Multiple hemodynamic, respiratory and metabolic parameters are now routinely and repetitively measured in critically ill patients. In our Shock Research Unit, measurements of arterial and central venous pressures, heart and respiratory rates, urine output, and skin temperatures are routinely taken every 5 mins and processed on a digital computer system (1). In addition, measurements of blood flow, such as cardiac output and circulation times, are routinely taken at 2-to 4-hr intervals. Other parameters of patient status, such as arterial blood gases, pH and lactate concentration, are also measured at regular intervals. In order to evaluate the status of the patient, the physician must have some means of assessing this wealth of data. When large numbers ofvariables are monitored, the interpretation of changes in their values by the physician himself, without the aid of analytical tools, becomes increasingly difficult.
The availability of a computer system makes possible the development of automatic procedures for data screening. The purpose of this study is to develop and apply such a procedure for use in real time at the bedside of critically ill patients. The absolute values of the monitored variables, their sequential changes, and their proportional changes are used as criteria to evaluate the data. To evaluate the simultaneous values of the variables, and their simultaneous changes during various intervals, the Mahalanobis distance D2, which incorporates the interdependencies among the variables, is used (2). The computation of D2 for unusual combinations of measurements or changes yields abnormally large values, even though individual variables or changes may be within normal limits.