Related Experiment Videos
Selected characteristics of nurses and physicians who have living wills
J K Stechmiller1, M Conlon, G C Anderson
1University of Florida, Gainesville.
Abstract:
The purpose of this report is to determine what characteristics are most closely associated with nurses and physicians who favor a living will. A statistical analysis called classification and regression trees (CART) was used on the data set from a previous study. The five predictor variables that were analyzed were chosen following incorporation of a univariate screening process in which variables were chosen by their association with the likelihood of signing a living will. In addition, variables that did not have complete data sets were not considered in the analysis. The CART procedure was then used to determine which of these five variables were considered significant predictors. Three variables were selected: religion, length of experience with the terminally ill, and nature of the relationship (professional or personal) with the terminally ill. The two nonsignificant and therefore rejected variables were education/occupation and current region of residence. Within the predictor variable religion, CART selected the grouping of values (agnostic/atheist and Christian/non-Christian) as the best predictor of nurses and physicians likely to sign a living will. The second best predictor variable identified was length of experience in providing care to at least one patient, family member, or friend who was terminally ill. The nature of the relationship (professional or personal) with the terminally ill was the next strongest predictor variable of signing a living will. No previous studies of the living will have used such multivariate techniques as CART analysis to predict the likelihood of a health care provider signing a living will.