OBJECTIVES: Bayesian network meta-analysis (NMA) has become standard practice in evidence synthesis. NMAs typically only include randomized clinical trials (RCT) based on the hierarchy of evidence. Real world evidence (RWE) is increasingly used in health economics and outcomes research. The objective of this research was to review methods combining different levels of evidence and to compare their benefits and drawbacks. METHODS: A systematic literature review was conducted to identify methodological papers and published NMAs combining different study designs. Searches were conducted in PubMed and Embase. Extensive hand searches were also conducted and consisted of reviewing citations found in included publications and searching conference proceedings, health technology agencies’ websites, and methodological guidelines. RESULTS:  Four main methods for combining evidence from different study designs were identified: naïve pooling of all types of evidence, conducting a design-adjusted analysis, using non-randomized evidence as prior information, and running a three level hierarchical model. These methods were associated with advantages such as optimizing precision and network connection through the inclusion of more evidence, modelling bias directly by accounting for between-study type variability, understanding the bias non-randomized data may introduce into the analysis, and generating more generalizable NMA outputs. These methods were also associated with the following drawbacks: introduction of bias by including non-randomized trials, challenges associated with evaluating the bias associated with RWE studies. CONCLUSIONS: Given the lack of published guidance in this research area, the methods reviewed are considered exploratory and their perception by health technology assessment agencies is uncertain. While the three level hierarchical modelling approach seems to best allow for bias adjustment, further research remains to be conducted to address the bias inherent to pooling data from different sources. Refining these methods would help develop tools for a more generalizable comparative effectiveness assessment of health technologies.