Operationalizing Multi‐Domain Electronic Health Record Data for Acute Hospital Rehabilitation Research: A Learning Health Systems Approach

Introduction Hospital‐based rehabilitation therapy electronic health record (EHR) documentation is uniquely rich in social and environmental content, yet has received little systematic attention in health services research. These EHR data encompass pre‐admission living situation, household composition, home accessibility, discharge barriers, and post‐acute care recommendations. Our technical report describes a systematic approach to operationalizing this multi‐domain data for acute hospital research within a Learning Health System (LHS) framework. Methods Working within an LHS embedded researcher model at Intermountain Health, we describe the data governance, pre‐data stakeholder engagement, and variable creation computational pipeline applied to acute hospital encounters. We negotiated a data use agreement for a limited dataset and engaged with inpatient therapists to map EHR documentation structures before submitting the data request. Sequential cleaning steps operationalized variables across seven analytic domains: clinical complexity, functional status, social and environmental factors, home environment, therapy delivery, procedures, and discharge planning. We processed the therapy documentation file (1 032 359 submissions across 41 form types) using a form prioritization algorithm, crosswalk harmonization across heterogeneous form vocabularies, and regular expression matching to classify free‐text entries. Results The pipeline produced a dataset of 156 557 hospital encounters between September 2022 and August 2023 and operationalized 74 analytic variables spanning clinical, functional, social, and environmental domains. Key structured variables included: Activity Measure for Post‐Acute Care (AM‐PAC) 6‐Clicks functional assessment scores (at least one assessment present for all 156 557 encounters); pre‐admission living situation across 10 harmonized categories (95.0% complete); barriers to safe discharge across 10 clinically grouped indicators (70.1% of encounters); post‐acute therapy recommendations (91.9% complete); and time to first therapy visit (mean 25.5 h; 99.7% complete). A master data dictionary documents all variables with technical specifications and plain‐language descriptions. Conclusion This work demonstrates that high‐completeness analytic variables spanning clinical, functional, social, and environmental domains can be systematically derived from heterogeneous multiform rehabilitation therapy EHR documentation at scale. The data cleaning scripts, crosswalk tables, and master data dictionary provide a transferable foundation for rehabilitation researchers and health systems seeking to leverage therapy documentation for outcomes research and quality monitoring as an LHS.

Authors

Institutions

Publication Details

Journal
Learning Health Systems
Published
2026-11-18
DOI
https://doi.org/10.1002/lrh2.70140
Primary Topic
Electronic Health Records Systems
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Operationalizing Multi‐Domain Electronic Health Record Data for Acute Hospital Rehabilitation Research: A Learning Health Systems Approach

Lucy A. Savitz, Rachel A. Prusynski, Kate I. Minick, Devyn Woodfield
Learning Health Systems
Electronic Health Records Systems
article

Operationalizing Multi‐Domain Electronic Health Record Data for Acute Hospital Rehabilitation Research: A Learning Health Systems Approach

Lucy A. Savitz, Rachel A. Prusynski, Kate I. Minick, Devyn Woodfield
article en

Abstract

Introduction Hospital‐based rehabilitation therapy electronic health record (EHR) documentation is uniquely rich in social and environmental content, yet has received little systematic attention in health services research. These EHR data encompass pre‐admission living situation, household composition, home accessibility, discharge barriers, and post‐acute care recommendations. Our technical report describes a systematic approach to operationalizing this multi‐domain data for acute hospital research within a Learning Health System (LHS) framework. Methods Working within an LHS embedded researcher model at Intermountain Health, we describe the data governance, pre‐data stakeholder engagement, and variable creation computational pipeline applied to acute hospital encounters. We negotiated a data use agreement for a limited dataset and engaged with inpatient therapists to map EHR documentation structures before submitting the data request. Sequential cleaning steps operationalized variables across seven analytic domains: clinical complexity, functional status, social and environmental factors, home environment, therapy delivery, procedures, and discharge planning. We processed the therapy documentation file (1 032 359 submissions across 41 form types) using a form prioritization algorithm, crosswalk harmonization across heterogeneous form vocabularies, and regular expression matching to classify free‐text entries. Results The pipeline produced a dataset of 156 557 hospital encounters between September 2022 and August 2023 and operationalized 74 analytic variables spanning clinical, functional, social, and environmental domains. Key structured variables included: Activity Measure for Post‐Acute Care (AM‐PAC) 6‐Clicks functional assessment scores (at least one assessment present for all 156 557 encounters); pre‐admission living situation across 10 harmonized categories (95.0% complete); barriers to safe discharge across 10 clinically grouped indicators (70.1% of encounters); post‐acute therapy recommendations (91.9% complete); and time to first therapy visit (mean 25.5 h; 99.7% complete). A master data dictionary documents all variables with technical specifications and plain‐language descriptions. Conclusion This work demonstrates that high‐completeness analytic variables spanning clinical, functional, social, and environmental domains can be systematically derived from heterogeneous multiform rehabilitation therapy EHR documentation at scale. The data cleaning scripts, crosswalk tables, and master data dictionary provide a transferable foundation for rehabilitation researchers and health systems seeking to leverage therapy documentation for outcomes research and quality monitoring as an LHS.

Learning Health SystemsVol. 11(1)
Intermountain Healthcare (US), University of Pittsburgh (US), University of Washington (US)
Patient-Centered Outcomes Research Institute, Agency for Healthcare Research and Quality
Good health and well-being, Reduced inequalities
Electronic Health Records Systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon