Add-Aspirin
The Add-Aspirin trial was conceived with the recognition that participants will require follow-up for at least 10 years [10]. This length of follow-up is required to assess the overall risk: benefit of regular aspirin use on the trial participants’ health. From the design stage of the trial, like for many trials [19], there was an intention to access data using routinely collected EHRs. When the trial was initially conceived in 2012, the Add-Aspirin trial team met with individuals from NCIN, the predecessor of NCRAS, to assess the feasibility of accessing data and also to ensure that an appropriate budget for this activity was incorporated into funding applications (Fig. 4). The protocol, patient information sheets and consent forms were designed to reflect the potential use of routinely collected healthcare data.
In 2017, after 2 years of recruitment and follow-up, there was a conversation with ODR to confirm the cost and current application process. In 2018, there was sufficient data to initiate the pre-defined methodology sub-study. A pre-application meeting with an ODR senior manager established the documentation that was needed going forward.
Following the implementation of the General Data Protection Regulation (GDPR) in the UK (2018), transparency of how exactly participant data would be used became a legal requirement. The previously agreed consent forms and patient information sheets did not meet the 2018 requirements of GDPR. The solution was for a privacy notice to be drafted and made publicly accessible on the trial’s website. The trial’s IG documentation also needed updating to ensure information security assurances (via the Data Security and Protection Toolkit) were in place within UCL.
Following submission of the data application (December 2018), ODR sent back revisions (January 2019) and confirmed the transparency statement (February 2019). For the application to proceed, an analyst needed to be allocated to check the defined data requirements. In April 2019, NCRAS unfortunately unassigned the analyst allocated to Add-Aspirin onto work on a project considered more critical. There was a meeting in May 2019, once further analytical support had been deployed, to discuss the data field requests. The new analysts suggested that a number of data fields should be expanded to give the best chance of capturing cancer recurrence as this is not, at present, collected sufficiently well within any single EHR dataset. They acknowledged at that time that algorithms were needed to identify data patterns indicative of tumour recurrence. ODR wanted to ensure that no unnecessary data from HES was provided for each participant. The MRC CTU therefore provided surgical/procedure codes (using Office of Population Censuses and Surveys (OPCS) definitions) and diagnosis codes (ICD-10 codes) to NCRAS to focus and limit the data extraction. In June 2019, it was agreed with ODR and NCRAS that, as this was a methodological project reviewing ways to gather trial outcomes in registry data, all HES data for these patients could be given to the MRC CTU.
The application then underwent an ODR internal moderation review, and a month later, a data sharing agreement (DSA) was sent from ODR to MRC CTU. Between August and October 2019, there were ongoing discussions between the MRC CTU contracts department and the ODR. The final DSA was signed on behalf of MRC CTU on 16 October 2019 and fully executed by ODR on 15 November 2019. A further new analyst was then assigned to the project who re-reviewed the data request. This new analyst advised an update to the data censor dates, since more up-to-date data was now available from NCRAS. The updated data request was sent back to ODR for re-signing. The DSA was re-signed and the MRC CTU checked the current consent status of patients before sending participants identifiable data to NCRAS on 23 December 2019. The one-off data extracts were successfully received at the MRC CTU on 06 February 2020. This 6-week interval before data receipt was due to NCRAS rewriting their standard filters to provide C44 (non-melanoma skin cancer) — a code that is not usually supplied but needed for this trial. In total, this application, excluding the planning and preparatory work, took approximately 13 months from submission of the application to receiving the data.
PATCH
The PATCH trial opened to recruitment in 2006 as a phase II feasibility trial, developing into a phase III RCT in 2013. The trial was not initiated with the use of EHR in mind but there was a statement included in the consent form to potentially allow information to be sought from the national registries in the future:
I agree that my details including my full name can be given to the MRC such that long-term follow-up information from the NHS Information Centre and the NHS Central Register or any applicable NHS information system.
With the assumption of valid consent for the use of EHR data, a methodological sub-study was devised to triangulate cardiovascular event data between HES, NICOR and trial data. There was an initial scoping of the project in 2014 with NICOR and HSC-IC advising data linkage before comparison at the MRC CTU (Fig. 5). During the initial conversations with NICOR and HSC-IC, the organisations stated that the consent statement was insufficient to acquire linked data from these two sources without first gaining approval from the Confidentiality Advisory Group (CAG). In 2016, the process to submit a CAG application was started. Several months of delays followed due to difficulty in acquiring the appropriate IG documentation for PATCH. CAG require detailed IG documentation for both the trial but also in this case from NICOR and NHS Digital (formerly HSC-IC until 2016). There were difficulties in identifying the appropriate person for this information within NHS Digital, taking most of 2016 to achieve (note: at this time, case officers were not assigned until after the application was formally submitted). During 2016, an alternative method of data access was explored via NCRAS, but as no cancer data was being sought, this option was deemed unviable. Consequently, in 2017, the project was put on hold.
In October 2018, the MRC CTU re-engaged with NICOR (which had moved to Barts Health NHS Trust following a European Union tender process) and NHS Digital. There were additional complexities for obtaining CAG approval as the PATCH trial at the time was in the process of changing sponsor and therefore the CAG application could not be approved.
As the explicit wording on the consent form was the main issue preventing access to the data, the MRC CTU asked the MRC Regulatory Support Centre for further guidance. They felt that the consent wording was sufficient. NICOR subsequently agreed that, if their data was not sent to NHS Digital for linkage, then CAG approval was not necessary. Therefore a further application was submitted and sent to NICOR for review (Fig. 3). NICOR’s review was completed in May 2019. The application was then submitted to HQIP by NICOR. The application was reviewed in June and amendments were returned to MRC CTU. HQIP issued a signed DSA on 19 July 2019, and a NICOR analyst was assigned. The analyst continued discussions with the MRC CTU on data extraction, and a one-off data extract was received at the MRC CTU on 17 October 2019.
As with NICOR, NHS Digital was re-engaged in October 2018, and it took several weeks to allow access to the DARS online system due to technical difficulties with the DARS system (Fig. 5). A new DARS application was submitted in February 2019, but this was initially rejected due to issues around consent and sponsorship and not meeting the DARS checklist criteria. After a phone call to DARS and changes to the application by the MRC CTU, it was accepted and a case officer allocated. The case officer reviewed and made extensive comments with required changes. A privacy notice was created for the project and circulated to participants once it was ethically approved. NHS Digital then advised that the application could not proceed until the NICOR DSA was signed, sponsorship clarified and the new protocol for the sub-study had been ethically approved.
Sponsorship was not resolved until September 2019, and at that point, the MRC CTU re-engaged with NHS Digital. On receipt of the revised application, NHS Digital returned it to the DARS triage service and a new case officer was allocated. Over the next few months, the case officer made amendments to the application and sent it internally to the data approvals officer (DAO). The DAO asked for further changes to the application to clarify certain points and was submitted to IGARD in December 2019 for final review. IGARD approved the application in January subject to one last data specification amendment. The DSA was signed on behalf of the MRC CTU in February 2020, and the MRC CTU uploaded identifiable data to NHS Digital in March. The NHS Digital production team made data available in May and data was received at the MRC CTU on 21 May 2020. When all efforts are taken into consideration, it has taken several years to obtain data from both of these providers. However, from the most recent effort, data was received approximately 8 and 15 months after submission of formal applications to NICOR and NHS Digital respectively.