Regulators have demonstrated increasing interest in and receptiveness to leveraging RWE for decision-making processes. Addressing continuity of coverage (e.g., enrollment and disenrollment) as relevant to data availability in candidate real-world data (RWD) sources is discussed in the FDA’s 2020 RWE guidance documents, but this topic has received little attention in the RWE-related methodological literature. Although there are considerations specific to the research question and the nuances of each data source, the development of customizable evaluation frameworks and planning tools can facilitate the conduct of clear, cogent analyses that demonstrate whether the candidate RWD source has the requisite coverage and comprehensiveness over time to be utilized for regulatory-grade studies.

In this conceptual paper, we provide a high-level framework for examining continuity of coverage related to data availability/comprehensiveness in EHR and administrative claims data. We also discuss analytic coverage assessment strategies specific to major components of study design (i.e., research question, study period/temporal anchors, and population).

Issues related to data coverage can arise in any phase of the study period when using EHR or claims data. For example, in retrospective cohort studies, the pre-index period may include a baseline covariate assessment window, eligibility assessment window, and washout periods, but data may be unavailable or inadequate to assess medical history. Conversely, in the follow-up period, some patients may reach an outcome during the study period, while others are lost-to-follow-up or otherwise censored, and the data may fail to represent the outcome. Describing patterns related to differential loss to follow-up is one important component of assessing coverage. Though challenging, linkage of multiple data sources to obtain a more comprehensive profile of each patient’s journey is possible through technological solutions, such as HIPAA-compliant tokenization approaches. We describe opportunities to identify and mitigate issues arising from lack of data continuity.