Transparency is important for creating trustworthy systems for data use. However, there is no common understanding of what transparency means in practice. This resource highlights the range of ways a commitment to transparency can be implemented.
Transparency matters because it lets people understand what data you're holding or using, how it's managed, who is accountable for its protection and where they can find out more. It is also a key principle of people's data rights under the General Data Protection Regulation.
Essentially, transparency means: say what you do, do what you say.
The practical interpretations listed here are the result of a workshop Understanding Patient Data convened to explore how data custodians and users interpret what it means to be transparent in practice. Upholding a commitment to transparency requires thinking both about the content ("what") of information about data use and also the context in which it is provided ("how").
Content: What information should be provided?
Answering the right questions: Information given neatly answers the questions your audience(s) may have.
- You explain what data is being used, by who, why and how it’s looked after.
- The information is wholly relevant, addresses known concerns and leads to better understanding, not just increased awareness.
- It is not a PR or marketing exercise to sell your approach, but answers what others may want to know.
Example: fair processing notices.
Checkable, assessable: Anyone else can easily check what you’ve done.
- There is sufficient information provided for independent scrutiny.
- Publishing information on the rules you follow and any protocols or decisions you’ve agreed as a result.
- If someone only had the information you provide, they could check whether you’ve adhered to the rules.
Examples: audit trails; feedback on research results.
Open, interoperable data: Open data is the default (where appropriate)
- Technical aspects including: open standards; open APIs; standardised languages and interoperability.
- Being open about what you make available, including: metadata; statistics or findings derived from the data; rationale for changes in data management.
- Details on why certain information is not available.
Examples: All Trials platform; ClinicalStudyDataRequest.com.
Context: How should information be conveyed?
Using the right language: The words and images used are clear and understandable.
- Plain language is the default, or language is tailored for the audience and their information and accessibility needs.
- Language used considers what others are doing. Vocabulary is shared and works together across initiatives to produce an understandable overview of data use.
Examples: plain language summaries; visualisations and imagery.
Accessible: Content is easily findable.
- Users don’t need in-depth knowledge of system structures, language and relationships to navigate information.
- Information needs are thought through, including timeliness, level of detail and ease of navigation to related information.
Examples: data release register; minutes of decision meetings.
Organisational values and behaviours: Openness and honesty are 'how we do business around here'.
- Acknowledgement when things go wrong and seeking to learn and improve.
- Time and resource are committed to transparency initiatives.
- Strong leadership demonstrating importance of transparency.
Examples: learning and improvement environment within teams; no-blame culture.