User experience and feasibility of CVD risk models: a study on clinician and patient expectations, and implementation in primary healthcare with P-CARDIAC pilot study in Hong Kong

Background A newly developed machine-learning-driven cardiovascular disease (CVD) risk prediction tool, the Personalised CARdiovascular DIsease risk Assessment for Chinese (P-CARDIAC), tailored for the Hong Kong population, has demonstrated superior predictive performance compared with existing risk prediction tool. This study aims to explore the acceptability and expectations of P-CARDIAC among clinicians and the general public, and to assess its feasibility within pharmacist-led services in local primary healthcare systems. Methods A cross-sectional study was conducted to investigate the awareness and expectations of risk prediction tools among clinicians and the general public. A pragmatic pilot study was carried out to implement P-CARDIAC in pharmacist-led services at local primary healthcare centres. Data analysis included descriptive statistics, univariate regression modelling and reliability assessments of validated measurement scales. Results For the cross-sectional study, a total of 113 of the general public and 17 clinicians responded to the questionnaire. The general public demonstrated low awareness but high health-seeking behaviour related to CVD risk prediction tools. Clinicians, especially cardiologists, reported limited experience with such tools due to unavailability and limited user-friendliness. In the pilot study, 15 participants engaged with pharmacist-led services incorporating P-CARDIAC. They expressed positive perceptions of the tool, believing that it could support better medication adherence and disease management. The pharmacist-led services incorporating P-CARDIAC were well-received by participants. Conclusion P-CARDIAC demonstrates promise as a local risk prediction tool to enhance cardiovascular care in Hong Kong. The study underscores the importance of improving awareness and adoption of such CVD risk tools.