Generating insights on the evolution of HIV care with the Terrence Higgins Trust
How do you analyse and understand eight years of valuable HIV online forum history while preserving the anonymity of the online members? The Health Economics Unit worked with Terrence Higgins Trust (a British charity that campaigns about and provides services relating to HIV and sexual health) to help them understand the themes and topics which users had discussed on their myHIV Forum. Using in-depth topic modelling we were able to build a valuable picture of the forum’s archive and how discussions changed over time and location, during a unique era in the history of people living with HIV.
A place for talk and support
Terrence Higgins Trust’s myHIV Forum was created in 2011 to allow people living with HIV to talk and support one another about everything from diagnosis and medication to experiences of relationships and stigma.
The forum was closed and replaced with a more modern one in 2019, but the trust recognised the conversations it contained were a valuable piece of history. They hoped the lessons learned from this archive around the benefits of peer support and how this impacts the lived experience of HIV could help shape future healthcare practices and better service the needs of those living with HIV and others with long-term health conditions. Forum members had also expressed their wish to preserve the data it contained.
The myHIV forum’s online services were in their early stages and people living with HIV faced significant levels of enacted stigma (stigma and unfair treatment from others) and internalised stigma (internal shame and the expectation of discrimination). This period also marked a significant transformation in HIV treatment and prevention with the development of PEP, PrEP, and improved treatment options.
Understanding discussions while preserving anonymity
Our expert analytical team was asked by Terrence Higgins Trust to help understand the themes discussed on the myHIV Forum.
Approaches to identifying common themes included:
To ensure the anonymity and confidentiality of forum members’ data, we collaborated with the trust to undertake a rigorous data protection and information governance process within both organisations.
Data was stripped of details including names, locations, age groups, co-morbidities and other health issues, before being securely moved to an access-restricted NHS server. We extracted themes using a machine learning algorithm, so no analyst directly read the already anonymised data.
领英推荐
Only IP addresses attached to posts were preserved, to show location related to themes. This was generalised to cities or countries to preserve anonymity.
Identifying topics and understanding themes
We used natural language processing to analyse thousands of posts from the myHIV forum, spanning the eight years it was active. The algorithm broke down written content into smaller segments and pulled out frequently appearing words for topic modelling.
Working with Terrence Higgins Trust, we grouped these keywords into 10 most-discussed topics, such as understanding HIV, medication and side effects, and emotional support. For example, the topic medication and side effects included frequently used words such as ‘drug’, ‘regimen’ and ‘dosage’.
The ten identified topic areas were:
Understanding the impact of HIV and treatments
Our analysis has allowed the Terrence Higgins Trust to see changes in the frequency of discussion topics over both time and geography, providing a valuable insight into:
The trust is now exploring how to communicate these findings further and how to use these key learnings to influence future approaches to peer support, self-management and treatment.