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)

Study design and population

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.

Sample collection and variant classification

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 performs RT-PCR tests for SARS-CoV-2 using Thermofisher TaqPath assay, targeting three regions of the SARS-CoV-2 genome: ORF1ab, N and S genes. Their samples are classified as BA.1 (no amplification) or B.1.617.2 (amplification), according to SGTF status. Only samples having both N and ORF1a positive signals and Ct values <=30 were considered. Secondly, we use Whole-Genome Sequencing (WGS) data provided by the National Health Institute (INSA) that routinely performs sequencing on random samples notified to SINAVE. Variants were classified first by whole genomic sequencing (WGS) and, if this information was unavailable, by detecting the S gene target failure (SGTF) (16).

Outcomes: hospitalization and death

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 (17). SICO platform allows the issuance of a death certificate for each person who dies in Portugal (18). 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 (17). SICO platform allows the issuance of a death certificate for each person who dies in Portugal (18).

Vaccination status and demographic covariates

We obtained COVID-19 vaccination status through the electronic national vaccination register (VACINAS). Vaccination exposure was indexed at the date of COVID-19 diagnosis and classified as: (i) unvaccinated: no register of vaccine administration prior to diagnosis; (ii) incomplete vaccination: SARS-CoV-2 infection/COVID-19 diagnosis anytime between the uptake of the first dose of a COVID-19 vaccine (ChAdOx1, BNT162b2 or mRNA-1273) and up to 14 days after the second dose uptake; (iii) complete vaccination: 14 or more days following 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.

Data source, record-linkage, and follow-up

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.

Statistical Analysis

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_(t|X) is the hazard at the time t of the outcome hospitalization conditional on covariates X (Omicron, age group, sex, previous infection and vaccine status), h₀(t) is the baseline hazard for the outcome, exp (β_{1 Omicron}) the hazard ratio for the outcome being infected with Omicron variant compared with Delta, exp (β_{2 age group}) are the hazard ratios for the outcome being 80 years or older, 65 years old to 80 compared with being 16 to 65 years old, exp (β_{3 sex}) the hazard ratio for the outcome being a female compared with being male, exp (β_{4 previous infection}) is the hazard ratio for the outcome of primary infection compared with having a previous infection, exp (β_{5 vaccination status}) is the hazard ratio for the outcome having incomplete, complete, or complete plus booster compared with unvaccinated, ε is the error term.

We tested the proportionality of risk assumptions using visual inspection of scaled Schoenfeld residuals (19). We used the R survival package to fit the proportional hazards Cox model (20). 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.

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, previous infection, 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 (21) (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 (22).

Sensitivity Analysis

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”(23) CRAN package.

Ethical Statement

The genomic surveillance of SARS-CoV-2 in Portugal is regulated by the Assistant Secretary of State and Health Executive Order (Despacho n.° 331/2021 of January 11, 2021). The research on genomic epidemiology of SARS-COV-2 received the clearance of the Ethics Committee of INSA on March 30, 2021.

Study participants characteristics

From 1^st to 29^th December 2021, we identified through WGS or SGTF with successful record linkage 22 960 SARS-CoV-2 cases in Portugal considering the inclusion criteria. 6 982 (30%) were excluded due to missing information about previous infection. Hence the study included 15 978 participants, 9 397 infected with Delta (B.1.617.2) and 6 581 with Omicron (BA.1) variant.

Study participants with Omicron (BA.1) variant were on average younger, 37.1(SD=14.8) for Omicron cases and 43.4 (SD=15.8) years for Delta (B.1.617.2). Participants with Omicron (BA.1) were more frequently female (51.1% versus 49.3%) and with complete plus booster vaccination status (4.5% versus 2.1%). Also, previous infection in participants with Omicron (BA.1) was four times higher than Delta (B.1.617.2), 6.8% versus 1.6%. Patient characteristics at baseline were not balanced between the groups for some potential confounders (Table 1 and Supplementary Materials figure S1).

Outcomes

We observed 164 hospitalizations and 26 deaths between 1^st and 29^th December. There were 148 (1.6%) hospitalizations in participants with Delta variant and 16 (0.2%) with Omicron (BA.1). All the 26 deaths were reported in participants with Delta (B.1.617.2) variant.

For hospitalized participants, the mean time from sample collection to hospitalization was 3.4 days (SD=3.7) for Delta (B.1.617.2) participants and 5.0 days (SD=3.9) for Omicron (BA.1). For the outcome of death, the mean time from diagnosis to death was for Delta (B.1.617.2) participants and 9.1 days (SD=5.3).

We observed no Intensive Care Unit (ICU) admission for Omicron (BA.1) cases; 17 patients were (11.5%) admitted to ICU for Delta (B.1.617.2) cases. The mean length of stay was also higher for patients with Delta (B.1.617.2), 8.6 days (SD=5.9) and 3.7 days (SD=2.9) for Omicron (BA.1), see table 2.

Hospitalization

The crude hazard ratio of hospitalization was 0.15 (95%CI 0.091 to 0.25) for Omicron compared with Delta. The age, sex, previous infection and, vaccination 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), figure 1. Hence the Omicron (BA.1) variant cases had a 75% risk reduction of hospitalization compared with Delta (B.1.617.2) (95%CI 57% to 85%). We report no statistically significant difference in the Ct values between Omicron (BA.1) and Delta (B.1.617.2) in hospitalized participants. The confounding-adjusted difference in length of stay of -4.0 days (95%CI -7.2 to -0.8), between Delta and Omicron is not explained by higher viral loads, measured as Ct Values at the time of diagnosis. In the stratified analysis, in the unvaccinated strata, the adjusted HR for hospitalization for the Omicron (BA.1) variant compared with B.1.617.2 was 0.21 (95%CI 0.08 to 0.51). In the vaccinated strata (complete and complete+booster), the adjusted HR for hospitalization for the Omicron (BA.1) variant compared with Delta (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).

• Download figure
• Open in new tab

Figure 1.

The hazard ratio for hospitalization in the fully adjusted cox model

COVID-19 death

All the 26 study participants who died were Delta (B.1.617.2) cases. The average age at the time of death was 83.1 (SD=8.9). In the group of participants that died, 20 (77%) were hospitalized, and no one had previous infection. Regarding vaccination status, 8 (31%) were not vaccinated, 16 (62%) had complete vaccination, and two complete boosters (8%). In the unadjusted model, the odds ratio of death was 0.026 (95% CI 0.00021 to 0.191), comparing patients infected with Omicron (BA.1) cases with patients infected with Delta (B.1.617.2). 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.

Sensitivity Analysis

After we matched for the week and age group of diagnosis, the study sample was 10 941 individuals, and in the adjusted HR for hospitalization for the Omicron (BA.1) variant compared with B.1.617.2 was 0.24 (95%CI 0.11 to 0.50) times the risk of Delta (B.1.617.2).