Stroke prevention in patients with atrial fibrillation (AF) and co-morbid physical and mental health problems

Lead Supervisor
Dr Fiona Gaughran
Reader in Psychopharmacology and Physical Health
Dept of Psychosis Studies, Institute of Psychology, Psychiatry and Neuroscience (IoPPN), King’s College London

Dr Mark Ashworth (King’s College London)

Project Details


1.To use observational data to identify the proportion of patients with co-morbid AF and severe mental illness(SMI; schizophrenia, bipolar disorder) who are at high risk of stroke and who would benefit from anticoagulation treatment but are not receiving such. 

2.To examine the feasibility and acceptability of utilising digital prompts to increase the offer and uptake of anticoagulation treatment where identification or implementation shortfalls are identified in the management of AF in this population


Anticoagulation can prevent stroke in patients at risk because of AF but is poorly implemented in both primary and secondary care. Up to half of patients with AF do not receive an anticoagulant despite being at high risk of stroke as determined by a CHAD2AD2-VASc score‚â•2. Only 47% of patients with AF who experienced a stroke had been anticoagulated beforehand(Royal College of Physicians, SSNAP,2017). 

Patients with co-morbid AF and SMI have increased rates of stroke compared to general population but are less likely to be prescribed anti-coagulation. Furthermore, people with AF and co-morbid schizophrenia have higher mortality following thromboembolic events when compared with those without SMI (Sogaard et al, 2017).

Anticoagulation is not without risk. The HAS-BLEDscore was developed to assess 1-year risk of bleeding in patients with AF(Pisters et al, 2010). Any uptake of anticoagulation in AF patients needs to be balanced by assessment of bleeding risk, to ensure that potential benefits of stroke preventative treatment are not outweighed by potential harms. 

Work has already taken place using Cogstack@KCH to identify people with AF who are not receiving indicated anticoagulation. This project will expand on the KCH work to identify people with SMI in both Lambeth primary care (EMIS) and SLaM Mental Health Secondary Care (Cogstack@Maudsley) to improve access to treatment in this at-risk group. 

Data sources: 

  1. In the first stage of the project, we will use Lambeth DataNet (LDN), which contains primary care records in the form of pseudonymised Read and SNOMED codes, for all patients registered at GP practices in Lambeth. We will look for patients with co-morbid AF and SMI, calculate their CHAD2AD2-VASc and HAS-BLEDscores and determine whether they are on anticoagulant. Secondly, we will look at number of prescriptions per year (proxy for treatment adherence) in those who are already prescribed anticoagulation.

    We will validate the results linking LDN to CRIS (Clinical Record Interactive Search), a coded summary of mental health records held by SLaM MHFT and A&E attendance data. Using pseudonymised NHS numbers, LDN has already been linked to CRIS and research database linkage has been conducted and used to identify heart failure treatment shortfalls in SMI patients (Woodhead et al.,2016). 
  2. Secondly, we will conduct the validated search in SLaM using Cogstack@Maudsley to identify SLaM service users with SMI and AF who are not receiving adequate treatment for AF. 


We will create a digital tool to inform the identified patients‚ GP of the potential shortfall in AF identification and/or management and prompt treatment where appropriate. 

Following this intervention, we will repeat the analysis, stratified according to mental and physical co-morbidities to identify determinants of persisting low uptake of anticoagulation.


LDN and CRIS linked data require local IG permission and Caldicott Guardian approval. However, HRA research ethics not required for the use of fully anonymised patient-level data (confirmed, Oct 2017, Health Research Authority)


Atrial Fibrillation (AF), Severe mental illness (SMI), anticoagulation