6. Overall considerations

Three key aspects require consideration:

• Risks: Risks pertain to potential events that may adversely affect project objectives, outcomes, or schedules. Managing risks entails identifying, evaluating, prioritising, and mitigating them to safeguard project success.

• Assumptions: These are accepted beliefs on which the Data Linkage model relies. Managing assumptions involves acknowledging, documenting, and validating these beliefs.

• Dependencies: Interdependencies among project components may exist. Managing dependencies entails identifying and monitoring these relationships and proactively addressing them during the planning phase.

🔴 RED:
Uncertainties are not recognised, managed or communicated.

🟡 AMBER:
Some uncertainties are recognised, but the communication and management process could be improved.

🟢 GREEN:
Uncertainties are comprehensively recognised, managed and communicated.

• Establish a roadmap for updates.

• Inform all stakeholders of planned changes to prevent disruption.

🔴 RED:
Absence of a roadmap for changes and ineffective engagement with stakeholders lead to changes being introduced without proper notification.

🟡 AMBER:
Partial roadmap for changes and non-timely engagement with stakeholders lead to changes being introduced poor notification.

🟢 GREEN:
Presence of a roadmap for changes and effective engagement with stakeholders ensure changes are introduced with proper notification.

• Regulations: Adhere to data protection regulations like GDPR and the Data Protection Act 2018. Supporting documents like Data Protection Impact Assessments (DPIAs) and Data Sharing Agreements (DSAs) should demonstrate compliance.

• Data minimisation: Separate personal identifiable information and clinical outcomes and ensure linked records don't reveal sensitive information beyond what's been agreed upon. Consider using Privacy Enhancing Techniques (PETs) to protect individual privacy (for example, encryption)

🔴 RED:
Data regulations and data minimisation principles are not followed. Unwanted disclosive information might be released.

🟡 AMBER:
Partial adherence to data regulations and data minimisation principles. Some risk of disclosive information been released exists.

🟢 GREEN:
Data regulations and data minimisation principles are followed comprehensively.

• Fairness: Ensuring that the project and its outcomes are non-discriminatory. Identifying issues such as bias or uneven error distribution among different population segments.

• Accountability: Ensuring that individuals in the Data Linkage workflow are accountable for the decisions made at each stage.

• Sustainability: Ensuring that the Data Linkage model remains relevant as real-world data evolve over time.

• Transparency: Offering clear access to processes and decisions.

🔴 RED:
No ethical considerations are made. Possible risks of bias or misuse of data are unaddressed.

🟡 AMBER:
Some ethical considerations, but without thorough implementation or follow-through. Risks of hidden bias or misuse of data may still exist.

🟢 GREEN:
Comprehensive ethical considerations. Risks are understood and disclosed.

• Roles: Clearly define and document key roles like Information Asset Owner (IAO) and Data Linkage Owner (DLO).

• Responsibilities: DLO is responsible for ensuring projects align with Quality Assurance Framework principles.

🔴 RED:
Lack of clear roles and responsibilities. THe principles set out in the Quality Assurance Framework are not followed.

🟡 AMBER:
Roles and responsibilities not clearly defined. Some awareness and adherence to Quality Assurance Framework principles, but gaps or inconsistencies in implementation.

🟢 GREEN:
Well-established roles and responsibilities. The Quality Assurance Framework principles are followed and there are regular reviews and updates.

• Collaborate to innovate: Join the Data Linkage conversation for knowledge exchange, technique comparison, and process optimisation. Collective wisdom drives the field forward.

• Contribute, elevate: Share your tools, code, and experiences to empower the Data Linkage community.

🔴 RED:
No engagement with the broader data linkage community.

🟡 AMBER:
Occasional engagement with the community but lacks systematic involvement.

🟢 GREEN:
Regular engagement with the data linkage community, promoting knowledge exchange and collaboration.

• Document decision making: refer to Transparency and Accountability in Ethics & Fairness

• Disseminate knowledge: Make your documentation user-friendly and readily available to all stakeholders. Code goes open-source if possible.

🔴 RED:
Lack of transparency and inadequate or inaccessible documentation of the project, making it difficult for others to understand and replicate. Code is not available.

🟡 AMBER:
Partially documented and accessible only by few. Code is internally available.

🟢 GREEN:
Transparent documentation that can be easily understood and accessed by others. Code is fully published and available on github.

• Frequency: Determining the frequency of reassessment, particularly for ongoing projects, is crucial to ensure the ongoing relevance of quality assurance procedures, especially in the face of potential alterations in data and requirements.

• Feedback: Provide feedback to the data providers regarding data quality matters, and capture feedback from the users of the linked data to enhance data linkage process.

• Backlog: Document lessons learned – methods, data, tools – and turn them into a backlog of actionable improvements with resource estimates.

• Adapt and evolve: Regularly review your quality assurance plan, especially for ongoing projects, to stay relevant as data and needs change.

🔴 RED:
No process is in place to capture and relay feedback. No actionable insights obtained to improve the Data Linkage.

🟡 AMBER:
Some feedback is captured but not systematically. Some of the insights are used to improve the Data Linkage.

🟢 GREEN:
A robust process is in place to capture and relay feedback. The insights are used to improve the Data Linkage.