Intrapulmonary shunt measurement (Qsp/Qt) is a basic assessment of oxygen transport data during mechanical ventilation treatment to avoid excessive inspired oxygen and ventilator pressures with subsequent pulmonary damage and cardiovascular commitment. Qsp/Qt have been considered by several investigators as the gold standard to assess oxygenation. However, direct measurement of Qsp/Qt requires a pulmonary artery catheter, which may not be available in some cases limiting its clinical use. For these reasons, various arterial blood gas-derived formulas have been proposed as convenient alternatives to reflect Qsp/Qt, and include alveolar-arterial oxygen tension difference [P(A-a)O₂], inspired oxygen concentration to arterial blood oxygen tension ratio (PaO₂FiO₂), arterial to alveolar oxygen tension ratio (PaO₂/PAO₂), and respiratory index [P(A-a)O₂/ PaO₂], which relate the arterial oxygen tension (PaO₂) to the driving force for oxygen transport to the blood vessels. These are commonly used in intensive care instead of Qsp/Qt. It is reasonable that a reliable index reflects the condition of the O₂ transport but not vary with changes of FiO₂.

Oxygen indices are widely used in clinical assessment of patient oxygen exchanges; however, some controversy exists concerning their efficacy or accuracy in mechanically ventilated patients, and few studies have evaluated their efficiency in children. Herrick et al. analyzed the stability and the means and magnitudes of the indices as they change after FiO₂ modifications in 10 postoperative mechanically ventilated adults, and FiO₂ varied from 0.30 to 1.0. As FiO₂ increased, the indices changed in the opposite direction from the corresponding changes in the physiological shunt; the magnitude of the opposite changes were 24% for PaO₂/PAO₂, 22% for PaO₂/FiO₂, 67% for IR, and 101% for P(A-a)O₂. These authors conclude that none of the indices reliably reflects the shunt in situations where FiO₂ is changing, and the use of any of the indices for the clinical assessment of a patient’s gas exchange when FiO₂ vary can be misleading.

Gowda and Klocke studied the distribution of the ventilation/perfusion relationship of 16 patients with adult respiratory distress syndrome (ARDS). The FiO₂ varied between 0.21 and 1.0 in a computer model of gas exchange. These researchers evaluated the variability of P(A-a)O₂, PaO₂/FiO₂, PaO₂/PAO₂, P(A-a)O₂/PaO₂, and venous admixture, under the different FiO₂. It was found that in patients with shunt, >30% had a greater PaO₂/FiO₂ at low FiO₂ that decreases to relative stable values at FiO₂ of >0.5 and PaO₂ <100 Torr, and it was determined that PaO₂/FiO₂ is a useful estimation of the degree of gas exchange abnormality under usual clinical conditions. Other indices in the same study exhibited less stability, as FiO₂ was varied; additionally, the venous admixture varied substantially with the change in FiO₂ proportional to the fraction of cardiac output-perfusing gas exchange units with ventilation/perfusion ratios of <0.1, concluding that the venous admixture is a poor indicator of the efficiency of pulmonary oxygen exchange.

Coetzee et al. retrospectively examined blood gas data from surgical and respiratory ICU, and prospectively examined blood gas data in 15 postoperative adults; the correlation of [P(A-a)O₂], PaO₂/FiO₂, PaO₂/PAO₂, with calculated pulmonary shunt, and the influence of FiO₂ modifications on the indices before, during, and after general anesthesia were determined. The indices did not correlate well with the calculated pulmonary shunt, r = 0.35, 0.08, and 0.40, respectively. Using a stepwise variable selection, Coetzee et al. found that FiO₂, PaCO₂, PAO₂, and shunt were the main components of the final models. The FiO₂ had an effect on all indices, varying directly with the change in FiO₂; the slope of this relationship was less steep during anesthesia than before and after, this finding explained by denitrogenation, the collapse of alveoli with low ventilation anesthesia, and the change in the mixed venous saturation during anesthesia. The authors conclude that oxygen indices perform poorly due to the fact that they do not take into account the mixed venous saturation or that they ignore the effect of alveolar ventilation.

The PaO₂/PAO₂ had been proposed as being able to predict the FiO₂ needed for a desired PaO₂, and a simple formula was developed by Maxwell et al.; however, no recent efforts in terms of its clinical evaluations have been made.

The objectives of this study are to evaluate the predictive value of P(A-a)O₂, PaO₂/FiO₂, PaO₂/PAO₂, and P(A-a)O₂/PaO₂ for pathological Qsp/Qt, and to evaluate the sensitivity and specificity of the FiO₂-required formula in mechanically ventilated, critically ill children.

This is a prospective study conducted between June and December 1996 in 50 children aged 1–14 years who were admitted to the Intensive Care Units (ICUs) of the National Institute of Pediatrics in Mexico City to receive assisted or controlled mechanical ventilation and subsequent hemodynamic stability. The Ethics Committee of the Institution approved the protocol, and the parents or guardians of children provided written consent. Patients with methemoglobinemia, hyperleukocytosis, heart diseases, active bleeding, or any pathology that contraindicates arterial or venous puncture for blood samples were excluded.

We measured body weight, height, heart rate, systemic blood pressure, hemoglobin (cyanomethemoglobin), arterial and venous blood gas data, pH, PaO₂, PaCO₂ and arterial hemoglobin saturation (SaO₂). Qsp/Qt, P(A-a)O₂, PaO₂/FiO₂, PaO₂/PAO_2, and P(A-a)O₂/PaO₂ were calculated by standard methods.

A PaO₂ of 90–100 Torr was considered as ideal for this study. If PaO₂ was outside of these limits, we estimated the FiO₂ required to obtain a desired PaO₂ on 90 Torr using the following formula: FiO₂ = [90/(PaO₂/PAO₂) + PaCO₂]/[Barometric pressure-alveolar water pressure]; we then carried out FiO₂ modifications. During this period, no modes of ventilation, respiratory frequency, level of PEEP, or mean airway pressure modifications were made. Thirty minutes after FiO₂ change, a second blood gas sample was obtained, Qsp/Qt and oxygen indices were again calculated, and the PaO₂ was evaluated to verify the accuracy of the FiO₂-prediction formula.

Blood gas samples were obtained after a skin cleansing technique with alcohol through direct arterial puncture of the radial artery, and venous blood samples were obtained from different sites after a modified Allen test. The blood gas samples were obtained with 1-mL hypodermic needles containing 0.05 mL of heparin 1:1000. Samples were transported on ice and processed within the following 15 min in a Corning 238 gasometer (Ciba Corning Diagnostics, S.A. de C.V., Mexico City).

Statistical analysis was done on a personal computer, utilizing the EPI6 program of the World Health Organization, and the SPSS version 6.0 for Windows (BMDP Statistical Packages, S.P.S.S.: México, S.A. de C.V., Mexico City). Initially, a descriptive analysis of all crude or derived variables was carried out to obtain means and standard deviations. The patients were separated into two groups according to their first PaO₂: >90 and <90 Torr. A comparative analysis was made between initial blood gas data and data after FiO₂ change by groups, using Student’s t test for normally distributed or the Krusskal-Wallis test for non-Gaussian-distributed. Differences <0.05 were considered significant. Qsp/Qt was correlated separately with each oxygen index by univariate analysis and Pearson product-moment lineal correlation; 95% confidence intervals were calculated. To evaluate the impact of age or body weight of the children on Qsp/Qt and P(A-a)O₂, PaO₂/FiO₂, PaO₂/PAO_2, and P(A-a)O₂/PaO₂ Pearson correlation, the tests were performed individually on children younger than and older than 12 years of age, and on groups of 20 k of body-weight blocks.