The Oxford–Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC) is a network of general practices and is now in its 54th season of surveillance and vaccine effectiveness in its long-standing collaboration with Public Health England (PHE) and predecessor bodies [1]. This season is complicated by the likely cocirculation of SARS-CoV-2 and its associated disease burden [2].

The RCGP RSC has been recruited as a nationally representative population [3] that provides pseudonymized data for surveillance of infectious diseases. However, the recent influx of practices may mean we have to adjust for practice distribution and ensure we recruit virology sampling practices that are evenly distributed. The disease surveillance program is commissioned by PHE and monitors 38 infectious diseases, including influenza; at the start of 2020, monitoring extended to include COVID-19 [4].

The recently expanded RCGP RSC extracts pseudonymized data from over 1700 general practices each week covering a nationally representative [3] population of over 13 million. Data from these practices are reported online in the RCGP’s Weekly Return [5], which includes monitoring weekly rates of influenza-like illness (ILI) and other communicable and respiratory diseases for England. We also produce an annual report [6]. The RCGP RSC data set includes all coded data and all prescribed items, including vaccine exposure [3]. To manage our extended network of practices, we have set up a new analytics hub: the Oxford–Royal College of General Practitioners Clinical Informatics Digital Hub (ORCHID) [7].

The RCGP RSC conducts virology surveillance each influenza season. Given the onset of COVID-19, virological sampling has expanded to approximately 300 general practices (see Figure 1). We have reported on which groups of patients were more likely to test positive for COVID-19 [8], on excess mortality [9], and the shift away from face-to-face consultations during the first wave of COVID-19 pandemic [10].

The RCGP RSC is also active in serosurveillance. During the 2018-2019 influenza season, we successfully conducted our first collection of serological samples from adults, linking them to their respective medical records [11,12]. Sentinel networks can provide a mechanism for systematic data collection and linkage to medical records and health outcomes [13], as well as sampling across the complete age range [14]. Current activity has included supporting PHE household studies and collecting convalescent sera from people previously infected with COVID-19.

Finally, the RCGP RSC network has supported trials and studies designated as high-priority public health research. These included recruiting to the PRINCIPLE (Platform Randomised trial of INterventions against COVID-19 In older peoPLE) trial [15] and supporting recruitment to RAPTOR-C19 (RAPid community Testing fOR COVID-19) [16].

This protocol sets out how the RCGP RSC will continue to provide enhanced passive surveillance, virological and serological surveillance from a subset of practices, and support high-priority public health research.

The aim of this protocol is to provide comprehensive surveillance of influenza, COVID-19, and other monitored conditions including health outcomes. The RCGP RSC network will also collect samples for virological and serological analysis, measure vaccine exposure and effectiveness, and support other high-priority public health research.

Our six objectives are as follows:

The methods include six components: (1) surveillance of influenza, COVID-19, other monitored conditions, and associated health outcomes; (2) virological surveillance and serosurveillance; (3) monitoring vaccine uptake and effectiveness; (4) supporting priority public health research; (5) data curation; and (6) governance.

We will collect pseudonymized, routinely collected, coded clinical data, which has been coded into patient records, using UK Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT) codes (SNOMED International), for enhanced passive surveillance. This will entail a wide range of clinical event data, including collection of monitored conditions, vaccine exposure, AEIs, hospital admissions, intensive care, and deaths, to monitor approved study outcomes and for any unexpected public health emergencies. Effective interoperability and communication will be facilitated by monthly pseudonymized linkage, facilitated by NHS (National Health Service) Digital and PHE [7], to secondary care and other data sources, including Secondary Uses Service, the continuously updated source of hospital data (see Textbox 1).

We have converted all our primary care coding to SNOMED CT coding and are using this throughout our data processing [17] (see Multimedia Appendix 1). We have been at the forefront of ensuring that primary care systems code cases correctly and that we comprehensively capture historic data [18]. This has included the development of database-linked tools to programmatically search SNOMED CT for relevant tools. The hierarchical nature of SNOMED CT is that it requires machine searching rather than manual searching (see Multimedia Appendix 2). This will continue to deliver effective COVID-19 surveillance in parallel with the annual influenza surveillance this season.

We will enhance surveillance through the provision of dashboards and observatories to provide feedback about levels of care compared to other practices in our network and to improve data quality, clinical care, and patient safety. There will be a particular focus on data quality, including the recording of ILIs, the use of the RCGP RSC case definition (see Textbox 2 [19-21]), recording of upper and lower respiratory tract infections, and recording of suspected COVID-19 cases, with an emphasis on recording episode type (ie, whether this is a new case or a follow-up). We will programmatically adjust for when episode type is not recorded [18,22].

This year we will extend our virology sampling window to 10 days. We will also report on health outcomes, including mortality [9] (see Figure 2).

A volunteer subset of practices will conduct virological surveillance (see Multimedia Appendix 3). They will conduct or arrange virological sampling of patients presenting with ILI or lower respiratory tract infection, including acute bronchitis and bronchiolitis in children under the age of 5 years. We will include those suspected to have, or have been exposed to, COVID-19 who have persistent cough, loss of taste and/or smell, shortness of breath, history of fever, and/or the presence of a wheeze. We will look for practices to collect 20 samples per week; additional samples can be collected if the spread that is present is within a household or a communal establishment. Practices will receive feedback via dashboards [23] (see Textbox 3 [5,24-28]). Our target is to collect 600 samples per week across all age bands.

The anticipated changes in the coming year are as follows:

A volunteer group of practices will also participate in serological surveillance. The largest project will be the opportunistic collection of an extra blood sample from patients undergoing a routine blood test. Practices will be asked to collect across age bands (aged 10 years and above) (see Figure 4), with feedback provided to general practices via a dashboard and overall via an observatory [30]. We are now recruiting practices from areas with more ethnically mixed populations to provide greater insight into disease spread in ethnic minority populations. Other serology projects include extended serosurveillance in a subset of patients with positive virology tests (n=63) as well as recruiting and sampling patients included in the PHE-led household contacts of confirmed COVID-19 cases (HOCO) study [30]. We are piloting responsive surveillance in areas with higher disease incidence.

Capturing vaccine exposure is critical to conducting vaccine effectiveness studies. With an increasing range of vaccines available, capturing exposure data when not recorded within the practice, beyond the fact that a vaccine has been given, presents significant challenges [31]. Further issues include difficulties capturing vaccine batch numbers [32]. We will ask practices to preload the vaccine brands and batch numbers online.

We plan to continue to report AEIs following vaccination using routinely collected data, though we would have the capability to provide patient AEI reporting cards if commissioned for COVID-19 or other vaccines [33,34]. Practices will be provided with prompts to improve coding. Good data quality regarding AEI recording may identify any differences between influenza vaccine types, to elucidate whether egg- or cell-based vaccines with or without adjuvants are associated with different rates of AEIs. There may also be up to six types of COVID-19 vaccines available for which we may be able to provide uptake and AEI data.

We strongly support the FAIR principles [46]. We are a founding member of the Health Data Research UK Data Alliance, and our data are included in their metadata, with our activities ranked as high priority [47]. We are creating a data set within ORCHID, an observational research platform that has themed databases that can be made readily available to researchers simultaneously; these will be used to support DECISION-COVID and other studies (see Multimedia Appendix 4) [48]. Additionally, we are linking over 50 years of virology and RCGP RSC data, as part of the Wellcome Trust–funded quinquagenarian database project, to make these data more accessible [49]. ORCHID [7] aims to provide a large, near-real-time–themed primary care health informatics hub for the use of data from consenting patients in clinical trials and to supplement existing disease surveillance using in situ network data without large-scale data extraction. We aim to make outputs available using common data models, which we will be providing to the European Health Data Evidence Network [50,51].

The surveillance for the coming year will be initiated by a joint letter to volunteer practices informing them of their role in the surveillance process. The legal basis for our surveillance with PHE is defined as Health Protection under Regulation 3 of the Health Service (Control of Patient Information) Regulations 2002 and is approved annually by the PHE Caldicott Guardian; other studies require appropriate ethical approval. The University of Oxford is compliant with the General Data Protection Regulation and the NHS Digital Data Security and Protection policy.

We want to minimize any risk associated with taking part in this surveillance. Infection prevention and control advice will follow extant national guidance. Our member practices already have most processes in place to meet Regulation 12 of the Health and Social Care Act 2008 (Regulated Activities) Regulations 2014, concerning safe care and treatment, and are periodically inspected by the Care Quality Commission [52]. Our training will include reminders about safe handling of specimens and will recommend revision of infection control measures, where needed.

The COVID-19 pandemic has led to considerable numbers of practices joining the Oxford–RCGP RSC network. The collaboration with EMIS (Egton Medical Information Systems) Health has made it much easier for practices to contribute pseudonymized patient data at rapid speed and at a large scale. The network has tripled in size since January 2020, from 500 practices to 1724. Of these practices, 1076 are using the new ORCHID platform being developed in collaboration with EMIS for direct data extract. Pseudonymized data from over 13 million patients has enabled the growth of a rich database from which researchers are developing a deeper understanding of the COVID-19 pandemic and other diseases in general practice. We have opened a Facebook page and Twitter account to provide more online information and help (see Textbox 3).

We formerly had 100 practices taking part in the national influenza surveillance scheme. This has expanded to 322 for COVID-19 surveillance. As of August 17, 2020, the surveillance system has collected 5357 swabs and detected 581 cases of COVID-19 through extensive virological sampling. A total of 273 practices have contributed toward serosurveillance, collecting a baseline of 2000 samples and an additional 1000 samples per week. A total of 19,761 samples have been collected for serology as of August 17, 2020. We are also piloting serosurveillance in two cities in the northwest of England with a high incidence of COVID-19.

We started collecting vaccine brands and batch numbers via our online portal on September 1, 2020, for the upcoming influenza season. We have made provisions in our dashboards and observatories for the changed vaccination policy for influenza and for virology sampling in the coming season. We have successfully piloted the secure transfer of data between ORCHID and PHE using the OxFile – Large File Exchange Service secure data transfer system [53].

We have recruited 517 practices into the PRINCIPLE trial. A total of 440 patients have had successful screening and have been randomized. As of August 13, 2020, 537 patients have been recruited via general practitioner practices set up as official recruiting sites. Oxford–RCGP RSC practices have recruited 81.9% of these patients.

Our key focus is the development of ORCHID and accessibility of our extended data set. Currently, 1076 practices use the ORCHID virtual server.

We will put in place processes to allow approved researchers access to ORCHID.

We have tripled the sign-up to the Oxford–RCGP RSC and are well on our way to operationalizing this new extended network ahead of the coming winter, in a season complicated by SARS-CoV-2 circulation and an extended flu vaccination program.

The Oxford–RCGP RSC now has in place the most comprehensive set of linked data in its history, supplemented by reference lab virology and serology. Over half of our network will have data available on a daily basis via unprecedented, timely, remote researcher access to our themed data sets.

We have also introduced self-swabbing to increase safety and convenience for patients and practices, to enable direct relay of COVID-19 results to patients, and to reduce the burden of form filling. We have also extended the number and sophistication of our dashboards and observatories, adding interactivity and more extensive help.

Safety of practices remains our primary concern and we are not aware of any increased risk to practice staff or other patients from involvement in surveillance. Personal protective equipment was difficult to source at the start of the pandemic, as were SARS-CoV-2 tests. Both are now much more available to primary care.

Pandemic preparedness is part of the role of the RCGP RSC. The RCGP RSC has operated for over 50 years and has been involved in collecting samples to monitor disease and vaccine effectiveness through the Hong Kong flu pandemic of 1968-1969, the Russian flu of 1977-1978, the 2009 swine flu pandemic [54,55], and the first wave of the COVID-19 pandemic earlier this year.

The introduction of self-swabbing kits this year will enable wider distribution, allowing for an increase in the number of people to be tested. People in self-isolation and quarantine can access testing kits without leaving the confines of their homes, negating the risk of transmission to others with whom they interact with in transit [6]. Self-swabbing appears to be reliable [56,57].

We have shown considerable adaptability so far this year, tackling a wide range of issues and producing publicly available dashboards as well as research outputs [6-8,10,22].

The strengths of our network lie in the willingness of practices to share data and the quality of those data. The Oxford–RCGP RSC team has a deep understanding of primary care data, the context of data recording, and the ability to process routine data. We have developed even closer relationships with computerized medical records system suppliers, especially EMIS. We have a long-term partnership with PHE and its predecessor bodies. There is a good understanding of the strengths and limitations of routine data and the capabilities of primary care.

The limitations of our system are those of all routine data, in that the data may not be complete or accurate. Ensuring we have good data quality is key for the reliability of our screening. There are few available comparators, but we know the network detects more new cases of some cardiovascular diseases than others [58]. Given our ambitious program of scaling, via a new pathway for virology surveillance based on patient-completed data, a further limitation is that these data are currently not linked back into their general practice record. We have also found that the success of open access testing means that individuals can be tested more quickly at a public testing station [59] than self-swabbing via a test through the surveillance system, thereby restricting the extent of possible collection.

While our network is large enough to report the relative incidence of common AEIs following influenza vaccination, comparing brands and vaccine types, it may be too small to comment on very rare events. Little is known about the anticipated AEIs with the new COVID-19 vaccine, so it is currently unclear what role we could play.

While we are part of many European and international networks, we did not have the opportunity to rapidly produce a shared international protocol. However, this document will include the learning from our decades of joint surveillance. It is being made available as a preprint to enable comments to be made and to allow learning to be shared.

The Oxford–RCGP RSC has had a groundswell of interest in membership during the first wave of the COVID-19 pandemic. We have grown in network size and functionality. Our challenges are to maintain data quality and to help support our practices in identifying sufficient volunteers to deliver the required number of samples as we take on more general practice members. Another challenge is the uncertainty of a winter season with cocirculating COVID-19. The pace of change in our understanding of COVID-19 requires us to move ahead rapidly, rather than in close coordination with others. This protocol sets out our approach to surveillance in the coming season and makes our approach widely available for comment or use by others involved in surveillance.