Definition and detection of alarms in critical care

Critical care medicine has developed enormously in complexity and even more so in cost over the past twenty years. There has been evidence of remarkable progress in improved outcomes from some conditions, particularly when severely ill patients are treated in well equipped and well managed intensive care units (ICU) which have clear directorship and comprehensive management guidelines and protocols (Zimmermann et al., Crit Care Med 1993; 21:1443-1451). Nevertheless, for some conditions such as severe acute respiratory failure and multiple organ failure, there is considerable debate as to whether there has been any improvement at all (Lee et al., Thorax 1994; 49:596-597. Artigas et al., Adult respiratory distress syndrome, Churchill Livingstone, Edinbugh, London, Madrid, Melbourne, New York, Tokyo, pp. 509-525). Developments in signal processing and monitoring and recording technology have resulted in a vast increase in the quantity of data that is available to clinicians trying to manage critically ill patients (Price, Bailliere's Clin Anaesthesiol 1987; 1:533-556) but there is little evidence that this apparent gain has lead to better clinical decisions or earlier warning of significant instability. One of the tasks of the European Union sponsored IMPROVE group was to attempt to identify significant downward trends in vital parameters sufficiently early to allow clinical intervention to be potent and effective and ultimately improve patient outcome from a wide range of life threatening conditions. The first stage of this task was to define examples of such life threatening deterioration and conduct a survey in representative intensive care units of the incidence of these conditions and the subsequent patient outcomes. This is a preliminary task, the next stage being the gathering of "real time' data from critically ill patients for 24-h sample periods to probe for deteriorating trends and to compile a comprehensive annotated data library of physiological data as a rich resource for future adaptations in signal processing technology and clinical decision support.