Abstract:
|
Rates of adverse events, such as falls and pressure ulcers, have been used as quality indicators for patient cares in US hospitals. With the overall improving quality of patient care, the rates of such events kept decreasing. The reliability of using these measures as quality indicators needs to be reassessed. Due to the existing time trends, the traditional measure of intra-class correlation coefficients based on repeated measures could be problematic. Alternative signal-and-noise assessment approaches based on one-time measurements were proposed to overcome this issue. However, these alternative approaches are based on some strong assumptions about the distribution of risk that may not be satisfied. In this study, the performances of the two types of approaches were compared under different scenarios with simulations. The results of the two approaches applied to real US hospital data were also obtained and assessed.
|