Unreliability of oxygen tension-based indices in reflecting intrapulmonary shunting in critically ill patients.

Measurement of intrapulmonary shunting (Qsp/Qt), a widely used method for monitoring disturbances of pulmonary oxygen transfer in critically ill patients, involves calculation of arterial and mixed venous oxygen contents. In circumstances where mixed venous blood samples are not readily available, oxygen tension-based indices such as the alveolar to arterial oxygen tension differences (P[A-a]O2), arterial oxygen tension to alveolar oxygen tension ratio (PaO2/PAO2), PaO2 to FIO2 ratio (PaO2/FIO2) and respiratory index (RI) are widely utilized to reflect Qsp/Qt. Oxygen content-based indices such as the estimated shunt are not as widely utilized as the oxygen tension indices. In 75 critically ill patients in whom a pulmonary artery catheter was being utilized to augment clinical care, comparisons were made between Qsp/Qt and P(A-a)O2, PaO2/PAO2, PaO2/FIO2, RI, and estimated shunt to determine which index best reflected Qsp/Qt. Correlations between Qsp/Qt and estimated shunt were good (r = .94) and poor for the P(A-a)O2 (r = .62), PaO2/PAO2 (r = .72), PaO2/FIO2 (r = .71), and RI (r = .74). We conclude that there are no real substitutes for venous oxygen contents in critically ill patients. When pulmonary artery blood is not available for analysis, oxygen tension-based indices are unreliable reflectors of Qsp/Qt while the estimated shunt, an oxygen content-based index, provides a more reliable reflection of Qsp/Qt.