BACKGROUND
The use of predictive models and decision support systems can potentially make a substantial difference in clinical management and patient care in cardiac surgery. However, in order to develop decision support systems that are fit for purpose, working with clinical domain experts and stakeholders from the start is essential to gather key requirements early on in the system design process, prototyping, testing, and evaluation.
OBJECTIVE
This study aimed to capture the current challenges in cardiac surgery, clinical processes to avoid these challenges, and cardiac surgeons’ and anesthetists’ priorities when developing new prediction models and decision support tools in cardiac surgery.
METHODS
In total, n=12 in-depth semi-structured interviews were conducted across 3 Scottish cardiac centers with cardiac surgeons and cardiac anesthetists. The interview transcripts were analyzed, following a thematic analysis framework and using three a priori themes: “challenges in cardiac surgery”, “current processes to avoid adverse outcomes in cardiac surgery”, and “clinicians’ priorities for new clinical risk prediction models”.
RESULTS
A thematic analysis identified challenges regarding adverse postoperative outcomes connected to the changing cardiac patient population and surgical procedures, challenges in communication between health professionals and patients, and data collection. It was found that preoperative assessment clinics play the most important role when reducing the risk of patients developing adverse postoperative outcomes. Risk prediction tools were largely not used to avoid adverse outcomes, but rather for auditing and documentation purposes.
CONCLUSIONS
The findings of this study suggest that there is a need to increase research efforts on more personalized risk prediction tools and expand the remit of predictions to postoperative complications rather than the more traditional focus on mortality. While there are numerous prediction models in cardiac surgery, the involvement of potential users of such models in the development process is rare. Hence, this study provides a contribution by presenting the cardiac surgeons’ and anesthetists’ priorities for a clinical prediction model.