Senior Director, Project Management PPD, part of Thermo Fisher Scientific, France
Background: Endpoints used for clinical trials cannot always translate into relevant endpoints in a real-world setting where control over clinical care is not possible. Therefore, identifying endpoints that are taking into account routine practice care, diagnostics, follow-up and other real-world parameters is key to ensure success in vaccine observational studies.
Objectives: To describe the methodological specificities of defining vaccine effectiveness and safety endpoints when implementing vaccine studies in a real-world (RW) setting.
Methods: We reviewed non-interventional study protocols we developed over the past 3 years and summarized the considerations from our assessment.
Results: In a RW setting, it is crucial that study endpoints account for the standard of care at the time of study implementation (for vaccination and disease diagnosis) as well as factors including geographical variability in data, participating site profiles and the evolving medical/therapeutic landscape. Therefore, the sensitivity and specificity of endpoints in a RW setting is expected to differ from the clinical trial setting, where study conditions are standardized and endpoints are defined to have a high sensitivity and specificity. However, they may be more relevant clinically or from a public health perspective. As an example of the impact of study timing on endpoints, observational studies on COVID-19 vaccines implemented during the peak pandemic period could rely on testing to detect positives cases. However, the frequency of testing has now decreased significantly globally; this change of standard of care has already impacted the level of reporting of positive cases, meaning that endpoints for vaccine effectiveness studies implemented now would need to be defined differently. Another example is the specificity of endpoints in longitudinal safety follow-up studies, where vaccinated subjects are followed over time but not tied to any particular site, and data is collected primarily from the subject. The definition of safety endpoints then implies a two-step approach: the first step being the safety event collected from the subject (ideally very sensitive and not specific) and the second step being a medical confirmation of the event (to increase endpoint specificity). In this case, safety endpoint collection relies mostly on subject compliance and physician judgement.
Conclusions: Real-world evidence studies of vaccine effectiveness and safety are run in a very different context than clinical trials, are very dependent on standard of care practices for diagnosis and treatment, and endpoint definition should therefore be clinically relevant while meaningful in a real-world setting, operationally feasible and justified in view of the study objectives.
