One of the main symptoms of dementia is memory loss. Other symptoms include agitation/aggression. These symptoms make living with dementia much more difficult so treating them is important. One drug used in the UK is called risperidone. There is evidence that risperidone helps some people, but it has severe side effects, like stroke. Stroke is caused when clots in blood vessels cut of the supply of blood to the brain. It is likely that risperidone causes hundreds of strokes per year in people with dementia. Minimising stroke risk during risperidone treatment is important because there aren’t many other treatment options for agitation/aggression in dementia. A big problem is that we do not know whether a specific person will have a stroke if they take risperidone. This is what we want to change in this project. We think we can use data about a person’s medical history to predict how likely they are to have a stroke if they take risperidone. To do this, we will analyse the anonymous GP records of thousands of people and compare two groups:
1: people with dementia who have been prescribed risperidone.
2: people with dementia who have never been prescribed risperidone.
We will then look at how often strokes happen in these groups. Then we will find out if parts of people’s medical history make it more likely that they will have a stroke if they take risperidone. We hope this information will be used by doctors to help make prescribing risperidone safer.
Research question
Can individual clinical history be used to estimate stroke risk in dementia patients prescribed risperidone?
Background
The antipsychotic risperidone is used for agitation in dementia but it several severe side effects including stroke. A notable gap in risk-benefit decision making is that doctors do not know how likely it is that risperidone will cause a stroke in a given individual. We know that risperidone effects cardiovascular disease (CVD) biological pathways, suggesting that CVD-related medical history might influence vulnerability to side effects. We will examine this link by employing individual risk prediction models that we developed in diabetes (eRAP 22-002000).
Aims and methods:
Aim:
To estimates heterogeneity in stroke risk if prescribed risperidone.
Objectives:
1) Quantify the absolute risk of stroke following risperidone prescription.
2) Estimate heterogeneity in individual excess stroke risk if prescribed risperidone.
We will use CPRD primary care data (linked with Hospital Episode Statistics, Office for National Statistics Mortality data and Index of Multiple Deprivation data), to compare stroke risk in those prescribed risperidone (Risperidone Group) and those never prescribed risperidone (Comparator Group). The primary outcome will be stroke within 1 year of first prescription. Propensity score matching will be used to match up to 5 Comparator Group patients with each person in the Risperidone Group. First, we will estimate the relative risk of stroke in the Risperidone Group vs the Comparator Group. Second, we will evaluate heterogeneity in stroke risk across pre-specified subgroups defined by a history of CVD using the same approach. Finally, we will develop a prediction model to estimate excess risperidone-associated stroke risk for individual patients based on their characteristics. We will use Cox proportional hazard models, incorporating interaction terms between treatment group and individual patient-level clinical features. We hope these stroke risk estimates can inform more judicious prescribing in future.
Primary: ischaemic and haemorrhagic stroke;
Secondary: transient ischaemic attack, heart failure, myocardial infarction, hypertension, angina, hypocholesteraemia, coronary or peripheral artery disease, infections, fall, fractures and all-cause mortality.
John Dennis - Chief Investigator - University of Exeter
Byron Creese - Corresponding Applicant - University of Exeter
Christoph Mueller - Collaborator - King's College London (KCL)
Nalamotse Joshua Choma - Collaborator - University of Exeter
Nefyn Williams - Collaborator - University of Liverpool
William Henley - Collaborator - University of Exeter
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation