[The effect of an intracardiac left-right shunt on thermodilution measurements of cardiac output. An extracorporeal circulation model].

Thermodilution measurements of cardiac output (CO) by means of Swan-Ganz catheters, in a strict sense, represent pulmonary arterial blood flow (PBF). In principle, this is also true in the presence of intracardiac left-to-right shunts due to atrial or ventricular septal defects. However, early recirculation of indicator may give rise to serious methodological problems in these cases. We sought to determine the influence of intracardiac left-to-right shunts on different devices for thermodilution measurements of CO using an extra-corporeal flow model. METHODS. Blood flow was regulated by means of a centrifugal pump that at the same time enabled complete mixing of the indicator after injection (Fig. 1). Pulmonary and systemic parts of the circulation were simulated using two membrane oxygenators and a systemic-venous reservoir to delay systemic recirculation of indicator. Control measurements of PBF (Qp) and systemic (Qs) blood flow were performed by calibrated electromagnetic flow-meters (EMF). Blood temperature was kept constant using a heat exchanger without altering the indicator mass balance in the pulmonary circulation. Left-to-right shunt was varied at different systemic flow levels applying a Qp:Qs ratio ranging from 1:1 to 2.5:1. Thermodilution measurements of PBF were performed using two different thermodilution catheters that were connected to commercially available CO computers. Additionally, thermodilution curves were recorded on a microcomputer and analysed with custom-made software that enabled iterative regression analyses of the initial decay to determine that part of the downslope that best fits a mono-exponentially declining function. Extrapolation of the thermodilution curve was then based on the respective curve segment in order to eliminate indicator recirculation due to shunt flow. RESULTS. At moderate left-to-right shunts (Qp:Qs < 2:1) all thermodilution measurements showed close agreement with control measurements. At higher shunt flows (Qp:Qs > or = 2:1), however, conventional extrapolation procedures of CO computers considerably underestimated PBF (Fig. 2). This was particularly true when a slow-response thermistor catheter was used (Fig. 3). The reason for this underestimation of Qp was an overestimation of the area under curve because of inadequate mathematical elimination of indicator recirculation by standard truncation methods (Fig. 4). However, curve-alert messages of the commercially implemented software did not occur. A high level of agreement could be consistently obtained using a fast-response thermistor together with individual definition of extrapolation limits according to logarithmic regression analyses. DISCUSSION AND CONCLUSION. Under varying levels of left-to-right shunt, both the response time of thermodilution catheters and the algorithms for calculation of flow considerably influenced the validity of thermodilution measurements of PBF in an extracorporeal flow model. The use of computer-based regression analyses to define the optimal segment for monoexponential extrapolation could effectively eliminate indicator recirculation from the initial portion of the declining thermodilution curve and showed the closest agreement with EMF measurements of Qp. The quality of thermodilution curves with respect to recirculation peaks in the flow model was slightly better than in clinical routine. Nevertheless, the clinical applicability of the modified extrapolation algorithm could be illustrated during pulmonary thermodilution measurements in an exemplary patient with a ventricular septal defect (Fig. 5). PBF at extremely high shunt ratios, however, cannot be assessed by monoexponential extrapolation in principle (Fig. 6). Insufficient elimination of indicator recirculation resulted in flow values that closely resembled systemic rather than PBF. This finding is in accordance with a mathematical analysis of the underlying Steward-Hamilton equation if an infinite number of recirculations would be