The use of routine healthcare data in the development of efficient personalised medicine trials in heterogeneous critical care populations (360G-Wellcome-216606_Z_19_Z)

I will use routine healthcare data to undertake feasibility work to inform novel clinical trial design and trial populations in heterogeneous populations. There is an unmet need for this in intensive care research. I will frame this around critically ill patients with CVD following on from my PhD, however this will be applicable to other disease/pathologies with heterogeneous populations. I hypothesise that critically ill patients with co- existing CVD may benefit from higher oxygen delivery to the myocardium during critical illness, preventing myocardial infarction and subsequent cardiovascular mortality. Key goals: 1\. Identification of current outcomes for critically ill patients with CVD. I will link high quality Scottish routine healthcare datasets for all admissions to Scottish ICUs 2010-2018. I will characterise cardiovascular and non-cardiovascular outcomes and healthcare trajectories for these patients. 2\. Identification of patients in whom CVD has the highest attributable mortality. This group would be the target population for a cardiovascular intervention trial. I will use prediction modelling, latent variable analysis and propensity score analysis to address this. 3\. Identification of intermediate short-term outcomes using both routinely collected time series ICU physiological data, national datasets and existing trial datasets. 4\. Development of efficient trial methodology, including Bayesian adaptive trial designs.