BACKGROUND: The term “real‐world evidence” (RWE) has become increasingly common in recent years. Yet little is known about what constitutes RWE in published literature and whether RWE is similar across therapeutic interventions, such as drugs and devices.
OBJECTIVES: We evaluated RWE in 2017 publications overall and for drugs and devices separately.
METHODS: Pubmed and Embase were reviewed for English language 2017 titles and abstracts in which the term “real world” appeared. The following were extracted based on information in the title/ abstract: therapeutic area, exposure type, study design, primary outcome, timing of outcome, country and data source. Descriptive analyses were performed.
RESULTS: There were 1045 hits for “real world” publications in 2017. Of these, 315 were excluded because they lacked an abstract (n = 93) or were not related to provision of health care (n = 222); 730 remained in the analysis. Overall, most studies were retrospective (67%) vs prospective (31%); used a data source of medical records (32%), primary data collection (28%), administrative or pharmacy data (12%), registries (10%), and other/not reported (19%); and, were focused in 3 therapeutic areas (TAs): cardiovascular (CV, 23%), oncology (19%) and infectious disease (ID, 14%). Eight percent focused on short‐term outcomes (<30 days). Real‐world data from over 50 countries were analyzed in the studies, with the US most frequently noted. Most studies assessed RWE of drugs (67%); 15% assessed devices and 18% other interventions. Drug RWE studies were mostly retrospective (72%) vs prospective (26%); used a data source of medical records (34%), primary data collection (23%), or administrative/pharmacy data (14%); evaluated efficacy (56%) or treatment patterns (13%); and were focused in one of 4 TAs (oncology [22%], ID [20%], CV [15%], or metabolic [10%]). Drug studies also tended to focus on longer term 410 ABSTRACTS outcomes (5% <30 days). Device RWE studies were split between retrospective (52%) and prospective (47%); used a data source of primary data collection (34%), medical records (31%) or a registry (17%); evaluated efficacy (53%) or safety (19%); were most often focused on CV TA (63%). Device studies also assessed shorter term outcomes more often than drug studies (19% <30 days).
CONCLUSIONS: In 2017, the published literature on RWE studies reported using a wide variety of outcomes and methods, with important differences between drugs and devices. Efforts to standardize reporting of RWE studies should cover this breadth and apply across therapeutic interventions.
