Intensive Care Medicine

, Volume 44, Issue 12, pp 2245–2247 | Cite as

Benefits and risks of the P/F approach

  • L. GattinoniEmail author
  • F. Vassalli
  • F. Romitti


The PaO2/FIO2 ratio represents the pressure exerted in the blood by the unbound molecules of oxygen, normalized to the fractional volume of inspired oxygen. The PaO2/FIO2 ratio is used to assess the lung’s capability to oxygenate the blood, primarily in ARDS, where its thresholds of 150, 200, and 300 are used/proposed to classify ARDS severity [1, 2]. Ideally, a given PaO2/FIO2 ratio value should correspond to a definite lung severity, independently of FIO2. In reality, the same severity may be associated with quite different PaO2/FIO2 values, depending on several factors, as previously described [3].

Alveolar PO2

Ideally, PaO2 should be normalized to alveolar PO2 (PAO2) instead of FIO2. Indeed, for the same PaO2/FIO2 ratio, the PaO2/PAO2 ratio may vary depending on barometric pressure (Pb), PaCO2, and the respiratory exchange ratio (R), as may be easily understood by examining the alveolar air equation:
$${\text{PAO}}_{2} = {\text{FIO}}_{2} \times \left( {{\text{Pb}} - 47} \right) - \frac{{{\text{PaCO}}_{2} }}{R}$$

Consequently, an identical PaO2/FIO2 ratio of 150 measured at the barometric pressure of Mexico City (2250 m) or Göttingen (150 m) in two patients breathing 30% O2, with identical PaCO2/R ratios, would result in a sharply different PaO2/PAO2 ratios: 0.32 in Göttingen, decidedly less than the 0.49 in Mexico. The impact of PaCO2/R ratio on PAO2 is less dramatic, unless extracorporeal CO2 removal is in use. In this case, the R may be very low, producing a consistent decrease in the alveolar PO2, if FIO2 is not adequately increased [4, 5, 6].

Arterial PO2

According to Riley’s model (two compartment lung, one ideally perfused and ventilated, one perfused and not ventilated) [7], the arterial oxygen content (CaO2) is the weighted mean of the oxygen contents blended from the two compartments. The blood from the perfused/ventilated compartment will have a PO2 equal to the alveolar PAO2 in equilibrium with the capillary oxygen content (CcO2), while the blood coming from the perfused/non-ventilated compartment will have a PO2 and oxygen content equal to the mixed venous blood (CvO2). The fraction of the cardiac output coming from the perfused/non-ventilated compartment (venous admixture) may be easily quantitated at the bedside:
$${\text{Venous admixture }} = \frac{{{\text{CcO}}_{ 2} {-}{\text{CaO}}_{ 2} }}{{{\text{CcO}}_{ 2} {-}{\text{CvO}}_{ 2} }}.$$
Although venous admixture is the variable that more accurately assesses oxygenation impairment, it nowadays is considered impractical and cumbersome; hence, the PaO2/FIO2 is used for severity assessment. The limits of the PaO2/FIO2 approach can be understood by considering Eq. 1 (which defines the PAO2) together with Eq. 2 (which defines the venous admixture). Indeed,
  1. 1.

    CcO2 strictly depends on PAO2, which is proportional to the FIO2 (Eq. 1), while the CaO2 is proportional to the PaO2 (through the oxygen dissociation curve) [8]. Therefore, the difference (CcO2 – CaO2) and the ratio (CaO2/CcO2) are strictly related and hold the same physiological meaning of PaO2/FIO2 ratio.

  2. 2.

    Because the (CcO2 – CaO2) difference equals the product: [venous admixture × (CcO2 – CvO2)], the same (CcO2 – CaO2), i.e., the same PaO2/FIO2, may derive from myriad combinations of venous admixture fraction and (CcO2 – CvO2). These range from extremely high venous admixture fraction and low (CcO2 – CvO2), i.e., high CvO2, or vice versa.

  3. 3.

    CcO2 primarily depends on FIO2; therefore, for a given FIO2 any change of (CcO2 – CvO2) only depends upon the CvO2.

  4. 4.

    CvO2, for a given arterial oxygenation, strictly depends on oxygen consumption (VO2) and cardiac output (Qt); indeed, CvO2 = CaO2 – VO2/Qt.

The consequence of these relationships are summarized in Fig. 1. Figure 1a shows PaO2 as a function of FIO2 at venous admixture levels from 10% to 40%, and a cardiac output range between 6 and 10 L/min, assuming an oxygen consumption of 200 ml/min. Two features are worth noting:
  • PaO2 is lower at higher venous admixture levels and increases non-linearly with FIO2 along the iso-venous admixture lines.

  • For a given oxygen consumption and venous admixture level, cardiac output exerts a tremendous effect on PaO2. It must be stressed, however, that the primary determinant is the CvO2 (see point 4 above).

Figure 1b presents the PaO2/FIO2 ratio as a function of FIO2 at venous admixture levels between 10% and 40% over a cardiac output range between 6 L/min (lover CvO2) and 10 L/min (higher CvO2). This figure underlines the limits of PaO2/FIO2 alone in the assessment of lung injury severity. As an example, at venous admixture 20% and 10 L/min of cardiac output, the PaO2/FIO2 always exceeds 300, i.e., no ARDS. However, for the same venous admixture (20%) with a lower cardiac output of 6 L/min, a given patient would be classified as “mild ARDS” across FIO2 values from 0.3 to 0.7 but classified as “no ARDS” at FIO2 values from 0.7 to 1.0. Another hypothetical patient at venous admixture of 30%, depending on FIO2 and cardiac output, may oscillate between no ARDS, mild ARDS, or moderate-severe ARDS.
Fig. 1

PaO2 (a) and PaO2/FIO2 (b) as a function of FIO2 at shunt of 10%, 20%, 30%, and 40%. Values computed at cardiac output 10 L/min (upper boundaries) and 6 L/min (lower boundaries), at VO2 200 ml/min, hemoglobin 10 g/dL, and alveolar PCO2 40 mmHg. PaO2 values were derived from oxygen content, by using the oxygen dissociation curve equation, proposed by Severinghaus [9]. The arteriovenous oxygen difference was 2 ml/dL at 10 L/min of cardiac output and 3.3 ml/dL at 6 L/min of cardiac output. Note that, for a given shunt, the upper boundary would move up and the lower boundary would move down if the arteriovenous oxygen difference was lower than 2 ml/dL and greater than 3.3 ml/dL, respectively. Values were chosen as proof of the concept

Clinical use

Assessment of severity

Although the PaO2/FIO2 ratio has limits as a surrogate of venous admixture, the PaO2/FIO2 ratio offers several advantages: first, it is easy to measure; second, when tested across large populations (but not necessarily in individual patients), the PaO2/FIO2 reflects reasonably well the severity of anatomical derangements measured by CT scanning [1]. Nonetheless, the accuracy of PaO2/FIO2 ratio for indexing ARDS severity (e.g., Berlin ARDS definition) would improve greatly if determined at a standard PEEP value. In previous work [10], we used 5 cmH2O to avoid the masking effect of higher PEEP on PaO2/FIO2 ratio, which may be due either to decreasing venous admixture or altering hemodynamics. Standardization of FIO2 would further improve the accuracy and comparability of severity among patients [11].

PEEP selection

Changes in PaO2/FIO2 ratio are frequently used to assess recruitability during ARDS, on the assumption that increases in PaO2/FIO2 ratio are due to lung recruitment [12]. Unfortunately, increasing PEEP often decreases cardiac output. Theoretically, if the venous admixture and oxygen consumption do not change, this would reduce the PaO2/FIO2 ratio. However, this seldom occurs, as the venous admixture usually changes in proportion to the cardiac output [12, 13, 14, 15]. Therefore, caution must be used when setting PEEP with the PaO2/FIO2 approach, as its apparent that improvement may be due to decreased cardiac output in the absence of recruitment—a principle long known but often forgotten.


  • PaO2/FIO2 ratio is a surrogate of venous admixture measurement for approximating ARDS severity and relates well to anatomical differences on the CT scan.

  • At a given venous admixture, the PaO2/FIO2 ratio may differ, depending on oxygen consumption and cardiac output. Conversely, for the same PaO2/FIO2, venous admixture may vary with FIO2.

  • To better assess severity of lung injury and follow its evolution, PaO2/FIO2 ratio should be measured at standardized levels of PEEP and FIO2. Selecting PEEP according to PaO2/FIO2 ratio may be misleading if hemodynamics are not taken into account.


Compliance with ethical standards

Conflicts of interest

On behalf of all authors, the corresponding author states that there is no conflict of interest.


  1. 1.
    ARDS Definition Task Force, Ranieri VM, Rubenfeld GD, Thompson BT, Ferguson ND, Caldwell E, Fan E, Camporota L, Slutsky AS (2012) Acute respiratory distress syndrome: the Berlin definition. JAMA 307:2526–2533Google Scholar
  2. 2.
    Maiolo G, Collino F, Vasques F, Rapetti F, Tonetti T, Romitti F, Cressoni M, Chiumello D, Moerer O, Herrmann P, Friede T, Quintel M, Gattinoni L (2018) Reclassifying acute respiratory distress syndrome. Am J Respir Crit Care Med. CrossRefPubMedGoogle Scholar
  3. 3.
    Aboab J, Louis B, Jonson B, Brochard L (2006) Relation between PaO2/FIO2 ratio and FIO2: a mathematical description. Intensive Care Med 32:1494–1497CrossRefPubMedGoogle Scholar
  4. 4.
    Kolobow T, Gattinoni L, Tomlinson T, Pierce JE (1978) An alternative to breathing. J Thorac cardiovascr Surg 75:261–266Google Scholar
  5. 5.
    Gattinoni L, Kolobow T, Tomlinson T, White D, Pierce J (1978) Control of intermittent positive pressure breathing (IPPB) by extracorporeal removal of carbon dioxide. Br J Anaesth 50:753–758CrossRefPubMedGoogle Scholar
  6. 6.
    Gattinoni L (2016) Ultra-protective ventilation and hypoxemia. Crit Care 20:130CrossRefPubMedPubMedCentralGoogle Scholar
  7. 7.
    Riley RL, Cournand A (1949) Ideal alveolar air and the analysis of ventilation–perfusion relationships in the lungs. J Appl Physiol 1:825–847CrossRefPubMedGoogle Scholar
  8. 8.
    Gabel RA (1980) Algorithms for calculating and correcting blood-gas and acid-base variables. Respir Physiol 42:211–232CrossRefPubMedGoogle Scholar
  9. 9.
    Severinghaus JW (1979) Simple, accurate equations for human blood O2 dissociation computations. J Appl Physiol Respir Environ Exerc Physiol 46:599–602PubMedGoogle Scholar
  10. 10.
    Caironi P, Carlesso E, Cressoni M, Chiumello D, Moerer O, Chiurazzi C, Brioni M, Bottino N, Lazzerini M, Bugedo G, Quintel M, Ranieri VM, Gattinoni L (2015) Lung recruitability is better estimated according to the Berlin definition of acute respiratory distress syndrome at standard 5 cm H2O rather than higher positive end-expiratory pressure: a retrospective cohort study. Crit Care Med 43:781–790CrossRefPubMedGoogle Scholar
  11. 11.
    Allardet-Servent J, Forel JM, Roch A, Guervilly C, Chiche L, Castanier M, Embriaco N, Gainnier M, Papazian L (2009) FIO2 and acute respiratory distress syndrome definition during lung protective ventilation. Crit Care Med 37(202–207):e204–206Google Scholar
  12. 12.
    Lemaire F, Harf A, Simonneau G, Matamis D, Rivara D, Atlan G (1981) Gas exchange, static pressure-volume curve and positive-pressure ventilation at the end of expiration. Study of 16 cases of acute respiratory insufficiency in adults. Ann Anesthesiol Fr 22:435–441PubMedGoogle Scholar
  13. 13.
    Freden F, Cigarini I, Mannting F, Hagberg A, Lemaire F, Hedenstierna G (1993) Dependence of shunt on cardiac output in unilobar oleic acid edema. Distribution of ventilation and perfusion. Intensive Care Med 19:185–190CrossRefPubMedGoogle Scholar
  14. 14.
    Dantzker DR, Lynch JP, Weg JG (1980) Depression of cardiac output is a mechanism of shunt reduction in the therapy of acute respiratory failure. Chest 77:636–642CrossRefPubMedGoogle Scholar
  15. 15.
    Lynch JP, Mhyre JG, Dantzker DR (1979) Influence of cardiac output on intrapulmonary shunt. J Appl Physiol Respir Environ Exerc Physiol 46:315–321PubMedGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature and ESICM 2018

Authors and Affiliations

  1. 1.Department of Anesthesiology, Emergency and Intensive Care MedicineUniversity of Göttingen (UMG)GöttingenGermany

Personalised recommendations