Joint modelling on clinical trials in clinical journals: a systematic review

BACKGROUND: Joint modelling (JM) provides an advanced statistical framework for the simultaneous analysis of longitudinal data and time-to-event outcomes. This review aims to assess clinical findings generated by JM in clinical trials and to identify common applications of JM within these studies. METHODS: We systematically searched electronic databases to identify clinical trial reports that applied JM and were published in clinical journals. From each report, we extracted information on JM applications and summarized the key characteristics of the studies. RESULTS: We identified 44 eligible clinical trials. Most studies used JM to evaluate how time-to-event outcomes were adjusted, mediated, or predicted based on longitudinal data, such as health-related quality of life or biomarkers. Six studies addressed methodological adjustments for missing-not-at-random data, informative dropout, and bias arising from competing risks. Additionally, three studies employing Bayesian JM demonstrated how complex models facilitated multivariate and dynamic predictions of survival benefits. CONCLUSIONS: Clinical application of JM remains limited, although statistical methodologies for JM have advanced considerably. Progress in translating JM into clinical research practice has been modest. To support its effective use, clinical researchers require a deeper understanding of how to interpret outcomes generated by JM and how to integrate them into an evidence-based framework.