2015 QPRC

Dynamic Control Limits for the Risk-adjusted Bernoulli CUSUM Chart for Monitoring Surgical Outcome Quality

The most recommended chart for monitoring surgical binary outcome quality is the risk-adjusted cumulative sum chart. The selection of control limits to maintain a specified in-control average run length, however, depends heavily on the underlying population of risk factors of the patients. These risk factors are used to determine the probability of death, or some other adverse event, within a given time window following surgery. In our paper we present a computationally intensive method for determining the control limit as each patient arrives for surgery. The probability of a false alarm following surgery for a given patient, given no previous false alarm, is controlled to be a fixed quantity. This method yields time-varying control limits that are based on the particular sequence of patients and does not require any assumptions about the population of risk factors.