Clinical evaluation of circulating blood volume in critically ill patients- contribution of a clinical scoring system

The circulating blood volume (CBV) of critically ill patients may be diffic ult to estimate on the basis of history and physical examination. The aim o f this study was to evaluate the ability of seven clinical signs and centra l venous pressure (CVP) to predict CBV in critically ill patients; CBV was evaluated with the [I-125]human serum albumin technique. A scoring system w as constructed using a combination of independence Bayes method and logisti c regression. Sixty-eight patients constituted a 'model development' sample and 30 patients a validation sample. Thirty-six patients (53%) in the mode l development sample were found to have a low CBV (measured CBV at least 10 % lower than the predicted mean normal CBV). Neither the haemodynamic varia bles monitored in ICU, nor the spot urinary sodium concentrations were diff erent between patients with and without a low CBV. Individually, none of th e clinical signs tested have a good positive or negative predictive value. For CVP, only extreme values seem to have clinical significance. To constru ct the score, the signs tested were ranked according to their discriminatin g efficacy. The probability of a low CBV was obtained by adding the weights of each sign tested and converting the score obtained into a probability. On a validation sample of 30 patients, the predictions are reliable as asse ssed by Z statistics ranging between -2 and +2. Our results suggest that: ( 1) individually, no clinical sign presented a clinical useful predictive va lue; and (2) a clinical scoring system may be helpful for the evaluation of CBV in critically ill patients.