Periodic Reporting for period 2 - unCoVer (Unravelling Data for Rapid Evidence-Based Response to COVID-19)
Période du rapport: 2022-05-15 au 2023-05-14
The outbreak of COVID-19, caused by SARS-CoV-2, was declared a public health emergency of international concern by the WHO on 30 January 2020 and a pandemic on 11 March 2020. Despite the deployment of public health measures, such as restrictions on movements and gathering, personal protection, as well as massive vaccination campaigns, the coronavirus is still largely affecting mortality and morbidity worldwide, including persisting symptoms after the infection, what has now been termed post-COVID-19 condition. Early epidemiological data on COVID-19 showed a higher risk of severe disease among older individuals, in particular those with chronic respiratory, cardiometabolic and other chronic diseases, including mental disorders and immunosuppressed individuals.
Common disease risk factors such as smoking and overweight/obesity have been identified as key predictors of hospitalisation and critical illness, even in young adults with no underlying conditions. While the pandemic is evolving and countries are adapting their health systems to new phases of preventive measures, the research community is trying to fully elucidate the transmission and progression of the infection, as well as the most effective ways of treating and preventing new cases in preparation for any new waves, particularly due to new variants of SARS-CoV-2.
Responding to the COVID-19 pandemic in real-time required a colossal effort from health systems worldwide and across Europe, where several countries have been severely affected. As a result, a wealth of data have been accumulated as part of health systems’ efforts to fight COVID-19. These real-world data (RWD) reflect the impact of COVID-19 on patients’ health and characterise the protocols of healthcare in different health system settings. These close-to-reality data allow for studies into patients’ characteristics, determinants of disease prognosis and effectiveness of potential strategies against COVID-19 in real-world settings. They also complement findings from ongoing efficacy/safety clinical trials, where vulnerable/heterogeneous populations and those most at risk of severe COVID-19 are often excluded.
The unCoVer network aims to provide a research platform and data infrastructure for the expert use of RWD by bringing together complementary data and medical and scientific expertise to address the still urgent questions regarding the determinants of COVID-19 prognosis to inform more effective medical and public health strategies.
Common disease risk factors such as smoking and overweight/obesity have been identified as key predictors of hospitalisation and critical illness, even in young adults with no underlying conditions. While the pandemic is evolving and countries are adapting their health systems to new phases of preventive measures, the research community is trying to fully elucidate the transmission and progression of the infection, as well as the most effective ways of treating and preventing new cases in preparation for any new waves, particularly due to new variants of SARS-CoV-2.
Responding to the COVID-19 pandemic in real-time required a colossal effort from health systems worldwide and across Europe, where several countries have been severely affected. As a result, a wealth of data have been accumulated as part of health systems’ efforts to fight COVID-19. These real-world data (RWD) reflect the impact of COVID-19 on patients’ health and characterise the protocols of healthcare in different health system settings. These close-to-reality data allow for studies into patients’ characteristics, determinants of disease prognosis and effectiveness of potential strategies against COVID-19 in real-world settings. They also complement findings from ongoing efficacy/safety clinical trials, where vulnerable/heterogeneous populations and those most at risk of severe COVID-19 are often excluded.
The unCoVer network aims to provide a research platform and data infrastructure for the expert use of RWD by bringing together complementary data and medical and scientific expertise to address the still urgent questions regarding the determinants of COVID-19 prognosis to inform more effective medical and public health strategies.
The activities of the unCoVer network have been developing on interlinked work packages and tasks:
WP1 - Data identification: Mapped all data sources available to unCoVer and reviewed their codebooks and variable dictionaries: information on the type of data available, the target group(s), time coverage, characteristics as well as variable types, were mapped for all databases. After identifying the most common variables, a standardised codebook was developed to assist data harmonisation. A total of 24 databases across Europe and elsewhere, and over 500 unique variables were identified for further processing.
WP2 – Ethics and data protection: An overall compliance framework for the network as a whole has been constructed. The data processing activities of the project have been defined and documented and data-related agreements have been concluded to comply with GDPR and national legal regulations. This comprehensive understanding of the data protection regulation informed subsequent workpackages.
WP3 – Data harmonization: Variables were recoded into unCover's standardized codebook. In addition, based on the heterogeneity of data governance schemes of individual partners, and aiming for full compliance with GDPR, the network opted to deploy a decentralized data system allowing for federated analyses, without sharing or disclosing any individual-level data. This infrastructure has been fully implemented and is continuously improving in terms of analytical capabilities.
WP4 - Data analysis and outcomes: This WP provided continuous updates on the research questions that will be addressed by the network, considering the available data and has developed a joint protocol for clinical and administrative follow-up of cohorts of COVID-19 patients.
WP5 - Exploitation and Dissemination: unCoVer has a website also hosting the unCoVer toolbox. The social media channels of the project were also set to ensure successful communication and digital presence of the network.
WP6 - Coordination: Several supporting structures were set including a General Assembly, Steering Committee, as well as internal and external advisory boards. This WP has also defined and established appropriate internal communication tools and procedures. Further, the Project Coordination team has established links of collaboration with several other EC data-related projects.
As a whole, the work performed across WP resulted in strategic thinking, prioritizing, and formulating solutions to provide rapid access to data that can be actioned for informing public health strategies, accomplishing unCoVer's overall aims.
WP1 - Data identification: Mapped all data sources available to unCoVer and reviewed their codebooks and variable dictionaries: information on the type of data available, the target group(s), time coverage, characteristics as well as variable types, were mapped for all databases. After identifying the most common variables, a standardised codebook was developed to assist data harmonisation. A total of 24 databases across Europe and elsewhere, and over 500 unique variables were identified for further processing.
WP2 – Ethics and data protection: An overall compliance framework for the network as a whole has been constructed. The data processing activities of the project have been defined and documented and data-related agreements have been concluded to comply with GDPR and national legal regulations. This comprehensive understanding of the data protection regulation informed subsequent workpackages.
WP3 – Data harmonization: Variables were recoded into unCover's standardized codebook. In addition, based on the heterogeneity of data governance schemes of individual partners, and aiming for full compliance with GDPR, the network opted to deploy a decentralized data system allowing for federated analyses, without sharing or disclosing any individual-level data. This infrastructure has been fully implemented and is continuously improving in terms of analytical capabilities.
WP4 - Data analysis and outcomes: This WP provided continuous updates on the research questions that will be addressed by the network, considering the available data and has developed a joint protocol for clinical and administrative follow-up of cohorts of COVID-19 patients.
WP5 - Exploitation and Dissemination: unCoVer has a website also hosting the unCoVer toolbox. The social media channels of the project were also set to ensure successful communication and digital presence of the network.
WP6 - Coordination: Several supporting structures were set including a General Assembly, Steering Committee, as well as internal and external advisory boards. This WP has also defined and established appropriate internal communication tools and procedures. Further, the Project Coordination team has established links of collaboration with several other EC data-related projects.
As a whole, the work performed across WP resulted in strategic thinking, prioritizing, and formulating solutions to provide rapid access to data that can be actioned for informing public health strategies, accomplishing unCoVer's overall aims.
As a Coordinated Support Action, unCoVer has been concerned with topics such as standardisation, dissemination, communication, networking, and mutual learning on COVID-19-related research. Activities under these topics have been developed internally across a sizeable number of partners, and externally through continuous interactions with other related EC-funded projects such as SYNCHROS, DRAGON, ORCHESTRA, ReCoDID, VERDI, to name a few, lately organized in the Cohorts United and the Cohorts Coordination Board with representatives of the projects and members of the EC, ECDC, EMA, EFPIA.
Through these interactions, and strategic planning unCoVer has streamlined the processes necessary for the development of a data access infrastructure allowing privacy-preserving analysis based on federated learning. The infrastructure is accessible via a dedicated toolbox, currently available at the different levels of accesibility.
1. The interaction with the public is guaranteed through the first level: ‘Meta-data dashboard’. At this level, the meta-data of the project is openly accessible to the general public through the website of unCoVer.
2. A second level has been defined (only accessible by unCoVer partners through user authentication) with pre-programmed analyses including descriptive tables, and regression models for associations between patients' characteristics or hospital variables and patients' outcomes.
3. The third level, limited to the data analysts and programmers of the consortium, is dedicated to the development of privacy-preserving scripts to inform the further development of models in level two.
The unCoVer's outreach provided a forum to disseminate the activities and results increasing impact at the scientific and societal levels. The network has published a number of scientific manuscripts including protocols, and guidance documents in collaboration with other projects, as well as the results of its own analysis, and currently works on the exploitation of the toolbox for additional analysis and considering its further expansion. Finally, the final conference of unCoVer took place as part of the 15th European Public Health Conference in Berlin, bringing together most of the consortium members, collaborating projects, and European experts on health data.
Through these interactions, and strategic planning unCoVer has streamlined the processes necessary for the development of a data access infrastructure allowing privacy-preserving analysis based on federated learning. The infrastructure is accessible via a dedicated toolbox, currently available at the different levels of accesibility.
1. The interaction with the public is guaranteed through the first level: ‘Meta-data dashboard’. At this level, the meta-data of the project is openly accessible to the general public through the website of unCoVer.
2. A second level has been defined (only accessible by unCoVer partners through user authentication) with pre-programmed analyses including descriptive tables, and regression models for associations between patients' characteristics or hospital variables and patients' outcomes.
3. The third level, limited to the data analysts and programmers of the consortium, is dedicated to the development of privacy-preserving scripts to inform the further development of models in level two.
The unCoVer's outreach provided a forum to disseminate the activities and results increasing impact at the scientific and societal levels. The network has published a number of scientific manuscripts including protocols, and guidance documents in collaboration with other projects, as well as the results of its own analysis, and currently works on the exploitation of the toolbox for additional analysis and considering its further expansion. Finally, the final conference of unCoVer took place as part of the 15th European Public Health Conference in Berlin, bringing together most of the consortium members, collaborating projects, and European experts on health data.