key: cord-0262207-okpq1vbg
authors: PERALTA SANTOS, A.; Pinto Leite, P.; Casaca, P.; Fernandes, E.; Freire Rodrigues, E.; Moreno, J.; Ricoca, V.; Gomes, J. P.; Ferreira, R.; Isidro, J.; Pinto, M.; Borges, V.; Vieira, L.; Duarte, S.; Sousa, C.; Menezes, L.; Almeida, J. P.; Ferreira, B.; Matias, A.; Pelorito, A.; Freire, S.; Grilo, T.; Medeiro Borges, C.; Moutinho, V.; Kislaya, I.; Rodrigues, A. P.; Leite, A.; Nunes, B.
title: Omicron (BA.1) SARS-CoV-2 variant is associated with reduced risk of hospitalization and length of stay compared with Delta (B.1.617.2)
date: 2022-01-23
journal: nan
DOI: 10.1101/2022.01.20.22269406
sha: 1e48163b19894b39b62a7a468bfdec26d29f3b44
doc_id: 262207
cord_uid: okpq1vbg

Introduction Early reports showed that Omicron (BA.1) SARS-CoV-2 could be less severe. However, the magnitude of risk reduction of hospitalization and mortality of Omicron (BA.1) infections compared with Delta (B.1.617.2) is not yet clear. This study compares the risk of severe disease among patients infected with the Omicron (BA.1) variant with patients infected with Delta (B.1.617.2) variant in Portugal. Methods We conducted a cohort study in individuals diagnosed with SARS-CoV-2 infection between 1st and 29th December 2021. Cases were individuals with a positive PCR test notified to the national surveillance system. SARS-CoV-2 variants were classified first by whole genomic sequencing (WGS) and, if this information was unavailable, by detecting the S gene target failure. We considered a hospitalization for all the patients admitted within the 14 days after the SARS-CoV-2 infection; after that period, they were censored. The comparison of the risk of hospitalization between Omicron and Delta VOC was estimated using a Cox proportional hazards model. The mean length of stay was compared using linear regression, and the risk of death between Omicron and Delta patients was estimated with a penalized logistic regression. All models were adjusted for sex, age, previous infection, and vaccination status. Results We included 15 978 participants aged 16 or more years old, 9 397 infected by Delta (B.1.617.2) and 6 581 infected with Omicron (BA.1). Within the Delta (B.1.617.2) group, 148 (1.6%) were hospitalized, and 16 (0.2%) were with the Omicron (BA.1). A total of 26 deaths were reported, all in participants with Delta (B.1.617.2) infection. Adjusted HR for hospitalization for the Omicron (BA.1) variant compared with Delta (B.1.617.2) was 0.25 (95%CI 0.15 to 0.43). The length of stay in hospital for Omicron (BA.1) patients was significantly shorter than for Delta (confounding-adjusted difference -4.0 days (95%CI -7.2 to -0.8). The odds of death were 0.14 (95% CI 0.0011 to 1.12), representing a reduction in the risk of death of 86% when infected with Omicron (BA.1) compared with Delta (B.1.617.2). Conclusion Omicron was associated with a 75% risk reduction of hospitalization compared with Delta (B.1.617.2) and reduced length of hospital stay.

Early reports showed that Omicron (BA.1) SARS-CoV-2 could be less severe. However, the magnitude of risk reduction of hospitalization and mortality of Omicron (BA.1) infections compared with Delta (B.1.617. 2) is not yet clear. This study compares the risk of severe disease among patients infected with the Omicron (BA.1) variant with patients infected with Delta (B.1.617.2) variant in Portugal.

We conducted a cohort study in individuals diagnosed with SARS-CoV-2 infection between 1 st and 29 th December 2021. Cases were individuals with a positive PCR test notified to the national surveillance system. SARS-CoV-2 variants were classified first by whole genomic sequencing (WGS) and, if this information was unavailable, by detecting the S gene target failure. We considered a hospitalization for all the patients admitted within the 14 days after the SARS-CoV-2 infection; after that period, they were censored.

The comparison of the risk of hospitalization between Omicron and Delta VOC was estimated using a Cox proportional hazards model. The mean length of stay was compared using linear regression, and the risk of death between Omicron and Delta patients was estimated with a penalized logistic regression. All models were adjusted for sex, age, previous infection, and vaccination status.

We included 15 978 participants aged 16 

Omicron was associated with a 75% risk reduction of hospitalization compared with Delta (B.1.617.2) and reduced length of hospital stay.

3

The Omicron (BA.1) SARS-CoV-2 has been a variant of concern (VOC) since late November 2020 (1) and swept South Africa, Europe and other regions, becoming rapidly dominant. The Omicron (BA.1) has mutations associated with increased transmissibility, immune escape, and a higher risk of previous infection (2, 3) . It also harbors a deletion (spike Δ69-70) leading to "S gene target failure" (SGTF) by some real-time PCR assays (e.g.

TaqPath COVID-19, ThermoFisher, Waltham, MA, United States). This distinctive feature compared with the Delta (B.1.617.2) can be used as a marker (4), as it has been done previously with the Alpha variant (5).

The first real-world studies in England (6), Scotland (7) and Denmark (8) Early reports showed that Omicron could be less severe (9) . In animal models, the Omicron showed less severe symptoms (10) and less viral load in the lungs; in vitro T cell response was also preserved with the Omicron variant (11) . In real-world human studies, reduced severity was confirmed in studies in Scotland (7), England (6, 12) , Canada(13) and the US (California) (14) ; while studies consistently showed a risk reduction the degree of risk reduction (from 80% to 45%) of hospitalization varied considerably. Furthermore, length of stay and risk of death resulting from Omicron cases have not been described.

We aim to study the risk of hospitalization and death from Omicron (BA.1) infections compared with Delta is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted January 23, 2022. ; https://doi.org/10.1101/2022.01.20.22269406 doi: medRxiv preprint

We designed a cohort study to assess the risk of hospitalization and death of the individuals infected with Omicron (BA.1) and Delta (B.1.617.2) variants. The study population was individuals eligible for vaccination (>16 years old) diagnosed with SARS-CoV-2 infection/COVID-19 by nasopharyngeal swab tested with RT-PCR from December 1 st and 29 th notified through the laboratory service of the existing surveillance system (SINAVE), in Portugal mainland. We included individuals with samples classified either by whole-genome sequencing or Spike Gene Target Failure (SGTF). Both symptomatic and asymptomatic individuals were included.

We used two different methods to identify which cases were Delta (B.1.617.2) or Omicron (BA.1) variants infections. We used information from a nationwide network group of laboratories (UNILABS, ABC and CVP) that 

We defined COVID-19 hospitalization as any admission to a public hospital in Portugal mainland within the 14 days following a positive sample SARS-CoV-2 collection, notified in the surveillance system. Admission data was obtained through the Central Hospital Morbidity Database and Integrated Hospital Information System. The Central Hospital Morbidity Database gathers data from all public hospitals in Portugal mainland, which account for most hospitals in Portugal and most hospitals admitting patients with COVID-19. Individuals with a SARS-CoV-2 diagnosis (by sample collection day) or notification after the admission date were excluded.

A COVID-19 death was defined as any record of death on the national Death Certificate Information System (SICO) with COVID-19 as the primary cause of death (ICD-10 code U.071) according to the WHO classification (16) . SICO platform allows the issuance of a death certificate for each person who dies in Portugal (17) . A COVID-19 death was defined as any record of death on the national Death Certificate Information System (SICO) with COVID-19 as the primary cause of death (ICD-10 code U.071) according to the WHO classification (16) . SICO platform allows the issuance of a death certificate for each person who dies in Portugal (17) .

We obtained COVID-19 vaccination status through the electronic national vaccination register (VACINAS). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted January 23, 2022. ; https://doi.org/10.1101/2022.01.20.22269406 doi: medRxiv preprint second dose vaccine uptake or first dose of Ad26.COV2.S or up to 14 days after the booster dose uptake; and (iv) complete vaccination plus booster: 14 or more days following booster dose vaccine uptake. Information about age, sex and date of diagnosis was routinely collected on surveillance system SINAVE and was extracted from it. Previous infection was defined as a PCR or rapid antigen SARS-CoV-2 notification for the same individual more than 90 days apart.

We used data from different sources: databases from the three laboratories with information about unique identifiers (national health service number), sample collection date, Ct value and SGTF classification; this was combined (full join) with data from sequencing was provided by the National Institute of Health Dr Ricardo Jorge.

Duplicates were excluded using the unique identifier. Afterward, we linked (left join) the former database with SINAVE, the national epidemiological and laboratory surveillance platform for SARS-CoV-2 infections, providing age and sex information. SICO, the electronic death certificate platform, provided information about COVID-19 basic cause of death and date of death. VACINAS, the electronic vaccination platform, provided information about vaccination status. SONHO, the hospital admission platform of the National Health System, provided data about hospitalization. Data from the laboratories were extracted on December 29 and followed up (hospitalization and death) until January 11, 2022.

We compared characteristics of SARS-CoV-2 cases using central tendency and dispersion measures for continuous variables and absolute and relative frequencies for categorical variables. To compare the mean, we used the T-test and the Chi-squared test for relative frequencies.

The confounding-adjusted hazard ratio of hospitalization, with 95% confidence intervals (CI), was calculated using a proportional hazards Cox model, adjusted for sex, age, previous infection and vaccination status (equation 1).

Where ℎ ( | ) is the hazard at the time t of the outcome hospitalization conditional on covariates X (Omicron, age group, sex, previous infection and vaccine status), ℎ 0 ( ) is the baseline hazard for the outcome, We tested the proportionality of risk assumptions using visual inspection of scaled Schoenfeld residuals (18) . We used the R survival package to fit the proportional hazards Cox model (19) . Due to the nature of our analysis assumed a type I right censoring, thus using a fixed end of follow-up date defined as 14 days after the sample collection date for non-hospitalized or the day of hospitalization. Analyses were based on a complete case analysis.

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted January 23, 2022. We calculated the length of stay as the difference between discharge and admission dates. We used a linear regression model to calculate the confounding-adjusted difference in length of stay between Omicron (BA.1) and Delta (B.1.617.2), adjusted for sex, age, previous infection, and vaccination status. We also performed a stratified analysis by vaccination status in the adjusted model.

The risk of death was assessed using logistic regression adjusted for sex, age, reinfection, and vaccination status.

To account for the fact that we did not observe deaths in Omicron study participants, we used a penalized maximum likelihood logistic regression (20) (supplementary material, equation S1).

All tests were two-sided, and a P-value < .05 was considered statistically significant. All statistical analyses were performed on R® using R Studio version 1.4 (21) .

As an additional analysis to account for potential biases regarding the week of diagnosis, we performed a matching based on the week and age group of SARS-CoV-2 infection diagnosis using variable ratio matching, at a ratio of 1 to k, where k is the maximum number of B.1.617.2 cases available on the respective week. We used a nearest-neighbor algorithm with the "Matchit"(22) CRAN package.

The is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted January 23, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted January 23, 2022. ; https://doi.org/10.1101/2022.01.20.22269406 doi: medRxiv preprint (B.1.617.2) was 0.34 (95%CI 0.13 to 0.87), see supplement materials, figures S2 and S2. We found no significant violation of the proportionality of hazards assumptions (see supplementary material, figure S4 ). However, after adjusting for age, sex, vaccination, and previous infection, the odds ratio of death was 0.14 (95% CI 0.0011 to 1.12), see supplement materials, table S1. That represents a reduction in the risk of death of 86% when infected with Omicron (BA.1) compared with Delta (B.1.617.2), although with a wide confidence interval. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted January 23, 2022. ; https://doi.org/10.1101/2022.01.20.22269406 doi: medRxiv preprint

We report a 75% risk reduction of hospitalization for participants infected with Omicron (BA.1) compared with Delta (B.1.617.2). The length of stay for hospitalized study Omicron (BA.1) participants was significantly shorter than for Delta (B.1.617.2) participants, mean difference -4.0 days (95%CI -7.2 to -0.8). However, there were no differences in mean Ct values between Omicron and Delta hospitalized participants. The risk of death was also lower; we show a reduction in the odds of death of 86% when infected with Omicron (BA.1) compared with Delta (B.1.617.2), OR 0.14 (95% CI 0.0011 to 1.12); however, the effect was highly uncertain.

Early assessment of the risk of hospitalization in South Africa founded an odds ratio (aOR) 0.2, 95% confidence interval (CI) 0.1-0.3) (9), adjusted for age, sex, reinfection, vaccination and comorbidities. In the Scottish study, the adjusted observed/expected ratio was 0.32 (95% CI 0. 19 (14) . Finally, the English study (12) reports an adjusted HR for hospitalization of 0.55 (0.51-0.59) for Omicron (BA.1) compared with Delta (B.1.617.2) ( "All Pillar 1 and Pillar 2 cases, ECDS Admitted" ). All these studies have different designs, but they were consistent with a risk reduction for hospitalized individuals infected with Omicron compared with Delta. Our study has a confounding-adjusted HR lower than most studies: HR 0.25 (95%CI 0.15 to 0.43); this could be due to residual confounding, the study participants with Omicron (BA.1) were younger than those with Delta (B.1.617.2), and we do not have data on comorbidities to adjust for. However, even after matching for the week of diagnosis and age group, the results are consistent with a risk reduction of a similar magnitude. In our study, the reduction in the risk of hospitalization for Omicron (BA.1) participants is maintained for not vaccinated individuals and vaccinated individuals, accounting for confounders. This result is consistent with the study from England (12) .

It is also important to highlight that those infected with Omicron (BA.1) had a shorter length of hospitalization than Delta(B.1.617.2), mean difference of -4.0 days (95%CI -7.2 to -0.8). The study in California also reported a median duration of hospital stay of 3.4 (2.8-4.1) days shorter for hospitalized cases with Omicron variant infections than hospitalized patients with Delta variant infections (14) . This fact has important implications in the management of hospital services and will reduce the impact of omicron waves on health systems, and to inform statistical models.

Our results also indicate a reduction in the risk of death of participants infected with Omicron, although with high uncertainty. Our study might be underpowered to study this risk reduction, but a reduction in the risk of death is plausible given the animal models showing less severe inflammation in the lungs (10) , the preserved T cell function in vitro (11) and a reduced risk in hospitalizations in humans. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted January 23, 2022. ; https://doi.org/10.1101/2022.01.20.22269406 doi: medRxiv preprint Our study has several strengths,1) we used an established method to identify Delta and Omicron variants, the SGTF and genome sequencing, 2) we also provided data from record linkage of study participants that were admitted to the hospital within 14 days of the diagnosis of the infection, that was used before (7), 3) we provide data from a short period (4 weeks), minimizing the probability of changes in the testing policies, clinical guidelines; 4) we adjusted for the main confounders associated with the probability of severe disease, especially vaccination and previous infection that could differ across time and variant.

Our study also had limitations; we did not have access to previous comorbidities, which could be an unaccounted confounder. Also, the study participants with Omicron and Delta had baseline differences that could influence the risk of hospitalization; although this was accounted for in the adjusted models, residual confounding could still be present. The adjusted risk increased the HR by 66%, hence the protection effect of omicron could be partially explained by age, sex, previous infection, and vaccination status.

Finally, we did not account for correcting hazard ratio estimates in unvaccinated individuals for under ascertainment of past infection status; that method was shown in the English study to decrease the risk reduction. Hence, we could be overestimating the risk reduction.

Omicron was associated with a 75% risk reduction of hospitalization compared with Delta (B.1.617.2); the risk reduction is maintained for unvaccinated and vaccinated participants. Additionally, Omicron was associated with a shorter length of hospital stay, mean difference -4.0 days (95%CI -7.2 to -0.8), and reduced risk of death, however the latter with high uncertainty. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted January 23, 2022. ; https://doi.org/10.1101/2022.01.20.22269406 doi: medRxiv preprint

7%) 3,215 (48.9%) Vaccination

Rapid epidemic expansion of the SARS-CoV-2 Omicron variant in southern Africa. medRxiv

Omicron SARS-CoV-2 variant: a new chapter in the COVID-19 pandemic. The Lancet

Increased risk of SARS-CoV-2 reinfection associated with emergence of the Omicron variant in South Africa

Track Omicron's spread with molecular data

Tracking SARS-CoV-2 VOC 202012/01 (lineage B. 1.1. 7) dissemination in Portugal: insights from nationwide RT-PCR Spike gene drop out data

Effectiveness of COVID-19 vaccines against the Omicron (B. 1.1. 529) variant of concern

Severity of omicron variant of concern and vaccine effectiveness against symptomatic disease: national cohort with nested test negative design study in Scotland

Epidemiological characterisation of the first 785 SARS-CoV-2 Omicron variant cases in Denmark

Early assessment of the clinical severity of the SARS-CoV-2 Omicron variant in South Africa. medRxiv

Variant leads to less severe disease than Pango B and Delta variants strains in a mouse model of severe COVID-19. bioRxiv

Omicron variant is preserved in most but not all prior infected and vaccinated individuals

Hospitalisation risk for Omicron cases in England

Early estimates of SARS-CoV-2 Omicron variant severity based on a matched cohort study

Clinical outcomes among patients infected with Omicron (B.1.1.529) SARS-CoV-2 variant in southern California

Changing composition of SARS-CoV-2 lineages and rise of Delta variant in England

Medical certification of cause of death for COVID-19

Tecnologias de informação e comunicação no registro de óbitos em Portugal

Proportional hazards tests and diagnostics based on weighted residuals

R survival package

RStudio: a platform-independent IDE for R and Sweave

Package 'MatchIt'. Version