Minimizing Missing Data in Clinical Trials

Abstract

Missing data in clinical trials remains an ongoing concern. With the expansion of data privacy efforts and the consequent inability to contact trial participants for follow-up, the magnitude and reasons of missing data in clinical trials have shifted. The impact of missing data on a clinical trial results largely depends on the reason why the data are missing. When data are missing at random, the influence on the study’s conclusions may be minimal. In contrast, when data are missing not at random, the integrity of the trial results can be compromised. For example, if participants are lost to follow-up or withdraw consent due to adverse reactions or side effects like bleeding, then the remaining participants may disproportionately represent those who can tolerate the therapy or are less frail, leading to biased conclusions regarding the drug’s safety and efficacy, a phenomenon referred to as differential censoring. The best strategy is to minimize missing data from the outset of the trial, which includes designing an informed consent form that sets the expectation that and the alternate methods by which outcomes will be tracked even if the participant elects to discontinue study treatment. Likewise, rather than waiting until the end of the study, missing data should be continually and proactively minimized during the trial by offering patients more convenient and infrequent visit strategies or follow-up through relatives or other health care professionals as needed. Also, it is critical to characterize the basis for data missingness so that its impact on trial interpretation can be better assessed. This article provides a roadmap to successfully implement all of these strategies to minimize missing data.