key: cord-0274305-u81wkldr
authors: Kuelper-Schiek, W.; Piechotta, V.; Pilic, A.; Batke, M.; Dreveton, L.-S.; Geurts, B.; Koch, J.; Koeppe, S.; Treskova, M.; Vygen-Bonnet, S.; Waize, M.; Wichmann, O.; Harder, T.
title: Facing the Omicron variant - How well do vaccines protect against mild and severe COVID-19? Third interim analysis of a living systematic review
date: 2022-05-27
journal: nan
DOI: 10.1101/2022.05.25.22275516
sha: 66a4058ee8883e5881341a931ecb591fa0df4bdc
doc_id: 274305
cord_uid: u81wkldr

Background: The SARS-CoV-2 Omicron variant is currently the dominant variant globally. This 3rd interim analysis of a living systematic review summarizes evidence on COVID-19 vaccine effectiveness (VE) and duration of protection against Omicron. Methods: We systematically searched the COVID-19 literature for controlled studies evaluating the effectiveness of COVID-19 vaccines approved in the European Union up to 14/01/2022, complemented by hand-searches of websites and metasearch engines up to 11/02/2022. We considered the following comparisons: full primary immunization vs. no vaccination; booster immunization vs. no vaccination; booster vs. primary immunization. VE against any confirmed SARS-CoV-2 infection, symptomatic, and severe COVID-19 (i.e. COVID-19-related hospitalization, ICU-admission, or death) was indicated providing estimate ranges. Meta-analysis was not performed due to high study heterogeneity. Risk of bias was assessed with ROBINS-I, certainty of evidence evaluated using GRADE. Results: We identified 26 studies, including 430 to 2.2 million participants. VE against any confirmed SARS-CoV-2 infection compared to no vaccination ranged between 0-62% after full primary immunization, and between 34-66% after a booster dose. VE-range for booster vs. primary immunization was 34-54.6%. Against symptomatic COVID-19, VE ranged between 6-76% after full primary immunization, and between 19-73.9% after booster immunization, if compared to no vaccination. When comparing booster vs. primary immunization VE ranged between 56-69%. VE against severe COVID-19 compared to no vaccination ranged between 3-84% after full primary immunization, and between 12-100% after a booster dose. One study compared booster vs. primary immunization (VE 100%, 95% CI 71.4-100). VE was characterized by a moderate to strong decline within three to six months for SARS-CoV-2 infections and symptomatic COVID-19. Against severe COVID-19 protection remained robust at least for up to six months. Waning immunity was more profound after primary than booster immunization. Risk of bias was moderate to critical across studies and outcomes. GRADE-certainty was very low for all outcomes. Author's conclusions: Under the Omicron variant, effectiveness of EU-licensed COVID-19 vaccines in preventing any SARS-CoV-2 infection or mild disease is low and only short-lasting after primary immunization, but can be improved by booster vaccination. VE against severe COVID-19 remains high and is long-lasting, especially after receiving the booster vaccination.

Introduction 50

The Omicron variant of SARS-CoV-2 (Phylogenetic Assignment of Named Global Outbreak 51 (Pango) lineage designation B.1.1.529) was first detected in South-Africa in November 2021. 52

Since then, the variant spread rapidly across countries and has largely replaced all other variants 53 globally (>99,8% of all sequences submitted to the Global Initiative on Sharing All Influenza 54

Data (GISAID) were Omicron in week 5 of 2022) (1). Evidence suggests that the Omicron 55 variant has a growth rate advantage compared to the previously dominant Delta variant, leading 56 to its overtake as the dominant variant globally, while the Delta variant now only represents 57 0.1% of collected samples (1). 58

High rates of asymptomatic infection and symptomatic COVID-19 disease among people 59

previously infected with SARS-CoV-2 or fully vaccinated with a COVID-19 vaccine raise 60 concerns that the currently available vaccines are less or no-longer effective against the 61

Omicron variant. To summarize the existing evidence on the effectiveness and the duration of 62 protection confered by Methods 69

The LSR follows the Preferred Reporting Items for Systematic Review and Meta-Analysis 71

(PRISMA) guideline (supplement material 1, part 1), and was registered in the Prospective 72

Register of Systematic Reviews (PROSPERO registration ID: CRD42020208935; updated on: 73 09 March 2022). All amendments since initial registration are available online. The methods 74

have been previously described in detail (2). In brief, we included studies of any design that 75 had a comparison group and investigated vaccine effectiveness (VE) against SARS-CoV-2 76 infection of any severity of COVID-19 vaccines approved by the European Medicine Agency 77 (EMA) in people ≥12 years of age (see supplement material 1, part 2 for complete PICO 78 question). For this 3 rd update of the LSR, we only considered studies that reported on outcomes 79

that are due to the SARS-CoV-2 Omicron variant or that occurred during a dominant circulation 80 of the Omicron variant. There were no restrictions with regard to publication language or status. 81

We searched the COVID-19 literature database constructed by the RKI library (see (2) after 2-4 and 5-9 weeks after vaccination, the estimate of 5-9 weeks was included in the depicted 106 effect range). Only for the time stratum "≈14 days", VE estimates closest to 14 days were 107

included. The latter stratum included furthermore VE data that was only assessed at "≥14 days" 108 after primary vaccination, "≥7 days" post booster vaccination or when the time point of 109 assessment after primary immunization was not reported at all. In the studies the SARS-CoV-2 variant causing the investigated outcome was identified either 135 from time periods with known dominant circulation of the Omicron (resp. Delta) variant in the 136 corresponding study location, from whole-genome sequencing (WGS), or through S-gene target 137 failure (SGTF) in PCR assays. Studies using the latter method subdivided the SARS-CoV-2 138 positive samples by the detection or non-detection of the S-gene in a three gene PCR assay. As 139 the SGTF is characteristic for the Omicron variant, but rare in the Delta variant, studies used 140 . CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 the detection of the S-gene as a proxy for the Delta variant, and the "non-detection" as a proxy 141 for the Omicron variant. As most studies report overall VE estimates without further 142 differentiating between WGS-or SGTF-variant assessment the data was synthesized only into 143 "dominance" and "sequencing/SGTF". 144 Studies were conducted in ten different countries and mainly used national electronic registries 145 or claims data from the general population for laboratory, immunization and patient Prevention of infection with SARS-CoV-2 Omicron variant (without differentiation 160 between asymptomatic or symptomatic cases) 161

Twelve studies (including between 1,220 and 2.2 million participants) reported the 162 effectiveness of COVID-19 vaccines in preventing infection of any type with SARS-CoV-2 163

Omicron variant (without differentiation between asymptomatic or symptomatic cases). After 164 full primary immunization VE across all studies ranged between 0 and 62% at "≈14 days" post 165

vaccination, compared to no vaccination. For the time-periods ">14 days up to 3 months", ">3 166 months up to 6 months" and ">6 months" after vaccination VE-ranges, assessed across all 167

reported VE estimates, were 4.2 to 42.8%, 0 to 23% and 0 to 8.6%, respectively (table 1, figure  168 2). The data from the two studies reporting VE for at least two time points show a decline of 169 VE between >14 days to up to 6 months by 16 to 34% after vaccination with mRNA-based 170 vaccines (figure 2) (7, 14). As there was no VE data over time identified for vector-based 171 vaccines, unspecified vaccine or heterologous schedules respective VE decline could not be 172 assessed. 173

After booster vaccination, VE estimates at "≈14 days" post vaccination across all studies ranged 174 between 34 and 76%, compared to no vaccination, and between 14 and 53%, compared to full 175 primary immunization (12, 21). Follow-up data were insufficient to evaluate waning of 176 immunity after booster vaccination. 177

The study investigating VE of four vs. three doses of mRNA-based vaccines reported VE 178 estimate of 47% (95% CI 44-47) against Omicron infection at 12 or more days after the fourth 179 dose (29). VE ranges against infection with the Delta variant for the different time points are 180 provided in supplement material 1, part 5. 181

Risk of bias was serious to critical for all assessed studies (see figure 3) . Key concern was no 182 or insufficient adjustment for confounders. 183

Prevention of symptomatic COVID-19 184 . CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10. 1101 /2022 Seven studies, including between 430 and 2.2 million participants, estimated the effectiveness 185 of COVID-19 vaccines in preventing symptomatic infection with the SARS-CoV-2 Omicron 186 variant (supplement material 1, part 4). Compared to unvaccinated individuals, VE at ≈14 187 days after vaccination summarized from all studies reporting on this time point ranged between 188 6 and 76%. For the time-period ">14 days up to 3 months", ">3 months up to 6 months" and 189 ">6 months" VE ranges were 12-54%, 6.1-20.8%, and 0-13%, respectively. VE estimates from 190 studies that report data for at least two time points suggest a decrease in protective immunity 191 against symptomatic COVID-19 between 45-63% for mRNA-based vaccine recipients, and of 192 50% for vector-based vaccine recipients over the time period >14 days up to >6 months after 193 vaccination (figure 4, (8, 25)). 194

After booster vaccination, VE against symptomatic infection ranged between 19 and 73.9% at 195 "≈14 days" post vaccination, compared to no vaccination, and between 50 to 68%, compared 196 to primary vaccination (6, 9). For the time point ">14 days up to 3 months" estimates were only 197

reported for the comparison with no vaccination and ranged between 43.7 and 65.4%. No study 198

reported VE for later observation periods. The two studies that provide data for both these time 199

periods show a 9-28% decrease in VE after booster vaccination with a mRNA-based vaccine 200

and between 32-34% after heterologous vaccination schemes (figure 4 (8, 25)). However, 95% 201

confidence intervals between observation periods were partially overlapping. 202

Risk of bias was serious to critical for all but one of the assessed studies (see figure 5 ). The 203

remaining study was rated to have a moderate risk. All factors considered relevant for 204 confounding were taken into account, however residual confounding could not be ruled out. 205

Prevention of severe COVID-19 (hospitalization, ICU-admission, or death) 206 VE against severe COVID-19 was assessed in seventeen studies including 1,220 to 2.2 million 207 participants (supplement material 1, part 4). VE estimates in primary vaccinated compared 208

to unvaccinated individuals ranged from 3 to 84% at "≈14 days" post vaccination, between 21 209 and 95% at ">14 days up to 3 months", between 0 and 91% at ">3 months up to 6 months" and 210 between 32.7 and 86% at ">6 months" after vaccination. Studies reporting VE for at least two 211

time points indicate a decline by up to 40% for mRNA-based vaccines, and 15-67% for vector-212 based vaccines between 14 days and ≥6 months after vaccination (see figure 6 ) (8, 20). 213

However, 95% confidence intervals were wide and overlapping across time points. One study 214 reported no difference in the first and last time point estimate (17) The study assessing the effect of a fourth vs. a third mRNA-based vaccine dose reported VE 227 against severe disease of 75% (95% CI 57-86) at ≥12 days after additional booster vaccination 228 (29). Data to assess duration of protection was not available. 229 . CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10. 1101 /2022 In all but one studies risk of bias was serious to critical (see figure 7) . Though evidence is uncertain about the exact level of protection against all investigated 255 outcomes, both after primary and after a booster vaccination, data suggest that VE was higher 256 after booster when compared to primary immunization. Results suggest a rapid decline of 257 vaccine-induced protection after completion of the primary vaccination series. The effect is 258 profound for infections of any type, but less pronounced for severe disease. This is in line with 259 the findings of a recently published meta-regression analysis on pre-Omicron variants (32). VE 260 could be restored to high levels of protection by the booster dose, though first follow-up data 261 do also suggest a waning effect (8, 19, 25 immunocompromised people a three-dose primary vaccination schedule is recommended by 286 the WHO Strategic Advisory Group of Experts on Immunization to improve immune response. 287

As this recommendation is implemented in many countries 1 , it is possible that VE estimates for 288 a booster schedule include those that received a 3 rd doses within an "optimized" schedule for 289 immunocompromised people. 290 291 This 3 rd update provides a comprehensive overview of the currently available evidence on VE 292

against infection with the SARS-CoV-2 Omicron variant. We provide an in-depth analysis of 293 the VE estimates extracted from the included studies that were identified following a pre-294 registered protocol. We conducted thorough risk of bias assessment of the studies, and 295 evaluated the certainty of evidence using the GRADE approach. 296

Due to the highly dynamic publication landscape in this field, additional studies have been 297 published since our last search that are not captured by this analysis. In fact, we are aware of at 298 least two studies reporting on adolescents, which were published after our data cut (35, 36). As 299 in adults, lower VE against infections and severe disease were observed for the Omicron 300 variant, when compared to the Delta variant. However, data on duration of protection are 301

contradicting. While one study reported a decrease in VE against hospitalizations by 6-19% at 302 more than 5 months after primary immunization (35). The second study did not identify a 303 decrease in VE against any infections but was based only on few events (36). In addition to 304 studies not included as they were published after our final search date, we noticed that some 305 studies updated data on preprint servers after initial publication including data on longer follow-306 up periods. We therefore cannot exclude that authors revised the available pre-print versions 307

including additional data after we completed data extraction for this update. Further, we cannot 308 exclude a risk of potential bias as real-life observation are based on retrospective analyses which 309

are not systematically registered. Thus, intent and possibility of publication might depend on 310 observed results. However, due to the increase of publications on pre-print servers, the risk of 311 publication bias is probably low. 312

Additional data limitations stem from the fact that VE estimates for more than one vaccine were 313

reported. 314 315 1 An additional dose is recommended for immunocompromised people in all eight countries where included studies that reported VE estimates on booster immunization were conducted (Canada, Denmark, Netherlands, Norway, Qatar, South Africa, United Kingdom, United States of America).

. CC-BY-NC-ND 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 May 27, 2022. . CC-BY-NC-ND 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 May 27, 2022. . CC-BY-NC-ND 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 May 27, 2022. . CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10.1101/2022.05.25.22275516 doi: medRxiv preprint CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10.1101/2022.05.25.22275516 doi: medRxiv preprint CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10.1101/2022.05.25.22275516 doi: medRxiv preprint Figure 6 . Forest plots of VE estimates against severe ICU 517 admission or death) due to SARS-CoV-2 infection of the Omicron variant after full primary 518

immunization and booster dose, as reported in the study for the defined time strata after 519 immunization. For booster immunization the COVID-19 vaccine used for primary 520 immunization is indicated. 521 522 523 . CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10.1101/2022.05.25.22275516 doi: medRxiv preprint Figure 7 . Risk of Bias assessment for severe ICU admission 524 or death) due to SARS-CoV-2 infection of the Omicron variant 525 526 . CC-BY-NC-ND 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 May 27, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 

Efficacy 355 and effectiveness of COVID-19 vaccines against SARS-CoV-2 infection: interim results of a 356 living systematic review

358 ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions

Severe COVID-19 (hospitalization, ICU admission, or death) Full primary immunization mRNA-based (any) e 3% (-114-56) to

22) -8% (-213-62) to 55.8% (34.1-70.3); h

Booster immunization mRNA-based (any) e 12% (-45-46) to 100% (71

Vector-based (any) f 78% (76-80) to 84%