Abstract
The Omicron (B.1.1.529) variant of SARS-CoV-2 rapidly achieved global dissemination following its emergence in southern Africa in November, 2021.1,2 Epidemiologic surveillance in multiple settings has revealed changes in COVID-19 case-to-hospitalization and case-to-mortality ratios following Omicron variant emergence.3–7 However, interpretation of these changes is difficult due to potential differences in immunity among cases with Omicron or Delta (B.1.617.2) variant SARS-CoV-2 infections, as well as possible longer-term changes in testing and other healthcare practices.8 Here we report clinical outcomes and characteristics of 228,594 cases with Omicron variant infections and 42,515 cases with Delta variant infections within the Kaiser Permanente Southern California healthcare system, who were followed longitudinally from an initial outpatient test during a period of mixed circulation of the Omicron and Delta variants. Adjusted hazard ratios of progression to symptomatic hospital admission, intensive care unit admission, mechanical ventilation, and death were 0.62 (95% confidence interval: 0.54-0.72), 0.59 (0.51-0.69), 0.45 (0.26-0.78), and 0.36 (0.18-0.74), and 0.21 (0.10-0.42) respectively, for cases with Omicron versus Delta variant infections. These differences in risk of severe clinical endpoints were most pronounced among individuals who had not received any vaccination against COVID-19, owing to partial vaccine protection against severe disease associated with both Omicron and Delta variant infections. Among hospitalized symptomatic cases, average duration of hospital stay was 70.7% (64.4-76.0%) shorter among patients with Omicron as compared to Delta variant infections. These findings demonstrate that the Omicron SARS-CoV-2 variant is associated with substantially severity of illness in comparison to the Delta variant.
Following its first detection in Gauteng Province, South Africa, the Omicron (B.1.1.529) variant of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was declared by the World Health Organization (WHO) to be a variant of concern on November 25, 2021.1 Rapid transmission of the Omicron variant fueled a fourth wave of SARS-CoV-2 infections in South Africa, during which daily diagnosed infections exceeded totals recorded during all previous periods in the country. Following its initial detection in the United States on 1 December, 2021,9 the Omicron variant rapidly became the dominant circulating lineage, accounting for 95% of all SARS-CoV-2 infections diagnosed nationwide by the week ending January 8, 2022.10 Similar patterns have unfolded globally, with the Omicron variant fueling a surge in newly-diagnosed cases worldwide exceeding all previous phases of the COVID-19 pandemic.11
Understanding the potential for new SARS-CoV-2 variants to differ in clinical severity or in susceptibility to vaccine-derived immunity is crucial to inform public health responses. Such concerns are pronounced with the Omicron variant, which harbors a constellation of mutations in the spike protein associated with altered cell entry and immune evasion. Reduced neutralization of the Omicron variant has been reported in studies using plasma specimens from individuals with complete (two- or three-dose) mRNA vaccine series,12 and from patients with prior SARS-CoV-2 infection.13,14 Epidemiologic data from South Africa have suggested higher rates of Omicron variant infections among persons with prior SARS-CoV-2 infection, as compared to observations with previous variants,15 while early observational studies in multiple settings have suggested reduced effectiveness of COVID-19 vaccines against Omicron variant infection.16–20 However, increases in SARS-CoV-2 infections following emergence of the Omicron variant have not been associated with parallel increases in hospitalizations and deaths to the extent observed during previous waves.3–7 While reduced risk of severe outcomes has been reported among individuals with Omicron variant infection in several large-scale studies linking case data across various nationwide surveillance platforms,19,21 resolving whether these differences reflect lower intrinsic severity of Omicron variant infections, or differences in cases acquiring Omicron and Delta variant infections, requires detailed information about individual-level clinical and epidemiologic risk factors.
Study setting
We aimed to compare clinical outcomes and risk factors among individuals with Omicron and Delta variant SARS-CoV-2 infections within the Kaiser Permanente of Southern California (KPSC) healthcare system. As an integrated healthcare organization serving 4.7 million members (∼19% of the population of southern California), KPSC provides comprehensive care to its members across virtual, outpatient, emergency department, and inpatient settings, captured in near real-time via patients’ electronic health records (EHRs). Our primary analyses included all cases first ascertained in via SARS-CoV-2 reverse transcription-polymerase chain reaction (RT-PCR) testing between November 1, 2021 and January 21, 2022, whose tests were processed using the ThermoFisher TaqPath COVID-19 Combo Kit (Figure 1; details on testing procedures are provided in the Methods). Previous evidence has indicated that the Δ69-70 amino acid deletion in the spike (S) protein of Omicron variant specimens causes a failure in PCR probes targeting the S gene, whereas the Orf1ab and nucleocapsid (N) probes retain sensitivity; in contrast, SGTF is rare in Delta variant SARS-CoV-2 infections.17,21,22 Among 1,477 cases whose SARS-CoV-2 isolates were selected at random for sequencing over the study period, the Omicron variant was detected in all SGTF samples (382/382) and the Delta variant was detected in 1,092 of 1,095 non-SGTF samples (99.7%; Table S1), validating the interpretation of SARS-CoV-2 infections with and without SGTF as Omicron and Delta variant infections, respectively.
Plots illustrate cumulative new outpatient SARS-CoV-2 detections with and without SGTF over the study period, cumulatively (a-b) and by date (c), as well as the age distribution of diagnosed cases with and without SGTF (d), dates of symptomatic hospital admission among these cases (e), and the age distribution of hospitalized cases (f).
Risk of severe outcomes associated with Omicron variant infection
Following an outpatient diagnosis, cumulative 30-day risks of symptomatic hospital admission (defined as new inpatient admission ≤14 days after new-onset acute respiratory symptoms) among cases with Delta and Omicron variant infections were 17.3 (95% confidence interval: 15.6-19.1) and 6.9 (5.4-8.7) per 1000 cases, respectively (Figure 2). Risks of any hospital admission over the same period were 18.2 (16.5-20.0) and 8.0 (6.2-9.9) for cases with Delta and Omicron variant infections, respectively. For endpoints of intensive care unit (ICU) admission, mechanical ventilation, and mortality, 30-day risks among cases with Delta variant infection were 1.9 (1.3-2.5), 1.1 (0.7-1.6), and 1.9 (1.4-2.5) events per 1000 cases, respectively, as compared to 0.4 (0.1-0.9), 0.2 (0.0-0.6), and 0.1 (0.0-0.4) events per 1000 cases with Omicron variant infection.
Plots illustrate cumulative 30-day risk of severe clinical outcomes among cases first ascertained in outpatient settings, stratified by SGTF status (as a proxy for Omicron vs. Delta variant infection), including any hospital admission (a), hospital admission on or <14 days after new-onset acute respiratory symptoms (b), intensive care unit admission (c), mechanical ventilation (d), and death (e). Shaded areas denote 95% confidence intervals around point estimates of daily risk. Inset panels within plots c-e present the data on a y-axis scale of 0-2% for comparison to panels a-b.
In analyses accounting for risk factors including cases’ age, sex, health status, history of documented SARS-CoV-2 infection and COVID-19 vaccination, and prior-year healthcare utilization across outpatient, emergency department, and inpatient settings, adjusted hazards ratios comparing cases with Omicron versus Delta variant infections were 0.62 (0.54-0.72) for any hospital admission, 0.59 (0.51-0.69) for symptomatic hospital admission, 0.45 (0.26-0.78) for ICU admission, 0.36 (0.18-0.74) for mechanical ventilation, and 0.21 (0.10-0.42) for mortality (Figure 3; Table S2). Similar results held within analyses not restricted to cases first ascertained in outpatient settings (Figure S1). While cumulative incidence curves comparing risk of severe outcomes among cases with Omicron and Delta variant infections revealed violations of the proportional hazards assumption, analyses comparing cumulative incidence of severe clinical outcomes in the first 30 days after SARS-CoV-2 detection (among cases ascertained by December 20, 2021) provided similar numerical estimates of the adjusted relative risk of each outcome associated with Omicron variant infection (aRR=0.58 [95% CI: 0.39-0.86] for hospital admission; 0.52 [0.34-0.79] for symptomatic hospital admission; 0.33 [0.07-1.60] for ICU admission; Table S3). No instances of mechanical ventilation or death were observed among cases with Omicron variant infections within this restricted sample.
Points and lines denote estimates and accompanying 95% confidence intervals for the adjusted hazard ratio of each endpoint, comparing cases with Omicron versus Delta variant infection, in case strata defined by history of COVID-19 vaccination. Analyses are restricted to individuals tested diagnosed in outpatient settings by RT-PCR testing using the ThermoFisher TaqPath COVID-19 combo kit; adjusted hazard ratios are estimated using Cox proportional hazards regression models, controlling for covariates listed in Table S2 and matching on testing date.
While the adjusted hazard ratios described above can be interpreted as a composite difference in risk of severe outcomes associated with Omicron variant infection under observed levels of COVID-19 uptake, intrinsic differences in variant-associated severity may be masked by differential vaccine protection against severe disease by variant. Indeed, vaccine-associated reductions in risk of severe outcomes associated were greater among cases with the Delta variant than the Omicron variant (aHR point estimates for 2 mRNA vaccine doses: 0.23-0.37 [Delta], 0.46-0.55 [Omicron]; Table S4). We therefore undertook analyses comparing risk of severe outcomes associated with Omicron variant infection, stratifying by prior vaccine exposure. Adjusted hazards ratios of hospital admission, symptomatic hospital admission, ICU admission, mechanical ventilation, and mortality were 0.45 (0.37-0.54), 0.41 (0.34-0.50), 0.32 (0.17-0.62), 0.25 (0.12-0.52) and 0.10 (0.04-0.25), respectively, for unvaccinated cases with Omicron versus Delta variant infections (Figure 3). In contrast, variant-specific differences in risk of hospital admission or symptomatic hospital admission were not apparent among individuals who received ≥3 mRNA vaccine doses or who received a second mRNA vaccine dose <90 days before their test date. Estimated aHRs for any hospital admission and symptomatic hospital admission associated with Omicron variant infection were 0.43 (0.25-0.75) and 0.37 (0.21-0.66), respectively, among individuals who had received one dose of Ad.26.COV2, and 0.44 (0.21-0.91) and 0.48 (0.22-1.03), respectively, among individuals who had received one mRNA vaccine dose. However, the aHR for death associated with Omicron variant infection among vaccinated individuals was 0.32 (0.12-0.83). Omicron variant infection was likewise associated with reduced risk of symptomatic hospital admission in subgroups defined by cases’ age, sex, and presence of comorbidities, and among cases without prior documented SARS-CoV-2 infection (Table S5). Among cases with prior documented SARS-CoV-2 infection, the aHR of symptomatic hospital admission associated with Omicron variant infection was 0.09 (0.01-0.85).
Other attributes of cases predicting progression to severe endpoints in adjusted analyses included older age, non-white race or Hispanic ethnicity, higher body mass index, and higher rates of healthcare utilization in the prior year (Table S4). Whereas the aHR of symptomatic hospital admission among outpatient-tested cases with Delta variant infections comparing the oldest (≥80 years) and youngest (<1 year) age groups was 84.8 (11.3-636), the corresponding aHR across this age span among cases with Omicron variant infections was only 15.7 (3.7-67.0; p=0.091 for a test of the null hypothesis of no difference in age effects). Additionally, whereas males had 1.49 (1.26-1.75)-fold higher rates of progression to symptomatic hospital admission than females among cases with Delta variant infections, sex-associated differences in progression were not apparent among cases with Omicron variant infections (aHR=1.00 [0.86-1.16]).
Variant associations with risk of acute respiratory symptoms
Among cases diagnosed in outpatient settings, 79.9% and 78.0% of those with Delta and Omicron variant infections, respectively, reported new-onset acute respiratory symptoms on or <14 days before their date of testing (Table S6). To examine whether selection on symptoms as a cause for testing could obscure differences in the natural history of infection associated with the Omicron and Delta variants within our primary analyses, we undertook subgroup analyses restricted to cases who reported no symptoms at the time they were tested. These analyses provided similar estimates of the aHR for severe clinical outcomes associated with Omicron variant infection (Table S7). Moreover, among cases tested in outpatient settings who reported no symptoms at their time of testing, 3.2% and 2.4% of those with Delta and Omicron variant infections were found to subsequently experience new-onset symptoms (aHR=0.68 [0.64-0.72], for Delta vs. Omicron variant infection; Table S8), suggesting Omicron variant infections were associated with reduced risk of any symptoms in addition to reduced risk of severe clinical outcomes. Among cases without symptoms at the time of their initial outpatient test, the aHR of subsequent new onset symptoms associated with prior vaccination was 0.77 (0.69-0.85) for cases with Omicron variant infections (2.4% vs. 3.0% of vaccinated and unvaccinated cases experiencing new-onset symptoms, respectively) and 0.82 (0.77-0.87) for those with Delta variant infections (3.8% vs. 3.2% of vaccinated and unvaccinated cases experiencing new-onset symptoms, respectively).
Changes in risk of severe outcomes following Omicron variant emergence
Because excluding cases whose tests were not processed using the ThermoFisher TaqPath COVID-19 combo kit could limit the generalizability of our primary analyses, we also assessed changes over the study period in the proportion of all cases diagnosed in outpatient settings who subsequently progressed to severe clinical outcomes. Fitting Cox proportional hazards models with up to two change points for the relationship between testing date and risk of severe clinical endpoints,23 we identified stable risk of severe outcomes from November through the first week of December, when Omicron variant detections were rare, followed by steep declines in risk through the end of December (Figure 4). This period of decline aligned with increases in the proportion of outpatient-diagnosed cases with the Omicron variant identified. While modest increases in the time-specific hazards of severe endpoints were apparent through January, this change coincided with increases in the reported time since symptom onset among diagnosed cases, suggesting that cases sought outpatient testing at a more advance stage of clinical illness by the end of the study period in comparison to late December.
Top row: Panels illustrate changes in the time-specific adjusted hazard ratios, among all cases ascertained in outpatient settings, for endpoints of any hospital admission (a), symptomatic hospital admission (b), ICU admission (c), and mortality (d). Liens and shaded areas in each plot delineate point estimates and accompanying 95% confidence intervals, as estimated by Cox proportional hazards models fitted to allow up to two changes in slope. Estimates are averaged across models allowing for changepoints at each day in the study period, according to posterior model weights defined by the Bayesian information criterion. Middle row: Panels illustrate the proportion of outpatient RT-PCR tested samples tested on the ThermoFisher TaqPath COVID-19 combo kit with SGTF identified (e), the proportion of all outpatient-diagnosed cases reporting symptoms on or before their testing date (f), the mean time from symptoms onset to testing among symptomatic outpatient tested cases (g), and the mean time from symptoms onset to hospital admission among cases first ascertained in inpatient settings (h). Black lines in panels f-h delineate 95% confidence intervals of the mean based on bootstrap resampling; red lines in panels a-h indicate the dates when SGTF was detected in 10%, 50%, and 90% of positive outpatient samples. Bottom row: Panels illustrate the distribution of times from symptoms onset to testing among outpatient-diagnosed cases (i) and from symptoms onset to hospital admission, among hospitalized patients (j), each stratified by SGTF or non-SGTF sample class and subset to cases tested using the ThermoFisher TaqPath COVID-19 combo kit. Last, we illustrate total newly diagnosed cases each day across all settings along with the number of new symptomatic hospital admissions, ICU admissions, patients beginning mechanical ventilation, and deaths (k).
To determine whether observed changes in the proportion of cases experiencing severe endpoints tracked with our primary estimates of the aHR of severe endpoints associated with Omicron variant infection, we considered the following relationship of variant-specific hazards (λDelta and λOmicron) to era-specific hazards (λ0 and λ1 for 1-30 November and 1-21 January, respectively):
with π0 and π1 defining the proportion of cases in the two eras associated with the Omicron variant. Defining π0≈0 for 1-30 November,
enabling comparison of the variant-specific hazards from the era-specific hazard ratio λ1/ λ0. Under this framework, we estimated the aHRs for hospital admission, symptomatic hospital admission, ICU admission, and death associated with Omicron variant infection were 0.48 (0.40-0.57), 0.47 (0.39-0.56), 0.39 (0.23-0.61), and 0.35 (0.19-0.57), respectively, confirming the ranges of our primary estimates (Figure 4; Table S9).
Lengths of hospital stay
As an additional measure of disease severity, we compared lengths of hospital stay among patients who experienced new inpatient admissions following symptoms onset with Omicron and Delta variant infections. To avoid bias due to censoring of incomplete hospitalizations, or selection on short hospital stays among patients admitted at the end of the study period, we restricted these analyses to individuals with completed hospitalizations who were admitted on or before 1 January, 2022, and thus had ≥20 days of hospital-based follow-up time. Within this sample, mean durations of stay were 5.6 (5.1-6.1) days and 1.6 (1.4-1.9) days among patients with symptomatic hospital admissions associated with Delta and Omicron variant infections, respectively (Figure 5; Table S10). Whereas 85.0% (78.6-91.0%) of Omicron variant-associated symptomatic hospitalizations lasted ≤2 days, only 30.3% (26.4-34.3%) of Delta variant-associated symptomatic hospitalizations lasted ≤2 days. The 3.9 (3.3-4.5) day difference in means represented a 70.7% (64.4-76.0%) shorter mean duration of hospital stay with Omicron variant-associated symptomatic hospitalizations.
Panels illustrate the distribution of the length of hospital stay among cases admitted by 1 January, 2022, whose admissions occurred on or after the date of symptoms onset, for cases tested using the ThermoFisher TaqPath COVID-19 combo kit and identified as Delta variant cases (a) or Omicron variant cases (b). Inset panels illustrate draws from the distribution of the difference in mean lengths of hospital stay within these samples, based on bootstrap resampling (c), and cumulative probabilities of hospital outcomes of being discharged alive (d) or death or hospice discharges (e), based on parametric competing-risk models fitted to the full data and accounting for censoring of incomplete hospitalizations. Panels d-e distinguish estimates for cases with Omicron (red) and Delta (black) variant infections.
Differences among cases with Omicron and Delta variant infections
As compared to cases diagnosed in outpatient settings with Delta variant infection, cases with Omicron variant infection were less likely to be infants or older adults, were more often white and of non-Hispanic ethnicity, and were less likely to live in very low-income communities (Figure 1; Table S11). Cases with Omicron variant infection also tended to have lower prior-year rates of healthcare utilization across outpatient and emergency department settings. Whereas comorbid conditions included in the Charlson comorbidity index were generally less prevalent among cases with Omicron variant infections as compared to Delta variant infections, these differences were not generally apparent after adjustment for other demographic and epidemiologic differences between cases in either outpatient or hospital-based samples (Table S12).
Adjusted odds of a prior documented SARS-CoV-2 infection ≥90 days before individuals’ first positive test during the study period were 2.48 (2.04-3.02) fold higher among cases with Omicron variant infections than among cases with Delta variant infections infection (Table S11); among unvaccinated cases, adjusted odds of a prior documented SARS-CoV-2 infection were 3.32 (2.53-4.35) fold higher among cases with Omicron variant infections than those with Delta variant infections (Table S13). Prior receipt of vaccine series associated with greater degrees of immune protection were more common among cases with Omicron variant infection; for instance, adjusted odds of prior receipt of a single mRNA vaccine dose (B162b2/mRna-1273) and a single Ad.26.COV2.S dose were 1.73 (1.62-1.85) and 1.72 (1.62-1.83) fold higher among cases with Omicron as compared to Delta variant infections, while adjusted odds of receipt of ≥3 mRNA vaccine doses were 5.17 (4.94-5.42) fold higher among cases with Omicron variant infections (Table S11).
Discussion
Among cases followed after an outpatient SARS-CoV-2 diagnosis from November 1, 2021 to January 21, 2022, infection with the Omicron variant was associated with substantially lower risk of progression to severe clinical outcomes including hospital admission, symptomatic hospital admission, ICU admission, mechanical ventilation, and mortality, in comparison to the Delta variant. These differences in risk of severe clinical outcomes among individuals with Omicron versus Delta variant infection were in agreement with reductions in the proportion of all SARS-CoV-2 cases that progressed to severe outcomes during the period of Omicron variant emergence in the study population. We also observed shorter durations of hospital stay following admission among cases with Omicron as compared to Delta variant infections. Taken together, our findings support the conclusion that the Omicron variant of SARS-CoV-2 intrinsically causes less severe illness than the Delta variant.
Our findings are broadly in agreement with several previous studies, which have estimated reductions in risk of hospital admission associated with Omicron variant infection across a range spanning 20-80%.19,21,24–26 Whereas this variability in prior estimates from database linkage studies may owe in part to intra- and inter-study differences in immunity, health status, and healthcare-seeking behaviors, highly-resolved data on cases’ clinical history provided several new insights within this study. First, we found that cases with Omicron variant infection had lower rates of prior-year healthcare utilization than those with Delta variant infection, suggesting better health status. Second, we found that Omicron variant infection was associated with reduced risk of new-onset acute respiratory symptoms, regardless of whether cases progressed to hospital admission and other severe clinical endpoints, and that prior vaccination was associated with reduced risk of new-onset symptoms among cases with both Omicron and Delta variant infections. Transient reductions in the average time from symptoms onset to diagnosis in late December further suggested that testing captured a less-severe spectrum of illness as the Omicron variant emerged, possibly in association with testing before holiday gatherings and travel. Last, we identified that differences in risk of severe clinical outcomes associated with Omicron versus Delta variant infections were attenuated among previously vaccinated individuals. Differences in vaccine-conferred protection against severe disease progression with the two variants led to similar risk of hospital admission or symptomatic hospital admission among cases with Omicron or Delta variant infections who had received 3 mRNA vaccine doses.
Our findings collectively suggest the reduced propensity to cause severe clinical illness is an intrinsic attribute of the Omicron variant. Recent ex vivo studies demonstrate higher replication of the Omicron variant in the human upper respiratory tract as compared to the small airways of the lung,27 consistent with animal experiments suggesting that disease associated with the Omicron variant might be confined to the large airway.28 Our findings also suggest epidemiologic consequences of differences in the Omicron variant’s binding affinity to the human angiotensis-conoverting enzyme 2 (ACE2) receptor.29,30 Whereas we identify that male cases with Delta variant infection had 49% higher risk of symptomatic hospital admission than female cases, male and female cases with Omicron variant infection did not differ in risk of this outcome. Additionally, we identified 85-fold increases in risk for cases with Delta variant infection to progress to symptomatic hospital admission between infancy and age ≥80 years, as compared to a 16-fold increase in risk across the same age span among cases with Omicron variant infection. As ACE2 expression is higher in males and in older age groups,31,32 reduced binding affinity of the Omicron variant (versus the Delta variant) with ACE2 may contribute to these observations.
In this analysis, prevalence of prior vaccination differed among cases with Omicron and Delta variant infections. While our analysis cannot infer absolute vaccine effectiveness against the distinct variants, our findings suggest vaccine protection against infection with the Omicron variant may be lower than protection against infection with the Delta variant. This result is consistent with studies showing reduced neutralization efficiency of two and three doses of BNT162b2 vaccine against the Omicron variant (versus non-Omicron variants).12,33 Similarly, in multiple settings, vaccination with two doses showed slightly lower effectiveness against hospitalization with Omicron than Delta variant infections.16–20 Our finding of higher relative protection against Delta variant infections is reassuring considering the greater severity observed in infections with the Delta variant as compared to the Omicron variant. However, evidence for a reduction in severe outcomes among vaccinated cases with both Delta and Omicron variant infections in our study suggests substantial public health benefits from continued COVID-19 vaccination.
Our study has at least three limitations. First, incidental detections of SARS-CoV-2 among hospitalized patients might also have been more commonly associated with Omicron variant than Delta variant infections, leading to underestimates of attenuated disease severity in analyses of all hospitalizations. This consideration supports our use of EHR data on symptoms (assessed for all patients tested at KSPC) to define an endpoint of hospitalizations associated with new-onset acute respiratory symptoms, and our use of a prospective cohort design monitoring for new inpatient admissions among cases first ascertained by outpatient testing. Second, restricting the primary analysis population to cases whose tests were processed using assays that detect SGTF could induce selection bias. However, results were aligned with changes in risk of severe clinical outcomes over the study period in a complete-case analysis (Figure 4; Table S9). Third, prior infections may have gone untested or unrecorded in the EHR (for instance, if individuals were tested in other settings). Regardless of the potential for false-negative classifications of this exposure, we observed substantially lower risk of symptomatic hospital admission associated cases with Omicron variant infection within the subgroup of cases with documented prior SARS-CoV-2 infection (Table S4), suggesting variant-specific differences in risk of severe outcomes are not simply attributable to Omicron variant infections occurring among individuals with greater immunity.
While attenuation of disease severity in Omicron variant infections is an encouraging finding, evidence of higher transmissibility of Omicron variant infections34 as well as immune evasion from prior infection and vaccination remain concerning. Rapid spread of the Omicron variant has led to unprecedented surges in COVID-19 among our study population and others globally. High rates of infection in the community have overwhelmed health-care systems and have translated to high absolute numbers of hospitalizations and deaths, even with lower severity of the Omicron variant. Thus, implementation of concurrent prevention strategies including vaccination, masking, and appropriate infection mitigation strategies remains important to curb transmission, decrease morbidity and mortality, and reduce burden on health systems.
Data Availability
Requests for access to data should be submitted to Sara.Y.Tartof{at}kp.org.
METHODS
Setting, procedures, and study population
Care delivery and EHR data capture in the KPSC healthcare system have been described previously.35 Briefly, members of KPSC receive care through employer-provided, pre-paid, or federally sponsored insurance plans and closely resemble the sociodemographic profile of the surrounding geographic area in terms of age, racial/ethnic composition, and community characteristics.36 Within-network care delivery is captured in real time through patients’ HER, while out-of-network care is captured through insurance reimbursements. COVID-19 vaccines were provided at no cost to KPSC members following emergency use authorization and were therefore captured in the HER. Vaccinations administered outside KPSC were captured via the California Immunization RRegistry, to which providers are required to report all COVID-19 vaccine administrations within 24 hours.
Polymerase chain reaction (PCR) testing for SARS-CoV-2 occurred in a variety of clinical settings within KPSC during the study period. A majority of tests conducted in outpatient settings are submitted to regional laboratories, where >90% of samples are processed using the ThermoFisher TaqPath COVID-19 Combo Kit. Samples collected in hospitals (including some tests conducted in emergency department settings) are processed using in-house tests, without SGTF readout. In total, 272,851 of 328,024 (83.2%) cases occurring during the study period had samples processed using the ThermoFisher TaqPath COVID-19 Combo Kit, including 271,109 of 322,705 (84.0%) of samples from cases ascertained in outpatient settings. Attributes of outpatient cases processed using the ThermoFisher TaqPath COVID-19 Combo Kit or other assays are presented in Table S14.
Our primary analyses included all cases diagnosed in outpatient settings with a positive PCR test outcome during the study period. Where specified, additional analyses included all cases diagnosed during the study period or all cases diagnosed in outpatient settings during the study period. All analyses excluded individuals who had received a positive PCR result for SARS-CoV-2 infection within 90 days prior to their first positive result during the study period.
Outcome measures
As primary endpoints, we considered five markers of clinically severe illness: any hospital admission, hospital admission associated with new-onset acute respiratory symptoms, intensive care unit (ICU) admission, mechanical ventilation, and mortality. Hospitalizations and ICU admissions were considered to be COVID-19-related if they occurred between 7 days before to 28 days after the date of each patient’s positive SARS-CoV-2 RT-PCR test. Symptomatic hospital admissions were those with acute respiratory infection symptoms beginning on or ≤14 days before the admission date; we ascertained presence of symptoms and dates of symptoms onset via natural language processing of open-text EHR fields including clinical notes and patient-provided questionnaire responses, which are submitted by all KPSC patients who seek SARS-CoV-2 testing regardless of test setting.35 We considered new-onset respiratory symptoms following a positive test as a secondary endpoint for further exploratory analyses among cases who were asymptomatic at the time of their original test. We recorded the first date that each study endpoint occurred, censoring observations at January 21, 2022 or patients’ dates of disenrollment, whichever was earliest.
Last, for a duration-of-hospital-stay analysis, we recorded dates of discharge, in-hospital mortality, or censoring for all hospitalized patients. Patients who died in hospital or were discharged to hospice were considered to have experienced fatal COVID-19 hospitalizations, while patients discharged for any other reason were considered to have experienced non-fatal COVID-19 hospitalizations. Living patients who had not yet been discharged by January 21, 2022 were considered censored for the duration-of-stay analysis. Duration-of-stay analyses included all patients admitted to hospital whose samples were processed using the ThermoFisher TaqPath COVID-19 Combo Kit, and who experienced acute new-symptoms onset respiratory symptoms on or before their admission date.
Exposures
Exposures of interest included demographic characteristics of patients (age, sex, race/ethnicity, and neighborhood-level median household income) as well as clinical risk factors recorded within the prior year (including history of smoking, body mass index, Charlson comorbidity index and its constituent conditions, and healthcare utilization across outpatient, emergency department, and inpatient settings). We additionally recorded patients’ history of a positive SARS-CoV-2 test result of any type or COVID-19 diagnosis ≥90 days prior to their first positive RT-PCR test during the study period, as well as the dates of receipt of any COVID-19 vaccine doses (BNT162b2 [Pfizer/BioNTech], mRNA-1973 [Moderna/National Institutes of Health], or Ad.26.COV2.S [Janssen]).
Association of Omicron variant infection with risk of severe clinical outcomes
Within the primary analysis population and among all cases whose samples were processed using the ThermoFisher TaqPath COVID-19 combo kit, we compared times from the first positive test to each outcome event or censoring among patients who tested positive for SARS-CoV-2 by RT-PCR, with and without SGTF. We used Cox proportional hazards models to estimate the adjusted hazard ratio (aHR) for each endpoint associated with SGTF, adjusting for all available demographic and clinical covariates: age (defined within bands of <1, 1-4, 5-9, 10-19, 20-29, 30-39, 40-49, 50-59, 60-69, 70-79, and ≥80 years); sex; race/ethnicity (white, black, Hispanic of any race, Asian, Pacific Islander, and other/mixed/unknown race); census-tract median household income (measured continuously on the log scale); smoking status (current, former, or never smoker); body mass index (BMI; underweight, normal weight, overweight, and obese); Charlson comorbidity index (0, 1-2, 3-5, and ≥6); prior-year emergency department visits and inpatient admissions (each defined as 0, 1, 2, or ≥3 events); prior-year outpatient visits (0-4, 5-9, 10-14, 15-19, 20-29, or ≥30 events); documented prior SARS-CoV-2 infection; and history of COVID-19 vaccination (unvaccinated, Ad.26.COV2.S as one dose or with a booster; and 1, 2, or 3 mRNA vaccine doses, disaggregating 2-dose recipients by time since receipt of the second dose). We accounted for strata (baseline hazards) by cases’ approximate testing date to account for potential secular changes in testing and healthcare practices over the study period. Testing dates were jittered by +/-1 or 0 days to preserve confidentiality of patient data. We additionally fit models allowing for interactions of SGTF sample status with cases’ vaccination history to assess variation in the association of SGTF with vaccination status, as described in the primary results above.
We repeated analyses of the symptomatic hospital admission endpoint within subgroups defined by patient age, sex, Charlson comorbidity index, and history of documented SARS-CoV-2 infection and vaccination, controlling for all other risk factors via covariate adjustment. In secondary analyses that included patients tested in inpatient settings, times to events were recorded as 0.5 days for patients tested on or before the date of an outcome event occurring.
We conducted multiple (m=5) imputation of missing covariate values and pooled results obtained with each imputed dataset via Rubin’s rules37 (Table S15). We verified proportional hazards assumptions by testing for slopes in Schoenfeld residuals.38 To assess whether violations of the proportional hazards assumption resulted in biased estimates, we estimated adjusted relative risks of each outcome based on cumulative occurrence over 30 days of follow-up among cases whose diagnoses occurred on or before 21 December, 2021. These sensitivity analyses used Poisson regression models with the same specifications described above, allowing distinct intercepts for each testing week.
Comparing risk of severe clinical outcomes during periods of Delta and Omicron variant predominance
Among all cases ascertained in outpatient settings, we fit aHRs relating the risk of severe clinical endpoints to cases’ testing date via Cox proportional hazards models. As the goal of these analyses was to relate changes in risk of severe clinical outcomes to the emergence of the Omicron variant in the study population, these models did not include strata for testing date and did not include SGTF status as a covariate. We fit candidate models allowing for up to two changepoints in the slope of associations between testing date and risk of clinical endpoints, with changepoints defined at all dates in the study period between 15 November, 2021 and 5 January, 2022. We averaged daily slopes across all candidate models according to posterior model weights defined by the Bayesian information criterion.23,39 As the sporadic occurrence of mechanical ventilation during the early study period hindered estimation of slopes in risk of this outcome (Figure 4K), analyses addressed endpoints of hospital admission, symptomatic hospital admission, ICU admission, and death only.
Hospital duration of stay analysis
We used bootstrap resampling to sample from the empirical distribution of the mean duration of hospital stay among cases with Omicron and Delta variant infections whose tests were processed using the ThermoFisher TaqPath COVID-19 combo kit. Analyses were restricted to cases with completed, symptomatic hospital admissions, with admission dates on or before 1 January, 2022 in order to allow ≥20 days of follow-up per case. We also fit parametric distributions of times to outcomes of mortality (defined by in-hospital death or discharge to hospice) or discharge alive, among cases with Omicron or Delta variant infections, within the full sample of hospitalized cases tested with the ThermoFisher TaqPath COVID-19 combo kit. These analyses assumed log-normally distributed event times and accounted for competing risk among the three outcomes of fatal, non-fatal, or censored discharge dispositions.
Comparing attributes of cases with Omicron and Delta variant infections
We used logistic regression to estimate adjusted odds ratios (aORs) of each exposure among cases with SGTF versus non-SGTF infections. Consistent with our primary analyses, we pooled results across multiple imputed datasets.
Software
We conducted all analyses using R (version 4.0.3; R Foundation for Statistical Computing, Vienna, Austria). We used the survival 40 and cfc 41 packages for time-to-event analyses, and the Amelia II 42 package for multiple imputation.
Points and lines denote estimates and accompanying 95% confidence intervals for the adjusted hazard ratio of each endpoint, comparing cases with Omicron versus Delta variant infection, in case strata defined by history of COVID-19 vaccination. Analyses are restricted to individuals tested diagnosed by RT-PCR testing using the ThermoFisher TaqPath COVID-19 combo kit, in contrast to Figure 3 which further limits analyses to cases diagnosed in outpatient settings. Adjusted hazard ratios are estimated using Cox proportional hazards regression models, controlling for covariates listed in Table S4 and matching on testing date.
Acknowledgments
This work was funded by the US Centers for Disease Control and Prevention. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC.
References
