What is real-world data? A review of definitions based on literature and stakeholder interviews

Background: Health Technology Assessment (HTA) agencies increasingly require robust evidence on relative effectiveness of products. Randomised controlled clinical trials (RCT's) provide robust evidence on efficacy, but not on effectiveness. Real-World Data (RWD) may provide an alternative source for effectiveness data. However, despite increasing recognition of the value of real-world data (RWD) for drug assessment, consensus on the definition of RWD is lacking. Objectives: In order to shed light on similarities and differences between available RWD definitions, we aimed to review definitions for RWD from the perspective of different stakeholders. Methods: A literature review and stakeholder interviews were used to compile data from 8 groups of relevant stakeholders. Outputs from documents and interviews were pooled and subjected to coding analysis. Definitions identified were classified into one of four pre-defined categories: 1-Data collected in a non-RCT setting, 2-Data collected in a non-interventional/non-controlled setting, 3-Data collected in a non-experimental setting (ie, not based on a pre-established study protocol), and 4-Other (ie, do not fit into three categories above). The frequency of definitions identified under each category was recorded. Results: In total 52 documents and 20 interviews were assessed. 38 definitions were identified: 20 out of 38 definitions (53%) were category 1 definitions, 9 (24%) were category 2 definitions, 5 (13%) were category 3 definitions and 4 (11%) were category 4 definitions. Differences were identified both between, and within, definition categories. For example, opinions differed on the aspects of intervention (treatment assignment, patient selection, and patient follow-up) with which non-interventional/non-controlled settings should abide. No definitions were provided in 2 interviews or identified in 32 documents. Conclusions: The majority of definitions assessed defined RWD as data collected in a non-RCT setting. However, a considerable number of alternative definitions diverged from this concept. Moreover, a significant number of authors and stakeholders did not have an official, institutional definition for RWD. Persisting variability in stakeholder definitions of RWD may lead to confusion among different stakeholders when discussing the use of RWD in decisionmaking.