Hospital-based surveillance of COVID-19 in Switzerland
Coronaviruses (CoV) are a large family of RNA viruses including important human and veterinary pathogens (1). Four CoVs are continuously circulating in the human population which cause only mild to moderate respiratory illness. Three CoVs with a zoonotic origin emerged in the last two decades; they are responsible for more severe disease. The Severe Acute Respiratory Syndrome (SARS-CoV) emerged in 2002/2003 (2), the Middle Eastern Respiratory Syndrome (MERS-CoV) in 2012 and COVID-19 in 2019 (3). In late December 2019, COVID-19 was identified by next-generation sequencing. The virus was identified in a cluster of Chinese adult patients with pneumonia of unknown cause (4). Initially, most infected people had visited a large seafood and wet animal market in Wuhan City, Hubei Province, which was considered as the source of the outbreak. It became clear that person-to-person transmission via droplets (by coughing or sneezing or direct contact) is occuring when infections spread to medical workers and family members who had not visited the Wuhan market. As of February 4th 2020, COVID-19 has been reported in 24 countries, with 20,630 confirmed cases, the vast majority of them being reported in China (20,471 cases) (5). Even though to date, no confirmed case was reported in Switzerland, a few cases have been confirmed in neighboring countries: 12 in Germany, 6 in France, and 2 in Italy (5). Those include possible or confirmed transmission outside China (6). With the recent notion of transmission from mildly diseased patients, and infectious virus shedding from the upper respiratory tract, containment of this virus seems a challenging, if not impossible task (10).
On January 30th 2020, the World Health Organisation (WHO) declared the COVID-19 epidemic outbreak a Public Health Emergency of International Concern (PHEIC) WHO emphasized the urgent need to coordinate international efforts to investigate and better understand this novel coronavirus, to minimize the threat in affected countries and to reduce the risk of further international spread (7). WHO designed various study protocols for early investigation of the COVID-19 outbreak, as well as a global COVID-19 Anonymized Clinical Data Platform (the “nCoV Data Platform”) to enable State Parties to the International Health Regulations (IHR) (2005) to share with WHO anonymized clinical data and information related to patients with suspected or confirmed infections with the COVID-19 (collectively “Anonymized nCoV Data”) (8).
Since the November 15th, 2020 season, both the COVID-19 surveillance project and the Influenza surveillance project have been merged into one single registry, modified to account for both diseases and easily include other respiratory viruses if needed.
The main goal of this project is to offer an emergency surveillance tool to the Federal Office of Public Health (FOPH) in order to monitor the COVID-19 spread in Switzerland. As a side goal, we wish to test the surveillance system implemented for Influenza to identify whether it can be easily used for other infectious diseases and outbreaks.
It will allow us to collect quality and trustworthy data and will be regularly checked for inconsistencies.
Currently, 21 Swiss hospitals (including university hospitals and cantonal hospitals) are participating in this surveillance system:
- Centre Hospitalier Universitaire Vaudois (CHUV) in Lausanne - VD
- Ente Ospedaliero Cantonale (EOC) in Lugano - TI
- Hôpital du Valais (HVS) in Sion - VS
- Hôpitaux Universitaires de Genève (HUG) in Geneva - GE
- Inselspital Bern (INSEL) in Bern - BE
- Kinderspital Basel (UKBB) in Basel - BS
- Kinderspital Zürich (KISPI USZ) in Zürich - ZH
- Kantonsspital Sankt Gallen (KSSG) in Sankt Gallen - SG
- Universitätspital Basel (USB) in Basel - BS
- Universitätspital Zürich (USZ) in Zürich - ZH
- Kantonsspital Garubuenen (KSGR) in Chur - GR
- Kindersspital St. Gallen (OKS) in Sankt Gallen - SG
- Hôpital de Fribourg (HFR - paediatrics only) in Fribourg - FR
- Kantonsspital Aarau (KSA - paediatrics only) in Aarau - AG
- Kantonsspital Winterthur (KSW - paediatrics only) in Winterthur - ZH
- Spitaeler Schaffhausen (Spitaeler SH) in Schaffhausen - SH
- Luzerner Kantonsspital (LUKS) in Luzern - LU
- Kantonsspital Münsterlingen (STGAG) in Münsterlingen - TG
- Hirslanden AG Zürich (HAGZH) in Zürich - ZH
- Kantonsspital Niedwald (KSNW) in Stans - NW
- Hirslanden Clinic St Ana - Luzern in Luzern
1. World Health Organisation (WHO). Coronavirus [Internet]. [cited 2020 Feb 4]. Available from:https://www.who.int/westernpacific/health-topics/coronavirus
2. Luk HKH, Li X, Fung J, Lau SKP, Woo PCY. Molecular epidemiology, evolution and phylogeny of SARS coronavirus. Infection, Genetics and Evolution. 2019 Jul 1;71:21–30.
3. Middle East respiratory syndrome coronavirus (MERS-CoV): A review, GERMS - Enabling the future [Internet]. [cited 2020 Feb 4]. Available from:http://www.germs.ro/en/Articles/Middle-East-respiratory-syndrome-coronavirus-MERS-CoV-A-review-886
4. Zhu N, Zhang D, Wang W, Li X, Yang B, Song J, et al. A Novel Coronavirus from Patients with Pneumonia in China, 2019. New England Journal of Medicine [Internet]. 2020 Jan 24 [cited 2020 Feb 4]; Available from:https://www.nejm.org/doi/10.1056/NEJMoa2001017
5. World Health Organisation (WHO). nCoV 2019 situation (public) [Internet]. [cited 2020 Feb 4]. Available from:http://who.maps.arcgis.com/apps/opsdashboard/index.html#/c88e37cfc43b4ed3baf977d77e4a0667
6. World Health Organisation (WHO). Novel Coronavirus (2019-nCoV): Situation Report-14 [Internet]. 2020 [cited 2020 Feb 4]. Available from:https://www.who.int/docs/default-source/coronaviruse/situation-reports/20200203-sitrep-14-ncov.pdf?sfvrsn=f7347413_2
7. 2019-nCoV outbreak is an emergency of international concern [Internet]. 2020 [cited 2020 Feb 4]. Available from:http://www.euro.who.int/en/health-topics/emergencies/pages/news/news/2020/01/2019-ncov-outbreak-is-an-emergency-of-international-concern
8. World Health Organisation (WHO). Novel Coronavirus (2019-nCoV) technical guidance: Early investigations [Internet]. [cited 2020 Feb 4]. Available from:https://www.who.int/emergencies/diseases/novel-coronavirus-2019/technical-guidance/early-investigations
It is possible to access the data contained within the database with some restriction. The procedure is described in the following document:
In a nutshell: the applicant must fill a concept sheet detailing the team in charge of the analysis, the analysis plan, and the variables needed for the analysis. The applicant then must send it the Organising Committee (OC) who will forward it to the Scientific Committee (SC). After acceptance by the SC, the concept sheet will be circulated among participating hospital to decide if they want to opt-out of the analysis. Note that we only accept one project per concept-sheet.
Before submitting a concept sheet, please make sure that it is not a duplicate from one of the accepted projects (listed below) and that everything is complete. If you need any additional information, please reach one of the contact person listed on the right.
The CH-SUR study group member list can be found here. This is the official co-author list for each participating hospital.
Cohort Profile: SARS-CoV-2/COVID-19 hospitalised patients in Switzerland
Lead: Amaury Thiabaud (ISG) & Anne Iten (HUG)
Status: Published on Swiss Medical Weekly
Background. SARS-CoV-2/COVID-19, which emerged in China in late 2019, rapidly spread across the world causing several million victims in 213 countries. Switzerland was severely hit by the virus, with 43’000 confirmed cases as of September 1st, 2020.
Aim. In cooperation with the Federal Office of Public Health, we set up a surveillance database in February 2020 to monitor hospitalised patients with COVID-19 in addition to their mandatory reporting system.
Methods. Patients hospitalised for more than 24 hours with a positive PCR test, from 20 Swiss hospitals, are included. Data collection follows a custom Case Report Form based on WHO recommendations and adapted to local needs. Nosocomial infections were defined as infections for which the onset of symptoms started more than 5 days after the patient’s admission date.
Results. As of September 1st, 2020, 3645 patients were included. Most patients were male (2168 - 59.5%),and aged between 50 and 89 years (2778 - 76.2%), with a median age of 68 (IQR 54-79). Community infections dominated with 3249 (89.0%) reports. Comorbidities were frequently reported: hypertension (1481 - 61.7%), cardiovascular diseases (948 - 39.5%), and diabetes (660 - 27.5%) being the most frequent in adults; respiratory diseases and asthma (4 -21.1%), haematological and oncological diseases (3 – 15.8%) being the most frequent in children. Complications occurred in 2679 (73.4%) episodes, mostly for respiratory diseases (2470 - 93.2% in adults, 16 – 55.2% in children), renal (681 – 25.7%) and cardiac (631 – 23.8%) complication for adults. The second and third most frequent complications in children affected the digestive system and the liver (7 - 24.1%). A targeted treatment was given in 1299 (35.6%) episodes, mostly with hydroxychloroquine (989 - 76.1%). Intensive care units stays were reported in 578 (15.8%) episodes. 527 (14.5%) deaths were registered, all among adults.
Conclusion. The surveillance system has been successfully initiated and provides a robust set of data for Switzerland by including about 80% of SARS-CoV-2/COVID-19 hospitalised patients compared to official statistics, with similar age and comorbidities distributions. It adds detailed information on the epidemiology, risk factors, and clinical course of these cases and, therefore, is a valuable addition to the existing mandatory reporting.
Risk factors for severe outcome for 3264 COVID-19 patients hospitalized in Switzerland, February to June 2020: prospective observational cohort study
Lead :Gertraud Schuepbach (VetSuisse/BAG) & Beatriz Vidondo (VetSuisse)
Content: Analysis of risk factors for patient with COVID-19 infection,
COVID-19: More than "a little flu"? Insights from the Swiss hospital-based surveillance of Influenza and COVID-19
Lead: Georg Fröhlich (INSEL) & Rami Sommerstein (INSEL)
Status: Preprint available at medRXiV; submitted at Eurosurveillance
Background and study aim
It has been a matter of ongoing debate, whether in-hospital outcomes of infection with SARS-CoV-2 are comparable to outcomes of infection with Influenza A/B virus. Therefore, this study aims to compare adverse outcomes in patients, who were hospitalized with SARS-CoV-2 and/or Influenza A/B virus infection. To the best of our knowledge, only a few small case series were reported to directly compare outcomes for Coronavirus disease 2019 (COVID-19) with Influenza infection.
Data are derived from the prospective registries “Hospital based surveillance of COVID-19 and Influenza” that are based at the Geneva University Hospitals/Institute of Global Health, University of Geneva. In total, 15 hospitals all over Switzerland collected data on patient characteristics, concomitant medication and outcomes in 2 separate cohorts with similar data collection of various patient-related variables, including a diverse group patients with Covid-19 and Influenza. Data collection included Influenza patients, collected from 7 hospitals. All registered patients ≥ 18 years will be included into the analysis. The study groups consist of hospitalized patients with either laboratory-proven Covid-19 OR Influenza infection. The primary outcome measure is the in-hospital all-cause mortality. Secondary outcome measures are need for ICU care, need for non-invasive or invasive ventilation, need for readmission, and complications like pneumonia, kidney or liver. For the analysis we will use mixed-effects Cox proportional hazards model. To adjust for differences in baseline characteristics and clustering effects, inverse probability weighted propensity score matching and multivariate mixed-effects models with subdistribution analysis of competing outcomes (discharge) will be applied.
Results and Outlook
The results will provide a robust estimate regarding risks in outcomes for hospitalized Covid-19 patients when compared to seasonal influenza patients. These results will provide valuable elements for the ongoing discussion on morbidity and mortality of Covid-19 vs seasonal influenza patients. Moreover, it may help clinicians, scientists, policy makers and the population to make evidence-based decisions on the level of prevention/actions
Patterns of antimicrobial prescribing in patients hospitalized for influenza virus infection or COVID-19
Lead: Dr. Danielle Vuichard-Gysin (STGAG/Swissnoso), Dr. Julia Anna Bielicki (UKBB), Prof. Sarah Tschudin-Sutter(USB), Dr. Catherine Plüss-Suard (IFIK/UniBE), Prof. Stephan Harbarth (HUG), Prof. Olivia Keiser (ISG/UniGE), Prof. Andreas Widmer(USB)
Background and objectives:
Widespread use of antimicrobials drives emergence of antibiotic resistance. Antibiotics are often prescribed for viral respiratory infections without clear indication representing inappropriate use. Designing effective interventions to improve antimicrobial use in patients admitted with viral respiratory infections requires understanding of prescribing practices. So far, antimicrobial prescribing in hospitalized patients with influenza has not been compared to patients with COVID-19. Our main goal is to characterize antimicrobial use in these patients.
We propose to use data from the prospective national surveillances on influenza and COVID-19, respectively, to investigate and compare antimicrobial use in these two cohorts. We will consider all adults, hospitalized for > 24 hours with either laboratory confirmed influenza virus infection or COVID-19. The primary outcome measure is the proportion of patients prescribed antimicrobial agents in each cohort. Secondary outcomes are the relative proportions of AWaRe group antibiotics among treated patients, the proportion of patients with single or combined antibiotic treatment, distribution of different antimicrobial classes between the two cohorts, proportion of broad-spectrum antibiotic and the proportion of patients with infectious complications.
For our primary analysis, we will use univariable logistic regression to calculate odds ratios for antimicrobial use among influenza and COVID-19 patients. To adjust for potential confounders, we perform multivariable logistic regression analysis. We will use Generalized Estimating Equations to adjust for clustering at the hospital level.
Results and outlook
Results derived from these analyses may eventually help to design new interventions for rational use of antibiotics in patients admitted with respiratory viral infections.
Characteristics and Outcomes of Patients who Developed a Healthcare Associated SARS-CoV-2 Infection
Lead: Dre. Gaud Catho (HUG), Dre. Stéphanie D’Incau (INSEL), Dr. Andrew Atkinson (INSEL), Dre. Anne Iten (HUG), Prof. Stephan Harbarth (HUG), Prof. Jonas Marschall (INSEL)
Background and objectives:
Healthcare associated COVID-19 (HA-COVID-19) infections have been described since the beginning of the COVID-19 pandemic (1). Since then, several healthcare-associated outbreaks have been reported (2-4) and have shown that hospitals are an important platform for viral transmission. Because hospitalized patients are usually more fragile and comorbid, HA-COVID-19 might cause significant morbidity and mortality. Nevertheless, an intensive screening strategy to break transmission chains inside healthcare facilities may also lead to increased detection of mild COVID-19, including asymptomatic cases, even among elderly. Thus, uncertainty remains on the clinical outcome of patients who contracted SARS-CoV-2 in healthcare facilities. The main objectives of the current work are (1) to describe and compare characteristics of patients who acquired COVID in the community, versus acute care hospital versus Long Term Care Facilities (LTCF), and (2) to assess and compare clinical outcomes of all these patients.
We will use data from the prospective national surveillance on COVID-19 (CH-SUR) which includes all hospitalized COVID-19 cases with a laboratory confirmed infection. COVID-19 cases are classified in three main categories, stratified by the likely place of SARS-CoV-2 acquisition: 1) community –acquired, 2) acute care hospital-acquired or 3) LTCF-acquired. Hospital-acquired cases are defined as those detected more than 5 days after in-hospital admission.
All routinely collected variables and available clinical features will be retrieved and described in the 3 subcohorts of patients: demographics characteristics (age, gender and comorbidities), transfer to intensive care or intermediate care unit, in-hospital length of stay, and in-hospital deaths.
In addition, for the subgroup of HA-COVID-19 cases, we will record the following items: time to nosocomial acquisition from admission, unit of acquisition (geriatric, medicine, surgery, obstetrics, intensive care unit, other), symptoms at the time of the diagnosis, complications, antiviral treatment, corticosteroids, time to detection if the patient was symptomatic before the day of the first positive test, time to onset of symptoms if the patient was asymptomatic at the time of the first positive test.
The primary outcome will be a composite outcome of in-hospital all-cause mortality and intensive care unit transfer.
In a first step, we will perform descriptive statistics between the different groups. In a second step we will evaluate mortality and ICU transfer in the different groups using complementary analytical methods (i.e., multilevel Poisson regression and failure-time models, accounting for clustering effect and competing risks).
Results and outlook
Results derived from these analyses may help clinicians to better understand and compare outcomes of HA-COVID-19. Moreover, results may help to target interventions for prevention of HA-COVID-19.
Evolution of mortality over time in CH-SUR
Lead: Alexis Martin (ISG) & Maroussia Roelens (ISG)
Status: Ongoing - Submitted to Swiss Medical Weekly
Background and objectives
When comparing the periods of time before and after the first wave of the ongoing SARS-CoV-2/COVID-19 pandemic, the mortality rate seems to be lower in the second period. However, many confounders could explain this decreasing mortality rate, especially changes in demographic characteristics of the population. The goal of this study is to analyse to what extent observed changes in mortality, intensive care units use (ICU), intermediate care use and duration of both hospitalizations and ICU use may be explained by various cofactors such as age, sex, and comorbidities.
We will conduct a Cox proportional hazard regression model to compare mortality and ICU/intermediate care use between three time periods (first wave, summer and second wave). We will account for competing risks.
Duration of hospitalization (overall and in ICU/intermediate care) will be analysed using Kaplan Meier graphs and Cox regression.
SUPER-COVID- Superinfections in patients with COVID-19
Lead: Dr Werner Albrich (KSSG) & Dre Carol Strahm (KSSG) - initially a mono-centric study, this was deemed of interest by several other centers.
Note: Due to the need of additional data, Ethics agreed to the use of further data (Project ID: 2021-00785, EKOS 21/060)
Co- and superinfections in COVID-19 patients represent a major concern for clinicians as a factor complicating clinical management and contributing to morbidity and mortality of COVID-19. While data from the first wave indicated a low overall incidence of coinfections, newer publications from the second wave report higher rates of bacterial and fungal coinfections, particularly in critically ill COVID-19 patients, and are consistent with the increase of respiratory coinfections in COVID-19 patients that we noticed in St. Gallen during the second wave. However, time-varying risk factors as well as epidemiology and clinical impact of such infections are understudied yet.
We aim to describe the epidemiology, predictors and clinical characteristics of co- and superinfections in COVID-19 patients and their changes over the course of the pandemic, with particular focus on respiratory co- and superinfections.
Retrospective observational cohort study of all patients with COVID-19 and suspected or confirmed co- and superinfection who were admitted to any of the acute care hospitals that are part of the infectious disease care network of the Cantonal Hospital of St. Gallen (OSKI) between 1st March 2020 and 1st March 2021.
This study will provide valuable evidence on nature, risk factors and impact of coinfections in COVID-19 patients in Switzerland, which will help to optimize management of COVID-19 patients and to guide antibiotic stewardship strategies in the hospital setting in Switzerland.
60-day readmissions after hospital discharge among COVID-19 patients in Switzerland
Lead: Kene David Nwosu (ISG) & Prof Olivia Keiser (ISG) & Dre Anne Iten (HUG)
Within a few weeks of discharge, COVID-19 patients often deteriorate and require readmission to the hospital. Understanding the incidence of such readmissions, and the associated risk factors, is crucial to facilitating effective healthcare delivery and optimising hospital capacity.
This study aims to assess the rate of 60-day readmission among COVID-19 patients discharged from participating hospitals, to describe the trajectory of readmission risk over time, and to determine the risk factors for readmission.
We will conduct a retrospective observational cohort study of all COVID-19 patients in participating hospitals who were discharged. We will describe the characteristics of patients readmitted versus those not readmitted, construct cumulative incidence curves of readmission, and use multivariable logistic regression models with hospital-level random intercepts to determine the risk factors for readmission.
This study will provide valuable evidence on the nature and risk factors of hospital readmissions for COVID-19 patients, helping inform healthcare resource allocation and clinical decisions on when to discharge COVID-19 patients.
Characteristics and outcomes of fully vaccinated, incompletely vaccinated and non-vaccinated COVID-19 hospitalized patients in CH-SUR
Lead: Dr. Philipp Jent (INSEL), Dr. Omar Al-Khalil (INSEL) & Andrew Atkinson (INSEL)
Background and Objectives
In the SARS-CoV-2 pandemic, vaccines have been developed rapidly. Approved vaccines in Switzerland have well established vaccine efficacy (1, 2), but with mass application breakthrough infections have been reported increasingly (9, 10). To date, little is known on the characteristics of individuals suffering from breakthrough SARS-CoV-2 infection with the need of hospitalization, as well as on their disease course and outcome compared to unvaccinated COVID-19 patients. Knowledge on this population is crucial in order to guide public health efforts aiming at avoiding hospital overload in coming surges of the pandemic.
We aim to investigate the differences in patient characteristics, disease severity, disease course and outcome between fully vaccinated, incompletely vaccinated and non-vaccinated individuals hospitalized with COVID-19 in Switzerland.
We will perform a retrospective cohort study to compare the characteristics and outcomes of the mentioned groups. Data will be derived from the prospective national surveillance on hospitalized COVID-19 patients (CH-SUR). Adult patients hospitalized for >24 hours in participating Swiss hospitals with laboratory-confirmed COVID-19 from December 23, 2020 onwards will be included.
In a first step, we will use descriptive statistics to establish demographic characteristics and outcomes, and identify differences between fully vaccinated, incompletely and non-vaccinated patients in the inpatient setting.
Due to the still limited number of vaccine breakthrough cases in the database, a complementary in-depth analysis is planned as a second step, when case numbers are sufficient for regression analysis. We will use competing risk models (Fine-Gray) in order to analyse the outcome of vaccine breakthroughs in comparison to other hospitalized SARS-CoV-2 cases in the cohort, and determinants of the outcome. The primary outcome will be time to hospital discharge, secondary endpoints further outcome parameters like in-hospital mortality, proportion of patients with oxygen support, non-invasively-ventilated and ventilated patients, time in ICU, amongst others.
Drug therapy decisions for COVID-19 during the 1st year of the pandemic in Switzerland - a retrospective analysis of the national hospital-based surveillance database
Lead : Michèle Birrer (INSEL), Vanja Piezzi (INSEL), Lukas Baumann (INSEL) & Christine Thurnheer (INSEL)
Background and objectives
The COVID-19 pandemic is a major challenge for clinicians due to the lack of evidence- based treatment recommendations. At an unprecedented speed, research data is being published, potentially effective drugs emerge, receive emergency authorizations and some of them are revoked soon after. The drug prescription practice for COVID- 19 patients in Switzerland is largely unknown, as well as the drivers behind treatment decisions.
The main objectives of this study are 1) to assess drug treatment diversity and investigate changes in therapeutic approaches for COVID-19 over the course of the first year of the pandemic in Switzerland in comparison to the recommendations of international and national authorities, and 2) to correlate drug choices with potential drivers of treatment decisions: the severity of COVID-19, demographic characteristics and other influencing variables.
Materials and Methods
Data is retrieved from the prospective registry “hospital-based surveillance of COVID- 19” which includes hospitalised COVID-19 patients from 21 hospitals in Switzerland. All registered patients aged >18 years, hospitalised from 01.03.2020 to 28.02.2021, will be included. Primary outcome will be a description of the proportion of patients receiving specific immunomodulatory/antiviral drugs and the frequency of each drug used against COVID-19 overall and during four time periods 03-05/20 (first wave), 06- 08/20 (summer, low incidence), 09-11/20 (second wave), and 12/20-02/21. Secondary outcome will be the drug use overall and during the four time periods stratified by severity of illness, demographic characteristics, time since symptoms / positive test, hospital acquisition of infection, type of hospital ward, Swiss language region (German, French/Italian).
Differential treatment effect of remdesivir on the mortality of hospitalized patients with COVID-19
Lead: Plamenna Venkova (ISG) and Janne Estill (ISG)
Background and objectives
Remdesivir was the first antiviral drug fully licensed for the treatment of COVID-19, but it has subsequently been suspended from the WHO prequalification list after the results of clinical trials showed no benefit. We analysed routinely collected data from the CH-SUR database to explore whether the treatment response differed by patient characteristics.
We included patients in the CH-SUR enrolled by 9 November 2020 who either did not receive any treatment, who received remdesivir, and who received a treatment other than remdesivir. Patients receiving both remdesivir and other drugs were excluded.
We used model-based recursive partitioning to group the patients according to the association between remdesivir use and the risk of death. In this method, a Cox regression is performed iteratively for subgroups defined based on one of the pre-selected partitioning variables at a time. At each round, the dataset is split according to the variable that caused most instability in the parameters of the Cox model; this process is repeated until the instability in all possible partitionings is above the significance level.
We conducted two analyses. In the first analysis, we adjusted the Cox model for treatment (none, remdesivir, other), sex and age, and included the following partitioning variables: body mass index; urea nitrogen level >19 mg/dl; respiratory rate >30/min; low blood pressure (diastolic <60 mmHg or systolic <90 mmHg); chronic obstructive pulmonary disease; chronic cardiovascular disease; chronic renal disease; oncological condition; and a composite variable of having at least two of the following: age >65 years, metal score <9, high urea nitrogen, high respiratory rate, low blood pressure (CURB65). In the second analysis, we adjusted the Cox model for treatment only, and used the same variables plus age and sex in the partitioning. To control for selection bias, we conducted both analyses twice: using either the treatment variables directly; or with a method called local centering where the treatment variable was modified according to the propensity to receive treatment.
Results and outlook
We will present the results graphically as trees, where the leave nodes represent subgroups of patients that differ by the effect of remdesivir on mortality. The results will help to identify groups of COVID-19 who may potentially benefit from remdesivir treatment.