Hospital adverse events and control charts: the need for a new paradigm

Control charts are being increasingly used to summarise sequential rates of hospital adverse events (AEs). They are designed to detect departures from stable, predictable processes and are therefore appropriate when information about the mean value and variability of the relevant time-series data is available, from when the process is or has been in a stable, predictable state. This is often the case with binary data AEs such as surgical site infections and surgical mortality. However, it may not always be possible to determine the stable predictable rate at which events such as patient falls, pressure ulcers, medication errors or new isolates of a multiresistant organism (MRO) occur. Furthermore, such a rate may sometimes not exist, as is frequently the case with antibiotic usage and MRO prevalence data. It may then be better to employ time-series methods to analyse and present the data. A convenient approach is to employ spline-regression or a generalised additive model.