Author: Kelly Brown (2024)
Advisory Committee: Angela Ross, DNP, MPH, RN, PMP, PHCNS-BC
PhD thesis: McWilliams School of Biomedical Informatics at UTHealth Houston.
ABSTRACT
Clinical patient registries, such as the pediatric congenital heart surgery registry, are designed to collect large amounts of varied clinical, administrative, and demographic data on specific patient populations. Data from these registries are used for multiple purposes, such as reporting, national benchmarking, research, quality improvement activities, and accreditation. Therefore, organizations must have accurate, complete, and timely data to meet these demands. However, manual data abstraction is the predominant method for populating a clinical patient registry. This method utilizes paid data abstractors, clinicians, and administrative staff and can be inefficient, incomplete, inaccurate, and costly.
The aim of this project was to demonstrate that an automated data solution could improve data abstraction efficiency, accuracy, and completeness and identify gaps in the data abstraction process. Participants were observed abstracting data pre- and post-intervention of an automated data solution, and observations were recorded using dimensions from Sittig and Singh’s Sociotechnical Model. Data were collected and analyzed pre- and post-intervention to evaluate data accuracy, completeness, and data abstraction efficiency.
Findings indicated that application of an automated solution improved data completeness, accuracy, and data abstraction efficiency. Additionally, observations using dimensions from the sociotechnical model identified gaps in the data abstraction process. These gaps could have impacted the data abstraction process, resulting in decreased data completeness, accuracy, and data abstraction efficiency. Findings also highlight the need to evaluate all aspects of health information technology in complex adaptive systems to identify gaps. However, findings from this translational project may not be applicable to other systems, and further studies may be necessary to better understand the value of data automation to solve the manual data abstraction burden.