Calculating follow-up completeness: a comparison of multiple methods under different simulated scenarios and a use case

Abstract

Background and Objectives

Completeness of follow-up is a crucial aspect of data quality in cohort studies and clinical trials. This study aims to provide an overview of different methods to calculate follow-up completeness. In addition, the performance of these methods is tested in several scenarios using simulated datasets and a use case, with the aim of guiding researchers in selecting the most appropriate method for their data.

Methods

The literature was searched for methods of quantification of follow-up completeness. These methods were investigated in simulated datasets, in which the true completeness of follow-up was known. A total of 27 different scenarios were investigated, based on different survival distributions, total proportions of dropout of participants, and different time points of dropout. The methods were also investigated using real-world mortality data from the population-based Rotterdam Study cohort. Kaplan–Meier curves were used to depict observed survival, and completeness of follow-up was calculated in percentages using a freely available GitHub package developed by our research group.

Results

In total, six methods were found in the literature for quantification of follow-up completeness. Overall, two methods (the simplified person-time method and the modified Clark's Completeness Index C∗) were closest to the true follow-up completeness in the 27 scenarios.

Conclusion

Researchers should make attempts to report follow-up completeness. This simulation study may assist researchers in selecting the most appropriate method to calculate follow-up completeness in different scenarios.