{"paper_id":"31c1e070-c37a-406f-a2b3-678dc6c1bd46","body_text":"1 | Page \n \nAccident and emergency (AE)  attendance in England following \ninfection with SARS-CoV-2 Omicron or Delta \n \nAuthors: Daniel J. Grint 1, Kevin Wing 1, Hamish P. Gibbs 1, Stephen JW Evans 1, Elizabeth Williamson 1, \nKrishnan Bhaskaran 1, Helen I McDonald1, Alex J. Walker 2, David Evans 2, George Hickman 2, Rohini \nMathur1, Anna Schultze1, Christopher T Rentsch1, John Tazare1, Ian J Douglas1, Helen J. Curtis2, Caroline \nE Morton2, Sebastian Bacon 2, Simon Davy 2, Brian MacKenna2, Peter Inglesby2, Richard Croker2, John \nParry3, Frank Hester3, Sam Harper3, Nicholas J DeVito2, Will Hulme2, Chris Bates3, Jonathon Cockburn3, \nAmir Mehrkar2, Ben Goldacre2, Rosalind M. Eggo1, Laurie Tomlinson1 \nCorresponding author: Daniel Grint. daniel.grint@lshtm.ac.uk \n \n1London School of Hygiene & Tropical Medicine, Keppel Street, London, UK \n2The DataLab, Nuffield Department of Primary Care Health Sciences, University of Oxford, UK \n3TPP, TPP House, 129 Low Lane, Horsforth, Leeds, UK \n \nAuthor contributions \nDJG, RME, and LT led the study. DJG, RME, KW, and LT drafted the study protocol. DJG, KW, HG, EW, \nHIMcD, KB, DE, SJWE, AJW, AS, CTR, HJC, CEM, RM, WH, and JT contributed to data preparation and \nvariable definitions. DE, GH, CB, JC, HJC, CEM, SB, SD, AM, LT, IJD, BMack, PI, RC, JP, FH, SH, NJDeV, \nWH, and BG contributed to building the analytical platform. DJG, KW, RME, KB, SJWE, and LT \ncontributed to study design. DJG performed the statistical analysis and drafted the manuscript. All \nauthors contributed to manuscript preparation and refinement. \n \nAbstract \nThe SARS-CoV-2 Omicron variant is increasing in prevalence around the world. Accurate estimation of \ndisease severity associated with Omicron is critical for pandemic planning. We found lower risk of \naccident and emergency (AE) attendance following SARS-CoV-2 infection with Omicron compared to \nDelta (HR: 0.39 ( 95% CI: 0.30 – 0.51; P<.0001). For AE attendances that lead to hospital admission, \nOmicron was associated with an 85% lower hazard compared with Delta (HR: 0.14 (95% CI: 0.09 – \n0.24; P<.0001)). \nKeywords \nSARS-CoV-2; Coronavirus; COVID-19; Omicron; Delta; Accident and emergency attendance \nConflicts of Interests  \nNothing to declare. \nFunding statement \nThis work was supported by the Medical Research Council MR/V015737/1. TPP provided technical \nexpertise and infrastructure within their data centre pro bono in the context of a national emergency. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \nNOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.\n\n2 | Page \n \nRosalind Eggo is funded by HDR UK (grant: MR/S003975/1), MRC (grant: MC_PC 19065), NIHR (grant: \nNIHR200908). \n \n \n \n \n \n  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \n\n3 | Page \n \nAbstract \nThe SARS-CoV-2 Omicron variant is increasing in prevalence around the world. Accurate estimation of \ndisease severity associated with Omicron is critical for pandemic planning. We found lower risk of \naccident and emergency (AE) attendance following SARS-CoV-2 infection with Omicron compared to \nDelta (HR: 0.39 ( 95% CI: 0.30 – 0.51; P<.0001). For AE attendances that lead to hospital admission, \nOmicron was associated with an 85% lower hazard com pared with Delta (HR: 0.14 (95% CI: 0.09 – \n0.24; P<.0001)). \nIntroduction \nThe SARS-CoV-2 variant B.1.1.529 (Omicron) was first identified in South Africa in late 2021. Analysis \nhas found that Omicron is more transmissible than the predominant B.1.617.2 vari ant (Delta) and it \nhas since become the dominant strain throughout the UK.(1) Only a small proportion of Omicron cases \nare identified by whole-genome sequencing. In PCR assays for SARS -CoV-2 processed by TaqPath \nlighthouse laboratories, missingness in one spike protein gene target occurs with the Omicron variant, \nbut not the Delta variant. Spike gene target failure (SGTF) is therefore a pro xy for Omicron \nidentification, and has been shown to have excellent sensitivity in England over the study period.(1)  \nWorking on behalf of NHS England, we estimate the risk of accident and emergency (AE) attendance \nfollowing confirmation of SARS-CoV-2 infection in England, comparing infection with Omicron to Delta, \nafter accounting for demographic factors and comorbidities (Supplement 3). \nStudy platform and population \nAll data were linked, stored and analysed securely within the OpenSAFELY platform \nhttps://opensafely.org/ (Supplement 1). The OpenSAFELY dataset is based on 24 million people \ncurrently registered with GP surgeries using TPP SystmOne software, covering 40% of England’s \npopulation. Pseudonymized data include coded diagnoses, medications and physiological parameters. \nAll code is shared openly for r eview and re -use under MIT open license \n(https://github.com/opensafely/SGTF-Omi). \nWe used linked GP, SARS -CoV-2 testing, vaccination and emergency care data (Supplement 2) to \ndefine the study cohort of  people first testing positive for SARS -CoV-2 between 5 th December 2021 \nand 1 st January 2022. The study was analysed according to the pre-define study protocol \n(https://github.com/opensafely/SGTF-Omi-research/tree/main/docs), in line with previous work. (2, \n3) \nSGTF status was known for 330,380/755,432 (44%) people with a first confirmed SARS-CoV-2 infection \nbetween 5th December 2021 and 1st January 2022 (237,430 Omicron; 92,950 Delta). A total of 660 (341 \nOmicron; 319 Delta) AE attendances were recorded with SARS-CoV-2 recorded as the patient diagnosis \nprior to 21st January 2022, when follow-up was administratively censored. The exposure groups were \nsimilar in terms of sex, ethnicity, and regional distribution (Table 1, Supplement 4). The median age of \nthe Omicron group was higher (35 years (interquartile range (IQR) 24 – 49)) vs. 32 (11 – 44), with more \ncomorbidities (2+ comorbidities: 2.4% vs. 1.4%). A lower proportion of Omicron cases were \nunvaccinated (17.1% vs. 43.0%) while a higher proportion had received a booster vaccination (23.2% \nvs. 5.1%) compared to Delta at the time of diagnosis. \nDelta diagnoses were more frequent in the first week of the study period, while Omicron diagnoses \npredominated thereafter. Consequently, median follow -up time was shorter among the  Omicron \ngroup (26 days (IQR: 23 - 31)) than the Delta group (39 days (34 - 43)) (Figure 1). \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \n\n4 | Page \n \nRelative hazard of AE attendance \nWe estimate d the relative hazard of AE attendance with Omicron compared to Delta using Cox \nproportional hazards regression models stratified by upper tier local authority area (UTLA). (4) \nCovariate adjustment was informed by a directed acyclic graph (DAG) (Supplement 5). Follow -up \nbegan at the date of positive SARS-CoV-2 test and was censored at the earliest of death, AE attendance \nwith diagnosis coded as SARS -CoV-2, or 7 -days prior to the emergency care data lock (28 th January \n2022). \nOmicron was consistently associated with lower hazard of AE attendance compared to Delta. In fully-\nadjusted analysis accounting for demographics, vaccination status, and comorbidities, the hazard of \nAE attendance was 60% lower for Omicron (hazard ratio (HR): 0.39 (95% confidence interval (CI): 0.30 \n– 0.51; P <0.0001) compared with Delta. \nThe hazard of AE attendance was consisten tly lower for Omicron across all subgroup analyses \nincluding epidemiological week, age group, vaccination status, comorbidity status, and ethnicity \n(Figure 2). \nThere was strong evidence for effect modification by vaccination status (P=.0004). While Omicron was \nassociated with lower hazards of AE attendance regardless of vaccination status, the effect was \nstrongest among the unvaccinated (HR: 0.20 (95% CI: 0.13 – 0.31)) (Figure 2). \n44 people who attended AE were excluded from these analyses as they at tended AE on the day of \ntesting positive for SARS-CoV-2. In sensitivity analysis, adding one day to all follow-up times to include \nthese outcomes, the relative hazard estimates were unchanged (data not shown). Estimates were also \nconsistent when restrict ing to people with at least 14 -days between testing positive and the data \ncensor, and with multiple imputation for missing ethnicity data (Figure 1). \nFor AE attendances which resulted in hospital admission, Omicron was associated with an 85% lower \nhazard compared to Delta (HR: 0.14 (95% CI: 0.09 – 0.24; P<.0001)) (Figure 1). \nAbsolute risk of AE attendance \nWe estimate the absolute risk of AE attendance by 14 -days after SARS-CoV-2 positive test by the \nmarginal means from a fully-adjusted logistic regression model, including an interaction term between \nSARS-CoV-2 variant and vaccination status. This analysis was restricted to  positive tests at least  14 \ndays before the censoring date. AE attendances beyond 14 days were censored. \nThe absolute risk of AE attendance was lower for people double or booster vaccinated for all age and \ncomorbidity subgroups, compared to those unvaccinated. The largest differential in absolute risk of \nAE attendance between Omicron and Delta was seen for unvaccinated people with two or  more \ncomorbidities over the age of 70 (62 AE attendances per 1000 diagnoses (36 - 88) vs. 17 per 1000 (8 - \n26)). However, even after booster vaccination, people with two or more comorbidities aged over 70 \nwith Omicron had more than twice the absolute risk  of AE attendance compared to unvaccinated \npeople aged over 70  without comorbidities (9 per 1000 (6 - 12) vs. 4 per 1000 (2 - 6)) (Figure 2, \nSupplement 6). \nDiscussion \nWe show that Omicron is associated with considerably lower risk of AE attendance and in p articular \nadmission to hospital following AE attendance than the Delta variant. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \n\n5 | Page \n \nThe consistency of the effect for all epidemiological weeks shows that the reduced severity with \nOmicron cannot be explained by other secular changes such as hospitals exceedin g capacity or \nbehavioural patterns. \nThere was strong evidence that the relative reduction in AE attendance for Omicron was largest \namong the unvaccinated. However, Omicron was consistently associated with a relative reduction in \nAE attendance regardless of  vaccination status and this difference is likely to reflect the greater \nefficacy of the vaccine against Delta. (5, 6)  For the avoidance of doubt, the absolute risk of AE \nattendance was lower for people double or booster vaccinated for all age and comorbidity subgroups, \ncompared to those unvaccinated. \nThe relative reduction in AE attendance for Omicron compared with Delta was largest when restricting \nthe outcome to AE attendances which resulted in hospital admission. \nAlthough there was no evidence of differential severity of Omicron compared to Delta by comorbidity \nstatus, in the fully-adjusted model those with two or more comorbidities were at greater than 4-fold \nincreased risk of AE attendance compared to those with no reported comorbidities (data not shown). \nFurther, while the absolute risks of AE attendance were consistently lower for Omicron, the risk among \npeople in older age groups living with multiple comorbidities after booster vaccination, remained \ndouble that of otherwise healthy unvaccinated older age groups. \nThis study has several limitations, AE attendan ce data does not include people who are admitted to \nhospital without going to AE. Our data include only people first testing positive for SARS -CoV-2. Re-\ninfection with Omicron is common and extra immunity from prior infection may reduce the severity \nof Omicron further.(7, 8) Our study includes few peo ple in the oldest age groups. Further detailed \nanalysis of which groups remain at greatest risk from Omicron will be essential for health provision \nplanning when societies move toward light touch restrictions in the presence of a high burden of \ncirculating SARS-CoV-2. \n \nEthical approval \nThis study was approved by the Health Research Authority (REC reference 20/LO/0651) and by the \nLSHTM Ethics Board (reference 21863). \n \nAcknowledgements \nWe are grateful for the support received from the TPP Technical Operations  team and for generous \nassistance from the information governance and database teams at NHS England / NHSX. \n \n \n  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \n\n6 | Page \n \nTable 1. Summary demographic and clinical characteristics of the study population \n Total \nN (%) \nDelta \nN (%) \nOmicron \nN (%) \nTotal population 330,380 92,950 237,430 \nAE attendances 660 (0.2) 319 (0.3) 341 (0.1) \nLeading to admission 190 (0.1) 125 (0.1) 65 (0.0) \nTime to AE attendance \nDays, median (IQR) 5.0 (2.0-8.0) 5.0 (3.0-8.0) 4.0 (2.0-8.0) \nFollow-up time \nDays, median (IQR) 29.0 (24.0-36.0) 39.0 (34.0-43.0) 26.0 (23.0-31.0) \nEpidemiological week of diagnosis \n05Dec-11Dec 42,691 (12.9) 39,317 (42.3) 3,374 (1.4) \n12Dec-18Dec 63,870 (19.3) 30,881 (33.2) 32,989 (13.9) \n19Dec-25Dec 96,025 (29.1) 15,980 (17.2) 80,045 (33.7) \n26Dec-01Jan 127,794 (38.7) 6,772 (7.3) 121,022 (51.0) \nSex    \nFemale 170,098 (51.5) 47,799 (51.4) 122,299 (51.5) \nAge group \n0-39 199,764 (60.5) 60,497 (65.1) 139,267 (58.7) \n40-54 81,761 (24.7) 23,262 (25.0) 58,499 (24.6) \n55-64 31,750 (9.6) 6,730 (7.2) 25,020 (10.5) \n65-74 11,607 (3.5) 1,729 (1.9) 9,878 (4.2) \n75-84 4,453 (1.3) 580 (0.6) 3,873 (1.6) \n85+ 1,045 (0.3) 152 (0.2) 893 (0.4) \nCategorical number of comorbiditiesa    \nNone 296,883 (89.9) 85,970 (92.5) 210,913 (88.8) \nOne 26,517 (8.0) 5,718 (6.2) 20,799 (8.8) \nTwo or more  6,980 (2.1) 1,262 (1.4) 5,718 (2.4) \nSARS-CoV-2 vaccination status at diagnosis \nUnvaccinated 80,499 (24.4) 40,007 (43.0) 40,492 (17.1) \nFirst dose >14-days prior 20,486 (6.2) 6,187 (6.7) 14,299 (6.0) \nSecond dose >14-days prior 169,684 (51.4) 42,027 (45.2) 127,657 (53.8) \nBooster 59,711 (18.1) 4,729 (5.1) 54,982 (23.2) \n \naComorbidities as defined in Supplement section 3.  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \n\n7 | Page \n \nFigure 1 Geographical distribution of study population in England \n \nPanel a) Geographical distribution of cohort SARS-CoV-2 diagnoses between 5th December 2021 \nand 1st January 2022. Panel b) Total number of Omicron and Delta cases by epidemiological week.  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \n\n8 | Page \n \nFigure 2. Hazard ratios for AE attendance comparing Omicron vs. Delta from C ox proportional \nhazards regression stratified by Upper Tier Local Authority (UTLA). All subgroup analyses were \nperformed on the fully-adjusted model. \n \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \n\n9 | Page \n \nFigure 1 legend \naSeperate models adjusted for age and vaccination status only. *Likelihood ratio test for interaction \nbetween exposure group and subgroup. Number of events marked X are masked to prevent \nidentifiability. All models are stratified on region by UTLA. \nDemographically adjusted model includes adjustment for: age, sex, vaccination status, IMD, \nethnicity, household size, rural urban classification, epidemiological week, and care home status. \nThe fully-adjusted model includes adjustment for: age, sex, vaccination status, IMD, ethnicity, \nsmoking status, obesity, household size, rural urban classification, comorbidities, epidemiological \nweek, and care home status. There was weak evidence of non-proportional hazards in this model \n(global test of Schoenfeld residuals, P=0.052). However, this was driven by the cubic spline terms for \nage, for the primary exposure SGTF there was no evidence of non-proportional hazards (P=0.17). \nThe first sensitivity analysis is restricted to people with a minimum of 14-days from testing positive \nfor SARS-CoV-2 to the follow-up censor. In the second sensitivity analysis missing data on ethnicity \nhas been imputed using multiple imputation. In the final sensitivity analysis the outcome is defined \nby admission to hospital following AE attendance.\n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \n\n10 | Page \n \nFigure 3 Absolute risk of AE attendance following diagnosis of Omicron or Delta stratified by age, vaccination, and comorbidity status \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint \n\n11 | Page \n \n1. SARS-CoV-2 variants of concern and variants under investigation in England. Technical \nbriefing 38. 11th March 2022. UK Health Security Agency (UKHSA). . \nhttps://www.gov.uk/government/publications/investigation-of-sars-cov-2-variants-technical-\nbriefings2022. \n2. Grint DJ, Wing K, Houlihan C, Gibbs HP, Evans SJW, Williamson E, et al. Severity of Severe \nAcute Respiratory System Coronavirus 2 (SARS-CoV-2) Alpha Variant (B.1.1.7) in England. Clinical \nInfectious Diseases. 2021. \n3. Grint DJ, Wing K, Williamson E, McDonald HI, Bhaskaran K, Evans D, et al. Case fatality risk of \nthe SARS-CoV-2 variant of concern B.1.1.7 in England, 16 November to 5 February. Eurosurveillance. \n2021;26(11):2100256. \n4. Cox DR. Regression methods and life tables. JR Stat Soc. 1972;34:187-220. \n5. Andrews N, Stowe J, Kirsebom F, Toffa S, Rickeard T, Gallagher E, et al. Covid-19 Vaccine \nEffectiveness against the Omicron (B.1.1.529) Variant. New England Journal of Medicine. 2022. \n6. Lauring AS, Tenforde MW, Chappell JD, Gaglani M, Ginde AA, McNeal T, et al. Clinical \nseverity of, and effectiveness of mRNA vaccines against, covid-19 from omicron, delta, and alpha \nSARS-CoV-2 variants in the United States: prospective observational study. BMJ. 2022;376:e069761. \n7. Altarawneh HN, Chemaitelly H, Hasan MR, Ayoub HH, Qassim S, AlMukdad S, et al. \nProtection against the Omicron Variant from Previous SARS-CoV-2 Infection. New England Journal of \nMedicine. 2022. \n8. Cele S, Jackson L, Khoury DS, Khan K, Moyo-Gwete T, Tegally H, et al. Omicron extensively \nbut incompletely escapes Pfizer BNT162b2 neutralization. Nature. 2022;602(7898):654-6. \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted May 3, 2022. ; https://doi.org/10.1101/2022.05.03.22274602doi: medRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}