Objective: To develop a quantitative method for designing appropriate longitudinal follow-up frequency. We demonstrate how to use this new method using a progression study of Parkinson’s disease (PD) cognition.

Background: Current practice of frequency of longitudinal assessments in PD cognition research is mainly based on experience and preference, tradition, and availability of resources. Although a couple of statistical methods have been proposed to offer guidance for selection of follow-up time-points, they require knowing the true distribution of the risk of developing the event of interest. Since in practice the true distribution is not known, the utility of these statistical methods is limited. As such, there is a need to develop an unbiased and robust method for justifying the choice of different follow-up frequencies to monitor progression from a non-demented state to PD dementia (PDD).

Method: We propose a novel method to evaluate the statistical appropriateness of follow-up frequency by evaluating the impact of varied follow-up intervals in longitudinal studies. Specifically, we evaluate the bias in estimating the hazard ratio of developing the event (e.g., conversion to PDD) using the Cox model under a variety of follow-up schedules.

Results: Using the proposed method, we validate the current longitudinal follow-up frequency among PD patients enrolled in our PD Center. We find that the current follow-up frequency (annually for the first four years post-baseline, and then biennially thereafter) is appropriate for studying conversion to cognitive impairment (i.e., mild cognitive impairment or dementia) in PD patients. Although we apply the method to PD patients, it can be used in other biomedical research areas (e.g. Alzheimer’s disease) that have sufficient historical data.

Conclusion: Our method can guide the selection of longitudinal follow-up frequency in an unbiased and robust way. Thus, it will help to save unnecessary study cost and reduce patient burden without sacrificing the accuracy in estimating the association of interest.

« Back to MDS Virtual Congress 2020

MDS Abstracts - https://www.mdsabstracts.org/abstract/quantitative-method-for-designing-appropriate-longitudinal-follow-up-frequency-with-application-to-parkinsons-disease-cognition-research/