MindKind study: Exploring how best to collect longitudinal youth mental health data

The study explores how to collect data in a way that balances the rights and privacy of participants in different geographies with the ability to support open data for researchers from a diverse set of disciplines and global regions.

It shares learnings from a randomised trial of different data stewardship models and qualitative research in India, South Africa and the UK. Data stewardship models describe different ways that data is collected, stored, and shared, and who gets to make key decisions about it.

It also shares learning from qualitative research exploring young people and researchers views on what are the best models of data stewardship for mental health data.

To note: we encourage all interested to read the MindKind report in full, as it is packed with rich insights, only a few of which are touched on below.

These insights include learning how to set up a project governance model that embeds youth; how to run a deliberative democracy as a research methodology in a multinational, entirely virtual context; how to address cross-country differences in data infrastructure and ethical governance; how to successfully recruit youth in different contexts; and an in-depth discussion of different typologies of data stewardship to inform future work in this area.

1. Different data stewardship models did not affect engagement.

When given a choice, participants in the MindKind study’s quantitative arm expressed preferences about key data governance questions. Interestingly, sometimes these preferences differed by country. However, surprisingly, these preferences did not impact observed participation rates in the app-mediated study: globally, youth participants demonstrated equal participation regardless of the data governance model to which they had been randomised during the consent process. 

2. Trust is a key factor in data stewardship.

Youth in the MindKind study’s qualitative arm engaged deeply in balancing their desire for data to be shared as openly as possible – to destigmatise and advance global understanding of mental health – with their concerns about privacy and protection of this data.

Conversations about who can access and how had a strong focus on global equity. At the same time, youth expressed general scepticism about digital anonymity and data control. There was broad recognition of the untrustworthiness of systems (such as for-profit industry and governments) and researchers with sensitive data, both historically and in recent times.

This led to participant trust in those collecting and sharing the data being one of the key factors data stewardship model preference. Regardless of region, charitable and independent organisations were seen to be most trustworthy.

3. Providing youth more choice in which data they collected reduced engagement.

Surprisingly, participants in the quantitative arm who got to choose what data they collected (for instance, which “active ingredient” activities they participated in data collection for) showed lower engagement with the study than participants who were assigned study topics. This was counter to the anticipated direction of effect since it was thought that giving young people more freedom of choice might lead to more engagement.

4. Retention was at or above industry standards for uncompensated data collection.

In the quantitative study arm the proportion of participants contributing at least 50% of the protocol surveys was 3.8%, 5.3%, and 10.1%, and the median days of retention were two, six and 11 days in India, South Africa and the UK, respectively. This retention rate matches, or is greater than, observed retention in similar uncompensated studies, but less than the 26-day median retention rate for a similar study where youth were for their participation.

5. Key data governance specifications suggested by youth.

These suggestions included:

• Data control boards/stewards should be paid for their work or, as a fallback, data should be democratically controlled by data contributors. 
• Data should be hosted on a server (“sandboxed”) rather than being available for download.
• Data should not be used for free to generate a profit, and almost never for commercial gain.

6. Key data access specifications suggested by external researchers.

These suggestions included:

• Metadata about data collection context is always available. For example, a catalogue of consent documents and assessment tools should be available on the platform. 
• Information orienting visitors to the study should be comprehensible to researchers and lay people.
• Data quality assessments by individual scientists should be shared with the whole researcher community to inspire confidence in quality.

For more information on this report please contact the mental health team at mentalhealth@wellcome.org, please put "MindKind Report" in the subject so we can deal with your query quickly.