key: cord-1050307-g2991gqd
authors: Gram, M. A.; Emborg, H.-D.; Schelde, A. B.; Friis, N. U.; Nielsen, K. F.; Moustsen-Helms, I. R.; Legarth, R.; Lam, J. U. H.; Chaine, M.; Malik, A. Z.; Rasmussen, M.; Fonager, J.; Sieber, R. N.; Stegger, M.; Ethelberg, S.; Valentiner-Branth, P.; Hansen, C. H.
title: Vaccine effectiveness against SARS-CoV-2 infection and COVID-19-related hospitalization with the Alpha, Delta and Omicron SARS-CoV-2 variants: a nationwide Danish cohort study
date: 2022-04-20
journal: nan
DOI: 10.1101/2022.04.20.22274061
sha: 10533b94a11bb1fad1f2d7a671c8d8b958d189b1
doc_id: 1050307
cord_uid: g2991gqd

Background The continued occurrence of more contagious SARS-CoV-2 variants and waning immunity over time require ongoing re-evaluation of the vaccine effectiveness (VE). This study aimed to estimate the effectiveness in two age groups (12-59 and 60 years or above) of two and three vaccine doses (BNT162b2 mRNA or mRNA-1273 vaccine) by time since vaccination against SARS-CoV-2 infection and COVID-19-related hospitalization in an Alpha, Delta and Omicron dominated period. Methods A Danish nationwide cohort study design was used to estimate VE against SARS-CoV-2 infection and COVID-19-related hospitalization with the Alpha, Delta and Omicron variants. Information was obtained from nationwide registries and linked using a unique personal identification number. The study included all residents in Denmark aged 12 years or above (18 years or above for the analysis of three doses) in the Alpha (February 20 to June 15, 2021), Delta (July 4 to November 20, 2021) and Omicron (December 21, 2021 to January 31, 2022) dominated periods. VE estimates including 95% confidence intervals (CIs) were calculated using Cox proportional hazard regression models with adjustments for age, sex and geographical region. Vaccination status was included as a time-varying exposure. Findings In the oldest age group, VE against infection after two doses was 91.0% (95% CI: 88.5; 92.9) for the Alpha variant, 82.2% (95% CI: 75.3; 87.1) for the Delta variant and 39.9% (95% CI: 26.4; 50.9) for the Omicron variant 14-30 days since vaccination. The VE waned over time and was 71.5% (95% CI: 54.7; 82.8), 49.8% (95% CI: 46.5; 52.8) and 4.7% (95% CI: 0.2; 8.9) >120 days since vaccination against the three variants, respectively. Higher estimates were observed after the third dose with VE estimates against infection of 86.0% (Delta, 95% CI: 83.3; 88.3) and 57.6% (Omicron, 95% CI: 55.8; 59.4) 14-30 days since vaccination. Among both age groups, VE against COVID-19-related hospitalization 14-30 days since vaccination with two or three doses was 94.8% or above for the Alpha and Delta variants, whereas among the youngest age group, VE estimates against the Omicron variant after two and three doses were 62.4% (95% CI: 46.3; 73.6) and 89.8% (95% CI: 87.9; 91.3), respectively. Conclusions Two vaccine doses provided high protection against SARS-CoV-2 infection and COVID-19-related hospitalization with the Alpha and Delta variants with protection waning over time. Two vaccine doses provided only limited protection against SARS-CoV-2 infection and COVID-19-related hospitalization with the Omicron variant. The third vaccine dose substantially increased the protection against Delta and Omicron.

observed relatively poor protection against infection but high VE against COVID-19-related hospitalization 67 after the third dose (13). However, previous studies have observed differences in the VE against SARS-CoV-2 68 infection with the Alpha, Delta and Omicron variants (4-10) and only few large-scale studies have compared 69 VE against all three variants (10). Despite variation between the variants, a previous study of the risk of 70 hospitalization and death associated with the Delta and Omicron variants observed a significant variation 71 with age (14) . The aim of this study was to estimate the effectiveness of two and three doses of the BNT162b2 performing SARS-CoV-2 testing to report electronically to MiBa (18) . Data on all positive laboratory-121 confirmed RT-PCR tests were extracted from MiBa (18). Information on rapid antigen tests was not included 122 in this study due to moderate sensitivity in asymptomatic patients compared with . (Fig 1) . 141

Separate models were fitted to estimate the VE for the two outcomes, SARS-CoV-2 infection and COVID-19-142 related hospitalization, and separately for the analysis of two and three doses. Event rates in the vaccinated 143 and unvaccinated exposure groups were compared using hazard ratios estimated in a Cox regression model 144 adjusted for age group, sex and geographical region, with calendar time as the underlying time scale to 145 control for temporal variations in the infection rate. VE was estimated as 1 minus the hazard ratio. 146

Vaccinated individuals were followed from the start of the study or the date of assumed protection after the 147 second or third vaccine dose, i.e. 14 days since vaccination. All individuals remained in follow-up until the 148 date of SARS-CoV-2 infection, booster vaccination (third dose), death, emigration or the end of study period, 149 whichever occurred first. Unvaccinated individuals remained in follow-up from the start of the study and until 150 the date of their first vaccination, death, emigration, SARS-CoV-2 infection or end of the study, whichever 151 occurred first. 152

Exposure status was categorized as either unvaccinated or vaccinated with the last dose administered in the 153 past 14-30 days, 31-60 days, 61-90 days, 91-120 days or >120 days. Time falling outside of these categories 154 was not included in the analysis. Data were analyzed using SAS version 9.4. 155 156 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 20, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 Ethical considerations 157 According to Danish law, ethical approval is not required for anonymized aggregated register-based studies. 158

The study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) 159 (Table S1 STROBE From a high proportion of all SARS-CoV-2 infections, with a median of 77% and 87% mid (<3,000 Ns) and high 172 quality (>150 Ns) genomes were obtained using whole-genome sequencing (WGS) during the Alpha and Delta 173 periods (Fig 1) as previously described (22). A lower median proportion was whole-genome sequenced during 174 the Omicron period (6%) (Fig 1) The elderly were the first to be vaccinated with both the second and the third dose (Fig 2) . Overall, the 182 vaccination coverage was high in the study population. The vaccination coverage on January 31, 2022 for two 183 doses was 85% among individuals aged 12-59 and 95% in those 60 years or older. The vaccination coverage 184 on January 31, 2022 for three doses was 64% and 90% among individuals aged 18-59 and 60 years or above, 185 respectively. Therefore, few individuals contributed with follow-up time as unvaccinated (reference group), 186 especially in the Delta and Omicron dominant periods (Fig 2) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 20, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 Among individuals aged 12-59 years, VE was 39.8% (95% CI: 38.4; 41.2) 14-30 days after vaccination. The 202 estimates decreased to 13.2% (95% CI: 12.5; 13.9) >120 days since vaccination. Similar estimates but with 203 wider CIs were observed among individuals aged 60 years or above (Table 2 and Fig 4) . 

Similar to the VE estimates after two doses, the VE after three doses was markedly lower against SARS-CoV-214 Table 3 . Adjusted vaccine effectiveness of three doses BNT162b2 mRNA or mRNA-1273 against SARS-CoV-2 226 infection with the Delta and Omicron variants by age groups (18-59 years and 60 years or above). 

A high VE against COVID-19-related hospitalization after two doses was observed for both the Alpha and 233

Delta variants. However, only limited protection was observed against COVID-19-related hospitalization 234 following infection with the Omicron variant (Table 4 ). Among individuals aged 60 years or above, VE against 235 COVID-19-related hospitalization with the Alpha and Delta variants after two doses was 96.4% (95% CI: 92.6; 236 98.3) and 100% (95% CI was not estimated as no hospital admissions were observed), respectively, 14-30 237 days since vaccination . The estimates decreased to 90.5% (Alpha, 95% CI: 67.0; 97.2) and 86.2% (Delta, 95% 238 CI: 84.2; 87.9) >120 days since vaccination. For the Delta variant, similar estimates were observed among 239 individuals aged 12-59 years (Table 4 ). Among individuals aged 12-59 years, VE against COVID-19-related 240 hospitalization following infection with the Omicron variant was 62.4% (95% CI: 46.3; 73.6) 14-30 days since 241 vaccination and 65.9% (95% CI: 62.0; 69.4) >120 days since vaccination. It was not possible to estimate two 242 dose VE against COVID-19-related hospitalization following infection with the Omicron variant among 243 individuals aged 60 years or above due to few cases, and since the majority of this group had already received 244 a third vaccine dose at this time (Table 4 and Fig 6) . 245 246 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 20, 2022. ; https://doi.org/10.1101/2022.04.20.22274061 doi: medRxiv preprint 7) . 269 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 20, 2022. ; https://doi.org/10.1101/2022.04.20.22274061 doi: medRxiv preprint Table 5 . Adjusted vaccine effectiveness of three doses BNT162b2 mRNA or mRNA-1273 against COVID-19-271 related hospitalization following infection with the Delta and Omicron variants by age groups (18- randomized studies including differences in behavior between unvaccinated and vaccinated individuals (24). 291 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 20, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 20, 2022. ; https://doi.org/10.1101/2022.04.20.22274061 doi: medRxiv preprint A test-negative case-control study from England observed limited protection against symptomatic disease 317 caused by the Omicron variant but reported that a third dose substantially increased protection (7). Their 318 observed VE estimates were higher and waned faster after the third dose than observed in our study. VE 319 against infection with the Omicron variant was 65.5% (95% CI: 63.9; 67.0) and 8.8% (95% CI: 7.0; 10.5) 2-4 320 weeks and ≥25 weeks since vaccination, respectively (7). After the third dose, VE against infection with the 321 Omicron variant was 67.2% (95% CI: 66.5; 67.8) and 45.7% (95% CI: 44.7; 46.7) 2-4 weeks and ≥10 weeks 322 since vaccination, respectively (7). An explanation to the lower estimates in our study may be that we 323 estimated VE against SARS-CoV-2 infection regardless of the symptom status. Denmark across the nationwide high-quality registries. The high sensitivity (97.1%) and specificity (99.98%) 330 observed for the RT-PCR test (21) minimize the risk of misclassification of the outcome. However, we were 331 not able to discriminate between asymptomatic and symptomatic infections. In addition, a previous study 332 has observed an inherent increased transmissibility of the Omicron sub-linage BA.2 (22) and we cannot rule 333 out that the VE differs between the Omicron sub-linages. BA.1 and BA.2 was the most frequent Omicron sub-334 linages in Denmark (25). BA.1 was most prevalent in the beginning of the included Omicron dominated 335 period. However, the prevalence of BA.2 has been increasing faster than BA.1 (25). Due to the short BA.1 336 dominated period, it was not possible to separate the VE analysis by BA.1 and BA.2. 337 COVID-19-related hospitalization was defined as an all-cause hospital admission lasting at least 12 hours and 338 occurring up to 14 days after or two days before a positive SARS-CoV-2 test. Some of the hospitalizations 339 included in our analysis will therefore have been caused by other factors than COVID-19. Assuming that the 340 rate of hospitalization due to other causes is similar in the vaccinated and unvaccinated populations, the lack 341 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 20, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 of specificity in the definition may have resulted in an underestimation of the VE. In addition, an effort was 342 made to ensure equal access to COVID-19 vaccination for all Danish residents. This was done through an 343 online booking system, special campaigns, offering vaccination in some workplaces, translating the 344 information about COVID-19 vaccination to several languages and arranging transport and pop-up 345 vaccination for those who were not able to reach the vaccination clinics on their own. However, the 346 populations initially prioritized for COVID-19 vaccination were the most vulnerable citizens and frontline 347 healthcare workers whereas the younger population was invited later. Therefore, it was not possible to 348 estimate VE of two and three doses for both age groups in all defined periods. No individuals had received 349 the third dose in the Alpha dominant period and the majority of the oldest age group had already received 350 third vaccine dose in the Omicron dominant period. 351

In general, non-randomized studies assessing COVID-19 VE can easily be flawed (24), which may also apply 352 to this study. Although we adjusted the Cox regression models for potential confounders (calendar time, age, 353 sex and geographical region), we cannot exclude time-varying factors such as test frequency and differences 354 in behavior or adherence to COVID-19 guidelines between vaccinated and unvaccinated, which may impact 355 infection exposure. Vaccinated individuals may be more frequently tested (and thus more likely to document 356 infection) if they are more health-conscious compared with unvaccinated individuals. The vaccinated 357

individuals may be less frequently tested, if the vaccination reduces the severity of the infection and thus 358 fewer infected people have symptoms. Furthermore, public health authorities may encourage more frequent 359 testing for the unvaccinated (24). Furthermore, vaccinated individuals may become more heavily exposed to 360 the virus after vaccination, if they feel liberated to engage in activities with more frequent and high-risk 361 exposure (24). This phenomenon of risk compensation decreases the benefit of vaccination (26). However, 362 some data have suggested little change in protective behavior early after vaccination (27) . 363

Overall, this study contributed with evidence of high vaccine protection against SARS-CoV-2 infection and 364

importantly against hospitalization with the Alpha and Delta variants after two doses of BNT162b2 mRNA or 365 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 20, 2022. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 20, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 .9 VE = vaccine effectiveness. CI = confidence intervals. VE estimates adjusted for five-year age groups, sex, and geographical region. Individuals were able to contribute follow-up time in more than one time category and (if vaccinated during the study period) to both the analysis of VE after two and three doses. * It was not possible within the model to estimate a 95% confidence interval for the estimated vaccine effectiveness against COVID-19-related hospitalization with the Delta variant 14-30 days after the second dose as no COVID-19-related hospitalization were observed. Table 5 . Adjusted vaccine effectiveness of three doses BNT162b2 mRNA or mRNA-1273 against COVID-19-related hospitalization following infection with the Delta and Omicron variants by age groups (18-59 years and ≥60 years)

Effectiveness of a third 380 after vaccination with the BNT162b2 or mRNA-1273 vaccine: A nationwide Danish cohort study

Comparative 414 analysis of the risks of hospitalisation and death associated with SARS-CoV-2 omicron (B.1.1.529) and delta 415 (B.1.617.2) variants in England: a cohort study

The Danish Civil Registration System as a tool in 417 epidemiology

The Danish vaccination register