Re: Hill-Rodriguez, D., Messmer, P. R., Williams, P. D., Zeller, R. A., Williams, A. R., Wood, M., & Henry, M. (2009). The Humpty Dumpty Falls Scale: A case-control study. Journal for Specialists in Pediatric Nursing, 14(1), 22-32. doi: 10.1111/j.1744-6155.2008.00166.x.

摘要

In a recent paper in Journal for Specialists in Pediatric Nursing, Hill-Rodriguez et al. tried to validate the Humpty Dumpty Falls Scale (HDFS) in identifying hospitalized kids who were prone to falling. The authors used a score of 12 on the HDFS as a cutoff point then reported the sensitivity and specificity of this instrument. Instead of focusing on a single threshold value, an alternative approach to evaluate the accuracy of the HDFS is through the receiver operating characteristics (ROC) curve, a very popular graphic tool that displays an instrument's true positive fraction (sensitivity) versus its false positive fraction (1 - specificity) across all possible cutoff thresholds. For the HDFS, a possible minimum total score is 7 and a maximum score is 23. So the ROC curve could be obtained as the plot of all observed pairs of false-positive rates and true-positive rates by varying the threshold of the HDFS from 7 to 23 with an increment of 1, and the usage of the ROC curve would avoid the loss of information that came from dichotomiza-tion as the paper did.