On 13 May 2019, coinciding with the UK Mental Health Awareness week, the RSS Merseyside Local Group hosted an event, 'Mental Health Statistics' which showcased work linked to the analysis of mental health data. The meeting attracted around 50 attendees, over half of whom were RSS members. Attendees from a range of departments and institutes were present, including the University of Liverpool, University of Manchester, University of Dublin, UCLAN, Liverpool John Moores University, Materials Innovation Factory, HMRC and HSE.
The meeting began with a talk from Dr Anais Rouanet of the MRC Biostatistics Unit, Cambridge University. Her talk highlighted the fact that dementia is a WHO recognised public health priority, with around 850,000 people living with dementia in the UK, and around 46.8 million worldwide.
Today, there is access to rich but complicated data. For example, many correlated longitudinal measures of cognitive function are available, and the data is complicated by competing risks between time to death and time to dementia. Research must account for the data complexity in order to better understand the condition.
Clustering and joint modelling methods were presented and applied to French and USA dementia datasets in order to identify subgroups of patients with distinct disease mechanisms, showing that profiles of cognitive decline over time could be separated into distinct groups, but that more individual level information could be useful to provide more individualised predictions.
After a break for tea and coffee, the meeting resumed with a talk from Professor Richard Emsley of King’s College London, who gave an interesting talk about causal pathways in data. Richard explained that analyses of trial data should show the full treatment pathway, which could consist of an intervention implemented in the trial, which affected a mechanism, which in turn affected an outcome. An example of this is Cognitive Behavioural Therapy (CBT) administered to psychosis patients. CBT could change the way the patient thinks, which in turn could change psychosis outcomes of interest. Mediation (a variable occurring in a causal pathway between an exposure and an outcome variable) was discussed.
A range of examples were given, demonstrating how data from trials could follow pathways more complicated than the treatment directly affecting the outcome. The presentation concluded with a series of tips for designing trials involving mediation, including thinking about all possible baseline and time varying confounders, and measuring all putative mechanisms in both treatment arms.
The next meeting of the RSS Merseyside local group will take place on Wednesday 2 October 2019, where we will welcome speakers from Chester Zoo to talk about research covering the following four areas:
- Behaviour and Welfare,
- Social Science,
- Population Health and Ecological Monitoring, and
- Conservation Physiology.
Further details of this, and all future talks, are available on the Merseyside local group website: sites.google.com/site/rssmerseyside/