In clinical studies, repeated measurements are frequently collected, such as a patient’s blood and urine samples or different imaging modalities (e.g., echocardiography), and linked to the patient’s risk of adverse outcomes. In clinical practice, it is also medically relevant to use all available information (baseline and follow-up) to accurately detect a disease’s dynamics and to profile an individual’s prognosis. Such dynamic prognostication could be integrated into a clinical decision-making process, and could be particularly useful for tissue-specific targeting of therapies. However, several features of these repeatedly measured data should be considered before making such inferences. We discuss these considerations in our article recently published in the Kidney International.

In the article, we describe how joint modeling of repeatedly measured and time-to-event (i.e., survival) data may help to assess a disease’s dynamics and to derive a personalized prognosis. For this purpose, joint modeling combines linear mixed-effects models and Cox regression model to relate patient-specific trajectory of a certain marker to the patient’s personalized prognosis. We describe several aspects of this relationship between different forms of time-varying markers, for example a marker’s value or its rate of change, and the endpoint of interest using real examples of  published clinical studies to illustrate the aforementioned aspects of the longitudinal data.

To find out more about the application of the dynamic prediction modeling in patients with chronic heart failure you can click here.