Globally, major research funders, governments, charities, and patient groups have emphasised the health inequalities experienced by people with severe mental illness (SMI). SMI is associated with high levels of morbidity and premature mortality, with a mean age of death 15 years younger than the general population. Schizophrenia, bipolar disorder and other non-organic psychotic illnesses have a lifetime prevalence of approximately 1% each, and are in the top twenty diseases in terms of disability. This burden has increased in the past 30 years.
SMI is challenging to treat; it can take a long time to find the optimum treatment for a particular patient and even then they are frequently left with residual symptoms and adverse effects. Approximately one-third of patients show a limited response to medication. Treatment options are similar across disorders: antipsychotic, mood stabiliser, antidepressant and anxiolytic medications. There have been almost no new effective drug regimens for these patients in the past 30 years. These challenges require novel approaches to advance the field, reduce illness burden for the individual and society and close the health inequality gap.
The aim of this project is to develop a series of prediction tools to guide medication selection across different illness stages and comorbidities. Existing effective drug treatment will be optimised for treatment of an individual’s symptoms, whilst minimising their risk of experiencing adverse events.
The aim of this project is to develop a series of personalised prediction tools to guide medication selection for people with severe mental illness (SMI: schizophrenia, bipolar disorder and other psychotic illness) across different illness stages. Existing effective drug treatment will be optimised for treatment of an individual’s symptoms, whilst minimising their risk of experiencing adverse events.
There are three key areas in which individual level prediction could dramatically improve illness trajectory: treatment choice for a first episode of SMI, early identification of likely treatment resistance and prediction of adverse events. We will develop prediction models with machine learning and epidemiological approaches for each of these scenarios.
We will use a cohort of adults newly diagnosed with SMI and first exposed to one of the treatments below.
First episode SMI: We will develop prediction models for time to treatment failure of common drugs of choice in first episode psychosis (aripiprazole, quetiapine, risperidone, olanzapine) and first episode bipolar disorder (quetiapine, olanzapine, sodium valproate and lithium). We will define treatment failure as any psychiatric hospitalisation, or switch to/add-on of an alternative psychotropic medication.
Treatment resistance: We will develop models to predict individuals at risk of treatment resistance in SMI, commonly defined as inadequate response to at least two drugs. In CPRD we will define this as exposure to the third psychotropic medication.
Adverse events: Weight gain trajectories will be predicted in all individuals commencing aripiprazole, quetiapine, risperidone, olanzapine, sodium valproate and lithium. We will also predict time to 5% weight gain, type II diabetes, hypercholesterolemia, sedation and akathisia. and hypertension.
We will develop prediction models using all available data at the treatment commencement and compare epidemiological methods (Cox, competing risks) with machine learning methods (Ensemble learners).
Pharmacological treatment response:
Drug treatment cessation; drug treatment augmentation (add in of another psychotropic medication); psychiatric hospitalisation; self-harm.
Adverse effects of pharmacological treatment:
Weight gain trajectory; weight gain; akathisia; sedation; non-psychiatric hospitalisation.
Joseph Hayes - Chief Investigator - University College London ( UCL )
Naomi Launders - Corresponding Applicant - University College London ( UCL )
Alvin Richards-Belle - Collaborator - University College London ( UCL )
David Osborn - Collaborator - University College London ( UCL )
Gareth Ambler - Collaborator - University College London ( UCL )
Glyn Lewis - Collaborator - University College London ( UCL )
Ian Wong - Collaborator - University College London ( UCL )
Irene Petersen - Collaborator - University College London ( UCL )
Kenneth Man - Collaborator - University College London ( UCL )
Magnus Boman - Collaborator - UCL Division of Psychiatry
Naomi Launders - Collaborator - University College London ( UCL )
Sarah Hardoon - Collaborator - University College London ( UCL )
HES Accident and Emergency;HES Admitted Patient Care;HES Outpatient;ONS Death Registration Data;Patient Level Index of Multiple Deprivation;Practice Level Index of Multiple Deprivation