Background:
Clinical trials are the primary mechanism for advancing clinical care and evidenced-based practice, yet challenges with the recruitment of participants for such trials are widely recognized as a major barrier to these types of studies. Data warehouses store large amounts of heterogenous clinical data that can be used to enhance recruitment practices, but multiple challenges exist when using a data warehouse for such activities, due to the manner of collection, management, integration, analysis, and dissemination of the data. A critical early step in leveraging the DW for recruitment purposes is being able to match trial eligibility criteria to discrete and semi-structured data types in the data warehouse, though trial eligibility criteria tend to be written without concern for their computability. We present the multimodal evaluation of a web-based tool that can be used for prescreening patients for clinical trial eligibility.
Results:
Clinical trials from the Division of Cardiology and the Department of Family Medicine were used for this multimodal evaluation, which included a validation study, usability study, and a heuristic evaluation. From the results of the validation study, the software demonstrated a positive predictive value (PPV) of 54.12% and 0.007%, respectively, and a negative predictive value (NPV) of 73.3% and 87.5%, respectively, for two types of clinical trials. Heuristic principles concerning error prevention and documentation were characterized as the major usability issues during the heuristic evaluation.
Conclusions:
This software is intended to provide an initial list of eligible patients to a clinical study coordinators, which provides a starting point for further eligibility screening by the coordinator. Because this software has a high `rule in` ability, meaning that it is able to remove patients who are not eligible for the study, the use of an automated tool built to leverage an existing enterprise DW can be beneficial to determining eligibility and facilitating clinical trial recruitment through prescreening. While the results of this study are promising, further refinement and study of this and related approaches to automated eligibility screening, including comparison to other approaches and stakeholder perceptions, are needed and future studies are planned to address these needs.
|