Short-term variability of pulse rate and blood pressure in cardiac surgery patients

The present state of trend detection and prediction in patient monitoring

An investigation has been carried out into the suitability of the following techniques for trend detection and forecasting in patient monitoring: Cusum; Trigg's Tracking Signal; The Patient Condition Factor; The Patient Alarm Warning System; Box-Jenkins models and the Harrison-Stevens Bayesian approach. The latter holds considerable promise since it is flexible and can be implemented on a microprocessor. Consideration has also been given to the need for a better knowledge of the statistical properties of the variables to be monitored and the problems of combining trends detected in severable variables.

Reliability of human and machine measurements in patient monitoring

In a study on 29 patients for 152 hours a continuous monitoring system has been shown to have a reliability of 90.53% for pulse rate and 82.20% for arterial blood pressure, but for approximately half of the period of unreliable monitoring a not obviously artifactual reading was being displayed. The relevance of this to possible incorrect diagnosis and management is discussed. The monitor and a nurse using traditional chart keeping methods gave equivalent charts with respect to average levels and trends, but the monitor assisted chart showed a much greater variability. The latter tended to obscure underlying trends and render clinical diagnosis from the chart more difficult.