• Cynthia Arslanian-Engoren and Milo Engoren.
• School of Nursing, University of Michigan, Ann Arbor, MI 48109, USA. cmae@umich.edu
• Nurs Res. 2007 Mar 1; 56 (2): 82-8.

BackgroundBecause rapid therapy can improve the unfavorable prognosis of individuals with acute coronary syndromes (ACSs), it is critical that nurses accurately associate the cues of ACS and quickly and aggressively initiate interventional strategies that reduce mortality.ObjectivesTo determine if genetic algorithms (GAs) can be used to decipher the prediction rules that emergency department (ED) nurses use to triage persons suspected of having ACS and to determine whether these rules differ based on patient gender.MethodsA nonexperimental, descriptive study was conducted. Three thousand ED nurses were selected randomly to receive a mailed clinical vignette questionnaire, and 840 questionnaires were returned. Data analysis included binary logistic regression (BLR), development of GA, sensitivity, specificity, receiver-operator characteristic curves, and Monte Carlo simulations.ResultsNurses use different prediction rules for triaging male and female vignette patients with possible ACS. Accuracies were similar between BLR and GA. Both suffered a loss of predictive accuracy when algorithms or equations developed on one sex were tested on the other sex. Monte Carlo simulations showed that similar cues were used in triaging both men and women but they were combined differently in producing GA.DiscussionUsing GA was as accurate as results found by BLR and can be used to predict nurses triage decisions for ACS. The GA presented as flow charts may be user-friendly.

44 keywords...

hide…