Predicting septic related mortality of the individual surgical patient based on admission host-defence measurements.

In order to assign a numerical probability of septic related death for a given patient, the following data were recorded for 461 patients on admission to the Royal Victoria Hospital in Montreal for major surgical treatment: age; delayed-type hypersensitivity (DTH) skin-test score; total serum protein level; serum albumin level; circulating hemoglobin level; neutrophil adherence; polymorphonuclear chemotaxis; and sex. Logistic regression was used to generate a predictive equation of the form: (formula; see text) (where P is the probability of death and exp is the exponential function) that could be used to calculate the actual probability of septic related death for an individual patient. The predictive accuracy of this equation was tested on a separate, prospectively collected data base of 625 surgical patients. There was an excellent fit of the model. There were 68 septic related deaths where 67.2 were predicted. It is now possible to calculate precisely the probability of a septic related death of a given surgical patient upon admission to the hospital. This should prove useful in clinical decision-making.