Evaluating Pulmonary Function: An Assessment
of Pao2/Fio2
John R. Feiner, MD; Richard B. Weiskopf, MD
Objectives: Pao2/Fio2 is used commonly for diagnosis of lung
injury (acute respiratory distress syndrome and transfusion-related
acute lung injury), for assessment of pulmonary disease course
and therapy, and in pulmonary transplantation for evaluation of
donor lungs and clinical outcome. It was developed for conve-
nience, without formal mathematical and graphic assessment to
validate its suitability for these purposes.
Design: We examined, mathematically and graphically, the rela-
tionship of Pao2/Fio2 to Fio2 at constant normal and several
degrees of increased intrapulmonary shunting (Qs/Qt), assessing
the impact of intra- and extrapulmonary factors on the relationship
and thus the reliability of Pao2/Fio2.
Measurements and Main Results: The relationship of Pao2/Fio2
varies at all shunt fractions but most with Qs/Qt from 0.1 to 0.3
with Fio2 approximately greater than 0.4. At higher Qs/Qt, the rela-
tionship is more constant and changes less with Fio2 more than
0.4. Hemoglobin concentration and arterial-venous oxygen con-
tent difference have large effects that can confound interpretation
of Pao2/Fio2. Barometric pressure has a substantial effect; Pco2,
base excess, and respiratory quotient have small effects.
Conclusions: At high Qs/Qt with Fio2 more than 0.4, the rela-
tionship of Pao2/Fio2 to Fio2 is relatively constant. However, with
All authors: Department of Anesthesia and Perioperative Care, University
of California, San Francisco, CA.
Dr. Feiner wrote and edited the article, developed the mathematical
analysis, and developed the computer programming for the mathematical
analysis. Dr. Weiskopf wrote and edited the article and developed the
mathematical analysis.
Supplemental digital content is available for this article. Direct URL cita-
tions appear in the printed text and are provided in the HTML and PDF
versions of this article on the journals website (http://journals.lww.com/
ccmjournal).
Dr. Feiner has received research funding from Masimo, Bluepoint Medical,
Nonin Medical, CAS Medical Systems, Covidien, Mespere Lifesciences,
Pacific Medico, Xhale, Nihon Koden, and Senspec. Dr. Weiskopf has a
relationship with or consults for the following organizations/companies:
U.S. Food and Drug Administration, National Heart, Lung, and Blood
Institute/National Institutes of Health, U.S. Department of Defense, Teru-
moBCT, and HbO2 Therapeutics. Dr. Weiskopf has also consulted for
Sangart, OPK Biotech, Octapharma USA, and CSL Behring, within the
past 3 years. His institution received funding from Masimo, Bluepoint
Medical, and Nonin Medical.
For information regarding this article, E-mail: John.Feiner@ucsf.edu
Copyright 2016 by the Society of Critical Care Medicine and Wolters
Kluwer Health, Inc. All Rights Reserved.
DOI: 10.1097/CCM.0000000000002017
e40 www.ccmjournal.org January 2017 Volume 45 Number 1
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Qs/Qt of 0.10.3, Pao2/Fio2 changes substantially with Fio2.
Understanding the important effects of nonpulmonary factors
(especially hemoglobin concentration and arterial-venous oxygen
content difference) should enhance appropriate clinical use, inter-
pretation of Pao2/Fio2, and interpretation of previous publications
and future studies (especially those seeking to assess effects of
anemia or transfusion on lung function). The ratio of Pao2/Fio2 is
a good tool for some, but not many clinical circumstances, and
is insufficiently robust for most research applications. (Crit Care
Med 2017; 45:e40e48)
Key Words: computer simulation; hemoglobin; hypoxemia; hypoxia;
oxygen; pulmonary gas exchange
A
ppreciation of oxygen transfer in the lungs is impor-
tant for understanding pulmonary physiology and any
pathologic impairment of oxygen transfer, as the latter
is a critical feature of pulmonary disorders and is an important
therapeutic consideration. The fraction of cardiac output (QT)
that traverses the lung without participating in gas exchange,
intrapulmonary shunt, may be calculated from estimates of
pulmonary capillary oxygen content and measured arterial
and mixed venous oxygen content (1). The difference between
alveolar partial pressure of oxygen (PAo2) and Pao2 (AaDo2)
had been the traditional measure for providing an estimate of
shunt in clinical circumstances, as it reflects shunt well when
Fio2 is 1.0 (2) and when the cardiac output and Ca-vo2 (AVDo2)
are within a reasonably normal range (3). More complicated
tests that are generally not available clinically are required to
evaluate the mismatching of pulmonary ventilation and perfu-
sion (V/Q) that may contribute to increased AaDo2 when the
Fio2 is less than 1.0 (46). Furthermore, for most patients with
clinically important oxygenation deficits, especially when Fio2 is
greater than 0.3, V/Q mismatch is of limited consequence (7).
The relationship between Pao2 and Fio2, expressed as
Pao2/Fio2 (P/F), is commonly used as a measure of disease pro-
gression, response to therapy, and to compare patients with
disparate Fio2. An analogous ratio between the two seems to
have been suggested first as Pao2/PIO2 by Lyons et al (8) in an
effort to compare oxygenation in patients with varying Fio2s.
Subsequently, the ratio was altered to the apparent first use of
P/F by Horovitz et al (9). Neither of these reports provided a

physiological basis for dividing one variable by the other, in
lieu of examining the more physiologically based difference in
Po2. Apparently, P/F was chosen later as a measure because of
its simplicity and practicality, as the data are generally readily
available in patients medical records (10) (M. Matthay, per-
sonal communication, 2014). Several reports have examined
the utility of this measure by assessing the relationship of
P/F with pulmonary shunt in clinical circumstances (1115).
Indeed, this variable has become so commonly used clinically
that it is the established oxygenation criterion for the diagno-
sis of adult respiratory distress syndrome (ARDS) (10, 16) and
transfusion-associated lung injury (TRALI) (17) and for suit-
ability of donor lungs for transplantation (18) and transplant
success (19).
In order for P/F to provide a valid assessment measure, it
would be necessary that P/F reflects intrapulmonary shunt
over a wide range independent of Fio2 and over a broad range
of variation of the nonpulmonary factors (cardiac output
or AVDo2, hemoglobin concentration, respiratory quotient
[RQ], Paco2, and barometric pressure [Pb]) that influence
oxygenation. A few mathematic and graphic analyses have
examined the relationship of P/F with intrapulmonary
shunt; however, they were limited, especially regarding the
influence of extrapulmonary factors (2022). We posed this
question: in the absence of change of pulmonary function,
is the value of P/F preserved when Fio2, AVDo2 (used as the
variable to model changes in QT with constant Vo2), hemo-
globin, Pco2, base excess (BE), or Pb are changed indepen-
dently? Herein, we examine the mathematical relationship
between P/F and Fio2 over a broad, but clinically relevant,
range of shunt (Qs/Qt) and the nonpulmonary variables
involved. To accomplish this, we used the concept of isos-
hunt graphics as developed by Nunn (23) and Benatar etal
(24) wherein the relationship between the independent
(Fio2) and dependent (P/F) variables is examined for the
condition of constant Qs/Qt.
MATERIALS AND METHODS
The local institutional review board (IRB) assessed the proj-
ect and determined that it did not meet the definition of
human subject research and therefore did not require IRB
approval.
Equations
We used the shunt equation (the fraction of Qt that does
not participate in pulmonary exchange of oxygen) as first
developed by Sackur (1) to calculate Pao2 for various combi-
nations of the factors that influence pulmonary gas exchange
for oxygen and the alveolar gas equation, as developed by
Fenn et al (25), to calculate Pao2 (for details, see supplemen-
tal data, Supplemental Digital Content 1, http://links.lww.
com/CCM/C57).
For all simulations, except where otherwise indicated, Pb
was set to 760mm Hg, RQ to 0.8, and water vapor pressure to
47mm Hg (the value at 37C), and PAco2 was assumed equal to
Paco2 and set to 40mm Hg.
Critical Care Medicine www.ccmjournal.org e41
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Online Clinical Investigations
Iterative Solutions
To create isoshunt figures for a desired shunt and a selected
AVDo2, it is not possible to calculate Pao2, and therefore P/F
ratio, from the equations described and the equations described
by Severinghaus (26) and Roughton and Severinghaus (27). We
used a computerized iterative approach for the purpose that
resulted in an accuracy for Qs/Qt that did not differ from the
desired Qs/Qt by more than 1010 (supplemental data, Sup-
plemental Digital Content 1, http://links.lww.com/CCM/C57).
Simulations
To explore the utility of P/F ratio among different patients
under different conditions or for a single patient where one
variable has been altered (e.g., Fio2 or hemoglobin), we sought
to examine the relationship of P/F with Fio2 for a range of
the variables that affect oxygenation and therefore P/F. This
required calculating the relationship between P/F and Fio2
while holding shunt and other variables, such as AVDo2 and
hemoglobin constant, over a range of those variables. That
in turn necessitated solving for Pao2, So2, Pvo2, and Svo2 for
an unchanging value of Qs/Qt. Simulations were calculated
for Qs/Qt ranging from normal ( 0.01) (2, 4, 28) to severely
increased (0.5).
The above process is satisfactory for determining the rela-
tionship of P/F for various shunt fractions for the full range of
Fio2 at the various values of AVDo2. However, that unmodified
process is not accurate when examining changes in hemoglobin
because of the physiological consequences and compensations
for changes in hemoglobin, such as decreased Pvo2 and Svo2 and
increased QT with decreasing hemoglobin (29). To correct for
these physiological changes, we performed an additional set of
calculations, similar to those above, for a range of hemoglobin
concentrations of 514g/dL, using the data for Svo2 and AVDo2
that were obtained in normal, unmedicated, healthy humans
at these hemoglobin concentrations (29). Calculations for
Svo2 and AVDo2 are in the supplemental data (Supplemental
Digital Content 1, http://links.lww.com/CCM/C57) and are
summarized in Table 1S (Supplemental Digital Content 1,
http://links.lww.com/CCM/C57). The values used were as
follows: for hemoglobin 14g/dL, AVDo2 = 4.25mL O2/dL;
for hemoglobin 10g/dL, AVDo2 = 3.5mL O2/dL; for hemoglo-
bin 7g/dL, AVDo2 = 2.75mL O2/dL; and for hemoglobin 5g/dL,
AVDo2 = 2.5mL O2/dL.
We examined the influence of other factors (RQ, Pco2, BE,
and Pb) by performing isoshunt analyses at hemoglobin 10g/dL
and AVDo2 3.5 mL O2/dL. For RQ, analyses were performed at
RQs 0.7 and 1.0 and compared with the main analyses at RQ
0.8. Analyses at Pco2 30 and 50 mm Hg were performed and
compared with Pco2 40 mm Hg. For Pb, an analysis was per-
formed at an altitude corresponding to Denver, Pb 625 mm Hg,
and compared with the analyses at Pb 760 mm Hg. Analyses of
BE of 10 and +10 were performed and compared with analy-
ses at BE 0. Calculations were performed using Microsoft Excel
for Mac 2011 version 14.4.6 (Microsoft Corporation, Redmond,
WA). Regression analyses were performed using JMP 11.0 (SAS
Institute, Cary, NC).
Feiner and Weiskopf

RESULTS compensation for the range of hemoglobin 514g/dL, on P/F

for the range of shunt fractions examined, is shown in Figures
Basic Relationship
3, 3S (Supplemental Digital Content 1, http://links.lww.com/
The variation of P/F as a function of Fio2, for a range of Qs/Qt,
CCM/C57), and 4S (Supplemental Digital Content 1, http://
with single values of hemoglobin (10 g/dL) and AVDo2
links.lww.com/CCM/C57). The conditions are the same as for
(3.5mL O2/dL) is shown in Figure 1. P/F is not constant and
Figure2, except that the altered AVDo2 resulting from changes
changes most prominently for Qs/Qt of 0.10.3 and Fio2
in hemoglobin (Table 1S, Supplemental Digital Content 1,
greater than or approximately 0.4. Figure 1S (Supplemental
http://links.lww.com/CCM/C57) are now included in the cal-
Digital Content 1, http://links.lww.com/CCM/C57) presents
culations. The general shape of the isoshunt curves is similar
an alternative analysis, with Qs/Qt on the x-axis.
to those of the uncompensated state (Fig.2), but the values for
Effect of Hemoglobin P/F and the inconsistency of the relationship of P/F to Fio2 dif-
The influence of a range of hemoglobin on P/F is shown in fer. Changing Fio2 without changing other variables can alter
Figure 2. Each panel depicts the range of hemoglobin at a spec- P/F substantially; for example, with a shunt of 0.2 at hemo-
ified Qs/Qt, demonstrating the impact of a change of hemo- globin 10g/dL, changing Fio2 from 1.0 to 0.5 changes P/F from
globin on P/F without any change in pulmonary function. approximately 400 to approximately 200. Increasing hemoglo-
These results do not include any physiological compensation bin now decreases P/F, especially at Fio2 exceeding approxi-
for alterations in hemoglobin concentration and thus are anal- mately 0.30.5. For example, with a Qs/Qt of 0.2, at an Fio2 of
ogous to the pioneering work by Nunn (23, 24). The impact 0.7, the P/F at hemoglobin of 5, 7, 10, and 14g/dL is respec-
of hemoglobin on the relationship of P/F with Fio2 is of little tively, 367, 341, 267, and 217mm Hg. This is the opposite of the
clinical importance in the presence of normal shunt ( 0.01). effect that occurs in the uncompensated state, where at high
However, at increased shunt fractions, decreasing hemoglobin Fio2, changes in hemoglobin have little effect and at lower Fio2,
concentration decreases P/F, especially at lower Fio2. Figure 2S increasing hemoglobin increases P/F. Figure 3S (Supplemental
(Supplemental Digital Content 1, http://links.lww.com/CCM/ Digital Content 1, http://links.lww.com/CCM/C57) shows the
C57) similarly shows the effect of hemoglobin but with panels effect of hemoglobin with panels of different hemoglobins and
at different hemoglobins and Qs/Qt changing. Qs/Qt changing. Figure 4S (Supplemental Digital Content 1,
The above analysis, while of substantial theoretical impor- http://links.lww.com/CCM/C57) shows P/F ratio plotted ver-
tance, may not represent accurately the true relationships in sus Qs/Qt, with each panel at a different Fio2.
most clinical circumstances, as it does not include changes The explanation for the impact of changes of hemo-
because of physiological compensation for changes of hemo- globin concentration can be appreciated by examining the
globin. The influence of changing hemoglobin concentrations influence of AVDo2 on P/F. Figures4 and 5S (Supplemental
(e.g., anemia or RBC transfusion) when including physiological Digital Content 1, http://links.lww.com/CCM/C57) demon-
strate the influence of AVDo2
on P/F at hemoglobin 10g/
700 dL, for the range of shunt
PaO2/FIO2 D
Ratio (mm Hg) fractions examined. In the
600 presence of a normal shunt
Shunt Fraction (Qs/Qt): ( 0.01) AVDo2 has little
impact. The influence is
1% 500
10% greater, and at times sub-
20% stantial, in the presence of
30% 400 increased Qs/Qt, especially
40%
50% at Fio2 greater than approxi-
300
mately 0.4. These figures are
also of use when examin-
ing nonpulmonary causes of
200 changes in P/F because of other
causes of changes in AVDo2,
100 such as an increase because of
Hb = 10 g/dL cardiogenic shock, or a decrease
AVDO2 = 3.5 mL O2/dL
because of septic shock.
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Influence of Other Factors
a) RQ: the effect of RQ (range,
Figure 1. Pao2/Fio2 ratio as a function of Fio2 from room air (0.21) to 1.0 for simulated constant intrapulmonary
shunt (Qs/Qt), from 0.01 to 0.5. The simulation was calculated for a hemoglobin (Hb) concentration of
0.71.0) on P/F is generally
10g/dL and arterial-venous oxygen content difference of 3.5mL O2/dL. small (Fig. 5A). The greater the

e42 www.ccmjournal.org January 2017 Volume 45 Number 1

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
 Online Clinical Investigations

A 700 B 700
PaO2/FIO2 PaO2/FIO2
Ratio (mm Hg) Ratio (mm Hg)
600 600
Hemoglobin Hemoglobin
Hb = 5 g/dL 500 Hb = 5 g/dL 500
Hb = 7 g/dL Hb = 7 g/dL
Hb = 10 g/dL Hb = 10 g/dL
Hb = 14g/dL 400 Hb = 14g/dL 400

300 300

200 200

100 100
Qs/Qt = 1% Qs/Qt = 10%
AVDO2 = 3.5 mL/dL AVDO2 = 3.5 mL/dL
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

C 700 D 700
PaO2/FIO2 PaO2/FIO2
Ratio (mm Hg) Ratio (mm Hg)
600 600
Hemoglobin Hemoglobin
Hb = 5 g/dL 500 Hb = 5 g/dL 500
Hb = 7 g/dL Hb = 7 g/dL
Hb = 10 g/dL Hb = 10 g/dL
Hb = 14g/dL 400 Hb = 14g/dL 400

300 300

200 200

100 100
Qs/Qt = 20% Qs/Qt = 30%
AVDO2 = 3.5 mL/dL AVDO2 = 3.5 mL/dL
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

E 700 F 700
PaO2/FIO2 PaO2/FIO2
Ratio (mm Hg) Ratio (mm Hg)
600 600
Hemoglobin Hemoglobin
Hb = 5 g/dL 500 Hb = 5 g/dL 500
Hb = 7 g/dL Hb = 7 g/dL
Hb = 10 g/dL Hb = 10 g/dL
Hb = 14g/dL 400 Hb = 14g/dL 400

300 300

200 200

100 100
Qs/Qt = 40% Qs/Qt = 50%
AVDO2 = 3.5 mL/dL AVDO2 = 3.5 mL/dL
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

Figure 2. Pao2/Fio2 ratio as a function of Fio2 at several hemoglobin concentrations (Hb), using a constant arterial to venous oxygen content
differences (AVDo2) value of 3.5mL O2/dL, representing no cardiovascular compensation. Each panel shows a different shunt fraction (Qs/Qt):
A, 0.01; B, 0.1; C, 0.2; D, 0.3; E, 0.4; F, 0.5.

RQ, the greater PAo2. The effect is largest at the lowest Qs/Q
and at lower Fio2.
b) Pco2: the effect of Pco2 on P/F is variable but never sub-
stantial at Qs/Qt 0.10.5 (Fig. 5B). For a normal Qs/Qt of
0.01, P/F is lower at higher Pco2. However, at Qs/Qt of 0.1
through 0.3, there is an Fio2 at which this effect is reversed.
For shunt of 0.1, 0.2, and 0.3, the Fio2 at which P/F becomes
t
higher at a lower Pco2 is approximately 0.38, 0.63, and 0.92,
respectively. For shunt fractions of 0.4 and 0.5, P/F is lower
with lower Paco2 at all Fio2s.
c) BE: in general, the effect of BE is relatively small (Fig. 5C)
and less than that of Pco2. Higher BE (alkalosis) results in
lower P/F that is more significant at higher shunt fractions.
The effect is more pronounced at lower Fio2, so that an

Critical Care Medicine www.ccmjournal.org e43

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Feiner and Weiskopf

A 700 B 700
PaO2/FIO2 PaO2/FIO2
Ratio (mm Hg) Ratio (mm Hg)
600 600
Hemoglobin Hemoglobin
Hb = 5 g/dL 500 Hb = 5 g/dL 500
Hb = 7 g/dL Hb = 7 g/dL
Hb = 10 g/dL Hb = 10 g/dL
Hb = 14g/dL 400 Hb = 14g/dL 400

300 300

200 200

100 100
Qs/Qt = 1% Qs/Qt = 10%
Compensated AVDO2 Compensated AVDO2
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

C PaO2/FIO2
700 D PaO2/FIO2
700

Ratio (mm Hg) Ratio (mm Hg)

600 600
Hemoglobin Hemoglobin
Hb = 5 g/dL 500 Hb = 5 g/dL 500
Hb = 7 g/dL Hb = 7 g/dL
Hb = 10 g/dL Hb = 10 g/dL
Hb = 14g/dL 400 Hb = 14g/dL 400

300 300

200 200

100 100
Qs/Qt = 20% Qs/Qt = 30%
Compensated AVDO2 Compensated AVDO2
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

E 700 F 700
PaO2/FIO2 PaO2/FIO2
Ratio (mm Hg) Ratio (mm Hg)
600 600
Hemoglobin Hemoglobin
Hb = 5 g/dL 500 Hb = 5 g/dL 500
Hb = 7 g/dL Hb = 7 g/dL
Hb = 10 g/dL Hb = 10 g/dL
Hb = 14g/dL 400 Hb = 14g/dL 400

300 300

200 200

100 100
Qs/Qt = 40% Qs/Qt = 50%
Compensated AVDO2 Compensated AVDO2
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

Figure 3. Pao2/Fio2 ratio as a function of Fio2 at hemoglobin (Hb) concentrations from 5 to 14g/dL, using arterial to venous oxygen content differences (AVDo2)
values for appropriate physiological cardiovascular compensation (for hemoglobin 14 g/dL, AVDo2 = 4.25 mL o2/dL, for 10 g/dL, AVDo2 = 3.5 mL o2/dL; for
7 g/dL, AVDo2 = 2.75 mL o2/dL; and for 5 g/dL, AVDo2 = 2.5 mL o2/dL). Each panel shows a different shunt fraction (Qs/Qt): A, 0.01; B, 0.1; C, 0.2; D, 0.3;
E, 0.4; F, 0.5.

effect remains at Fio2 of 1.0 only for shunt fractions of 0.3
and more.
d) Pb: lower Pb results in lower Pio2 and PAo2 and conse-
quently lower Pao2 and P/F (Fig. 5D). The effect is lesser
at higher Qs/Qt and larger at higher Fio2. For example,
at the approximate altitude of Denver, CO (Pb = 625mm
Hg), P/F is lower by 135, 124, 131, 85, 15, and 5mm Hg for
Qs/Qt of 0.01, 0.1, 0.2, 0.3, 0.4, and 0.50, respectively.
DISCUSSION
P/F was instituted as a convenient measure using readily avail-
able data, for assessment of changes in pulmonary gas exchange
as a measure of disease progression or result of therapy and to
be able to compare patients with disparate Fio2. The ideal vari-
able for diagnostic and comparative purposes would be sensi-
tive and specific for changes in pulmonary gas transfer and be
insensitive to changes of Fio2 and nonpulmonary effects.

e44 www.ccmjournal.org January 2017 Volume 45 Number 1

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
 Online Clinical Investigations

A PaO /F O I
700 B PaO2/FIO2
700
2 2
Ratio (mm Hg) Ratio (mm Hg)
600 600
AVDO2 AVDO2

1.5 mL O2/dL 500 1.5 mL O2/dL 500

2.5 mL O2/dL 2.5 mL O2/dL
3.5 mL O2/dL 3.5 mL O2/dL
6.0 mL O2/dL 400 6.0 mL O2/dL 400

300 300

200 200

100 100
Qs/Qt = 1% Qs/Qt = 10%
Hb = 10 g/dL Hb = 10 g/dL
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

C 700 D 700
PaO2/FIO2 PaO2/FIO2
Ratio (mm Hg) Ratio (mm Hg)
600 600
AVDO2 AVDO2

1.5 mL O2/dL 500 1.5 mL O2/dL 500

2.5 mL O2/dL 2.5 mL O2/dL
3.5 mL O2/dL 3.5 mL O2/dL
6.0 mL O2/dL 400 6.0 mL O2/dL 400

300 300

200 200

100 100
Qs/Qt = 20% Qs/Qt = 30%
Hb = 10 g/dL Hb = 10 g/dL
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

E PaO /F O
2 I 2
700
F PaO2/FIO2
700

Ratio (mm Hg) Ratio (mm Hg)

600 600
AVDO2 AVDO2

1.5 mL O2/dL 500 1.5 mL O2/dL 500

2.5 mL O2/dL 2.5 mL O2/dL
3.5 mL O2/dL 3.5 mL O2/dL
6.0 mL O2/dL 400 6.0 mL O2/dL 400

300 300

200 200

100 100
Qs/Qt = 40% Qs/Qt = 50%
Hb = 10 g/dL Hb = 10 g/dL
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

Figure 4. Pao2/Fio2 ratio as a function of Fio2 at values of arterial to venous oxygen content differences (AVDO2) of 1.5, 2.5, 3.5, and 6.0 mL O2/dL at a
hemoglobin concentration of 10 g/dL. Each panel shows a different shunt fraction (Qs/Qt): A, 0.01; B, 0.1; C, 0.2; D, 0.3; E, 0.4; F, 0.5.

Our analyses indicate that for some clinical condi-
tions, P/F meets these objectives, whereas for others,
it does not. In examining the relationship of P/F with
Fio2, we have identified important issues (1): the vari-
ability of P/F at differing Fio2 and differing Qs/Qt (2),
the substantial impact of hemoglobin and AVDo2 and
their relationship, and (3) the effect of other nonpul-
monary factors. In general, P/F is not independent of
Fio2. The extent to which P/F varies differs depending on
Qs/Qt, even when all other variables are constant. In general,
the deviation of P/F from a constant value is greatest for
Qs/Qt values of 0.10.3, at Fio2 of approximately greater
than or equal to 0.4, where perhaps an assessment of pul-
monary derangement of oxygen transfer would have its
greatest use. At Qs/Qt below 0.1 and Fio2 less than 0.4 and at
Qs/Qt of greater than or equal to 0.4, the pulmonary disor-
der is so small and great, respectively, that differences in P/F
would not likely be of significant diagnostic importance.

Critical Care Medicine www.ccmjournal.org e45

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Feiner and Weiskopf

A PaO2/FIO2
700 B PaO2/FIO2
700

Ratio (mm Hg) Ratio (mm Hg)

600 600
Shunt Fraction (Qs/Qt): Shunt Fraction (Qs/Qt):

1% 500 1% 500
10% 10%
20% 20%
30% 30% 400
400
40% 40%
50% 50%

300 300
Respiratory Quotient (RQ) Carbon Dioxide (CO2)

0.7 200 30 mm Hg 200

0.8 40 mm Hg
1.0 50 mm Hg
100 100
Hb = 10 g/dL Hb = 10 g/dL
AVDO2 = 3.5 mL O2/dL AVDO2 = 3.5 mL O2/dL
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

C PaO2/FIO2
700 D PaO2/FIO2
700

Ratio (mm Hg) Ratio (mm Hg)

600 600
Shunt Fraction (Qs/Qt): Shunt Fraction (Qs/Qt):
1% 500 1% 500
10% 10%
20% 20%
30% 400 30% 400
40% 40%
50% 50%
300 300
Base Excess (BE) Barometric Pressure (Pb)
-10 200 760 mm Hg 200
0.0 625 mm Hg
+10
100 100
Hb = 10 g/dL Hb = 10 g/dL
AVDO2 = 3.5 mL O2/dL AVDO2 = 3.5 mL O2/dL
0 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Inspired Oxygen (FIO2) Fraction of Inspired Oxygen (FIO2)

Figure 5. Influence of respiratory quotient (RQ) (0.7, 0.8, and 1.0; A), Paco2 (30, 40, and 50mm Hg; B), base excess (BE) (10, 0, and +10; C), and
barometric pressure (Pb) (625 and 760mm Hg; D) on Pao2/Fio2 (P/F) ratio at shunt fractions 0.01, 0.10, 0.20, 0.30, 0.40, and 0.50.

The deviation of P/F from constancy in the Qs/Qt range
of 0.10.3 can be substantial and of important diagnos-
tic impact. For example, with Qs/Qt of 0.2, with Fio2 0.7,
changes in hemoglobin from 7 to 10 g/dL changes P/F from
341 to 267 and thus could alter the diagnosis of acute lung
injury or ARDS or assessment of progress of disease. These
issues, and those below, also likely limit the utility of P/F
as a robust research tool, in the presence of changes of the
variables affecting P/F.
Gowda and Klocke (20) first recognized the curvilinear
relationship between P/F ratio and Fio2 in an analysis that was
compared with patient data. They used an assumed hemoglo-
bin, with a few simulations, but recognized that they did not
consider AVDo2. Similar to Gowda and Klocke, Karbing et al
(21) developed a model focused on comparing patient data with
their predicted P/F ratio. Both groups were particularly focused
on whether a pure shunt model or one that incorporated a com-
ponent of V/Q mismatch best fit patient data. Aboab et al (22)
used a mathematical model of pure shunt without V/Q mis-
match, which also showed the curvilinear relationship between
P/F and Fio2. However, none of these models developed a com-
plete understanding of hemoglobin, AVDo2, and their interrela-
tionship, which likely have a greater impact than V/Q mismatch,
particularly at higher, clinically relevant Fio2.
In examining the ability of P/F to reflect shunt, indepen-
dent of Fio2 and over a broad range of variation of the nonpul-
monary factors (cardiac output or AVDo2 of oxygen content,
hemoglobin concentration, RQ, Paco2, BE, and Pb) that influ-
ence P/F, we used the concept of isoshunt as mathematically
determined and its graphic display, as developed by Nunn (23),
and Benatar et al (24). They developed isoshunt graphics to
enable prediction of Pao2 from Fio2, in order to avoid oxy-
gen toxicity. Their figures displayed Pao2 as a function of Fio2.
Their work predated the use of P/F ratio and thus did not ana-
lyze P/F as a function of Fio2. Most of their analyses used a fixed
AVDo2 of 5mL O2/dL, with a few analyses at a very high AVDo2
of 8mL O2/dL, and none at a more commonly found, lesser
AVDo2 that is associated with lower hemoglobin concentra-
tion. In addition, while recognizing the impact of hemoglobin
concentration because of the nonlinear relationship between
Po2 and So2, their analyses were limited to hemoglobin concen-
trations of 14 and 10g/dL as the former represented normal
and the latter the lower acceptable clinical limit at that time,
as red cell transfusion was commonly instituted at that hemo-
globin concentration, and thus, very few patients had lower
hemoglobin concentrations. Our results demonstrate a much
greater impact of lesser, but not uncommon, concentrations of
hemoglobin. Importantly, when examining the contribution

e46 www.ccmjournal.org January 2017 Volume 45 Number 1

Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

of venous admixture, it is necessary to allow AVDo2 to change
to accommodate changes of cardiac output to maintain Vo2
unchanged. Other analyses have maintained AVDo2 constant,
thus underestimating the contribution of decreased hemoglo-
bin concentrations.
It is important to recognize that changes in hemoglobin
alone can alter P/F and thus give the erroneous impression
of a change in pulmonary function. For example, changes in
hemoglobin in the presence of moderate shunt can change P/F
from above to below 200 or 300 (or the reversefrom below to
above) and thus alter the diagnosis ARDS (16) or TRALI (17).
Consequently, it is possible that some reports of pulmonary
adverse events with anemia or transfusion could benefit from
reassessment. Similarly, future reports in this area should take
note of this effect.
The analyses using physiological compensation for acute
anemia, that is changes of AVDo2, may also be used to esti-
mate the effects of altered AVDo2 because of factors other
than changes of hemoglobin: for example, that of high AVDo2
because of low cardiac output of cardiac failure or low AVDo2
with septic shock. Here, too, reports of changes in P/F asso-
ciated with conditions of abnormal AVDo2 should take this
point into account. To the extent that people with pathophysi-
ological limitations might not be able to fully compensate for
acute anemia, the clinical results would lie somewhere between
those we depicted for uncompensated states and those shown
for full compensation. Of note, previous model development
does not appear to have recognized this point.
The effects of RQ, Pco2, and BE are relatively small, espe-
cially for the clinical circumstances where the assessment of
P/F is important, that of clinically significant impaired oxy-
genation and high Fio2. Pb, of course, by decreasing PAo2, has a
substantial impact on Pao2. The isoshunt curves at decreased
Pb are very similar to those at higher Pb, and thus, when main-
taining Pb unchanged, as is the normal clinical circumstance,
the utility of P/F should be similar. However, attempts to com-
pare P/F of patients at differing Pb would likely be problematic.
The limitations of some aspects of our examination of P/F
should be understood. We made several assumptions, of neces-
sity, as it would be unrealistic to examine all possible physi-
ological perturbations that might influence the results. We
assumed that capillary Po2 is equal to PAo2, a standard, neces-
sary assumption of the shunt equation. This issue has been
examined by others, with the conclusion that states of disequi-
librium between the two are rare (30, 31). We also assumed
that Paco2 is equal to PAco2. Even substantial changes of Paco2
have relatively small effect on P/F (Fig. 5B). Changes in Paco2
have two effects on P/F. First, Paco2 alters PAo2 and therefore
Pao2. The second effect is its influence on the association of
hemoglobin and O2 (32).
Lastly, we have not modeled the effects of V/Q mismatch.
We sought to evaluate P/F because of its common use for
evaluating severe abnormalities of oxygenation, and its use as a
criterion for the definitions of ARDS and TRALI and for lung
transplantation. In 20 spontaneously breathing patients in an
ICU, V/Q mismatch had a negligible effect above an Fio2 of
Critical Care Medicine www.ccmjournal.org e47
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Online Clinical Investigations
0.35 (7). We have shown that nonpulmonary factors, which
were not assessed in previous models, could have even larger
influence.
CONCLUSIONS
The utility of P/F for assessment of pulmonary transfer of
oxygen is variable. It is not independent of Fio2 and var-
ies with Fio2, Qs/Qt, and several extrapulmonary factors.
Thus, the use of P/F can be a reasonable assessment of
alterations of pulmonary transfer of oxygen in some cir-
cumstances but not others. Understanding the conditions
that influence P/F and the effects of nonpulmonary factors
may enhance appropriate use and interpretation of P/F for
diagnosis, assessment of therapy, and comparison among
patients. Particularly important are the substantial effects
of changes of hemoglobin and of AVDo2. P/F ratio may be
adequate for some but not for many clinical uses; it is likely
problematic for research applications. The widespread use
of electronic systems could allow the adaptation of more
complex models.
REFERENCES
1. Sackur P: Weiteres zur Lehre vom Pneumothorax. Virchow Arch fr
pathologische Anatomie und Physiologie und fr kllinische Medicin
1897; 155:151160
2. Hedley-Whyte J, Laver MB, Bendixen HH: Effect of changes in tidal
ventilation on physiologic shunting. Am J Physiol 1964; 206:891
897
3. Bendixen HH, Egbert LD, Hedley-Whyte J, et al: Respiratory Care.
Saint Louis, The C. V. Mosby Company, 1965
4. Farhi LE, Rahn H: A theoretical analysis of the alveolar-arterial O2
difference with special reference to the distribution effect. J Appl
Physiol 1955; 7:699703
5. Rahn H, Sadoul P, Farhi LE, et al: Distribution of ventalation and perfu-
sion in the lobes of the dogs lung in the supine and erect position.
JAppl Physiol 1956; 8:417426
6. West JB: Regional differences in gas exchange in the lung of erect
man. J Appl Physiol 1962; 17:893898
7. Petros AJ, Dor CJ, Nunn JF: Modification of the iso-shunt lines for
low inspired oxygen concentrations. Br J Anaesth 1994; 72:515
522
8. Lyons JH, Lee C-J, Aquino C, et al: Biochemical patterns in septic
shock. In: Septic Shock Proceedings of a Workshop. Strawitz JG,
Grossblatt N (Eds). Washington, DC, National Academy of Sciences
- National Research Council, 1965, pp 223230
9. Horovitz JH, Carrico CJ, Shires GT: Pulmonary response to major
injury. Arch Surg 1974; 108:349355
10. Murray JF, Matthay MA, Luce JM, et al: An expanded definition of
the adult respiratory distress syndrome. Am Rev Respir Dis 1988;
138:720723
11. Covelli HD, Nessan VJ, Tuttle WK 3rd: Oxygen derived variables in
acute respiratory failure. Crit Care Med 1983; 11:646649
12. Cotev S, Perel A, Katzenelson R, et al: The effect of PEEP on oxygen-
ating capacity in acute respiratory failure with sepsis. Crit Care Med
1976; 4:186192
13. Martyn JA, Aikawa N, Wilson RS, et al: Extrapulmonary factors influ-
encing the ratio of arterial oxygen tension to inspired oxygen concen-
tration in burn patients. Crit Care Med 1979; 7:492496
14. Bredenberg CE, James PM, Collins J, et al: Respiratory failure in
shock. Ann Surg 1969; 169:392403
15. Gilbert R, Keighley JF: The arterial-alveolar oxygen tension ratio. An
index of gas exchange applicable to varying inspired oxygen concen-
trations. Am Rev Respir Dis 1974; 109:142145
Feiner and Weiskopf
16. Bernard GR, Artigas A, Brigham KL, et al: The American-European
Consensus Conference on ARDS. Definitions, mechanisms, relevant
outcomes, and clinical trial coordination. Am J Respir Crit Care Med
1994; 149(Pt 1):818824
17. Toy P, Popovsky MA, Abraham E, et al; National Heart, Lung and
Blood Institute Working Group on TRALI: Transfusion-related acute
lung injury: Definition and review. Crit Care Med 2005; 33:721
726
18. Weill D: Donor criteria in lung transplantation: An issue revisited.
Chest 2002; 121:20292031
19. Christie JD, Carby M, Bag R, et al; ISHLT Working Group on Primary
Lung Graft Dysfunction: Report of the ISHLT Working Group on Pri-
mary Lung Graft Dysfunction part II: Definition. A consensus state-
ment of the International Society for Heart and Lung Transplantation.
J Heart Lung Transplant 2005; 24:14541459
20. Gowda MS, Klocke RA: Variability of indices of hypoxemia in adult
respiratory distress syndrome. Crit Care Med 1997; 25:4145
21. Karbing DS, Kjaergaard S, Smith BW, et al: Variation in the Pao2/
FiO2 ratio with FiO2: Mathematical and experimental description, and
clinical relevance. Crit Care 2007; 11:R118
22. Aboab J, Louis B, Jonson B, et al: Relation between Pao2/FIO2 ratio
and FIO2: A mathematical description. Intensive Care Med 2006;
32:14941497
23. Nunn JF: The lung as a black box. Can Anaesth Soc J 1966; 13:8197
e48 www.ccmjournal.org January 2017 Volume 45 Number 1
Copyright 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
24. Benatar SR, Hewlett AM, Nunn JF: The use of iso-shunt lines for con-
trol of oxygen therapy. Br J Anaesth 1973; 45:711718
25. Fenn WO, Rahn H, Otis AB: A theoretical study of the composition of
the alveolar air at altitude. Am J Physiol 1946; 146:637653
26. Severinghaus JW: Simple, accurate equations for human blood O2
dissociation computations. J Appl Physiol Respir Environ Exerc
Physiol 1979; 46:599602
27. Roughton FJ, Severinghaus JW: Accurate determination of O2 dis-
sociation curve of human blood above 98.7 percent saturation with
data on O2 solubility in unmodified human blood from 0degrees to
37degrees C. J Appl Physiol 1973; 35:861869
28. Cole RB, Bishop JM: Variation in alveolar-arterial O2 tension dif-
ference at high levels of alveolar O2 tension. J Appl Physiol 1967;
22:685693
29. Weiskopf RB, Viele MK, Feiner J, et al: Human cardiovascular and
metabolic response to acute, severe isovolemic anemia. JAMA 1998;
279:217221
30. Weiskopf RB, Severinghaus JW: Diffusing capacity of the lung for
CO in man during acute acclimation to 14,246 ft. J Appl Physiol
1972; 32:285289
31. Staub NC: Alveolar-arterial oxygen tension gradient due to diffusion.
J Appl Physiol 1963; 18:673680
32. Severinghaus JW: Blood gas calculator. J Appl Physiol 1966;
21:11081116