Automated data extraction: merging clinical care with real-time cohort-specific research and quality improvement data

Abstract

Background/purpose: Although prohibitively labor intensive, manual data extraction (MDE) is the prevailing method used to obtain clinical research and quality improvement (QI) data. Automated data extraction (ADE) offers a powerful alternative. The purposes of this study were to 1) assess the feasibility of ADE from provider-authored outpatient documentation, and 2) evaluate the effectiveness of ADE compared to MDE.

Methods: A prospective collection of data was performed on 90 ADE-templated notes (N=71 patients) evaluated in our bowel management clinic. ADE captured data were compared to 59 MDE notes (N=51) collected under an IRB-exempt review. Sixteen variables were directly comparable between ADE and MDE.

Results: MDE for 59 clinic notes (27 unique variables) took 6months to complete. ADE-templated notes for 90 clinic notes (154 unique variables) took 5min to run a research/QI report. Implementation of ADE included eight weeks of development and testing. Pre-implementation clinical documentation was similar to post-implementation documentation (5-10min).

Conclusions: ADE-templated notes allow for a 5-fold increase in clinically relevant data that can be captured with each encounter. ADE also results in real-time data extraction to a research/QI database that is easily queried. The immediate availability of these data, in a research-formatted spreadsheet, allows for rapid collection, analyses, and interpretation of the data.

Level of evidence: IV.

Type of study: Retrospective Study.

Keywords: Automated data extraction; Clinical documentation; Electronic data capture; Electronic documentation; Templated notes.