Date Presented 04/22/2023
The purpose of this presentation is to describe potential quality issues with using rehabilitation data extracted from electronic health records (EHRs) for research. Challenges, opportunities, and strategies for data cleaning and management will be discussed in the context of an ongoing study of rehabilitation outcomes among adults with acquired brain injury in inpatient rehabilitation at a large, urban, academic medical center.
Primary Author and Speaker: Alison M. Cogan
Contributing Authors: Pamela S. Roberts, Trudy Mallinson
PURPOSE: To describe potential quality issues with rehabilitation data extracted from electronic health records (EHR) and approaches to data cleaning and management.
DESIGN: Secondary analysis of EHR and billing data. Data included demographic, diagnostic, functional assessment at admission and discharge, time billed, and CPT codes for rehabilitation services. Setting was an inpatient rehabilitation service at a large, urban academic hospital. Participants were adults with acquired brain injury (n=799) admitted between 2011 and 2016.
RESULTS: Demographic, diagnostic, functional measures, and billing data were obtained for the initial inpatient rehabilitation discharges for 799 adults with acquired brain injury from 2011-2016. Demographic, diagnostic, and functional data had been pre-cleaned during initial extraction procedure. Minimal missing or erroneous entries were identified in these data; 100% of the sample had complete FIM data for both admission and discharge. Billing data presented the greatest variability. Specifically, some billing entries included both rehabilitation units billed (e.g., 8-22 minutes) and actual minutes of treatment (e.g, 15 minutes), whereas others included only units billed. Treatment minutes were not required to be recorded in all years of data. Since units billed was a required entry, it was the more reliable measure of service delivery. All entries with rehabilitation units billed had valid CPT code descriptors. It was necessary to understand data entry processes and how they had changed over time to interpret raw data about service delivery and develop a coding approach that maximized reliability.
CONCLUSIONS: EHR may provide useful data for large observational studies of rehabilitation processes and outcomes. It is essential to have team members who are familiar with institution-specific front-end documentation and billing procedures.
IMPACT: EHR data presents challenges and opportunities for occupational therapy research.
References
Wolpert, M., & Rutter, H. (2018).Using flawed, uncertain, proximate and sparse (FUPS) data in the context of complexity: learning from the case of child mental health. BMC Med 16, 82 https://doi.org/10.1186/s12916-018-1079-6
Lyons, M., Dimas, J., Richardson, S. J., & Sward, K. (2022). Assessing EHR data for use in clinical improvement and research. American Journal of Nursing 122(6), 32-41. https://doi.org/10.1097/01.NAJ.0000832728.09164.3f
