OBJECTIVES: There is an increased demand for oncology RWD to support decision-making by health-technology assessment (HTA) bodies, particularly when there is a lack of long-term comparative data. The extent to which insights generated from US RWD should be used to address uncertainty in ex-US markets is commonly questioned. This study aimed to identify challenges in assessing the transportability of evidence derived from real-world US electronic health records (EHR) and proposes a framework for mitigating risks to HTA decision-makers.

METHODS: We focused on treatment patterns and outcomes published from German markets between 2015 and 2020 to those in the Flatiron Health US EHR-derived databases with a goal of enumerating challenges in replicating results between countries. We identified studies in four disease states (multiple myeloma, non-small cell lung cancer, breast cancer, and bladder cancer) published using German real-world databases. We categorized observable and non-observable data elements which could lead to dissimilarities between the insights from the German and US data studies.

RESULTS: We structured our findings into two ranks - a set of three core themes and four to five factors within those themes affecting the representativeness, and hence transportability, of real-world data. The identified themes were patient-characteristic differences, setting-of-care differences, and treatment pattern differences. Accounting for these in the pre-specification process allowed for a clearer understanding of whether inferences generated from the Flatiron Health EHR-derived data source may be transportable to other countries of interest for the purposes of HTA.

CONCLUSIONS: Differences in a given target population may impact the transportability of causal inferences generated from RWD. Some differences may be adjusted for while other, potentially unknowable differences, may not. Clearly characterizing these differences in a consistent framework promotes a more systematic approach during the pre-specification process, allowing for increased representativeness in the sample population and more transparency during a given HTA submission.