Patient-directed data capture for clinical research: A pilot study testing two platforms

Patient-directed data capture enables individuals to access and share their electronic health record (EHR) data, potentially supporting patient engagement while also reducing data fragmentation. Although many vendors offer patient-directed data capture, little is known about how feasible or reliable these approaches are in practice.

We conducted a single-site pilot study to evaluate the feasibility of different patient-directed data capture platforms. Thirty-four participants hospitalized for heart failure at Duke Health participated; 26 engaged with Hugo Health, which used application programming interfaces (APIs) to retrieve data, and 17 with Datavant, which relied on records requests. Nine participants used both platforms.

Data retrieval success varied by method: the API approach retrieved 88 % of requested records, while the records request approach retrieved 20 %. Concordance analysis showed the API approach captured additional terms from external systems (e.g., sacubitril/valsartan, 75.0 %) but missed some local EHR diagnoses (e.g., coronary heart disease, 50.0 %). The records-request method captured less external data and had higher rates of data omission. Only 22 % of participants reported willingness to reuse their assigned vendor, and 33 % required onboarding support. Technical challenges included the need to process JavaScript Object Notation (JSON) files (API approach) and to manually abstract documents from PDF files (records request approach).

This pilot study demonstrates that patient-directed data capture is technically possible but shows substantial variability by method. While API-based retrieval achieved high success rates, records request achieved limited success, underscoring the need for improved workflows and user support to make this approach broadly feasible.