Impacts of digital media use on children and young people's mental health and wellbeing: evidence review

Appendices

7.1 Appendix I: Search strategy

Only one research database was consulted, the Scottish Government library Knowledge and Evidence (KandE) database. KandE draws from academic databases including PubMed Central and MEDLINE.

The search strategy was informed by an initial scoping search undertaken by Scottish Government library staff. Searches were limited to August 2023 – July 2026, English language, peer-reviewed articles.

Screen time and screen use search:

SU ("screen time" OR "screen use") AND SU (child* OR adolescen* OR youth OR teenage* OR "young people") AND SU ("mental health" OR wellbeing)

Social and digital media search:

TI ("social media" OR "digital media") AND SU (child* OR adolescen* OR youth OR teenage* OR "young people") AND SU ("mental health" OR wellbeing)

Include:

• Peer-reviewed academic primary research and reviews considering associations between the use of digital and social media with mental health and wellbeing outcomes.
• Literature focusing on children and young people age 0-25.
• Literature published between August 2023 and July 2026.
• Evidence from Scotland and the wider UK prioritised, with consideration of studies from comparable high-income countries with similar access to social media/digital platforms, and similar education and health systems e.g. Australia, New Zealand, Canada and European countries.

Exclude:

• Children and young people’s mental health where this is not described in relation to digital media/online harm.
• Studies focusing on subgroups or highly specific populations of children or young people that won’t be representative e.g. young athletes.
• Adults over 25.
• Primary research from countries with different education/health systems and access to the internet and social media. Research from different national contexts but with large sample sizes (>1000) will be considered for inclusion.
• Non-peer reviewed editorials or opinion pieces.
• Studies looking at digital media use specifically in relation to the COVID-19 pandemic.

A total of 573 records were identified though the search strategy (223 records identified on screen time and screen use, and 350 on social and digital media use). The title and abstracts were screened according to the inclusion and exclusion criteria, reducing the two sets of database search results to 64 (screen time and use) and 107 (social/digital media use). The search results were combined and 10 duplicates were removed.

161 articles were taken forward for full-text review. On full-text reading, articles were excluded where the outcomes were not mental health and/or wellbeing (e.g. educational outcomes), where they considered policy approaches or interventions, validation studies, or those describing predictors of social media engagement rather than its impacts. Several commentaries, editorials and narrative review articles or essays were also excluded at this stage. This led to a further 38 being excluded, resulting in 123 taken forward for the report.

Snowballing of reference lists and discussions with policy subject experts resulted in the identification of additional relevant evidence. This included academic articles not retrieved through the database searches, grey literature, policy documents and research reports. These 18 sources were included where they met the inclusion criteria or provided relevant policy context, totalling 141 articles and grey literature.

Included sources were read and summarised by a single reviewer, with article details and key findings recorded in a spreadsheet. Each article was assigned an indicative rating (3, 2, or 1). This was based on reviewer judgement and considered sample size and study design. Systematic reviews and large longitudinal studies assigned the highest rating, followed by small longitudinal and cross-sectional studies. Qualitative studies provided important experiential insight, and narrative reviews were drawn on for broader overviews of evidence. The ratings were considered throughout the process of interpreting and synthesising the evidence but the report is not organised according to study quality, instead taking a thematic approach.

7.2 Appendix II: AI use statement

Microsoft Copilot was used during the review process. The reviewer independently identified, read, screened and interpreted all sources of evidence, extracted key findings and developed the report's conclusions.

Copilot supported initial categorisation of article summaries authored by the reviewer, facilitating a rapid approach to reviewing the large body of literature. These categories were adapted by the reviewer and developed into overarching themes as the review progressed. It was also used to support inclusion and exclusion decisions where there was uncertainty, and as a tool to aid understanding of methods sections in studies using complex statistical methodologies.

Copilot was used to conduct quality assurance tasks including identifying duplicate or missing references, checking the final reference list for errors, and supporting a count of included studies by country and region. Copilot was used to edit the Executive Summary authored by the reviewer, helping to make this more concise for a policy audience.

The reviewer verified all Copilot outputs, made all decisions regarding study inclusion, interpretation of study findings and final conclusions, and authored the report.