Outcome prediction in adult respiratory distress syndrome using discriminant analysis of cardiorespiratory data

In order to examine the prognostic value of different cardiopulmonary variables in adult respiratory distress syndrome the data of 30 patients with this illness were studied retrospectively. The patients were divided into 3 groups: Group A: survivors (9 cases, 40 examinations), Group B: early stage nonsurvivors (8 cases, 37 examinations), Group C: late stage nonsurvivors (19 cases, 89 examinations). In 6 nonsurvivor patients a few measurements were done in the early and late stage, too. There were highly significant differences between Groups A and C (mean pulmonary arterial pressure, pulmonary arterial diastolic pressure minus pulmonary capillary wedge pressure, left ventricular stroke work index, systemic and pulmonary vascular resistance, inspired oxygen fraction, arterial oxygen tension per inspired oxygen fraction, mixed venous oxygen saturation, pulmonary shunt fraction, and oxygen delivery, but the differences in relation to other groups were less prominent. Using a step-wise discriminant analysis, it was found that the oxygenation parameters alone determined the outcome correctly in 68-75%. Extending the analysis to haemodynamic variables the result improved (72-80%). Similar prediction was obtained when parameters potentionally measurable by noninvasive methods were analysed (69-80%). These results suggest that it is possible to predict the outcome of ARDS correctly without any invasive monitoring technique.