OBJECTIVES: Randomized controlled trials (RCTs) are regarded as the highest level of evidence. However, due to stringent participant selection criteria and controlled academic environments, RCTs are sometimes criticized for a lack of external validity. In contrast, real-world evidence (RWE) is more representative of real-world clinical practice and offers superior external validity. The need to incorporate RWE alongside RCTs in regulatory or decision-making processes is increasingly recognized. However, methods to quantitatively integrate evidence from randomized and non-randomized studies remain a growing area of research. This scoping review aims to identify available methods for evidence integration across heterogeneous study designs and qualitatively evaluate their strengths, limitations and appropriateness.

METHODS: PubMed, EMBASE, Medline and Cochrane Library were systematically searched using relevant keywords and MeSH terms, supplemented by grey literature searches and reference scanning. A two-stage screening process was conducted based on pre-determined eligibility criteria: title and abstract screening by the primary reviewer, followed by a collaborative full-text review by multiple reviewers. Data extraction focused on study characteristics, objectives, methods used or reviewed, strengths and limitations of the method.

RESULTS: The search yielded 3633 abstracts. After initial screening and full-text review, 68 studies were included. Data extraction produced 14 different evidence integration strategies, including but not limited to naive pooling, confidence profile method, propensity score-integrated power prior approach, Bayesian hierarchical modelling, Bayesian network meta-analysis with informative priors, target trial emulation, and etc. The strengths, limitations, assumptions, data requirements and other feasibility considerations of each method have been discussed in detail.

CONCLUSIONS: This scoping review serves as a comprehensive guide for selecting the most appropriate evidence synthesis method. Future work in this area should focus on promoting the application and uptake of RWE and RCT evidence synthesis methods in the policy space, developing methodological guidance and standards, and exploring the application of advanced machine learning methods in this area.