Development of an objective measure of clinical recovery after cardiac surgery for use in Randomised Controlled Trials

Abstract

Randomised controlled trials (RCTs) play a vital role in the evaluation of cardiac surgery interventions. Identifying clinically relevant, objective outcomes is important, given the challenges with blinding many interventions. Hospital length of stay (LoS) is often reported but at risk of bias in unblinded trials due to a potential lack of equipoise. Alternative outcomes to describe patient recovery over the entire post-operative period may be beneficial.

In this thesis, I used data from eleven cardiac surgery RCTs that each captured fifteen complications that are expected after cardiac surgery. Routinely collected Intensive Care Unit (ICU) data for trial participants were obtained when possible. Associations between complications were explored to identify those commonly experienced together and patients were grouped using Latent Class Analysis according to the profile of complications experienced during their hospital stay.

I developed two continuously scaled outcome measures for use in future RCTs: an overall complication score (including information about all expected complications between surgery and discharge) and an overall recovery score. The complication score combined two separate scores which I developed to (1) quantify the impact of complications on LoS and (2) quantify patients’ and surgeons' rating of perceived severity of each complication. The overall recovery score combined this complication score with an early recovery score, calculated using measurements taken over the first 24 hours in ICU. The early recovery score quantified the association of these measurements with LoS and the final overall recovery score also included patient vital status at discharge.

Information used to calculate the scores can be assessed against documented criteria, supporting their use in unblinded trials. Good agreement was shown between the scores and LoS, used as a proxy measure of recovery throughout, in an independent dataset. Further work is required to assess the construct validity of the scores.

Date of Award	18 Mar 2025
Original language	English
Awarding Institution	University of Bristol
Supervisor	Chris A Rogers (Supervisor) & Barnaby C Reeves (Supervisor)