An international observational study to assess the impact of the Omicron variant emergence on the clinical epidemiology of COVID-19 in hospitalised patients

Background: Whilst timely clinical characterisation of infections caused by novel SARS-CoV-2 variants is necessary for evidence-based policy response, individual-level data on infecting variants are typically only available for a minority of patients and settings. Methods: Here, we propose an innovative approach to study changes in COVID-19 hospital presentation and outcomes after the Omicron variant emergence using publicly available population-level data on variant relative frequency to infer SARS-CoV-2 variants likely responsible for clinical cases. We apply this method to data collected by a large international clinical consortium before and after the emergence of the Omicron variant in different countries. Results: Our analysis, that includes more than 100,000 patients from 28 countries, suggests that in many settings patients hospitalised with Omicron variant infection less often presented with commonly reported symptoms compared to patients infected with pre-Omicron variants. Patients with COVID-19 admitted to hospital after Omicron variant emergence had lower mortality compared to patients admitted during the period when Omicron variant was responsible for only a minority of infections (odds ratio in a mixed-effects logistic regression adjusted for likely confounders, 0.67 [95% confidence interval 0.61 - 0.75]). Qualitatively similar findings were observed in sensitivity analyses with different assumptions on population-level Omicron variant relative frequencies, and in analyses using available individual-level data on infecting variant for a subset of the study population. Conclusions: Although clinical studies with matching viral genomic information should remain a priority, our approach combining publicly available data on variant frequency and a multi-country clinical characterisation dataset with more than 100,000 records allowed analysis of data from a wide range of settings and novel insights on real-world heterogeneity of COVID-19 presentation and clinical outcome.

Funding: This works was made possible with the support of UK Foreign, Commonwealth and Development Office and Wellcome [215091/Z/18/Z and 225288/Z/22/Z] and the Bill & Melinda Gates Foundation [OPP1209135]; CIHR Coronavirus Rapid Research Funding Opportunity OV2170359 and was coordinated out of Sunnybrook Research Institute; Wellcome Trust fellowship [205228/Z/16/Z] and the National Institute for Health Research Health Protection Research Unit (HPRU) in Emerging and Zoonotic Infections (NIHR200907) at the University of Liverpool in partnership with Public Health England (PHE), in collaboration with Liverpool School of Tropical Medicine and the University of Oxford; Institute for Clinical Research (ICR), National Institutes of Health (NIH) supported by the Ministry of Health Malaysia; a grant from foundation Bevordering Onderzoek Franciscus; MUGK acknowledges funding from the Branco Weiss Fellowship, Google.org, the Oxford Martin School, the Rockefeller Foundation, and the European Union Horizon 2020 project MOOD (#874850). The contents of this publication are the sole responsibility of the authors and do not necessarily reflect the views of the European Commission.

The COVID-19 Clinical Information Network (CO-CIN) data was collated by ISARIC4C Investigators. Data provision was supported by grants from: the National Institute for Health Research (NIHR; award CO-CIN-01), the Medical Research Council (MRC; grant MC_PC_19059), and by the NIHR Health Protection Research Unit (HPRU) in Emerging and Zoonotic Infections at University of Liverpool in partnership with Public Health England (PHE), (award 200907), NIHR HPRU in Respiratory Infections at Imperial College London with PHE (award 200927), Liverpool Experimental Cancer Medicine Centre (grant C18616/A25153), NIHR Biomedical Research Centre at Imperial College London (award IS-BRC-1215-20013), and NIHR Clinical Research Network providing infrastructure support.

22 Makati Medical Center, Makati City, Philippines 48 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Introduction 136   137 The emergence of novel SARS-CoV-2 variants represents a threat to the long-term control of 138 COVID-19 1 . Whilst efforts to develop vaccines that protect against severe disease have been 139 successful 2-4 , mutations in the viral genome that lead to ability to escape immunity, and 140 increased transmissibility and/or clinical severity, either via intrinsic virulence or reduced 141 vaccine effectiveness 5 , have the potential to cause substantial disease burden despite high 142 vaccine coverage in many countries 6 . 143 144 These concerns motivated the prompt reporting, initially from South Africa 7,8 , of clinical 145 characteristics of infection with the Omicron variant only weeks after its emergence 9-11 , which 146 provided key information for risk assessment and health policies worldwide. Early data from 147 South Africa showed reduced severity of Omicron lineage BA.1 and similar results were 148 reported in the United Kingdom and the United States 9,12,13 . However, the impact, in terms of 149 clinical consequences (i.e., disease severity), of new variants has been shown to be context-150 specific, due to regional differences in disease epidemiology, including local circulation of 151 previous variants and their cumulative incidences, variable vaccination coverages, and 152 heterogeneity in population-level frequencies of risk factors (e.g. frequency of comorbidities) 153 for severe disease and mortality. For this reason, international studies with standardised 154 protocols are necessary to allow comparative assessments across different countries and 155 epidemiological contexts. 156

157
To understand the impact of the emergence of the Omicron variant of SARS-CoV-2 on the 158 clinical epidemiology of COVID-19 at the global level, in this study, we report multi-country 159 data, from all six World Health Organization regions, on clinical characteristics and outcomes 160 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 of Omicron variant infections in hospitalised patients and compare with infections in patients 161 admitted with other SARS-CoV-2 variants. For that, we use publicly available population-level 162 data on relative frequencies of the Omicron variant to determine periods when infections were 163 likely to be caused by Omicron versus other variants/lineages and compare infections 164 descriptively and using multivariable statistical models. In addition, we present an analysis that description of the data collection protocol and database can be found here 177 https://isaric.org/research/covid-19-clinical-research-resources/. In short, data collection for 178 this initiative was standardised, using the ISARIC case report forms, and pivoted into pandemic 179 mode in January 2020 to enable rapid characterisation of the clinical presentation and severity 180 of COVID-19. After the emergence of the Omicron variant, first reported in November 2021 14 , 181 a call was launched to encourage international investigators partnering with ISARIC to rapidly 182 share data on patients with confirmed or suspected COVID-19 to describe the clinical 183 characteristics of Omicron variant infection in different settings; recommendations on possible 184 hospitalised population sampling approaches were shared. Patients admitted to hospital from 185 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 1 st October 2021 to 28 th February 2022 were included in this analysis. More information on 186 ISARIC can be found in [15][16][17]  Omicron variants (Figure 1). Since individual-level data on the infecting variant were limited 196 to a few countries, these data are presented for comparison with the analysis performed using 197 population-level variant data. 198 199 For the analysis that required information on population-level variant frequency, for countries 200 contributing clinical data to this analysis, data from the Global Initiative on Sharing All 201 Influenza Data (GISAID) on each of the main SARS-CoV-2 variants were collated. These data 202 were aggregated by sample collection date and variant using a computational pipeline available 203 here: https://github.com/globaldothealth/covid19-variants-summary. The GISAID data were 204 downloaded on 11 April 2022; Pango lineage designation v1.2.133 was used 18 . We used these 205 data to define calendar time periods when the Omicron variant represented the majority of 206 infections in each country, and also periods during which the Omicron variant represented only 207 a small (<10%) fraction of infections. For each country, the period during which infections 208 were assumed to be caused by other variants ended in the epidemiological week before the 209 Omicron variant relative frequency crossed a low threshold percentage (e.g., 10%) (see Figure  210 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 1). The first epidemiological week when Omicron variant frequency, as a proportion of all 211 circulating variants, was higher than a given threshold percentage (90% in analyses presented 212 in the Results section and 80% in sensitivity analyses) was used as the start date of the period 213 during which all admissions were considered to be caused by the Omicron variant. Note (i) that 214 amongst different countries these two study periods started in different calendar weeks, 215 depending on when the Omicron variant was introduced to the location and on the rate of its 216 local spread, and (ii) that in this analysis all Omicron sub-lineages are included (e.g., BA. We report the frequencies of symptoms, comorbidities and vaccination status stratified by 222 country and time periods (before and after Omicron emergence). We also assessed the case 223 fatality risk and the frequency of a composite outcome that combined death and invasive 224 mechanical ventilation use during the two study periods; in this analysis, patients who were 225 discharged from hospital before the end of the follow-up period used in the definition of the 226 outcome (14 or 28 days) were assumed to have been alive at the end of that period. When 227 estimating risk of death by day 14 after admission or onset of symptoms, whichever happened 228 later, numerators were numbers of patients who died before or on day 14 after admission; 229 denominators in this calculation included those who died by day 14, those discharged at any 230 time during follow-up, and those who were followed at least for 2 weeks, regardless of final 231 outcome, including those who died after 14 days. The same approach was used to analyse the 232 28-day fatality risk. Note that for 35.5% of patients admitted to hospital during the two study 233 periods defined by Omicron variant frequency, date of onset of symptoms was missing; for 234 these patients we assumed onset of clinical disease happened before admission -i.e. that these 235 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 were not hospital acquired infections. Furthermore, for 7.2% of patients, outcome date (date of 236 death or discharge or latest date with follow-up information) was missing and 0.4% had an 237 outcome date that was earlier than date of admission or of symptoms onset; except for those 238 who were discharged and had missing outcome date, these two groups of patients were not 239 included in analyses on the frequencies of clinical outcomes but were included in analyses 240 describing distributions of symptoms and comorbidities. As described in the Results section, 241 some patients included in this study were admitted for treatment of a medical condition other 242 than COVID-19 but tested positive incidentally during hospitalisation. 243

244
We used mixed-effects logistic regression models to assess the association between study 245 period, i.e. periods defined by the Omicron variant frequency at the population level, and 14-246 day death risk, adjusting for age, sex, and vaccination status. Age was included with the 247 following categories: patients younger than 18 years, aged between 18 and 60 years, and older 248 than 60 years. Random intercepts were used to account for potential variation in the risk of 249 death between study sites in different countries. We also present models that adjust for the most 250 commonly reported comorbidities; for each comorbidity included in the analysis, a binary 251 variable was used to indicate presence or absence of the condition. Cox proportional hazards 252 models on time to death, adjusted for age and sex and stratified by country and previous 253 vaccination, were also fit; results of survival analyses are shown in the Supplementary 254 Appendix. Note that vaccination status was used as a binary variable in these models, without 255 dose counts or timing of vaccination, and due to limited information on dates of doses we did 256 not adjust for time since the most recent vaccination. 257 258 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 R and Python were used for data processing and descriptive analyses 19,20  In addition to the clinical data contributed by the collaborating centres, population-level variant 274 frequency data were used to define time periods when most infections in a country were 275 assumed to be caused by Omicron versus other lineages. As presented in Figure 1, different 276 countries reached the threshold relative frequencies of 10 and 90% of infections being caused 277 by the Omicron variant at different times. Similar plots are presented in Figures S1A and S1B 278 for other threshold frequencies. In Table S4, we list limitations in the use of these data to define 279 time periods when infections were more likely caused by Omicron versus previous variants. 280 281 Using information presented in Figure 1, 103,061 patients, from 28/30 countries, were 282 admitted either in the two months before the Omicron variant represented 10% of infections at 283 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 the country-level (N = 22,921;henceforth, or in the two months after 284 (N = 80,140) the Omicron variant was responsible for at least 90% of the infections; for ease 285 of reference, the latter period will be referred to as the Omicron period. Note that 12,085 286 patients were admitted during weeks between the end of the pre-Omicron period and the start 287 of the Omicron period and are not included in analyses presented in the following subsections 288 ( Figure 2); and 12,560 records of patients admitted two months after Omicron variant 289 represented 90% of infections were not analysed. All patients from South Africa, the United 290 Kingdom and Malaysia were assumed to be SARS-CoV-2 positive, as this is one criterion for 291 inclusion in their databases. Of the 2,296 records from other countries, information on SARS-292 CoV-2 diagnostic testing was available for 1,999 observations; whilst patients with negative 293 PCR test result (N=10) were excluded from the rest of the analysis, those with missing PCR 294 data (N=297) were assumed positive (see Table S5 for distribution by country). Of note, 295 clinical data from Laos were not included in comparative analyses as there was only limited 296 evidence of increase in local Omicron variant relative frequency during the study period 297 The median (IQR) ages of patients during the pre-Omicron and Omicron periods were 62 (43 303 -76) and 50 (30 -72) years, respectively; however, country-specific medians suggest that the 304 younger age of patients after Omicron variant emergence in the combined dataset is at least 305 partially explained by an increase in the proportion of data contributed by South Africa, relative 306 to the proportion of data contributed by other countries (Table S6). 48.3% and 54.8% of 307 patients admitted during these periods, respectively, were female. 5.2% and 9.1% of patients 308 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 in the pre-Omicron and Omicron periods, respectively, had the date of disease onset after 309 admission date. In some countries, information on whether COVID-19 was the main reason 310 for hospitalisation was also collected: 70.1% (N = 2,248) and 69.0% (N = 27,804) of patients 311 during the pre-Omicron and Omicron periods respectively were admitted to hospital due to 312 COVID-19; patients for whom this information was available were primarily from South 313 Africa (94.9%). There was no consistent pattern of within-country changes related to this 314 variable (  CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Temporal changes in frequencies of symptoms and comorbidities 344 345 Figure 3 shows age distributions of hospitalised patients before versus after Omicron variant 346 emergence; only countries with at least 10 observations in each period are included. Despite 347 similar medians of age in the two periods within countries, in some, but not all, country-specific 348 datasets, an increase in the proportion of the study population from younger ages was observed, 349 although the number of patients in some age categories is small. Furthermore, there were 350 differences between countries with regard to age distribution of cases, which could reflect 351 either epidemiological differences between settings or else differences in recruitment of 352 patients for this analysis. 353

354
The frequencies of the five most commonly reported symptoms and comorbidities in the 355 combined (all countries) dataset during the two study periods are presented in Figure 4A and 356 4B, by country and study period. When analysing the combined dataset, there was a decrease 357 in the percentage of patients with at least one of the comorbidities listed in Table S3  respectively); however country-specific data show variable patterns (Table S8). With a total 360 of 14 comorbidities being considered, median (IQR) numbers of comorbidity variables with 361 non-missing information in the pre-Omicron and Omicron periods were 11 (0 -12) and 9 (1 -362 11), respectively. Whilst the directions of changes (increase or decrease) in frequencies of 363 comorbidities were not consistent across countries, for many symptoms frequencies were lower 364 during the Omicron period versus the pre-Omicron period. As can be seen in Figure S2, this 365 pattern was consistent after stratifying frequencies of symptoms by age groups. The percentage 366 of patients during the pre-Omicron period with at least one of the symptoms in Table S2 was 367 96.6% (N = 11,683); this percentage was 88.6% (N = 17,859) during the Omicron period (see 368 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ; https://doi.org/10.1101/2022.06.22.22276764 doi: medRxiv preprint Table S9 for country-specific numbers). These numbers refer to records from countries other 369 than South Africa, where data on symptoms were not systematically available. The median 370 (IQR) numbers of variables with non-missing data on symptoms were 14 (0 -19) and 17 (0 -371 19) for the pre-Omicron and Omicron periods, respectively. 372 373 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022.  proportions; y-axes, age groups) when Omicron variant relative frequency was below 10 % 375 (blue bars) and when the frequency was 90% or higher (red bars). Data from different countries 376 are shown in different panels; only countries with 10 or more records in each period are 377 presented. Numbers of observations with age information are shown for each study period next 378 to country names. For all countries except Spain and Malaysia, x-axes range from 0 to 0. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Figure 4. Frequencies of symptoms and comorbidities by study period and country. 388
Frequencies of the five most common symptoms (A) and comorbidities (B) during the pre-389 Omicron (blue bars) and Omicron (red bars) periods. 95% confidence intervals are shown. 390 Note that South Africa is included in panel B but not panel A. For panel A, only data from the 391 pre-Omicron period were used to identify the most frequent symptoms; for panel B, as data on 392 comorbidities were available in the two countries contributing most records, the United 393 Kingdom and South Africa, and since their relative contributions to the study population 394 changed in the two study periods, the dataset including both the pre-Omicron and Omicron 395 periods was used to identify most common comorbidities. Only countries with at least 10 396 observations during each study period are included. For each symptom or comorbidity, 397 whenever fewer than 5 observations without missing data were available, bars were not shown 398 and the text "NS" (not shown) was included. 399 400 401 402 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. during the pre-Omicron period and 33,898 during the Omicron period). In Table 1, we present 418 vaccination status for study participants in each of the two periods by country. As expected, 419 there is considerable inter-country variation in the frequency of vaccination. Age-stratified 420 vaccination frequencies are shown in Figure S3 and suggest increases in frequency of previous 421 vaccination during the period after Omicron variant emergence. However, as shown in Figure   CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. times to death were 10 (5 -17) and 6 (3 -12) days for the periods before and after Omicron 446 emergence, respectively; similar information, on time from admission or symptoms onset to 447 death, stratified by country is shown in Table S10. In some countries (see Figure 5 for 448 comparisons on 14-day fatality risk, and Figure S5 for comparisons using the 28-day period), 449 during the Omicron period, a lower proportion of patients died during hospitalisation, 450 compared to the period before Omicron emergence; in India, the opposite pattern was observed 451 although numbers for that country were limited. 452

453
In a mixed-effects logistic model on 14-day fatality risk that adjusted for sex, age categories, 454 and vaccination status, hospitalisations during the Omicron period were associated with lower 455 risk of death (see Table 2). The inclusion of common comorbidities in the model did not change 456 the estimated association. Similar results were obtained when using 28-day fatality risk as the 457 outcome. We repeated the 14-day fatality risk analysis excluding patients who reported being 458 admitted to hospital due to a medical condition other than COVID-19; the estimated odds ratio 459 for the association between study period and the outcome was similar to those reported in Table  460 2. Cox proportional hazards models were also fit, and similar results were obtained (Table  461 S11). 462 463 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ; https://doi.org/10. 1101 In addition to using fatality risk in our analyses, we also considered the composite outcome of 464 death or invasive mechanical ventilation (IMV). Data on IMV were available in 74,563 records. 465 3,111/74,563 patients required IMV during hospitalisation; the date when IMV was initiated 466 was reported for 1,070/3,111 patients. Of those patients with data on IMV, 10,049/67,383 467 patients either died or required IMV. Figure S6 shows proportions of patients with this 468 outcome by country and study period. Since date of IMV initiation was only available for 469 1,070/3,111 records, we do not present graphs by time since admission date. 470 471 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

486
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Comparison with individual-level variant data 500 501 Whilst our approach of using population-level variant composition information allowed 502 inclusion in this analysis of data from settings where it was not feasible to systematically 503 identify the infecting SARS-CoV-2 variant, the use of aggregated data to infer the infecting 504 variant has limitations, including the possibility of misclassification (see Table S4   CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 We also performed sensitivity analyses using different population-level threshold frequencies 523 for the Omicron variant (10% and 80%, rather than 10% and 90%); these are shown in Figure  524 S8 and are consistent with findings described in the Results section. 525 526 527 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. COVID-19 symptoms were lower compared to the pre-Omicron period. In most but not all 538 countries, patients presenting to hospital during the Omicron period had better outcomes (lower 539 fatality risk), compared to those hospitalised before Omicron emergence, which could be 540 related to lower variant virulence, prior immunity or residual confounding. In summary, our 541 approach, which was consistent with analyses that used individual-level variant data from a 542 subset of the study population, suggest clinical differences in patients hospitalised with the 543 Omicron variant versus those admitted before this variant spread, and these differences vary 544 by country. 545 546 Our finding that mortality was generally lower during the period when the Omicron variant 547 was dominant is consistent with data from South Africa reported earlier this year 9 . In that study, 548 which included more than 30,000 patients with individual-level information on the infecting 549 variant, individuals infected with the Omicron variant had a lower risk of disease progression 550 that required hospital admission than individuals infected with other variants; amongst 551 hospitalised patients, the odds ratio for the association between Omicron variant infection and 552 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 severe disease was 0.7 (95% confidence interval [CI] 0.3 -1.4), which is similar to that 553 observed in this study using death as the outcome. A lower risk of death in Omicron variant-554 infected versus Delta variant-infected patients was also observed in a recent study in the United 555 Kingdom, although that analysis did not assess risk of death conditional on hospitalisation but 556 rather on infection 13 . In our analyses, statistical models were adjusted for vaccination history, 557 which is a potential confounder of the association between dominant variant period and risk of 558 death. However, the simplistic approach of using vaccination as a binary variable may be 559 infections that were not the primary reason for hospitalisation, were more frequent during the 574 Omicron period; the high transmissibility of this variant, and the consequent peaks in numbers 575 of infections, together with its reported association with lower severity, provides support for 576 this hypothesis. However, in the subset of patients with data on the reason for hospitalisation 577 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ; https://doi.org/10.1101/2022.06.22.22276764 doi: medRxiv preprint there was no decrease in the proportion of admissions thought to be directly caused by  19. An alternative and less plausible explanation would be that some of these patients 579 developed symptoms other than those presented here, and which are severe enough to prompt 580 hospital admission. Finally, it is also possible that the question on the primary reason for 581 hospitalisation might have been interpreted differently in different countries and even in 582 different hospitals in the same country, which would complicate its use in identifying incidental 583 infections. 584 585 We also observed that history of COVID-19 vaccination was more frequent during the Omicron 586 period. Whilst this would be expected if current vaccines were less effective against the 587 Omicron variant compared to previously circulating variants, as suggested by a recent study in 588 England analysing symptomatic disease 24 , there were changes in vaccination coverage in many 589 settings during the second half of 2021 and early 2022, including in response to the reports of 590 Omicron variant cases. Since non-COVID-19 patients (e.g., patients with respiratory infections 591 caused by other pathogens) were not systematically recruited for this multi-country study, it is 592 not possible to estimate vaccine effectiveness during the two study periods and assess its 593 change 25 . 594

595
The major strength of our study relates to inclusion of data from all WHO geographic regions, 596 collected with standardised forms, with over 100,000 records. However we note that 96.6% of 597 patients were from two countries -South Africa and the United Kingdom -and that the relative 598 contributions of these countries to the study data were different in the two study periods (Table  599 S5); to avoid misinterpretations linked to changes in country-specific contributions to data in 600 the pre-Omicron and Omicron periods we present descriptive analyses by country and use 601 statistical models that adjust for country-level variation. Other limitations of our study relate, 602 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. is associated with lower severity and if samples used to inform population-level frequency were 605 often from community cases, then these aggregated data might not represent variant frequency 606 in the hospitalised population. However, despite potential weaknesses in this approach, our 607 results are consistent with reports from South Africa and elsewhere 9 , and individual-level 608 variant data available for this study population often matched the two study periods defined by 609 Omicron variant frequency. 610

611
In conclusion, we believe our approach of comparing changes in clinical characteristics of 612 COVID-19 using multi-country standardised data, especially when combined with smaller 613 scale studies that collect individual-level data on infecting variants for validation, will be useful 614 in understanding the impact of new variants in the future. Another application will be in using CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ; https://doi.org/10. 1101 This work uses data provided by patients and collected by the NHS as part of their care and 628 support #DataSavesLives. The data used for this research were obtained from ISARIC4C. We 629 are extremely grateful to the 2648 frontline NHS clinical and research staff and volunteer 630 medical students who collected these data in challenging circumstances; and the generosity of 631 the patients and their families for their individual contributions in these difficult times. We also 632 acknowledge the support of Jeremy J Farrar and Nahoko Shindo. 633 634 635 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ; https://doi.org/10.1101/2022.06.22.22276764 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Manuscript
An international observational study to assess the impact of the Omicron variant emergence on the clinical epidemiology of COVID-19 in hospitalised patients

Conflict of Interest Declarations
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Acknowledgements (Partner institutions)
This work was supported by endorsement of the Irish Critical Care-Clinical Trials Group, coordinated  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Epidemiology of Omicron variant in Laos
Population-level variant data from Laos were not available in the Global Initiative on Sharing All Influenza Data (GISAID) platform that covered the period between October 2021 and

Clinical data from Pakistan
In Pakistan, there was an increase in the relative frequency of Omicron variant during the period from October 2021 to February 2022. However, despite causing 96.1% of infections in the GISAID data from the country in mid-January 2022, throughout February this percentage fluctuated. Data from Pakistan were thus not included in the Results section. Here, we discuss clinical data from this country; for that, we used as the start of the Omicron period the first week when this variant was responsible for more than 90% of infections, regardless of whether this percentage was lower in the following weeks.
Data from 929 patients from Pakistan were contributed to the study; 249 records were from the pre-Omicron period, and 478, from the Omicron period. The percentage of patients in the country with at least one symptom was 83.9% in the pre-Omicron period, and 57.9%, in the Omicron period. 52.2% and 59.2% had at least one comorbidity during these two periods, respectively. Vaccination data were available for 474 patients admitted during the study periods: 37.7% and 62.9% had history of COVID-19 vaccination during the pre-Omicron and Omicron periods. The 14-day fatality risk for hospitalised patients during the pre-Omicron period was 52.5%, and during the Omicron period, 45.4%.

Sensitivity analysis that excludes patients with other primary reason for hospitalisation
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 For 30,052 patients admitted during the two study periods, information was available on whether COVID-19 was the primary medical reason for hospitalisation; most of these patients were from South Africa. As a sensitivity analysis, we fit a mixed-effects logistic regression model on the 14-day fatality risk excluding patients who had reported that COVID-19 was not the reason for hospitalisation; patients for whom this information was missing were included.
The odds ratio for the association between study period and 14-day fatality risk was 0.68 (95% confidence interval 0.61 -0.75).
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Country
Number . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ; https://doi.org/10.1101/2022.06.22.22276764 doi: medRxiv preprint Table S2. Missing data on symptoms. Note that this information was not systematically recorded in South Africa, and for this reason data from that country are not included in this . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ; https://doi.org/10.1101/2022.06.22.22276764 doi: medRxiv preprint Table S4. Potential limitations of population-level variant data used to determine time periods when Omicron variant was dominant.

Potential limitation
Likely impact on analyses Population-level data come from a range of sources in each country, and for most samples it is not possible to determine whether patient was hospitalised or was a community (mild) case If different variants are associated with different severities upon infection and if a large fraction of samples used in the estimation of population-level frequency of variants are from community cases, then it is possible that this frequency does not fully represent the frequency in the hospitalised population. In particular, if Omicron variant infection is linked to lower risk of hospitalisation, as previous studies suggest, it is possible that even during periods when community-level frequency of Omicron variant was high, the frequency of Omicron variant in the hospitalised population might have been relatively low.
Use of country-level data, rather than data on variant frequency in the catchment areas of clinical centres contributing data If Omicron variant spreads asynchronously in a country, with some regions reaching high relative frequency faster than others, it is possible that country-level data, rather than data at a finer geographical level, might not reflect Omicron variant frequency in the population from which patients were recruited.

Delay between infection, onset of symptoms and hospitalisation
Depending on the data source used to define population-level frequency of variants, if clinical samples were obtained early during the infection, hospitalised cases might only have the same variant composition after a time lag, corresponding to average time from infection, or onset of symptoms, to hospital admission.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Table S11. Cox proportional hazards model, stratified by country, on time to death in the first 28 days since hospital admission or onset of symptoms, which happened latest. For this analysis, if follow-up duration was longer than 28 days, it was set to 28 days, and patients who were discharged were censored on the day of discharge. The assumption of proportional hazards was violated for the variable on previous vaccination; for this reason, the model was also stratified by this variable. An alternative analysis assumed that patients discharged from hospital were censored on day 28; in this analysis, the hazard ratio for the variable corresponding to study period was 0.68 (0.63 -0.74); for this model, the proportional hazards assumption did not hold for the study period variable.

Hazard ratio Variables
Omicron period 0. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Figure S1. In this figure, population-level variant data are presented for countries with clinical data included in our analysis. The same structure of Figure 1 was used but different cut-off frequencies for Omicron variant were applied: in A, the lower and upper threshold frequencies were 10% and 80%; in B, these frequencies were 5% and 90%.

Supplementary Figures
A . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 B . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Figure S2. Distributions of the five most common symptoms during the period before (blue bars) and after (red bars) Omicron variant frequency reached 10% and 90%, respectively. 95% confidence intervals are also shown. In A, data from individuals aged between 18 and 60 years; and B shows the same information for individuals older than 60 years. Data from children are not presented.
A . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 B . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Figure S3. Frequency of previous vaccination by study period, age category and country. Only data from countries with at least 10 observations during both study periods defined by Omicron variant frequency are shown. In each panel, the x-axis shows different age categories, with blue bars corresponding to the pre-Omicron period and red bars, to the period after Omicron variant frequency, relative to other variants, reaches 90%. Above each bar, the total number of records included in the calculation of the proportions (y-axes) are presented.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Data used to generate this figure were downloaded from https://ourworldindata.org/.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ; https://doi.org/10.1101/2022.06.22.22276764 doi: medRxiv preprint Figure S5. Risk of death in the first 28 days after hospital admission or disease onset, whichever occurred latest, during pre-Omicron and Omicron periods. In each panel, the x-axis shows countries, with different periods represented by circles with different colours (blue circles for the pre-Omicron period; red circles, for period after Omicron variant frequency reaches 90%). 95% confidence intervals are presented. The top panel shows data for individuals of all ages; the bottom panels, data for patients aged less than 18 years, between 18 and 60 years, and older than 60 years. Only countries with at least 10 observations in each study period and corresponding age group are included.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ; https://doi.org/10.1101/2022.06.22.22276764 doi: medRxiv preprint Figure S6. Risk of death or invasive mechanical ventilation by study period. In each panel, the x-axis shows countries, with different periods represented by circles with different colours (blue circles for pre-Omicron period; red circles, for the Omicron period). 95% confidence intervals are presented. The top panel shows data for individuals of all ages; the bottom panels, data for patients aged less than 18 years, between 18 and 60 years, and older than 60 years.
Only countries with at least 10 observations in each study period are included. Different from Figures 5 and S5, time since hospital admission or onset of symptoms was not used since for most patients who required invasive mechanical ventilation the start date of the therapeutic approach was not available. Only patients with information on invasive mechanical ventilation use and who were either discharged or died were included.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  Figure S3, panel D, Figure 5, panel E. The legends of those figures apply to the corresponding panels in this figure, except that instead of referring to time periods, the panels below show data for Delta and Omicron variants.
A . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)  Here the upper threshold frequency used to define Omicron variant dominance was 80% rather than 90%.
A . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Lim, W.S. declares his institution has received unrestricted investigator-initiated research funding from Pfizer for an unrelated multicentre cohort study in which he is the Chief . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Streinu-Cercel, Adrian has been an investigator in COVID-19 clinical trials by Algernon Pharmaceuticals, Atea Pharmaceuticals, Regeneron Pharmaceuticals, Diffusion Pharmaceuticals, and Celltrion, Inc., outside the scope of the submitted work.

Streinu-Cercel, Anca has been an investigator in COVID-19 clinical trials by Algernon
Pharmaceuticals, Atea Pharmaceuticals, Regeneron Pharmaceuticals, Diffusion Pharmaceuticals, Celltrion, Inc. and Atriva Therapeutics, outside the scope of the submitted work.
Summers, C. reports that she has received fees for consultancy for Abbvie and Roche relating to COVID-19 therapeutics. She was also the UK Chief Investigator of a GlaxoSmithKline plc sponsored study of a therapy for COVID, and is a member of the UK COVID Therapeutic Tedder, R. reports grants from MRC/UKRI during the conduct of the study. In addition, R.
Tedder has a patent United Kingdom Patent Application No. 2014047.1 "SARS-CoV-2 antibody detection assay" issued.
Turtle, L. reports grants from MRC/UKRI during the conduct of the study and fees from Eisai for delivering a lecture related to COVID-19 and cancer, paid to the University of Liverpool.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)