Toward near real-time acuity estimation: a feasibility study.

Journal Article

BACKGROUND: Patient acuity estimation provides the basis of the nursing workload determination. It is critical to estimate patient acuity in real-time to capture a patient's need for nursing care more consistently and accurately. OBJECTIVES: To explore the feasibility of using patient data documented in an electronic nursing flowsheet in computerized nursing decision making; that is, near real-time acuity estimation. METHODS: Decision algorithms to determine values of acuity indicators using the flowsheet data were developed based on input from experienced nurse decision-makers and implemented as a rule-based system (RBS) with the Java Expert Shell System provided in Protégé. The RBS was tested with randomly selected patient data by comparing the observed agreements with previously recorded nurses' decisions. RESULTS: The observed agreements varied. The agreement rates exceeded 60% in nine indicators and the RBS tended to agree more on assigning false values. No consistent association was observed among the agreement rates, number of observations, and number of data items included in the study. The agreement rates increased slightly when the values determined by the RBS were aggregated following the nurses' time frame. The agreement rates were higher than chance agreement in the majority of the indicators included in this study. CONCLUSIONS: Two major factors that influenced the accuracy of the RBS were noted:limitations in source data and incompleteness of decision rules. Replicating the study with more complete data sets and enhanced decision rules was identified as the next step.

Full Text

Duke Authors

Cited Authors

  • Kim, H; Harris, MR; Savova, GK; Speedie, SM; Chute, CG

Published Date

  • July 2007

Published In

Volume / Issue

  • 56 / 4

Start / End Page

  • 288 - 294

PubMed ID

  • 17625469

Electronic International Standard Serial Number (EISSN)

  • 1538-9847

International Standard Serial Number (ISSN)

  • 0029-6562

Digital Object Identifier (DOI)

  • 10.1097/01.nnr.0000280617.21189.c3

Language

  • eng