Retrospective Chart Review Methodology: Piloting an abstraction tool to describe family engagement in the hospital setting (Preprint)

Abstract

BACKGROUND Family engagement refers to the active involvement of family members in the care and decision-making processes of patient care. The current literature lacks clear insight into family engagement in hospital settings because it does not adequately describe the specific domains experienced, the roles families play during inpatient care, or the extent to which these roles are documented in electronic medical records (EMR). Retrospective Chart Review (RCR's), a foundational methodology of clinical research for over eighty years, extracts both structured and unstructured data from the medical record. It can offer insight into which clinicians initiate and sustain engagement, as evidenced in the unstructured narratives such as a daily progress note. RCR's pre-pilot and pilot testing of an abstraction tool are critical steps in the developing the tool used for data extraction. OBJECTIVE This article aims to explain the stages before piloting (pre-piloting) and during piloting of an abstraction tool used for a Retrospective Chart Review (RCR) study. It also explores how these stages influenced the development of the final abstraction tool that was used to extract data on family engagement in hospital settings from the EMR. METHODS Using a centralized clinical data registry system, the Research Patient Database Registry (RPDR), and the electronic medical record (EMR), a list of 2053 medical records meeting study inclusion and exclusion criteria was generated. A draft abstraction tool was tested on 9 records until it became stable in a pre-piloting phase. After stratifying the records by race, a convenience sample of 135 Black patients, a randomized sample of 135 White patients, and 135 Asian and other patients were selected using a random number generator. The modified abstraction tool then was piloted on 10% of the sample frame (n=39). RESULTS During the pre-pilot stage, the tool was tested on a total of 9 records from the aggregate 2032 records. During the pilot phase, the abstraction tool was applied to 39 patient records taken from the sample population.Two-hundred and ninety-three patients were eligible for inclusion. CONCLUSIONS The pre-pilot stage standardized the abstraction tool and aligned the flow of the tool to the EMR. It also addressed possible data extractor assumptions. The pilot phase provided insight into the available information, potential problems, and necessary modifications and refinements before the abstraction tool was finalized for the remaining 405 records.