Value of elementary, combined, and modeled hemodynamic variables.

PURPOSE: It has been well recognized that the usefulness of the clinical examination and simple hemodynamic variables in the critically ill is limited. Modelization for hemodynamic analysis may improve the diagnostic performance by a systematic and multivariate analysis. This requires a rigorous formalization that may otherwise expand the usefulness of hemodynamic data, both as predictors and as therapeutic targets. Our study was designed to test the value of a model for assessing the pathophysiology of circulatory disorders and for establishing the diagnosis. METHODS: We tested all available variables using survival as the end point. A population of 223 patients (652 measurements) with compromised circulatory status was studied. We evaluated traditional variables: (1) morphological and physical data, (2) elementary right heart catheterization data, and (3) usually calculated variables, versus (4) new modeled variables. These new modeled variables were derived from a previously validated computer program for hemodynamic evaluation. They expressed differences between observed hemodynamic performance and estimated needs. RESULTS: Among traditional variables, major prognostic factors were: (1) in all patients, lactate level elevation, physical signs of hypoperfusion, and a decreased systemic arterial pressure; (2) in septic patients, a high PaO2/SaO2 ratio; (3) in nonseptic patients, low left ventricle work indices. In all cases, modeled hemodynamic variables assessing performance-needs adequacy enhanced the prognostic value of hemodynamic monitoring. CONCLUSIONS: Compared with traditional variables, modeled variables were found of greater interest to quantify pathophysiology of shock. These results enabled us to validate the initial step of the hemodynamic reasonning formalization and to develop "new" diagnostic criteria that more closely fit the interrelationship between pathophysiology, diagnosis, and prognosis.