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: Bronner P. Gonçalves, Peter Horby, Gail Carson, Piero L. Olliaro, Valeria Balan, Barbara Wanjiru Citarella, and research costs were supported by the UK Foreign, Commonwealth and Development Office (FCDO) and Wellcome [215091/Z/18/Z, 222410/Z/21/Z, 225288/Z/22/Z]; and Janice Caoili and Madiha Hashmi were supported by the UK FCDO and Wellcome [222048/Z/20/Z]. Peter Horby, Gail Carson, Piero L. Olliaro, Kalynn Kennon and Joaquin Baruch were supported by the Bill & Melinda Gates Foundation [OPP1209135]; Laura Merson was supported by University of Oxford's COVID-19 Research Response Fund - with thanks to its donors for their philanthropic support. Matthew Hall was supported by a Li Ka Shing Foundation award to Christophe Fraser. Moritz U.G. Kraemer was supported by 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. Contributions from Srinivas Murthy, Asgar Rishu, Rob Fowler, James Joshua Douglas, François Martin Carrier were supported by CIHR Coronavirus Rapid Research Funding Opportunity OV2170359 and coordinated out of Sunnybrook Research Institute. Contributions from Evert-Jan Wils and David S.Y. Ong were supported by a grant from foundation Bevordering Onderzoek Franciscus; and Andrea Angheben by the Italian Ministry of Health "Fondi Ricerca corrente-L1P6" to IRCCS Ospedale Sacro Cuore-Don Calabria. The data contributions of J.Kenneth Baillie, Malcolm G. Semple, and Ewen M. Harrison were 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. All funders of the ISARIC Clinical Characterisation Group are listed in the appendix.

Introduction

The emergence of novel SARS-CoV-2 variants represents a threat to the long-term control of COVID-19 (Fontanet et al., 2021). Whilst efforts to develop vaccines that protect against severe disease have been successful (Polack et al., 2020; Voysey et al., 2021; Baden et al., 2021), mutations in the viral genome that lead to ability to escape immunity, and increased transmissibility and/or clinical severity, either via intrinsic virulence or reduced vaccine effectiveness (Lopez Bernal et al., 2021), have the potential to cause substantial disease burden despite high vaccine coverage in many countries (Our World in Data, 2021).

These concerns motivated the prompt reporting, initially from South Africa (Wolter et al., 2021; World Health Organization, 2022), of clinical characteristics of infection with the Omicron variant only weeks after its emergence (Wolter et al., 2022; Ulloa et al., 2022; Veneti et al., 2022), which provided key information for risk assessment and health policies worldwide. Early data from South Africa showed reduced severity of Omicron lineage BA.1 and similar results were reported in the United Kingdom and the United States (Wolter et al., 2022; Lewnard et al., 2022; Nyberg et al., 2022). However, the impact, in terms of clinical consequences (i.e. disease severity), of new variants has been shown to be context-specific, due to regional differences in disease epidemiology, including local circulation of previous variants and their cumulative incidences, variable vaccination coverages, and heterogeneity in population-level frequencies of risk factors (e.g. frequency of comorbidities) for severe disease and mortality. For this reason, international studies with standardised protocols are necessary to allow comparative assessments across different countries and epidemiological contexts.

To understand the impact of the emergence of the Omicron variant of SARS-CoV-2 on the clinical epidemiology of COVID-19 at the global level, in this study, we report multi-country data, from all six World Health Organization regions, on clinical characteristics and outcomes of Omicron variant infections in hospitalised patients and compare with infections in patients admitted with other SARS-CoV-2 variants. For that, we use publicly available population-level data on relative frequencies of the Omicron variant to determine periods when infections were likely to be caused by Omicron versus other variants/lineages and compare infections descriptively and using multivariable statistical models. In addition, we present an analysis that only includes patients with individual-level data on the infecting variant and paired clinical information.

Methods

ISARIC clinical characterisation protocol

Analyses presented in this manuscript use the ISARIC (International Severe Acute Respiratory and Emerging Infections Consortium) COVID-19 database, which includes prospectively collected data from countries where ISARIC partner institutions are located (see a global map of all ISARIC partner institutions here https://isaric.org/about-us/membership/). A full description of the data collection protocol and database can be found here https://isaric.org/research/covid-19-clinical-research-resources/. In short, data collection for this initiative was standardised, using the ISARIC case report forms, and pivoted into pandemic mode in January 2020 to enable rapid characterisation of the clinical presentation and severity of COVID-19. After the emergence of the Omicron variant, first reported in November 2021 (Viana et al., 2022), a call was launched to encourage international investigators partnering with ISARIC to rapidly share data on patients with confirmed or suspected COVID-19 to describe the clinical characteristics of Omicron variant infection in different settings; recommendations on possible hospitalised population sampling approaches were shared. Patients admitted to hospital from 1st October 2021 to 28th February 2022 were included in this analysis. More information on ISARIC can be found in ISARIC Clinical Characterisation Group, 2021; Hall et al., 2021; Reyes et al., 2022.

Population-level SARS-CoV-2 variant data

Two statistical analysis plans (SAPs) were developed in December 2021 with approaches to be used in the characterisation of Omicron variant infection. Analyses described in the first SAP required individual-level data on the clinical presentation and paired data on the variant causing the infection. In the second SAP, we used population-level frequencies of SARS-CoV-2 lineages to infer individual infecting variant during different time periods as Omicron or non-Omicron variants (Figure 1). Since individual-level data on the infecting variant were limited to a few countries, these data are presented for comparison with the analysis performed using population-level variant data.

Figure 1.

Population-level relative frequency of Omicron variant infections by country and time.

Here, data aggregated by epidemiological week and country were used to calculate the proportions of infections caused by the Omicron variant. For analyses reported in the Results section, two epidemiological periods were defined: the first corresponds to the two months before the Omicron variant reaches a threshold frequency of 10% (blue area in each panel; the pre-Omicron period); the second period corresponds to the two months after Omicron variant frequency reaches 90% (red area in each panel; the Omicron period). Sensitivity analyses, using other relative frequencies for defining periods, are presented in the Appendix 1. Each panel presents data for a country (ISO3 code as title) contributing clinical data for this analysis; y-axes represent proportions in each epidemiological week (x-axes). Data for Laos are not shown as, at the time of the analysis, samples were not included in the database that informed population-level frequency of Omicron variant during the study period. In Pakistan, due to fluctuations in Omicron variant frequency in the dataset, study periods were not defined. More information on the spread of the Omicron variant in Laos and analysis of the clinical data from Pakistan are presented in the Appendix 1.

For the analysis that required information on population-level variant frequency, for countries contributing clinical data to this analysis, data from the Global Initiative on Sharing All Influenza Data (GISAID) on each of the main SARS-CoV-2 variants were collated. These data were aggregated by sample collection date and variant using a computational pipeline available here: https://github.com/globaldothealth/covid19-variants-summary, (Dasgupta and Kraemer, 2022, copy archived at swh:1:rev:8adf2f756b182711ad1d0b8707c44d3703786d23). The GISAID data were downloaded on 11 April 2022; Pango lineage designation v1.2.133 was used (pango-designation, 2022). We used these data to define calendar time periods when the Omicron variant represented the majority of infections in each country, and also periods during which the Omicron variant represented only a small (<10%) fraction of infections. For each country, the period during which infections were assumed to be caused by other variants ended in the epidemiological week before the Omicron variant relative frequency crossed a low threshold percentage (e.g. 10%) (see Figure 1). The first epidemiological week when Omicron variant frequency, as a proportion of all circulating variants, was higher than a given threshold percentage (90% in analyses presented in the Results section and 80% in sensitivity analyses) was used as the start date of the period during which all admissions were considered to be caused by the Omicron variant. Note (i) that amongst different countries these two study periods started in different calendar weeks, depending on when the Omicron variant was introduced to the location and on the rate of its local spread, and (ii) that in this analysis all Omicron sub-lineages are included (e.g. BA.1.1, BA.2). Only patients admitted to hospital in the two months before country-level Omicron variant frequency reached the lower threshold and those admitted in the first two months after Omicron variant relative frequency reached 90% were included in the primary analysis; the reason for restricting the study population to those admitted during these time windows was to reduce confounding by unmeasured factors whose frequencies in the hospitalised population also changed over time and which might be associated with clinical outcomes.

Statistical analysis

We report the frequencies of symptoms, comorbidities and vaccination status stratified by country and time periods (before and after Omicron emergence). We also assessed the case fatality risk and the frequency of a composite outcome that combined death and invasive mechanical ventilation use during the two study periods; in this analysis, patients who were discharged from hospital before the end of the follow-up period used in the definition of the outcome (14 or 28 days) were assumed to have been alive at the end of that period. When estimating risk of death by day 14 after admission or onset of symptoms, whichever happened later, numerators were numbers of patients who died before or on day 14 after admission; denominators in this calculation included those who died by day 14, those discharged at any time during follow-up, and those who were followed at least for 2 weeks, regardless of final outcome, including those who died after 14 days. The same approach was used to analyse the 28 day fatality risk. Note that for 35.5% of patients admitted to hospital during the two study periods defined by Omicron variant frequency, date of onset of symptoms was missing; for these patients we assumed onset of clinical disease happened before admission – that is that these were not hospital acquired infections. Furthermore, for 7.2% of patients, outcome date (date of death or discharge or latest date with follow-up information) was missing and 0.4% had an outcome date that was earlier than date of admission or of symptoms onset; except for those who were discharged and had missing outcome date, these two groups of patients were not included in analyses on the frequencies of clinical outcomes but were included in analyses describing distributions of symptoms and comorbidities. As described in the Results section, some patients included in this study were admitted for treatment of a medical condition other than COVID-19 but tested positive incidentally during hospitalisation.

We used mixed-effects logistic regression models to assess the association between study period, that is periods defined by the Omicron variant frequency at the population level, and 14-day death risk, adjusting for age, sex, and vaccination status. Age was included with the following categories: patients younger than 18 years, aged between 18 and 60 years, and older than 60 years. Random intercepts were used to account for potential variation in the risk of death between study sites in different countries. We also present models that adjust for the most commonly reported comorbidities; for each comorbidity included in the analysis, a binary variable was used to indicate presence or absence of the condition. Cox proportional hazards models on time to death, adjusted for age and sex and stratified by country and previous vaccination, were also fit; results of survival analyses are shown in the Appendix 1. Note that vaccination status was used as a binary variable in these models, without dose counts or timing of vaccination, and due to limited information on dates of doses we did not adjust for time since the most recent vaccination.

R and Python were used for data processing and descriptive analyses (R Development Core Team, 2022; The pandas development team, 2020). Code used for analyses and aggregated data used to generate figures are available (ISARIC Data Platform, 2022) (see also Data availability statement). Stata 17 was used to fit mixed-effects logistic models and perform survival analysis.

Results

Description of study population and study periods

Overall, 129,196 records from patients admitted to hospital between 1st October 2021 and 28th February 2022 were included in this analysis. Clinical centres in 30 countries contributed data (median 53 observations per country, interquartile range [IQR] 18–162); 11 countries contributed data on more than 100 hospitalised clinical cases (Appendix 1—table 1). A total of 54.0% and 42.6% of records were from South Africa and the United Kingdom, respectively. Appendix 1—table 2 and Appendix 1—table 3 show information on missing data for both symptoms and comorbidities.

Appendix 1—table 1.

Numbers of records contributed by partner institutions in different countries between 01/10/2021 and 28/02/2022.

Country	Number of records
South Africa	69766
United Kingdom	55049
Pakistan	929
Canada	919
Nepal	504
Laos	456
India	409
Romania	166
Saudi Arabia	151
Spain	151
Netherlands	134
Malaysia	90
Norway	67
Turkey	57
Brazil	54
Colombia	52
New Zealand	46
Kuwait	35
United States	32
Philippines	26
Ghana	21
Ireland	20
Israel	17
Italy	12
Estonia	7
Australia	7
Indonesia	6
Portugal	5
Germany	4
Argentina	4

Appendix 1—table 2.

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 table.

Symptoms	Yes	No	Missing data
Any cough	20431	13726	25273
Fever	16045	19465	23920
Headache	3896	28398	27136
Confusion	5960	28548	24922
Seizures	570	33424	25436
Sore throat	2394	29353	27683
Runny nose	1639	30279	27512
Vomiting	6956	27734	24740
Wheezing	2042	31191	26197
Diarrhoea	4418	29989	25023
Chest pain	5488	28732	25210
Conjunctivitis	106	32606	26718
Myalgia	3686	28195	27549
Rash	476	32877	26077
Fatigue	11339	22150	25941
Ageusia	1682	28341	29407
Inability to walk	252	3797	55381
Anosmia	1393	29040	28997
Shortness of breath	20490	14030	24910
Lymphadenopathy	145	32795	26490

Appendix 1—table 3.

Missing data on comorbidities.

In this table, data from all countries are included.

Comorbidities	Yes	No	Missing data
Liver disease	1786	40992	86418
Diabetes	12956	68743	47497
Chronic cardiac disease	13546	73423	42227
Hypertension	32052	57401	39743
Current smoking	5090	26674	97432
COPD	9304	77794	42098
Active TB	1579	45731	81886
Asthma	8720	79175	41301
Chronic kidney disease	8441	78453	42302
Malignant neoplasm	5062	81465	42669
Dementia	4646	38530	86020
HIV	5925	79121	44150
Chronic neurological disorder	5615	37740	85841
Obesity	5723	45367	78106

In addition to the clinical data contributed by the collaborating centres, population-level variant frequency data were used to define time periods when most infections in a country were assumed to be caused by Omicron versus other lineages. As presented in Figure 1, different countries reached the threshold relative frequencies of 10 and 90% of infections being caused by the Omicron variant at different times. Similar plots are presented in Appendix 1—figure 1 for other threshold frequencies. In Appendix 1—table 4, we list limitations in the use of these data to define time periods when infections were more likely caused by Omicron versus previous variants.

Appendix 1—figure 1.

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%.

Appendix 1—table 4.

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.

Using information presented in Figure 1, 103,061 patients, from 28/30 countries, were admitted either in the two months before the Omicron variant represented 10% of infections at the country-level (N=22,921; henceforth, the pre-Omicron period) or in the two months after (N=80,140) the Omicron variant was responsible for at least 90% of the infections; for ease of reference, the latter period will be referred to as the Omicron period. Note that 12,085 patients were admitted during weeks between the end of the pre-Omicron period and the start of the Omicron period and are not included in analyses presented in the following subsections (Figure 2); and 12,560 records of patients admitted two months after Omicron variant represented 90% of infections were not analysed. All patients from South Africa, the United Kingdom and Malaysia were assumed to be SARS-CoV-2 positive, as this is one criterion for inclusion in their databases. Of the 2296 records from other countries, information on SARS-CoV-2 diagnostic testing was available for 1,999 observations; whilst patients with negative PCR test result (N=10) were excluded from the rest of the analysis, those with missing PCR data (N=297) were assumed positive (see Appendix 1—table 5 for distribution by country). Of note, clinical data from Laos were not included in comparative analyses as there was only limited evidence of increase in local Omicron variant relative frequency during the study period (additional information is provided in the Appendix 1). For Pakistan, population-level data available at the time of the analysis indicate increasing Omicron variant frequency during the study period, but the proportion of local infections caused by this variant fluctuated; analyses of clinical data from that country are described in the Appendix 1.

Figure 2.

Study flowchart.

In this figure, we present the numbers of observations included in analyses in the different subsections of the Results section.

Appendix 1—table 5.

Numbers of records in the pre-Omicron and Omicron periods by country.

	Omicron emergence
Country	Before 10%	After 90%	Total
South Africa	4180	51929	56109
United Kingdom	18124	26479	44603
Canada	61	763	824
Nepal	197	204	401
India	89	212	301
Netherlands	60	65	125
Saudi Arabia	2	121	123
Romania	1	100	101
Spain	21	56	77
Malaysia	42	11	53
Norway	5	45	50
Brazil	15	33	48
New Zealand	34	6	40
Colombia	26	5	31
Turkey	0	27	27
Philippines	16	5	21
United States of America	14	7	21
Kuwait	0	19	19
Ghana	4	15	19
Ireland	14	3	17
Israel	0	14	14
Australia	0	6	6
Portugal	3	2	5
Indonesia	1	4	5
Germany	2	2	4
Italy	3	0	3
Argentina	0	3	3
Estonia	1	0	1

The median (IQR) ages of patients during the pre-Omicron and Omicron periods were 62 (43 – 76) and 50 (30 – 72) years, respectively; however, country-specific medians suggest that the younger age of patients after Omicron variant emergence in the combined dataset is at least partially explained by an increase in the proportion of data contributed by South Africa, relative to the proportion of data contributed by other countries (Appendix 1—table 6). A total of 48.3% and 54.8% of patients admitted during these periods, respectively, were female. 5.2% and 9.1% of patients in the pre-Omicron and Omicron periods, respectively, had the date of disease onset after admission date. In some countries, information on whether COVID-19 was the main reason for hospitalisation was also collected: 70.1% (N=2248) and 69.0% (N=27,804) of patients during the pre-Omicron and Omicron periods respectively were admitted to hospital due to COVID-19; patients for whom this information was available were primarily from South Africa (94.9%). There was no consistent pattern of within-country changes related to this variable (Appendix 1—table 7). Of note, 465/36,761 (1.3%) individuals reported a history of previous SARS-CoV-2 infection before the acute episode leading to hospitalisation included in this analysis (128/15,563 [0.8%] and 337/21,198 [1.6%] in the pre-Omicron and Omicron periods, respectively).

Appendix 1—table 6.

Medians (interquartile ranges [Q1 - Q3]) of age by study period and country.

Only countries with 10 or more observations in both study periods are shown.

	Before 10%			After 90%
Country	Median	Q1	Q3	Median	Q1	Q3
Brazil	59	50	70	55	48	70
Canada	63	50	71	62	35	76
Spain	68	63	75	76	59	84
United Kingdom	66	48	78	67	38	81
India	63	47	72	70	60	76
Malaysia	63	52	68	59	55	63
Netherlands	74	64	80	70	55	77
Nepal	63	42	77	64	42	75
South Africa	45	30	62	41	27	63

Appendix 1—table 7.

Numbers of hospitalised patients admitted due to COVID-19.

For country-time period combinations with less than 10 observations, numbers are not presented.

		Before 10%			After 90%
Country	COVID-19 as reason (N)	COVID-19 as reason (%)	Total	COVID-19 as reason (N)	COVID-19 as reason (%)	Total
Australia				-	-	<10
Argentina				-	-	<10
Brazil	14	100	14	32	96.7	33
Canada	12	52.2	23	514	67.5	761
Colombia	2	7.7	26	-	-	<10
Germany	-	-	<10	-	-	<10
Ghana	-	-	<10
India	0	0	12	2	8.3	24
Indonesia				-	-	<10
Israel				8	66.7	12
Kuwait				0	0	18
Malaysia	-	-	<10	-	-	<10
Nepal	-	-	<10	0	0	15
Netherlands	49	81.7	60	39	61.9	63
New Zealand	30	90.9	33	-	-	<10
Norway	-	-	<10	34	75.5	45
Philippines	16	100	16	-	-	<10
Romania	-	-	<10	100	100	100
Saudi Arabia	-	-	<10	68	68.7	99
South Africa	1433	71.1	2015	18306	69.0	26512
Spain	11	64.7	17	37	66.1	56
Turkey				27	100	27
USA	0	0	11	-	-	<10

Temporal changes in frequencies of symptoms and comorbidities

Figure 3 shows age distributions of hospitalised patients before versus after Omicron variant emergence; only countries with at least 50 observations in each period are included. Despite similar medians of age in the two periods within countries, in some, but not all, country-specific datasets, an increase in the proportion of the study population from younger ages was observed, although the number of patients in some age categories is small. Furthermore, there were differences between countries with regard to age distribution of cases, which could reflect either epidemiological differences between settings or else differences in recruitment of patients for this analysis.

Figure 3.

Age distributions by study period and country.

Age distributions (x-axes show proportions; y-axes, age groups) when Omicron variant relative frequency was below 10% (blue bars) and when the frequency was 90% or higher (red bars). Data from different countries are shown in different panels; only countries with 50 or more records in each period are presented. Numbers of observations with age information are shown for each study period next to country names.

The frequencies of the five most commonly reported symptoms and comorbidities in the combined (all countries) dataset during the two study periods are presented in Figure 4A and B, by country and study period. When analysing the combined dataset, there was a decrease in the percentage of patients with at least one of the comorbidities listed in Appendix 1—table 3 before versus during Omicron variant dominance (78.9% [N=15,574] and 59.6% [N=60,625], respectively); however, country-specific data show variable patterns (Appendix 1—table 8). With a total of 14 comorbidities being considered, median (IQR) numbers of comorbidity variables with non-missing information in the pre-Omicron and Omicron periods were 11 (0–12) and 9 (1 – 11), respectively. Whilst the directions of changes (increase or decrease) in frequencies of comorbidities were not consistent across countries, for many symptoms frequencies were lower during the Omicron period versus the pre-Omicron period. As can be seen in Appendix 1—figure 2, this pattern was consistent after stratifying frequencies of symptoms by age groups. The percentage of patients during the pre-Omicron period with at least one of the symptoms in Appendix 1—table 2 was 96.6% (N=11,683); this percentage was 88.6% (N=17,859) during the Omicron period (see Appendix 1—table 9 for country-specific numbers). These numbers refer to records from countries other than South Africa, where data on symptoms were not systematically available. The median (IQR) numbers of variables with non-missing data on symptoms were 14 (0–19) and 17 (0–19) for the pre-Omicron and Omicron periods, respectively.

Figure 4.

Frequencies of symptoms and comorbidities by study period and country.

Frequencies of the five most common symptoms (A) and comorbidities (B) during the pre-Omicron (blue bars) and Omicron (red bars) periods. 95% confidence intervals are shown. Note that South Africa is included in panel B but not panel A. For panel (A), only data from the pre-Omicron period were used to identify the most frequent symptoms; for panel (B), as data on comorbidities were available in the two countries contributing most records, the United Kingdom and South Africa, and since their relative contributions to the study population changed in the two study periods, the dataset including both the pre-Omicron and Omicron periods was used to identify most common comorbidities. Only countries with at least 50 observations during each study period are included. For each symptom or comorbidity, whenever fewer than five observations without missing data were available, bars were not shown and the text ‘NS’ (not shown) was included.

Appendix 1—table 8.

Percentages of patients with at least one comorbidity by country and study period.

Only countries with at least 10 patients in each study period are included.

	Before 10%		After 90%
Country	% with one or more comorbidities	Total	% with one or more comorbidities	Total
Brazil	78.6	14	81.8	33
Canada	76.7	60	74.2	760
India	44.9	89	56.9	209
Malaysia	64.3	42	72.7	11
Nepal	46.2	197	55.4	204
Netherlands	86.7	60	78.5	65
South Africa	53.9	3170	44.7	37412
Spain	76.2	21	76.8	56
United Kingdom	86.5	11820	84.6	21501

Appendix 1—figure 2.

Frequencies 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 are shown; and (B) shows the same information for individuals older than 60 years. Data from children are not presented.

Appendix 1—table 9.

Percentages of patients with at least one symptom by country and study period.

Only countries with at least 10 patients in each study period are included.

	Before 10%		After 90%
Country	% with one or more symptoms	Total	% with one or more symptoms	Total
Brazil	100.0	14	100.0	32
Canada	91.7	60	91.6	754
India	28.1	89	18.6	210
Malaysia	64.3	42	90.9	11
Nepal	97.0	197	86.3	204
Netherlands	96.7	60	96.9	65
Spain	100.0	21	87.5	56
United Kingdom	97.4	11104	89.3	16157

Vaccination history in hospitalised patients

Data on vaccination status were available for 42,850/103,051 hospitalised patients (8,952 during the pre-Omicron period and 33,898 during the Omicron period). In Table 1, we present vaccination status for study participants in each of the two periods by country. As expected, there is considerable inter-country variation in the frequency of vaccination. Age-stratified vaccination frequencies are shown in Appendix 1—figure 3 and suggest increases in frequency of previous vaccination during the period after Omicron variant emergence. However, as shown in Appendix 1—figure 4, with population-level vaccination coverage from before Omicron variant emergence up to the end of February 2022, in many countries contributing data to this study there was an increase in vaccination coverage over time, including in the periods during and after the emergence of the Omicron variant. Note that 55.8% of vaccinated patients received two or more doses before hospital admission.

Table 1.

Vaccination status by country and study period.

Data for period-country combinations with less than 10 observations are not presented. Data on vaccination status were not available for patients from Saudi Arabia.

	pre-Omicron period		Omicron period
Country	% Vaccinated	Total N	% Vaccinated	Total N
Brazil	84.6	13	87.9	33
Canada	32.2	59	57.3	686
Colombia	42.1	19	-	<10
Estonia	-	<10
Germany	-	<10	-	<10
India	34.8	23	84.8	33
Malaysia	79.3	29	80.0	10
Nepal	25.3	190	39.3	183
Netherlands	60.0	60	51.0	51
New Zealand	5.9	34	-	<10
Norway	-	<10	82.2	45
Philippines	78.6	14	-	<10
Portugal	-	<10	-	<10
Romania	-	<10	78.6	98
South Africa	15.1	1605	27.9	24752
Spain	45.0	20	70.9	55
United Kingdom	65.4	6865	70.3	7846
United States of America	-	<10	-	<10
Argentina			-	<10
Australia			-	<10
Indonesia			-	<10
Israel			54.5	11
Kuwait			66.7	18
Turkey			74.1	27

Appendix 1—figure 3.

Frequency of previous vaccination by study period, age category and country.

Only data from countries with at least 50 observations with information on previous vaccination 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.

Appendix 1—figure 4.

Population-level vaccination coverage.

Data from different countries are presented in different panels; x-axes show epidemiological weeks since the first epidemiological week of 2020. As in Figure 1, continuous black lines represent frequency of Omicron variant relative to the other variants. In addition to information on Omicron variant frequency, each panel also shows data on vaccination: the dashed line shows the proportion of population vaccinated with at least one dose relative to the maximum number vaccinated in each country at the time of the analysis (March 2022). Data used to generate this figure were downloaded from https://ourworldindata.org/.

Clinical outcomes

Overall, 11,314 patients admitted during the two study periods died during hospitalisation: 8517/94,524 by day 14 after hospital admission or disease onset, whichever occurred latest, and 10,530/94,461 by day 28; 738 patients died after day 28 and 46 patients who died did not have an outcome date recorded. As explained in the Methods section, denominators for fatality risks included patients who were discharged or still in hospital by day 14 or 28. Median (IQR) times to death were 10 (5 – 17) and 6 (3 – 13) days for the periods before and after Omicron emergence, respectively; similar information, on time from admission or symptoms onset to death, stratified by country is shown in Appendix 1—table 10. In some countries (see Figure 5 for comparisons on 14-day fatality risk, and Appendix 1—figure 5 for comparisons using the 28-day period), during the Omicron period, a lower proportion of patients died during hospitalisation, compared to the period before Omicron emergence; in India, the opposite pattern was observed although numbers for that country were limited.

Appendix 1—table 10.

Medians (interquartile ranges [Q1 - Q3]) of time from admission or disease onset to death by study period and country.

Only countries with 10 or more observations in both study periods are presented.

	pre-Omicron period				Omicron period
Country	Median	Q1	Q3		Median	Q1	Q3
Canada	10	6	21		10	5	18
United Kingdom	11	6	19		11	6	19
India	6	3	8		7	3	12
Nepal	6	5	12		4	2	8
South Africa	6	2	12		5	2	10

Figure 5.

Risk of death (y-axes) in the first 14 days after hospital admission or disease onset, whichever occurred latest, during the pre-Omicron and Omicron periods.

In each panel, the x-axis shows countries (ISO3 codes are presented), with different periods represented by circles with different colours (blue circles for the pre-Omicron period; red circles, for the Omicron period). 95% confidence intervals are also 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 50 observations in both study periods are included in the figure; for panels presenting age-specific estimates (bottom row), a further requirement for inclusion was outcome data for at least 10 patients in the corresponding age range in both periods.

Appendix 1—figure 5.

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 50 observations in both study periods are included in the figure; for panels presenting age-specific estimates (bottom row), a further requirement for inclusion was outcome data for at least 10 patients in the corresponding age range in both periods.

In a mixed-effects logistic model on 14-day fatality risk that adjusted for sex, age categories, and vaccination status, hospitalisations during the Omicron period were associated with lower risk of death (see Table 2). The inclusion of common comorbidities in the model did not change the estimated association. Similar results were obtained when using 28-day fatality risk as the outcome. We repeated the 14-day fatality risk analysis excluding patients who reported being admitted to hospital due to a medical condition other than COVID-19; the estimated odds ratio for the association between study period and the outcome was similar to those reported in Table 2. In an additional sensitivity analysis, estimates from a model that only included data from countries with at least 50 records per study period were also similar (OR 0.65, 95% CI 0.61–0.69, adjusted for covariates included in model I, Table 2). Survival analysis was also performed, and similar results were obtained (Appendix 1—table 11).

Table 2.

Odds ratio for the association between study period and mortality outcome.

Results of multivariate logistic models, with random intercepts for countries, on 14-day fatality risk are presented. Different models were fit that included different variables. Model III adjusts for all variables in the table, however due to missing data in the vaccination and comorbidity variables, less than a third of the study population was included in the estimation of that model; models I and II were thus fit that did not adjust for these variables and included more individuals. In model IV, a category for missing data was created for the variable on previous vaccination; individuals in that category had an odds ratio of 0.74 (0.69–0.80; reference group in this comparison is the non-vaccinated group). Note that similar results were obtained when finer categorisation of the age variable, 10-year intervals, was used. As previous SARS-CoV-2 infection has been shown to reduce severity of COVID-19 (Altarawneh et al., 2022), a multivariable model that also adjusted for this variable was fit; in that model, the odds ratio for the association between study period and fatality risk was 0.70 (0.61–0.80). As in other epidemiological studies, estimates for covariates other than the primary exposure (study period) should be carefully interpreted (Westreich and Greenland, 2013).

Model	I	II	III	IV
Number of observations	94,077	39,950	26,728	56,329
	Odds ratio (95% CI)	Odds ratio (95% CI)	Odds ratio (95% CI)	Odds ratio (95% CI)
Variables
Omicron period*	0.65 (0.62–0.69)	0.67 (0.61–0.75)	0.68 (0.60–0.77)	0.64 (0.59–0.69)
Sex (male)	1.32 (1.26–1.38)	1.33 (1.23–1.43)	1.36 (1.24–1.49)	1.33 (1.25–1.42)
Age
Older than 60 years	Reference	Reference	Reference	Reference
Aged between 18 and 60 years	0.26 (0.25–0.27)	0.24 (0.22–0.26)	0.27 (0.25–0.30)	0.30 (0.27–0.32)
Younger than 18 years	0.06 (0.05–0.07)	0.06 (0.05–0.07)	0.07 (0.05–0.09)	0.06 (0.05–0.08)
Previous vaccination	-	0.60 (0.55–0.65)	0.53 (0.48–0.59)	0.59 (0.54–0.65)
Comorbidities
Hypertension	-	-	1.29 (1.16–1.42)	1.26 (1.17–1.35)
Diabetes	-	-	1.22 (1.09–1.38)	1.22 (1.12–1.32)
Chronic cardiac disease	-	-	1.50 (1.31–1.71)	1.51 (1.39–1.65)

Odds ratio in univariate analysis 0.65 (0.61–0.69) (N=94,524).

Appendix 1—table 11.

Survival models.

Results of a 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, are shown in the Hazard ratio column. 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. We also fit a competing risk model, with hospital discharge as competing event; estimates from this model are presented in the Subhazard ratio column. In this model, previous COVID-19 vaccination was included as a covariate (subhazard ratio 0.55, 95% CI 0.52–0.59). We also fit a competing risk model using only data from the six countries included in Figures 3—5 and that included country as a dummy variable; in this model, the subhazard ratio for the Omicron period variable was 0.68 95% CI (0.63–0.74).

		Hazard ratio	Subhazard ratio
Variables
	Omicron period	0.77 (0.71–0.84)	0.79 (0.73–0.84)
	Sex (male)	1.24 (1.17–1.32)	1.32 (1.24–1.40)
	Age
	Older than 60 years	Reference
	Aged between 18 and 60 years	0.41 (0.38–0.44)	0.26 (0.24–0.28)
	Younger than 18 years	0.13 (0.11–0.17)	0.06 (0.04–0.07)

In addition to using fatality risk in our analyses, we also considered the composite outcome of death or invasive mechanical ventilation (IMV). Data on IMV were available in 74,563 records. Of 74,563 patients, 3111 required IMV during hospitalisation; the date when IMV was initiated was reported for 1070/3111 patients. Of those patients with data on IMV, 10,049/67,383 patients either died or required IMV. Appendix 1—figure 6 shows proportions of patients with this outcome by country and study period. Since date of IMV initiation was only available for 1070/3111 records, we do not present graphs by time since admission date.

Appendix 1—figure 6.

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 the 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 50 observations in both study periods are included in the figure; for panels presenting age-specific estimates (bottom row), a further requirement for inclusion was outcome data for at least 10 patients in the corresponding age range in both periods. Different from Figure 5 and Appendix 1—figure 5, 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.

Comparison with individual-level variant data

Whilst our approach of using population-level variant composition information allowed inclusion in this analysis of data from settings where it was not feasible to systematically identify the infecting SARS-CoV-2 variant, the use of aggregated data to infer the infecting variant has limitations, including the possibility of misclassification (see Appendix 1—table 4 for a list of limitations of this approach). To assess whether patterns described in previous subsections are generally consistent with analyses using individual-level variant information, we repeated comparisons for countries where information on the infecting variant was collected; data on variant were available for 1275 records. Of these, 852 patients were admitted either during the pre-Omicron period or the Omicron period: whilst only 1.9% (16/827) of those admitted during the Omicron period were infected by a variant other than Omicron, 4.0% (1/25) of patients during the pre-Omicron period had Omicron as the causative virus variant; for the calculation of these percentages data from a participating institution that prioritised contributing Omicron variant cases were not included. Except for six clinical cases in South Africa and Saudi Arabia, all infections were caused either by Delta or Omicron variants, and for this reason only data on these two variants are presented (Appendix 1—table 12). Figures similar to Figures 3—5 but stratified by infecting variant, rather than study period, are shown in the Appendix 1 (Appendix 1—figures 7–10). The numbers of participants included in the latter comparisons are lower than the numbers included in the comparisons using population-level variant data; for countries with ten or more observations of both Omicron and Delta variants, the patterns observed are broadly consistent with results obtained using the population-level approach.

Appendix 1—table 12.

Distribution of infections with individual-level variant information by country and variant.

Only countries with at least 10 observations for Delta and Omicron variants are listed. Note that other countries had limited numbers for both or one of the two variants.

Country	Delta	Omicron
Canada	26	303
Netherlands	12	52
Norway	15	22
South Africa	17	720
Spain	10	16

Appendix 1—figure 7.

This figure shows similar information to that presented in Figure 3.

The legend of that figure applies to this figure, except that instead of referring to time periods, the figure shows data for Delta and Omicron variants. Only countries with at least 10 observations for Delta and Omicron variants are included.

Appendix 1—figure 10.

This figure shows similar information to that presented in Figure 5.

The legend of that figure applies to this figure, except that instead of referring to time periods, the figure shows data for Delta and Omicron variants. Only countries with at least 10 observations for Delta and Omicron variants are included. Age-stratified panels are not shown due to the limited number of observations with individual-level variant data.

We also performed sensitivity analyses using different population-level threshold frequencies for the Omicron variant (10% and 80%, rather than 10% and 90%); these are shown in Appendix 1—figures 11–14 and are consistent with findings described in the Results section.

Appendix 1—figure 11.

This figure shows similar information to that presented in Figure 3.

The legend of that figure applies to this figure. Here, the upper threshold frequency used to define Omicron variant dominance was 80% rather than 90%.

Appendix 1—figure 14.

This figure shows similar information to that presented in Figure 5.

The legend of that figure applies to this figure. Here, the upper threshold frequency used to define Omicron variant dominance was 80% rather than 90%.

Discussion

When new variants of SARS-CoV-2 emerge during the COVID-19 pandemic, several critical questions are asked by public health authorities as to differences in disease severity and risk factors, and vaccine protection. Here, we leveraged data from multiple sources, from population-level variant frequency information to individual-level data on the clinical journey of hospitalised patients with COVID-19, and from multiple countries, to compare characteristics of patients with infection during periods before Omicron emergence versus when this variant became locally dominant. We observed that when the relative frequency of the Omicron variant was high, the proportions of patients with some of the most common COVID-19 symptoms were lower compared to the pre-Omicron period. In most but not all countries, patients presenting to hospital during the Omicron period had better outcomes (lower fatality risk), compared to those hospitalised before Omicron emergence, which could be related to lower variant virulence, prior immunity or residual confounding. In summary, our approach, which was consistent with analyses that used individual-level variant data from a subset of the study population, suggests clinical differences in patients hospitalised with the Omicron variant versus those admitted before this variant spread, and these differences vary by country.

Our finding that mortality was generally lower during the period when the Omicron variant was dominant is consistent with data from South Africa reported earlier this year (Wolter et al., 2022). In that study, which included more than 30,000 patients with individual-level information on the infecting variant, individuals infected with the Omicron variant had a lower risk of disease progression that required hospital admission than individuals infected with other variants; amongst hospitalised patients, the odds ratio for the association between Omicron variant infection and severe disease was 0.7 (95% confidence interval [CI] 0.3–1.4), which is similar to that observed in this study using death as the outcome. A lower risk of death in Omicron variant-infected versus Delta variant-infected patients was also observed in a recent study in the United Kingdom, although that analysis did not assess risk of death conditional on hospitalisation but rather on infection (Nyberg et al., 2022). In our analyses, statistical models were adjusted for vaccination history, which is a potential confounder of the association between dominant variant period and risk of death. However, the simplistic approach of using vaccination as a binary variable may be subject to residual confounding by time since vaccination, number of doses, or vaccine type. Moreover, as part of the effort to characterise Omicron variant infection, information on whether COVID-19 was the main reason for hospitalisation was collected during the study period and suggests that for a non-negligible proportion of patients other clinical conditions might have prompted hospital admission. All these factors might have contributed to the observed association, possibly to different degrees in different countries, reason for which this result should not be assumed to necessarily relate to the differences in variant virulence previously suggested by mechanistic studies (Shuai et al., 2022; Halfmann et al., 2022). Of note, data from India (see Figure 5) suggest slightly higher fatality risk during the Omicron period compared to the pre-Omicron period for patients older than 60 years, which could be potentially explained by confounding unrelated to age, residual age-related confounding, not controlled by the categorisation used in our analysis, or alternatively by the limited sample size and consequent uncertainty.

During the period of Omicron variant dominance, fewer patients presented with the symptoms most commonly reported earlier. For example, we observed in the United Kingdom that shortness of breath was present in about three-quarters of patients before Omicron variant emergence and in about half of patients during the Omicron period. Notably, a similar pattern was observed in Nepal, where patients were more often recruited from critical care settings. One possible explanation for this finding would be if incidental SARS-CoV-2 infections, that is infections that were not the primary reason for hospitalisation, were more frequent during the Omicron period; the high transmissibility of this variant, and the consequent peaks in numbers of infections, together with its reported association with lower severity, provides support for this hypothesis. However, in the subset of patients with data on the reason for hospitalisation there was no increase in the proportion of admissions thought to be incidental infections and indeed proportions in both study periods were consistent with frequencies of incidental infections in recent studies in the United States (Klann et al., 2022) and the Netherlands (Voor In ’t Holt et al., 2022), although in the latter, non-incidental infections included patients for whom COVID-19 was a contributing but not the main cause of hospitalisation. An alternative and less plausible explanation for the lower frequency of symptoms during the Omicron period would be that some of these patients developed symptoms other than those presented here, and which are severe enough to prompt hospital admission. Finally, it is also possible that the question on the primary reason for hospitalisation might have been interpreted differently in different countries and even in different hospitals in the same country, which would complicate its use in identifying incidental infections.

We also observed that history of COVID-19 vaccination was more frequent during the Omicron period, although for most countries the number of patients with vaccination information was limited, especially after stratification by age. Whilst this pattern would be expected if current vaccines were less effective against the Omicron variant compared to previously circulating variants, as suggested by a recent study in England analysing symptomatic disease (Andrews et al., 2022a), there were changes in vaccination coverage in many settings during the second half of 2021 and early 2022, including in response to the reports of Omicron variant cases. Since non-COVID-19 patients (e.g., patients with respiratory infections caused by other pathogens) were not systematically recruited for this multi-country study, it is not possible to estimate vaccine effectiveness during the two study periods and assess its change (Andrews et al., 2022b).

The major strength of our study relates to inclusion of data from all WHO geographic regions, collected with standardised forms, with over 100,000 records. However, we note that 96.6% of patients were from two countries - South Africa and the United Kingdom - and that the relative contributions of these countries to the study data were different in the two study periods (Appendix 1—table 5); to avoid misinterpretations linked to changes in country-specific contributions to data in the pre-Omicron and Omicron periods, we present descriptive analyses by country and use statistical models that adjust for country-level variation. It is also important to consider the relative contributions of these countries when interpreting descriptive analyses that refer to the combined dataset. Other limitations of our study relate, as mentioned in Appendix 1—table 4, to the use of population-level variant data to define periods when infections were likely caused by Omicron variant. For example, if infection by Omicron variant is associated with lower severity and if samples used to inform population-level frequency were often from community cases, then these aggregated data might not represent variant frequency in the hospitalised population. Another weakness of our study is that recruitment procedure was not standardised and was defined locally. Whilst this likely affected the generalisability of our descriptive estimates (fatality risk and frequencies of symptoms and comorbidities) to local populations of hospitalised COVID-19 cases (Lash et al., 2021; Rothman et al., 2013), it might not have affected the association between study period and fatality risk, at least not beyond the well-described potential for collider bias in hospital-based studies on COVID-19 outcomes (Griffith et al., 2020). Finally, missing information on symptoms for patients from South Africa prevented our descriptive analysis of changes in clinical presentation in an African setting. However, despite potential weaknesses in this approach, our results are consistent with reports from South Africa and elsewhere (Wolter et al., 2022), and individual-level variant data available for this study population often matched the two study periods defined by Omicron variant frequency.

In conclusion, we believe our approach of comparing changes in clinical characteristics of COVID-19 using multi-country standardised data, especially when combined with smaller scale studies that collect individual-level data on infecting variants for validation, will be useful in understanding the impact of new variants in the future. Another application will be in using routinely collected health data for cross-country comparisons of variant characteristics. Equally importantly, the successful conduct of this study, and the lessons learned, including the potential weaknesses discussed above, shows that multi-country efforts to study emerging SARS-CoV-2 variants are feasible, improvable and can generate insights to inform policy decision making.

This manuscript compares COVID-19 mortality during the pre-Omicron and Omicron emergence periods in several countries. It finds evidence suggesting the Omicron variant was associated with lower mortality than previous dominant variants in multiple countries, though other factors than changing variant virulence might explain these observations, as discussed by the authors. This paper will be of interest to infectious disease scientists both for its content and its methods, as it validates that population-level variant frequency can be a good proxy for individual-level variant data to derive insights on variant biology with population data.

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "An international observational study to assess the impact of the Omicron variant emergence on the clinical epidemiology of COVID-19 in hospitalised patients" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Matthew Whitaker (Reviewer #3).

As is customary in eLife, the reviewers have discussed their critiques with one another. What follows below is the Reviewing Editor's edited compilation of the essential and ancillary points provided by reviewers in their critiques and in their interaction post-review. Please submit a revised version that addresses these concerns directly. Although we expect that you will address these comments in your response letter, we also need to see the corresponding revision clearly marked in the text of the manuscript. Some of the reviewers' comments may seem to be simple queries or challenges that do not prompt revisions to the text. Please keep in mind, however, that readers may have the same perspective as the reviewers. Therefore, it is essential that you attempt to amend or expand the text to clarify the narrative accordingly.

Essential revisions:

1. Please address the various requests for clarifications brought up by reviewers #1 and #2, which will help readers better understand the methods of the paper, especially in regard to inclusion/exclusion criteria and recruitment of patients.

2. Please consider performing a sensitivity analysis restricted to countries with large numbers of cases as suggested by reviewer #2 to assess whether results were impacted by the inclusion of countries with few cases; it is likely that the cases from these countries are highly selected and therefore that their inclusion may have led to a selection bias.

Reviewer #1 (Recommendations for the authors):

• Figure 2: it is not clear to me why there were 12,665 records excluded from before or after the study periods, as to my mind these would have fit the definition of pre- and post- Omicron. Please explain.

• Table S7: please add percentages of patients hospitalized for COVID as well as numbers as these are easier to compare between countries and periods.

• Table S11: This analysis shows the results from a standard Cox model where discharge is treated as a censored observation. However, discharge is informative censoring, as presumably, the risk of death is much lower in a discharged patient. The authors may want to consider performing a competing risk analysis (Fine & Grey model) instead of a Cox model for this sensitivity analysis. The results would be more comparable to the logistic regressions, where discharges are included in the denominators but not the numerators.

• Figure S4: it is not clear how vaccination coverage is defined here (1 dose? 2 doses? 3 doses?)

• The authors may want to comment on the results from individual countries where mortality increased in the Omicron era (ex. India, Brazil, Spain), and the factors that might explain this. Presumably, these results may be due to some confounding by age and clinical profile of patients in both eras.

Reviewer #2 (Recommendations for the authors):

This is a great attempt for collaborative work, and we do know and appreciate how difficult it is to set up such a framework.

However, when reading the paper, it looks as if the project is not mature yet. Patients' enrolment per country is extremely disproportional and discredits the project. Can you just include the countries that sent at least 100 patients's data and name all countries as part of the consortium? countries that have sent very few forms should not be mentioned.

Patients' data sources seem to differ a lot per country, at the end it looks like a patchwork of random data/patients that is being analyzed to see if it could be of any use.

To make the paper clearer I would suggest that you:

1. Clarify the goals:

If the goal is to assist policy makers when a new variant emerges it will always be too late as variants are not introduced in different countries on the same calendar week. For Omicron (a bad example) policy makers reacted within 24h or less after the first announcement of this new variant.

Such collaborative programme could be very useful to identify unique variant characteristics compared to previous variants and identify specific country patterns linked to different variant exposure, different vaccination level, comorbidity rates, access to care etc…such global initiative doesn't exist yet.

2. Clarify which patients are included in the data base. All COVID-19 patients hospitalized, only the ones with COVID-19 as main reason for hospitalization, critical care patients, all patients with respiratory symptoms? How are patients with COVID-19 as the main reason for hospitalization?

3. Standardize the data so that the analysis is not done on different populations for each criteria. Currently symptoms analysis seems to mostly reflect UK patients, vaccination status mostly South Africa patients.

4. Again I would keep the top 8 or 10 countries, the small number of data provided by the other countries introduces confusion and alter the impact of your paper.

Other questions brought up by the reviewer to be addressed by authors (editor's note: originally these questions were in the public review but have been moved to the comments to authors to be addressed):

• The study is presented as a multi-center international study that includes more than 100,000 patients from 30 countries, however, 96.6% of the study patients originated from 2 countries, South Africa (54%) and the United Kingdom (42.6%). Can this still qualify as a multicenter study?

• Country specific medians suggest that the younger age of patients after the Omicron variant experience in the combined dataset is at least partially explained by an increase of data contributed by South Africa. Could the high proportion of South African patients' data have impacted on other measures too?

• Are the patients recruited COVID-19 proven patients? Are incidental cases included or excluded? How was the primary reason for hospitalization identified? Interpretation of mortality rate could be influenced by the recruitment of patients.

Reviewer #3 (Recommendations for the authors):

I only have one specific suggestion for further analysis:

In the Results section on vaccination history, it is implied that more granular detail on vaccination status of patients (ie number of vaccines received) is available but is not used in the models. If this is the case, it might be interesting to see a version of the main model adjusted on number of vaccine doses, even if this is restricted to countries where this data is available.

Essential revisions:

1. Please address the various requests for clarifications brought up by reviewers #1 and #2, which will help readers better understand the methods of the paper, especially in regard to inclusion/exclusion criteria and recruitment of patients.

To address comments from reviewers, we made several changes in the manuscript that clarified methods used and explained implications for the interpretation of results (see answers to comments from reviewers #1 and #2).

2. Please consider performing a sensitivity analysis restricted to countries with large numbers of cases as suggested by reviewer #2 to assess whether results were impacted by the inclusion of countries with few cases; it is likely that the cases from these countries are highly selected and therefore that their inclusion may have led to a selection bias.

In our response to comments from reviewer #2, we have modified the main figures of the manuscript, and only present data from countries with at least 50 observations in each study period. Furthermore, we now report in the Results section a sensitivity analysis using mixed-effects logistic regression that only included data from countries meeting this criterion (see answer to reviewer #2 for estimates from this analysis).

Reviewer #1 (Recommendations for the authors):

• Figure 2: it is not clear to me why there were 12,665 records excluded from before or after the study periods, as to my mind these would have fit the definition of pre- and post- Omicron. Please explain.

The reason why records of patients admitted to hospital more than two months before or after country-level frequencies of the Omicron variant reached the thresholds used in the definition of the study periods were excluded was to improve comparability between patients who were admitted during the pre-Omicron period versus those admitted during the Omicron period. For example, patients admitted to hospital several months before the emergence of the Omicron variant most likely differed from patients hospitalised with the Omicron variant not only with regard to the infecting variant but also unmeasured (potential) confounders (e.g. time since last vaccination dose or since previous infection). In epidemiological terms, our objective in restricting the analytical population to relatively narrow time windows was to increase the plausibility of the exchangeability assumption (Greenland and Robins, Identifiability, Exchangeability, and Epidemiological Confounding. International Journal of Epidemiology 1986; VanderWeele, Rothman, Lash, Confounding and Confounders. in Modern Epidemiology. 4th Edition, 2021). We have now included the following statement in the Methods section to explain why data from these patients were not analysed (here and throughout this document, changes in the manuscript are underlined):

“For each country, the period during which infections were assumed to be caused by other variants ended in the epidemiological week before the Omicron variant relative frequency crossed a low threshold percentage (e.g., 10%) (see Figure 1). The first epidemiological week when Omicron variant frequency, as a proportion of all circulating variants, was higher than a given threshold percentage (90% in analyses presented in the Results section and 80% in sensitivity analyses) was used as the start date of the period during which all admissions were considered to be caused by the Omicron variant. Note (i) that amongst different countries these two study periods started in different calendar weeks, depending on when the Omicron variant was introduced to the location and on the rate of its local spread, and (ii) that in this analysis all Omicron sub-lineages are included (e.g., BA.1.1, BA.2). Only patients admitted to hospital in the two months before country-level Omicron variant frequency reached the lower threshold and those admitted in the first two months after Omicron variant relative frequency reached 90% were included in the primary analysis; the reason for restricting the study population to those admitted during these time windows was to reduce confounding by unmeasured factors whose frequencies in the hospitalised population also changed over time and which might be associated with clinical outcomes.”

Based on this comment, we also fit, as a sensitivity analysis, a logistic regression model that included the 12,665 records that were excluded from the primary analysis and obtained similar estimates of the association between study period and fatality risk.

• Table S7: please add percentages of patients hospitalized for COVID as well as numbers as these are easier to compare between countries and periods.

We have now included two columns that show percentages for the two study periods.

• Table S11: This analysis shows the results from a standard Cox model where discharge is treated as a censored observation. However, discharge is informative censoring, as presumably, the risk of death is much lower in a discharged patient. The authors may want to consider performing a competing risk analysis (Fine & Grey model) instead of a Cox model for this sensitivity analysis. The results would be more comparable to the logistic regressions, where discharges are included in the denominators but not the numerators.

We would like to thank the reviewer for this suggestion. We agree that a competing risk analysis might be appropriate in this context. We now present results of a competing risk model, fit using the stcrreg command in Stata (https://www.stata.com/manuals/ststcrreg.pdf) (see below). Results of the two different approaches are presented (Latouche et al. A competing risks analysis should report results on all cause-specific hazards and cumulative incidence functions. Journal of Clinical Epidemiology 2013).

“Table S11. Survival models. Results of a 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, are shown in the Hazard ratio column. 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. We also fit a competing risk model, with hospital discharge as competing event; estimates from this model are presented in the Subhazard ratio column. In this model, previous COVID-19 vaccination was included as a covariate (subhazard ratio 0.55, 95% CI 0.52 – 0.59). We also fit a competing risk model using only data from the six countries included in Figures 3 – 5 and that included country as a dummy variable; in this model, the subhazard ratio for the Omicron period variable was 0.68 95% CI (0.63 – 0.74).

• Figure S4: it is not clear how vaccination coverage is defined here (1 dose? 2 doses? 3 doses?)

In Figure S4, vaccination coverage refers to any vaccination, i.e. at least one dose. Below is the updated legend of the figure:

“In addition to information on Omicron variant frequency, each panel also shows data on vaccination: the dashed line shows the proportion of population vaccinated with at least one dose relative to the maximum number vaccinated in each country at the time of the analysis (March 2022).”

• The authors may want to comment on the results from individual countries where mortality increased in the Omicron era (ex. India, Brazil, Spain), and the factors that might explain this. Presumably, these results may be due to some confounding by age and clinical profile of patients in both eras.

We have now modified the following paragraph in the Discussion section to comment on these results:

“… All these factors might have contributed to the observed association, possibly to different degrees in different countries, reason for which this result should not be assumed to necessarily relate to differences in variant virulence. Of note, data from India (see Figure 5) suggest slightly higher fatality risk during the Omicron period compared to the pre-Omicron period for patients older than 60 years, which could be potentially explained by confounding unrelated to age, residual age-related confounding, not controlled by the categorisation used in our analysis, or alternatively by the limited sample size and consequent uncertainty.”

Note that to address a comment from reviewer #2, regarding presenting data from countries with relatively small study sample sizes, estimates for the Brazilian and Spanish study populations were removed from Figure 5.

Reviewer #2 (Recommendations for the authors):

This is a great attempt for collaborative work, and we do know and appreciate how difficult it is to set up such a framework.

However, when reading the paper, it looks as if the project is not mature yet. Patients' enrolment per country is extremely disproportional and discredits the project. Can you just include the countries that sent at least 100 patient's data and name all countries as part of the consortium? countries that have sent very few forms should not be mentioned.

Patients' data sources seem to differ a lot per country, at the end it looks like a patchwork of random data/patients that is being analyzed to see if it could be of any use.

The reason why numbers of records contributed by different countries are so variable is that datasets from both the United Kingdom and South Africa are part of country-wide, geographically representative epidemiological efforts, whilst in some of the other participating countries, a limited number of ISARIC partner centres recruited patients. As mentioned in another comment by this reviewer (see below), a major strength of this research project is its international and collaborative character, and for this reason, it is important to describe the data contribution of all participating countries. Note that the number of records contributed by each partner institution could be influenced by a multitude of factors: local incidence of SARS-CoV-2 infection during the study period, logistical constraints (e.g. number of staff collecting data), and spread of the Omicron variant locally; the latter determined the distribution of records in the two study periods.

To address this comment, we modified Figures 3, 4 and 5 to include only countries with at least 50 observations in each study period (see updated figures below); figures in the Supplementary Appendix presenting country-level data were also updated. Furthermore, in our multivariate analysis of the association between study period and fatality risk, we performed a sensitivity analysis that included this same set of countries; the modified paragraph in the Results section is also presented below.

“In a mixed-effects logistic model on 14-day fatality risk that adjusted for sex, age categories, and vaccination status, hospitalisations during the Omicron period were associated with lower risk of death (see Table 2). The inclusion of common comorbidities in the model did not change the estimated association. Similar results were obtained when using 28-day fatality risk as the outcome. We repeated the 14-day fatality risk analysis excluding patients who reported being admitted to hospital due to a medical condition other than COVID-19; the estimated odds ratio for the association between study period and the outcome was similar to those reported in Table 2. In an additional sensitivity analysis, estimates from a model that only included data from countries with at least 50 records per study period were also similar (OR 0.65, 95% CI 0.61 – 0.69, adjusted for covariates included in model I, Table 2). Survival analysis was also performed, and similar results were obtained (Table S11).”

To make the paper clearer I would suggest that you:

1. Clarify the goals:

If the goal is to assist policy makers when a new variant emerges it will always be too late as variants are not introduced in different countries on the same calendar week. For Omicron (a bad example) policy makers reacted within 24h or less after the first announcement of this new variant.

Such collaborative programme could be very useful to identify unique variant characteristics compared to previous variants and identify specific country patterns linked to different variant exposure, different vaccination level, comorbidity rates, access to care etc…such global initiative doesn't exist yet.

We would like to thank this reviewer for the valuable comments and for highlighting the public health value of our work. Global collaborations are needed when responding to international public health threats, and we believe the work reported in this manuscript will motivate research groups to establish similar initiatives. We modified the final paragraph of the manuscript based on this comment:

“In conclusion, we believe our approach of comparing changes in clinical characteristics of COVID-19 using multi-country standardised data, especially when combined with smaller scale studies that collect individual-level data on infecting variants for validation, will be useful in understanding the impact of new variants in the future. Another application will be in using routinely collected health data for cross-country comparisons of variant characteristics. Equally importantly, the successful conduct of this study, and the lessons learned, including the potential weaknesses discussed above, shows that multi-country efforts to study emerging SARS-CoV-2 variants are feasible, improvable and can generate insights to inform policy decision making.”

2. Clarify which patients are included in the data base. All COVID-19 patients hospitalized, only the ones with COVID-19 as main reason for hospitalization, critical care patients, all patients with respiratory symptoms? How are patients with COVID-19 as the main reason for hospitalization?

These questions were addressed in the responses to other comments. Only hospitalised patients with SARS-CoV-2 infection were recruited, and local investigators were responsible for the recruitment procedure. We included the following sentences in the Discussion section:

“Another weakness of our study is that recruitment procedure was not standardised and was defined locally. Whilst this likely affected the generalisability of our descriptive estimates (fatality risk and frequencies of symptoms and comorbidities) to local populations of hospitalised COVID-19 cases (Lash and Rothman, Selection Bias and Generalizability. in Modern Epidemiology 4 Edition 2021; Rothman et al. Why representativeness should be avoided. International Journal of Epidemiology 2013), it might not have affected the association between study period and fatality risk, at least not beyond the well-described potential for collider bias in hospital-based studies on COVID-19 outcomes (Griffith et al. Collider bias undermines our understanding of COVID-19 disease risk and severity. Nature Communications 2020).”

3. Standardize the data so that the analysis is not done on different populations for each criteria. Currently symptoms analysis seems to mostly reflect UK patients, vaccination status mostly South Africa patients.

As mentioned in the Results and Discussion sections, most patients in this study were from the United Kingdom and South Africa. In our analyses, we accounted for this by presenting data stratified by country, including in the main figures of the manuscript. Our regression analysis does include data from all study countries, but variation in risk is accounted for by the statistical method used. We included a statement about this:

“The major strength of our study relates to inclusion of data from all WHO geographic regions, collected with standardised forms, with over 100,000 records. However we note that 96.6% of patients were from two countries – South Africa and the United Kingdom – and that the relative contributions of these countries to the study data were different in the two study periods (Table S5); to avoid misinterpretations linked to changes in country-specific contributions to data in the pre-Omicron and Omicron periods we present descriptive analyses by country and use statistical models that adjust for country-level variation. It is also important to consider the relative contributions of these countries when interpreting descriptive analyses that refer to the combined dataset.”

In addition to including this statement in the manuscript, we have also included a short sub-section in the Supplementary Appendix that discusses the frequency of symptoms after excluding data from the United Kingdom:

“Frequency of symptoms outside the United Kingdom and South Africa

Most, 82.5% (N = 579), patients admitted to hospital during the pre-Omicron period outside the United Kingdom and South Africa had at least one symptom; this percentage is lower than the frequency estimated including the United Kingdom data (96.6%), possibly due to the low frequency of symptoms in India (Table S9). The corresponding frequency during the Omicron period was 81.5% (N = 1702).”

4. Again I would keep the top 8 or 10 countries, the small number of data provided by the other countries introduces confusion and alter the impact of your paper.

To address this and another comment from the same reviewer, we have now modified the main figures of the manuscript to include only countries with at least 50 hospitalised patients in each study period. We have also included a sensitivity analysis that estimates odds ratio after excluding countries with limited study data (see above).

Other questions brought up by the reviewer to be addressed by authors (editor's note: originally these questions were in the public review but have been moved to the comments to authors to be addressed):

• The study is presented as a multi-center international study that includes more than 100,000 patients from 30 countries, however, 96.6% of the study patients originated from 2 countries, South Africa (54%) and the United Kingdom (42.6%). Can this still qualify as a multicenter study?

Response to comment

Eight countries contributed at least 100 records during the two study periods, and three additional countries recruited more than 100 patients but due to the timing of the local spread of the Omicron variant some of these patients were not included in the analysis. We believe this justifies referring to this study as an international, or multi-country, epidemiological study. It is also important to note that data from multiple countries, including from the United Kingdom and South Africa, were collected in multiple centres.

• Country specific medians suggest that the younger age of patients after the Omicron variant experience in the combined dataset is at least partially explained by an increase of data contributed by South Africa. Could the high proportion of South African patients' data have impacted on other measures too?

As there is considerable between-country variation in vaccination coverage, we did not report aggregated frequencies for the history of vaccination, and most of the statements in the Results sub-section Vaccination history in hospitalised patients refer to Figure S3 and Table 1, which present data stratified by country. For this reason, we believe that data from South Africa did not impact reporting of the vaccination outcome. Regarding the fatality risk, our multivariate analysis uses random intercepts that account for differences in risk between countries; furthermore, our analysis is adjusted for factors that might vary in distribution between countries, such as age and history of vaccination.

• Are the patients recruited COVID-19 proven patients? Are incidental cases included or excluded? How was the primary reason for hospitalization identified? Interpretation of mortality rate could be influenced by the recruitment of patients.

As described in the Results section, patients with negative PCR test result for SARS-CoV-2 detection were excluded from the analysis. The statement is the following:

“All patients from South Africa, the United Kingdom and Malaysia were assumed to be SARS-CoV-2 positive, as this is one criterion for inclusion in their databases. Of the 2,296 records from other countries, information on SARS-CoV-2 diagnostic testing was available for 1,999 observations; whilst patients with negative PCR test result (N=10) were excluded from the rest of the analysis, those with missing PCR data (N=297) were assumed positive (see Table S5 for distribution by country).”

Furthermore, information on the primary reason for hospitalisation was available for nearly 30,000 patients; ~70% reported that the reason for hospitalisation was COVID-19. In the Discussion section, we mention that incidental infections might have affected our findings, although the inclusion of patients with incidental infections in hospital-based COVID-19 epidemiological studies is common and not specific to our design. The following paragraph was modified:

“One possible explanation for this finding would be if incidental SARS-CoV-2 infections, i.e. infections that were not the primary reason for hospitalisation, were more frequent during the Omicron period; the high transmissibility of this variant, and the consequent peaks in numbers of infections, together with its reported association with lower severity, provides support for this hypothesis. However, in the subset of patients with data on the reason for hospitalisation there was no increase in the proportion of admissions thought to be incidental infections and indeed proportions in both study periods were consistent frequencies of incidental infections in recent studies in the United States (Klann et al. Distinguishing Admissions Specifically for COVID-19 From Incidental SARS-CoV-2 Admissions: National Retrospective Electronic Health Record Study. J Med Internet Res) and the Netherlands (Voor in ’t holt et al. Admissions to a large tertiary care hospital and Omicron BA.1 and BA.2 SARS-CoV-2 polymerase chain reaction positivity: primary, contributing, or incidental COVID-19. International Journal of Infectious Diseases 2022), although in the latter, non-incidental infections included patients for whom COVID-19 was a contributing but not the main cause of hospitalisation.”

Reviewer #3 (Recommendations for the authors):

I only have one specific suggestion for further analysis:

In the Results section on vaccination history, it is implied that more granular detail on vaccination status of patients (ie number of vaccines received) is available but is not used in the models. If this is the case, it might be interesting to see a version of the main model adjusted on number of vaccine doses, even if this is restricted to countries where this data is available.

We performed an analysis that adjusted for the number of vaccine doses received; 19,360 patients, 18.8% of the study population, were included. In the model, also adjusted for sex and age, the odds ratio quantifying the association between study period and fatality risk is 0.72 95% confidence interval (0.63 – 0.82).