How can NCS healthcare data be connected with wastewater surveillance of COVID-19 in a privacy-preserving fashion to inform epidemiological models and democratise data access?

COVID-19 is a serious disease caused by a virus called SARS-C0V-2. To protect communities, there is a need to find out how many people have this virus, and who they might infect. Most people get tested when they start to feel ill. However, some people never feel ill when they are infected with the virus. But they can pass it on to their families without knowing. So, there is development in new ways to identify how many people have the virus but don't have symptoms, so communities can be protected.

Fortunately, traces of the genetic material of the virus can be found in the poo of everyone who is infected, even if they don't feel ill. The poo travels through the sewerage network to sewage treatment centres in their local area. Here, samples of the sewage are taken to measure the amount of genetic material it contains. This allows an estimation of how many people in the community are infected.

Imperial College London want their estimates to be as accurate as possible. The study team therefore need to compare them with information from local hospitals to understand how many people get sick and need medical help. The study team will collect the information needed for the comparisons in this project. In the end, it will allow the team to use the sewage measurements to predict whether hospitals will get busy and need extra help to keep the community safe.

The main objectives of the proposal are;

- to develop the methods required to aggregate healthcare records to wastewater catchment areas.

- to generate data products that can be shared with researchers working on wastewater-based epidemiology.

- to calibrate wastewater-based epidemiological models and better predict the pandemic.

The Article 6 (1)(e) justification is that data will be processed in the public interest to aid the national COVID-19 response through wastewater-based surveillance.

The Article 9(2)(j) justification for processing special category data is that ‘processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes …’ as the data are required for research purposes in the public interest.

The data requested will achieve the aim identified. Healthcare data pertaining to COVID-19 related symptoms, tests and hospital admissions will be collected from the requested datasets. All records include:

- temporal information, e.g. the date of a COVID-19 test or a hospital admission.

- geospatial information, e.g. the super lower output area that a patient is resident in.

- health-related information, e.g. symptoms, tests or hospital admissions.

These data will be combined with geospatial wastewater infrastructure data (such as the list of data zones or the area served by each wastewater treatment plant) to aggregate healthcare records. Summary statistics for each catchment area and date will be recorded as the output of the analysis, e.g. the number of cases associated with a particular catchment area.

Only datasets and data fields strictly required for this analysis have been requested. The following datasets are requested:

- Hospital Episode Statistics Admitted Patient Care. This database contains details of all hospital admissions in England.

The requested data fields and justification are as follows:

Diagnosis: required to determine whether the patient is admitted for complications arising from COVID-19.

Date of admission: required to evaluate the number of new admissions on a given day.

Date of discharge: required in conjunction with Start Date (Hospital Provider Spell) to evaluate the number of hospitalised patients on a given day.

Discharge method: required to evaluate whether the patient is discharged back to the community where they may continue to shed viral RNA (ribonucleic acid).

Provider code: required to evaluate the number of patients in hospitals in a given catchment area in case the patient's usual address is in a different catchment area than the hospital they stay in.

Lower Super Output Area: required to geolocate patients so the corresponding catchment area can be identified. Also required to calibrate SUS data (smaller lag but lower data quality) against HES data (longer lag but higher data quality) because SUS does not provide statistical reporting units from the census (such as output areas or lower-layer output areas).

Output area: required to geolocate patients so the corresponding catchment area can be identified. The team would like to access output areas so the team can check whether future requests for data could be made at output area or lower-layer super output area level.

Encrypted HES ID/Token Person ID - this will enable the team to group multiple records that are for one individual (to mitigate duplication).

- Secondary User Service (SUS) Plus - Admitted Patient Care. This provides a complete record of hospital admission data in England.

The requested data fields and justification are as follows:

Diagnosis (This is the diagnosis (ICD) codes from the episode which contains the dominant procedure or if no dominant procedure is determined - this will be the dominant diagnosis (looking across all the primary diagnosis in the episodes) based on the diagnosis hierarchy.): required to determine whether the patient is admitted for complications arising from COVID-19.

Start Date (Hospital Provider) (This identifies the admission date of a Hospital Provider Spell): required to evaluate the number of new admissions on a given day.

End Date (Hospital Provider) (This identifies the discharge date of a Hospital Provider Spell). This value is taken from the last episode for multi-episode spells.): required in conjunction with Start Date (Hospital Provider Spell) to evaluate the number of hospitalised patients on a given day.

Discharge Method (Hospital Provider) (This identifies the method of discharge from a Hospital Provider Spell. This value is taken from the last episode for multi-episode spells.): required to evaluate whether the patient is discharged back to the community where they may continue to shed viral RNA.

Provider site code: required to evaluate the number of patients in hospitals in a given catchment area in case the patient's usual address is in a different catchment area than the hospital they stay in.

Encrypted ID - this will enable the team to group multiple records that are for one individual (to mitigate duplication).

These datasets will be used to evaluate the number of patients admitted to hospital that are resident in each of the English wastewater catchment areas.

Individuals of all ages with a positive diagnosis or related symptom are to be included in the analysis. To minimise the data requested, the team would only like to access records where the diagnosis involves COVID-19. Excluding children or other vulnerable individuals from the analysis would lead to biased results, limiting the potential benefit of wastewater-based surveillance of COVID-19. The age of patients associated with healthcare records will not be available to researchers.

The time period required for the data is from 01/01/2020 to the most recent data available. This is to obtain an exhaustive picture of disease prevalence in the United Kingdom. The datasets will be used to evaluate summary statistics aggregated to the level of wastewater catchment areas, i.e. the area served by a given wastewater treatment works.

The sole data controller is Imperial College London who will also process data and the Office fo National Statistics (ONS) will be a joint Data Processor.

The project is funded by HDR UK under their Rapid funding programme for National Core Studies.

Once the required variables have been extracted from the datasets, data will be transferred from NHS Digital to the Office for National Statistics (ONS) Secure Research Services (SRS). The ONS SRS was chosen because it provides the software required for geospatial analysis. HES data will be transferred by Secure File Transfer Protocol (SFTP) and SUS Plus data will be transferred via Message Exchange for Social Care and Health (MESH). The ONS SRS team will liaise with NHS Digital once this application has gained approval from NHS Digital and also, from the ONS Research Accreditation Panel. The ONS Secure Research Services will be used for all analysis.

Transfer of data into the TRE will be arranged by the organisation administering the TRE (e.g. the Office for National Statistics Secure Research Services) and the data controller (e.g. NHS digital) without involvement of the researchers.

The ONS will make the data available to Imperial College London researchers via a Trusted Research Environment (TRE). Imperial College London researchers will process the data within the TRE. Only data fields strictly required for the analysis have been requested and no data linkage is required.

Imperial College London researchers will be granted access to the datasets via a remote connection to Trusted Research Environments (TREs) in the United Kingdom. No record-level data will be transferred out of or stored outside the TREs which are secure environments administered by the Office for National Statistics Secure Research Services.

Within the ONS TRE the Lower Super Output Area (NHS Digital data) will be used to identify the wastewater treatment catchment area. The number of patients with COVID-19 in that catchment area identified and the data will then be aggregated. The NHS Digital data is not being linked to any other data.

These data will be combined with geospatial wastewater infrastructure data (such as the list of data zones or the area served by each wastewater treatment plant) to aggregate healthcare records. Summary statistics for each catchment area and date will be recorded as the output of the analysis, e.g. the number of cases associated with a particular catchment area.

NHS Digital reminds all organisations party to this agreement of the need to comply with the Data Sharing Framework Contract requirements, including those regarding the use (and purposes of that use) by “Personnel” (as defined within the Data Sharing Framework Contract ie: employees, agents and contractors of the Data Recipient who may have access to that data).

Outputs will be disseminated in two formats:

First, methods and high-level results will be published in peer-reviewed journals. The intention is to publish the findings in one of the following peer-reviewed journals. Science of the Total Environment, Environmental Science and Technology or Water Research.

Second, aggregated healthcare data will be released to other research groups, government departments, and citizen scientists subject to satisfying disclosure checks. These data will be disseminated in a machine readable format, such as CSV or JSON, together with a technical report describing the data. Aggregated (with small numbers suppressed) datasets will be disseminated via the Health Data Research Innovation Gateway (http://healthdatagateway.org) which provides an access hub for health data in the context of the COVID-19 response. The data will allow Imperial College London, other research groups, government and interested citizen scientists to develop and calibrate wastewater-based epidemiological models.

Only health data aggregated to the level of wastewater infrastructure will be disseminated. Prior to being made available for download from the Trusted Research Environment (TRE), both accredited researchers and a member of the TRE team (ONS Research Services) will assess the outputs for disclosure risks following the NHS Digital Statistical Disclosure Control Protocol. Therefore, all outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide. Assessing the data products for disclosure risks is key to the success of the project so the team can share the aggregated data with other research groups to calibrate wastewater-based epidemiological models.

The study team will report summary statistics at both daily and weekly cadences. Daily summary statistics can provide a signal for the pandemic with high temporal resolution but may have to be censored frequently due to low counts. Weekly summary statistics provide a coarser temporal resolution, but they will be less likely to suffer from censoring due to low counts.

The team want to ensure members of the public are informed of this research. The team aim to do this by conducting a small public engagement study to understand attitudes and perceptions regarding the use of healthcare data combined with measuring genetic traces of the virus in poo at sewage treatment works. This will involve working together with 4 focus groups to gain a better idea of the current level of knowledge regarding wastewater-based epidemiology and where a greater understanding would be beneficial to improve public communication.

The target date for the production of outputs is 31/12/2021.

This project will deliver a public benefit to the UK by providing an evidence base to improve public service delivery. The data generated as part of this project will allow us to calibrate wastewater-based epidemiological models. Current approaches can identify changes in the level of infection in the community, but they cannot be used to estimate the number of people infected. Having access to high-quality healthcare data at the level of wastewater treatment works will allow us to calibrate epidemiological models and use the information extracted from wastewater samples to better understand the pandemic.

Subject to satisfactory statistical disclosure checks, the summary statistics at the level of wastewater treatment plants will be made available to research groups, government departments, and citizen scientists for wastewater-based epidemiology. This will also provide an evidence base for decisions likely to benefit society or quality of life for people in the UK. Well-calibrated models should allow these groups to infer disease incidence and predict healthcare needs so resources can be better allocated (e.g. whether hospitals will get busy and need extra help to keep the community safe).

This project will help to provide an evidence base for public policy decision making. The team will conduct a small public engagement study to understand attitudes and perceptions regarding the use of healthcare data combined with measuring genetic traces of the virus in poo at sewage treatment works. The recommendations from the focus group discussions will be used to inform the development of an information resource to communicate the aims of the research project in a clear and accessible manner to members of the general public. This research will make wastewater-based epidemiology more accessible to the general public.

Yielded Benefits is not a requirement for new applications.