Using SmartTools to Capture Structured Data From Text: The LINCC Case Study

Authors

Tyler R. Ross, Group Health Research InstituteFollow
June BlueSpruce, Group Health Research InstituteFollow
Erin Hertel, Group Health Research Institute
Woody Favinger, Group Health Research Institute
Clarissa Hu, Group Health Research Institute

Publication Date

8-15-2016

Keywords

informatics, electronic health records

Abstract

Background/Aims: The widespread adoption of modern electronic heath records (EHR) has digitized medical information previously captured on paper. However, research and analysis requires use of narrative text and involves resource intensive chart review and/or natural language processing. An emerging alternative solution is to use SmartTools in Epic’s EHR. We present a case study of a research project using linked SmartTools in Epic to create structured data from text and describe the strengths and challenges of these tools.

Methods: The Learning to Integrate Neighborhoods and Clinical Care (LINCC) study collaborated with clinic staff to develop SmartPhrases with embedded SmartLists to be used by community resources specialists (CRS). The CRS were trained to use these SmartPhrases to document their patient interactions and capture critical information for process improvement and evaluation. The LINCC study worked with information technology (IT) to preserve SmartList selections in discrete fields in the EHR reporting database. The study team retrieved both the CRS notes and the linked SmartList selections and created an abstraction system in which values for relevant variables could be coded.

Results: The CRS began documenting patient encounters using SmartPhrases before their embedded SmartLists were linked to the reporting database. As a result, only free-text data were available from the early part of the study period. As the CRS position evolved, the SmartPhrases were modified and new SmartLists required linkage. This iterative process created inconsistency in measurement over the study period and occasional loss of information due to linkage timing. Inconsistency in how SmartPhrases were used within and across the CRS led to further measurement error. Ultimately, the team abstracted variable values from free-text notes for the final set of measures for analysis.

Conclusion: In this case study, despite clinic leadership and IT support and the opportunity to use SmartTools, abstraction from each CRS note was necessary to obtain some analytic measures. Although SmartTools did not eliminate the need for note review, they helped structure documentation, reduced the number of abstracted measures and provided supplemental information that aided the abstraction process. Extensive planning, coordination and training are critical for SmartTools to be used effectively for gathering research quality metrics.

Recommended Citation

Ross TR, BlueSpruce J, Hertel E, Favinger W, Hu C. Using SmartTools to capture structured data from text: the LINCC case study. J Patient Cent Res Rev. 2016;3:224.