: 0.71, 0.75, and 0.74, with 95% confidence intervals (Cis) ranging from 0.68 to 0.76. The results did not substantially vary by vaccine type or age category, but when the analyses were restricted to the Omicron wave, only the subjects who received three doses showed some degree of protection against infection (OR: 0.81; 95% CI: 0.80-0.83), while those who received two doses did not differ from the unvaccinated (OR: 0.98; 0.97-1.00). On the contrary, during the pre-Omicron waves, the risk of infection was significantly lower among the subjects who received any dose of vaccine (online Table S3).
Overall, the frequency of COVID-19 hospitalization was considerably higher among the unvaccinated (0.41% of the population; 2.12% of the infected) as compared to those receiving two (0.06% and 0.44%, respectively) or three (0.06% and 1.00%, respectively) vaccine doses. Multivariable analyses confirmed a significantly and substantially lower risk of COVID-19 hospitalization for the subjects who received one (OR: 0.53), two (0.12) or three (0.21) vaccine doses, as compared with the unvaccinated (Table 3; all p < 0.001). The results were similar when the analyses were restricted to the infected persons, and stratified by vaccine type or age category. In contrast, during the Omicron wave (Table 3), and after six months from the last vaccine dose (Table 4), the effectiveness of two doses decreased to approximately 66% and 69%, respectively, although remained highly significant.
Overall, both mortality and lethality were considerably higher among the unvaccinated (0.28% and 1.46%, respectively) as compared to those receiving two (0.02% and 0.12%) or three (0.01% and 0.19%) vaccine doses. Multivariable analyses confirmed a significantly and substantially lower risk of death for the subjects who received one (OR: 0.49), two (0.06) or three (0.24) vaccine doses, as compared with the unvaccinated (Table 3; all p < 0.001). The results were similar when the analyses were restricted to the infected persons, and stratified by vaccine type or age category. A lower, yet considerable protection of two vaccine doses was observed six months after the last dose (OR: 0.25; 95% CI: 0.17-0.35; Table 4) and during the Omicron wave, where the number of deaths was too low to determine a lack of statistical power. Only the group receiving three doses showed a significantly lower risk of death compared to the unvaccinated (OR: 0.42; 95% CI: 0.26-0.68; Table 3).

Vaccine Safety
The regional surveillance system communicated that, between 2 January 2021 and 31 December 2022, the passive surveillance system received 179 spontaneous reports of possibly vaccine-related adverse events, 64 of which were defined as "severe", with no Vaccines 2022, 10, 662 9 of 14 deaths from 2,306,000 doses administered during the year 2021 (2.8 severe adverse event ×100,000 doses). The system did not provide any other data, including the proportion of severe adverse events whose correlation with vaccination was assessed and verified [36].     (6 months) from the start of follow-up, which was censored at 182 days; the second only included the events that occurred >6 months after the start of follow-up. The subjects with follow-up shorter than 6 months, as well as subjects that had a positive swab within 6 months of follow-up, were excluded from the analysis of the events occurring after the 6 months of follow-up.

Discussion
This study evaluated the effectiveness of some of the most commonly used SARS-CoV-2 vaccines in the general population of an Italian region, followed for more than six months on average, in the context of a mass vaccination campaign. The main findings were: (a) compared with the unvaccinated, persons who received two or three vaccine doses showed a 80% to 90% lower risk of COVID-19 hospitalization or death; (b) during the Omicron wave, and six months after the last dose, protection against COVID-19 decreased but remained considerable; (c) the effectiveness against infection was modest at any dose; (d) the risk of a lethal disease was very small during the Omicron wave and in the young population, even among the unvaccinated; and (e) overall, the reported incidence of serious adverse events was very low.
Our results on the effectiveness of current vaccination schedule are consistent with the existing literature on pre-Omicron waves [18,21]. Since the first months of 2021, two doses of vaccines were able to confer a high real-world protection against severe or lethal COVID-19 in several population studies [10][11][12][13][37][38][39]. With regard to the increase in the effectiveness that may derive from a third dose, which has been extensively recommended worldwide, some preliminary reports from Israel, the UK, and USA have recently been published, indicating the very high effectiveness of booster doses against severe or fatal disease [23,40,41]. In this study, we observed a limited or null increase in the protection after a third dose, but the effectiveness was already very high after two doses, and the follow-up was relatively short. Concerning the potential decrease in protection caused by the Omicron variant, our findings were reassuring, in line with those from a few recent studies that also reported a high vaccine effectiveness against severe or lethal COVID-19 during Omicron predominance [23,40,41]. As both present and previous studies could only consider the first weeks of the Omicron wave, further data are inevitably required to confirm whether the current vaccines should still be considered as the pivotal strategy to control the pandemic, or whether new vaccines or policy changes are required.
The first reports of waning vaccine immunity emerged in October 2021, when Tartof et al. hypothesized that vaccine effectiveness against infection decreases with time from immunization, irrespective of the dominant virus variant [42]. In December 2021, another study reported no residual effectiveness against severe or lethal disease after more than seven months [16], and data from Israel suggested that waning immunity may also incur after booster doses [43]. In our study, we observed a reduction of 30% in effectiveness after six months from the last vaccine dose, but subjects who received two doses still showed a 70% lower risk of COVID-19 hospitalization compared with the unvaccinated, during both pre-Omicron and Omicron waves. Although the biological mechanisms of the waning of vaccine-induced immunity remain to be elucidated [44], these findings remain concerning, suggesting that a strategy of periodical immunization should be evaluated if the pandemic does not decrease its morbidity.
Italy, as well as several other countries, has established a different set of restriction measures according to vaccination status, and unvaccinated subjects have been denied access to a number of public or private facilities [45]. As a consequence, vaccinated subjects may have been exposed to a higher risk of SARS-CoV-2 transmission than unvaccinated individuals [21]. This clear bias may explain, at least in part, why the effectiveness of current vaccines against infection was modest at any dose.
In line with an ample amount of the literature, the number of COVID-19 hospitalizations was very low, with no deaths among the individuals younger than 30 years [14,18,21,41,46]. Given that the incidence of COVID-19 hospitalization was as low as 1 out of 2000 infections for unvaccinated individuals, and even though two vaccine doses were able to reduce the risk of hospitalization by 90%, the risk-benefit profile of multiple vaccine doses for this population should be carefully evaluated.
The overall number of serious adverse events following vaccination that was reported by the surveillance system of the Abruzzo region was very low, in agreement with the acceptable safety profile reported by all previous studies [47,48]. However, we had no access to individual-level data and, as correctly admitted by the operators of the system, the overall number of reports was extremely scarce, suggesting a low efficacy of the current, passive pharmacovigilance system for monitoring vaccine safety [49]. Furthermore, more detailed data are strongly needed to properly assess the harm-benefit profile of vaccination for specific clusters of the population [50].
The strengths of this study are the use of official, routinely collected electronic health databases from the entire, general population of an Italian region, followed for a maximum of 12 months. Importantly, individual-level data were available on several comorbidities that may influence the risk of COVID-19 hospitalization and death, and the analyses were accordingly adjusted for. Finally, although the follow-up was inevitably short, we had the chance to evaluate vaccine effectiveness during the first seven weeks of the Omicron wave, when more than 140,000 new infections were recorded.
This study also has limitations that must be considered when interpreting the results. Some of them have been mentioned: a relatively short follow-up after the booster dose and a higher contagion risk of the vaccinated vs. unvaccinated subjects, resulting from the less stringent restrictions, that may lead to an underestimation of vaccine effectiveness against infection. However, on the other side, under the regulations in force in Italy at the time of this study, unvaccinated individuals were more likely to undergo multiple testing (requested to attend social gatherings) [45], which may have enhanced their likelihood of a testing positive for COVID-19, potentially resulting in an overestimation of vaccine effectiveness for this outcome [51]. Finally, it may also be argued that the negative test case-control design is more appropriate for assessing real-world vaccine effectiveness in the context of a pandemic [52]; however, a good concordance with findings from cohort studies was observed by previous research, which directly compared the two methodologies [37]. Likewise, the observational study design was suboptimal for performing head-to-head comparisons of the diverse administered vaccines, all of which showed a similar effectiveness profile.

Conclusions
This large population-based cohort study confirmed that two or three doses of all current vaccines were able to confer a strong protection against COVID-19 hospitalization or death. During the Omicron wave, and six months after the last dose, protection decreased but remained high. The effectiveness against infection was modest at any dose, likely as a result of the differences in restriction policies by immunization status. Finally, the rate of reported serious adverse events was very low.