Identifying Patient Satisfaction from Electronic Health Record Data - A Retrospective Register Study

The aim of this study was to review the potential of electronic health record (EHR) data to automatically identify key concepts related to patient satisfaction in cardiac care. A randomised sample of EHR data from 500 cardiac patients were screened for feature extraction, identifying 187 terms describing patient satisfaction. The predictive positive value (PPV) for positive descriptions was high, indicating the value of EHRs and narrative documentation when developing tools to measure patient satisfaction. The high number of false positives resulting from the automated screening and the low PPVs for negative descriptions call for more sophisticated methods to identify not only the used terms, but also their context to increase the reliability of future tools aimed at automated monitoring of patient satisfaction.