Mini-epidemic of Erroneous Central Venous Pressure Measurements Resulting from the Malproduction of a Specific Part of a Pressure Transducer System

ELECTROMANOMETRIC blood pressure measurements are routine in critically ill patients, and many interventions are based on correct measurements. We report a mini-epidemic of erroneous central venous pressure measurements due to faulty pressure transducer manufacturing resulting in wrong therapeutic decisions. Case Report For invasive pressure monitoring in patients undergoing cardiovascular surgery, we use a color-coded multitransducer system with a single pressurized bag of normal saline for line rinsing. On one day three patients undergoing cardiac surgery attracted attention because central venous pressure (CVP) measured via the atrial port of a pulmonary artery catheter (model 680100, 7.5-French, 110-cm; Becton Dickinson Critical Care Systems, Singapore) was much higher than mean pulmonary artery pressure. This persisted after transducer recalibration and assurance of normal gain settings and also when the pressure recording line was connected to an indwelling trilumen CVP catheter (model UE-24703-ND, 7-French, 30-cm; Arrow Deutschland, Erding, Germany). A change of the pressure transducer system and of connecting cables to the monitor did not alter these findings. However, CVP readings were considered implausible because there were no history or clinical signs of venous congestion or of tricuspid valve regurgitation, and transesophageal echocardiography revealed no right ventricular distention. After clamping the transducer’s rinsing line, CVP decreased substantially in all patients (27 mmHg 4t o 10 4, mean SD), and a typical CVP tracing was seen on the monitor’s display. After declamping, measured CVP again increased to abnormal values (fig. 1). Moreover, rinsing solution leaked continuously out off the transducer’s stopcock during transducer zeroing. Accordingly, false pressure readings resulting from excessive flow of rinsing solution were suspected. After withdrawal of the pulmonary artery catheter tip to a right atrial position, CVP measured via its distal lumen showed normal CVP values similar to the measurements obtained with the CVP transducer with its rinsing line clamped (fig. 1). Only one of the system’s three pressure transducers (the CVP transducer) was affected in each patient. Four additional dysfunctional transducers were identified after they had already been connected to patients, and further nine faulty transducers were detected during an implemented check of the systems before connection. Accordingly, we mailed alert warnings to all intensive care units of our hospital, the transducer system’s manufacturer (Smiths Medical Deutschland GmbH, Kirchseeon, Germany), and to regulatory institutions. Chart review of patients treated before problem detection revealed at least one patient treated inappropriately because of erroneously high CVP readings. Treatment by diuretics and fluid restriction had been stopped only in a preshock state, when transesophageal echocardiography had demonstrated cardiac volume depletion rather than suspected volume overload or cardiac tamponade. Subsequent investigations showed that all affected transducer systems derived from the same charge and had been delivered to various German hospitals. Upon activation of a transducer’s trigger flush knob, 60 ml/min of isotonic saline is normally delivered with the rinsing solution’s pressure maintained at 200 mmHg. With the trigger flush knob in its resting position, rinse solution continuously flows through a laser drilled hole only at 3 ml/h and flow does not interfere with correct pressure measurements. Microscopic examination of the dysfunctional transducers’ inside by the manufacturer revealed a roughened surface and additional holes resulting from excess amounts of solvent agent applied manually during production. This was responsible for an increased flow of rinsing solution (up to 120 ml/h) and for subsequently and artifactually increasing CVP measurements, as outlined schematically in figure 2.