Intensive Care Med (1999) 25: 1212±1214© Springer-Verlag 1999 EDITORIAL ma. In principle, this result fortunately confirmed what would have been predicted from the theory of indicator dilution [7]. Measurements of circulating blood volume by ICG seem therefore to be feasible in critically ill patients. This is even more true and could become, beyond scientific purposes, a clinically more relevant monitoring technique with the availability of less cumbersome bedside methods for assessment of ICG kinetics [4, 8] than the in vitro method used in this study. In a previous investigation, the same authors demonstrated by a simultaneous measurement of the initial distribution volume of glucose that in the case of a general capillary leakage an overestimation of plasma volume may occur by the ICG method [9]. It is questionable whether this observation also applies to the more recent bedside ICG techniques. Usually with the fiberoptic method the time span of registration is only 4 or 5 min, being considerably shorter than with the serial blood sampling in vitro technique used by Ishihara et al. By only including the compartments of circulation which actively contribute to the circulation within the time of measurement, the fiberoptically measured results will be less sensitive to capillary leakage. However, already in 1984 Shippy et al. observed an enormous variability of blood volume in critically ill patients [10]. In fact, it is difficult to define the normal value of circulating blood volume for a critically ill patient [11]. Obviously, a systemic inflammatory response is associated with systemic vasodilation not only within the arteriolar resistance vessels. The need for an increased blood volume in these patients also reflects an increased compliance of the low pressure system. Physiologically, ventricular filling rather than the volume status of the total circulation is crucial for cardiac performance [12]. Therefore, it is questionable whether circulating blood volume is really helpful for the therapeutic choice between volume resuscitation or vasopressors. In this respect, assessment of physiologic parameters which are more closely related to cardiac filling seem to be more advisable. In clinical routine, so called ªfilling pressuresº, such as central venous pressure or pulmonary capillary wedge pressure, are often used for this purpose. In particular, rapid changes of cardiac filling±like a sudden hypovolemia caused by acute bleeding±are mostly sufficiently reflected by concomitant changes of these pressures. However, it is well established now that less acute changes of cardiac filling cannot be monitored by pressure measurements, since these are also influenced by many other factors like mechanical ventilation, application of positive end-expiratory pressures and variation of ventricular compliance [13, 14, 15]. Obviously, more direct methods of volume measurements should be applied. One possibility, which is increasingly used in intensive care medicine, is transesophageal echocardiography (TEE). Still, it is difficult to generate reliable quantitative numbers for cardiac volumes from echocardiographic images. Moreover, TEE currently cannot really be applied as a monitoring tool but used only for intermittent investigations. Last not least, TEE requires trained and experienced personnel, which is not always or everywhere available.

The next best surrogate parameter for cardiac filling could be the filling of the intrathoracic circulation [16]. In fact, if circulating blood volume is measured at the bedside with ICG and an arterial fiberoptic catheter, then both circulating blood volume and intrathoracic blood volume can be determined from the indicator dilution curves [4]. In this case, fluid therapy is …