# Methods of Piloting an Abstraction Tool to Describe Family Engagement in the Hospital Setting: Retrospective Chart Review

## TL;DR

This paper outlines how a tool was developed and tested to extract family engagement data from hospital records to better understand how families are involved in patient care.

## Contribution

The study introduces a systematic approach to refining an abstraction tool for retrospective chart reviews focused on family engagement documentation.

## Key findings

- The prepilot and pilot phases improved the abstraction tool's reliability and alignment with electronic medical records.
- Only about one-third of patient records documented shared decision-making involving families.
- More than three-quarters of records included documentation of family presence and communication.

## Abstract

Family engagement in hospitals is crucial for improving outcomes and ensuring holistic, patient-centered care. However, there is limited understanding of how providers document family engagement in electronic medical records (EMR) and how factors such as race and health disparities influence engagement practices. The absence of standardized EMR templates complicates tracking engagement and assessing its impact on patient outcomes. Retrospective chart review (RCR) is an effective method for investigating clinical practice and how family engagement is documented, using both structured and unstructured data from patient records. Despite its potential, gaps remain in the literature regarding distinctions between the prepilot and pilot phases in RCR studies.

This article describes the prepiloting and piloting stages in the development of an abstraction tool for an RCR study, highlighting how these phases refined the tool for extracting family engagement data from the EMRs.

A cohort of 2032 medical records was selected using the Research Patient Database Registry and EMRs. Initially, a draft tool was tested during the prepilot phase to assess its stability. To optimize diversity, the sample was then stratified by race. The modified tool was subsequently piloted on a subset of the sample.

The prepilot phase tested the tool on 9 records. In the pilot phase, the tool was applied to 39 records, representing approximately 10% of the sample. After the prepiloting and piloting phases, 293 of the 405 patient records were deemed eligible for inclusion. More than three-quarters of patients had documentation of presence and communication; whereas, only about one-third had documentation of shared decision-making involving families.

The prepilot phase helped standardize the abstraction tool, align it with the EMRs, and address potential biases. The pilot phase provided insights into data availability and highlighted areas for refinement before finalizing the tool for the remaining records. Together, these phases ensured the tool’s effectiveness for use in large-scale RCR studies.