Social and Economic Predictors of the Severe Mental Disorders: Data Linkage between the 2011 UK Census and Electronic Mental Health Records in London: Metadata and Documentation, 2007-2019

People with severe mental illness (SMI) have significantly worse outcomes than the wider population. Despite this, there has been limited research on the impacts of social and environmental conditions on psychoses and other severe mental health conditions, and few studies have focused specifically on the experiences of minority ethnic groups. The conclusions from many of these studies are affected by people being lost to follow up, sample sizes being too small, and with too few people of an ethnic minority background taking part, in order to enable robust conclusions. Traditional studies which follow people with SMI over time are time-intensive, expensive to run and logistically challenging to conduct, as tracing people over time is difficult.

In order to address this gap in knowledge, the Social and Economic Predictors of the Severe Mental Disorder (SEP-MD) data linkage individually links data extracted from electronic mental health records, mortality records and the 2011 UK Census for England. Electronic mental health records were sourced from the South London and Maudsley NHS Foundation Trust (SLaM), which is the main provider of secondary mental health care for a catchment area of approximately 1.3 million individuals in four London boroughs. This area is highly urban and ethnically diverse. Using these sources of administrative and routine data allows for a larger sample and reduces the impact of sampling biases.

In the SEP-MD data linkage, 220, 864 mental health records matched to records in the 2011 Census after data cleaning (see Cybulski et al (2024) BMJOpen 14(1):e073582. for more details). Alongside these records, a further 581,209 population controls with only Census data were sampled from the same locality. The data sources were also linked to the Office for National Statistics mortality files (cause-specific mortality with date of death).

The SEP-MD data includes variables from the Census, such as employment status; self-ascribed ethnicity and housing situation, alongside detailed mental health variables, such as psychiatric diagnoses; prescribed psychiatric medications and admissions to inpatient secondary mental heath services.