Background:
Routinely collected clinical data provide a powerful resource for evaluating interventions for real-world patient populations because they are large, naturalistic resources; however, internal validity is frequently impacted by confounding. Hernan and colleagues have proposed a trial emulation framework which mimics a target trial using observational data. In this context propensity scoring approaches are often used to handle multiple confounders.
Novelty & Importance:
Propensity scoring is not a single method and applied researchers find that there are many questions that need answering in practice. This project aims to provide clarity and guidance on using propensity scoring for trial emulation with observational data. To this end a tool will be developed that can be applied across different types of EHR data, advancing methodological capability and supporting decision-making for trial emulations.
Aims & Objectives:
– To develop a decision support tool for principled propensity scoring in target trial emulation using EHRs
– To develop software tools to enable propensity scoring in practice, e.g. create wrappers to convert software output into clinically meaningful effect size estimates
– To develop exemplar target trial emulation studies using the Maudsley’s Clinical Record Interactive Search (CRIS) EHR data resource
The Maudsley’s CRIS platform has supported extensive research since 2008 by allowing access to de-identified mental health EHRs on over 500,000 cases from a diverse south London geographic catchment of 1.3m residents. Case studies might include antipsychotic polypharmacy and/or lithium discontinuation in late life, but these will be informed by both a review and patient/carer input.
References:
Scola G., Chis-Ster A., Bean D., Pareek N., Emsley E. & Landau S. (2023) Implementation of the trial emulation approach in medical research: a scoping review, BMC Medical Research Methodology 23: 186; DOI: 10.1186/s12874-023-02000-9
Kadra G, Stewart R, Shetty H, MacCabe JH, Chang CK, Kesserwani J, et al. (2017) Antipsychotic polypharmacy prescribing and risk of hospital readmission. Psychopharmacology 235: 281–289. DOI: 10.1007/s00213-017-4767-6

