How to improve electronic medical record data extraction and validation for secondary use: a proposed framework for a streamlined process

There is little research addressing best practice for extracting Real World Evidence (RWE) from Electronic Medical Records (EMRs). This study aims to identify key barriers and enablers across people, processes, information, and technology to propose a practical framework that supports an effective and streamlined approach to RWE extraction and validation. From February to October 2024, we conducted a mixed-methods study with clinicians, technical users, and administration staff in three phases: a pilot survey ( n = 3), a questionnaire survey ( n = 50), and stakeholder interviews ( n = 12). Cronbach’s Alpha assessed response reliability (α > 0.7). Group differences in competence and familiarity were assessed using the Kruskal–Wallis test with Dunn’s post hoc comparisons. Findings were mapped to the three-horizons framework, analyzed, and used to develop a framework aimed at facilitating and streamlining EMR data extraction and validation processes. This study identified 28 challenges and 13 enablers across people, process, information, and technology within the three-horizon framework. Horizon 1, focused on establishing digital infrastructure, was already achieved at the study site and was not relevant to data extraction and validation, so no results were reported. Horizon 2 revealed people- and process-related challenges including resource constraints, unclear data ownership and lack of standardization. Information and technology challenges included knowledge gaps, integration issues, and limited tools. Key enablers were strong privacy and security awareness. Horizon 3 involved unclear clinical and technology ownership and limited privacy awareness and highlighted enablers in data landscape and security knowledge. Key challenges included documentation gaps, inconsistent request and validation procedures, fragmented catalogue/tooling, and unclear clinical–technical ownership. Clinicians reported significantly lower awareness of data request support and lower confidence in data validation compared to other roles (Kruskal–Wallis p < 0.05; Dunn p < 0.05). The current state of data extraction processes presents several challenges and includes several notable enablers in the extraction of RWE from EMRs. Mapping the findings to the three-horizon model informed the development of a framework that can be utilized to implement clear, consistent data management processes, standardize toolsets, and integrate platforms, with the aim of improving operational effectiveness. These findings informed a practical, staged framework intended to standardize request and validation artifacts, clarifies clinical/technical ownership, and incorporates currently used tools, with the potential to reduce resubmissions and manual corrections.