mathematical modelled approach to gas exchange monitoring.
overview of one parameter models and description of Sapsford and Kjaergaard's two parameter models.computer algorithm in assessing gas exchange at the bedside in a MIGET fashion. prediction of hypoxemia and full description of gas exchange through models that fit perfectly to patient data.
4. ο§ Bohr equation
VDanat/VT =
π·π¨πͺπΆπβπ·π¬πͺπΆπ
π·π¨πͺπΆπ
ο§ Enghoff modification
VDphysiol/VT =
π·ππͺπΆπβπ·π¬πͺπΆπ
π·ππͺπΆπ
ο§ Alveolar dead space by substraction
VDalv/VT = Enghoff β Bohr
VDalv/VT alv =
π·ππͺπΆπβπ·πππͺπΆπ
π·ππͺπΆπ
Dead Space
-fractions-
PACO2-alv mixed-
EtCO2(mean)
EtCO2=mean!-
(min+max)/2
5. Dead Space Caveats
ο§ Shunt dependence ( βshunt dead spaceβ β Suter,1975)
because assuming that PaCO2=PACO2 is flawed
ο§ Shunt dependence will spuriously elevate VD
ο§ Regions with β Va/Q are poorly set apart from regions with
Va/Q = β(true dead space) because CO2 solubility is rather
modest in comparison to acetone solubility, which is used in
MIGET and distingushes VD as regions with Va/Q>100)
ο§ Severe V/Q mismatch => βsloping alveolar plateauβ
ο§ Severe heterogeneity in Ο => βsloping alveolar plateauβ
ο§ Sometimes PetCO2 > PaCO2
6. Shunt dependence of VD
Bull Eur Physiopathol Respir 1984
Effect of right-to-left shunting on alveolar dead space.
Mecikalski et al
7. Negative β CO2
Great heterogeneity in Rβ’C product or/and severe V/Q mismatch
-sloping alveolar plateau-
8. Negative β CO2
ο§ ETCO2 is continuously estimated while PaCO2 is a mean value.
ο§ ETCO2 can be regarded as a regional and temporal specific parameter while PaCO2 is a global, mean
parameter with no regional or temporal attributes
ο§ A low Ο CO2(FRC/VCO2) as in IACS or a non-homogeneous Ο lung(Rβ’C) will facilitate negative
differences
9. ο§ Slow alveoli are characterized by a high RC and this assigns them a constant,
moderate sloping.
ο§ Fast alveoli are characterized by a low RC and this gives them a 2 phase sloping, the
second being responsible for the overshoot ( high FRC/VCO2 )
ο§ Eg. Obese patients ( Ecw high )
Negative β CO2
10. Dead Space as risk factor
-Enghoffβs dead space-
PULMONARY DEAD-SPACE FRACTION AS A
RISK FACTOR FOR DEATH IN THE ACUTE
RESPIRATORY DISTRESS SYNDROME, NEJM
2002, Nuckton et al
11. Dead Space as a PEEP setter
Optimum end-expiratory airway pressure in
patients with acute pulmonary failure, Suter et
al, NEJM 1975
12. Dead Space as a PEEP setter
OL-PEEP
OL-PEEP
Monitoring dead space during recruitment and PEEP titration in an
experimental model, ICM 2006, Suarez-Sipmann et al.
Recruitment=β βEELV=βSTRAINst+dyn=βVD
13. Dead Space as a PEEP setter
- VD as an image of respiratory mechanics more than of gas excahnge -
ο§ Best PEEP=lowest dynamic and static STRAIN
ο§ Best E=Best Vd
ο§ VD obeys Hickling model (1998 )
ο§ VD shows histeresis
ο§ VD is mechanics as well as E and is decoupled from gas
excange ( PaO2 )
Compliance and Dead Space Fraction Indicate an Optimal Level
of Positive End-Expiratory Pressure After Recruitment in
Anesthetized Patients, Anesth Analg 2008, Maisch and Tusman
14. Volumetric Capnography
Ξ²
ο§ Integrating the CO2 and volume signals
ο§ The abscissa is represented by volume
ο§ 3 phases, 2 slopes, one inflection point-the curve changes sign-
on SII
15. Volumetric Capnography
-phases and derived variables-
ο§ Phase I begins with the start of expiration and is completed after
βCO2>0.1% from baseline
ο§ Phase II starts at the end of phase I and ends at the intersection point of
slopes SII and SIII. Its inflection point (changes sign) is pretty much its
midpoint and likely represents the interface between Vdaw and alveolar gas,
that is the interface between convection and diffusion. It contains both
alveolar gas as well as Vdaw gas. RC influences phase II.
ο§ Phase III begins at the aforementioned intersection and ends with expiration.
This is gas inside the alveoli.
ο§ Slope II is an image of acini expiratory times. The more homogeneous the
expiration, the more the slope increases.
ο§ Slope III is again influenced by mechanical time constants but mostly by V/Q
mismatch. The slope increases with heterogeneity.
18. ο§ Ay=ABCD=PNCD
ο§ AMP=MNB=Ax
ο§ Ay=VTCO2
ο§ PNCD=PDΓ(PN+CD)/2=VTPDΓmeanCO2alv
ο§ VTCO2=mean expCO2ΓVT=mean expCO2(VTPD + VTOP)
ο§ ( VT β VTOP) Γ mean CO2alv= mean expCO2ΓVT
ο§ VTOP/VT= (meanCO2alv-mean expCO2 )/meanCO2alv
D
o
A P
M
N
B
O
C
Ax
Ax
Ay
22. Vdana and Vdalv can be read simultaneously on the abscissa
Volumetric Capnography
TANG 2006 β all VDs
23. =225 ml
=160 ml
=65 ml
Volumetric Capnography
TANG 2006 β all VDs
Vdana and Vdalv can be read simultaneously on the abscissa
24. ο§ We draw perpendiculars so that AOJA = AHJI (Fowler)
and AOKB = AFEDK (Tang)
ο§ VT = OC ; VDanat = OA (Fowler) ; VDphys = OB (Tang)
ο§ PECO2 = AODC /VT
ο§ VDphys Enghoff = VT Γ (1-PECO2/PaCO2) =
= VT Γ (1-AODC/ (PaCO2ΓVT))
G F E
D
H
I
K
J
C
J
A B
C
D
EFG
H
I
K
ο§ AODC = AOKB + ABKDC = ABKDC+AFEDK = ABCEF
ο§ VDphys Enghoff = VTΓ[1-ABCEF/(PaCO2ΓVT) ]=
= VTΓ[1-(PaCO2ΓBC)/(PaCO2ΓVT)]
= OB = VDphys Tang
28. How Tusman et al have confused Graf
Bohr equation
VDphysiol/VT =
π·π¨πͺπΆπβπ·π¬πͺπΆπ
π·π¨πͺπΆπ
PACO2-alv mixed-
EtCO2(mean)
30. Diffusion Limitation
- one compartment modeling -
V T
VDana water
FEO2
FECO2
FIO2
FICO2
CvO2
Q
CcO2PcO2
CaO2
Q
VCO2 VO2
VA
PAO2
PACO2
ο§ πππππππ ππππππππππ π·π¨πͺπΆ π = π¬π»πͺπΆ π ππππ
ο§ PAO2=PIO2 - PACO2 Γ ππ°πΆ π +
πβππ°πΆ π
πΉ
ο§ RDIFF =
π
π«π³πΆ π
ο§ RDIFF =
π·π¨πΆ πβπ·ππΆ π
π½πΆ π
RDIFF, when computed through a one compartment model, is nothing but a global parameter,
It does not set apart any of the gas-exchange abnormalities.
35. Fitting one parameter models to data
84
86
88
90
92
94
96
98
100 SaO2
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
FEO2
Shunt=21%
fA2=0.27
Shunt=11%
fA2=0.19
Shunt=10%
fA2=0.14
Shunt=8%
fA2=0.14
Dashed red line
ο Alveolar dead space model
Vdalv/VT=53%
ο Diffusion limitation model
Rdiff=42kPa/(l/min)
ο V/Q model
fA2=0.27
Solid orange line
ο Shunt model
Shunt=15.5%
stands for SaO2/FEO2 for the same
patient.
ο§ using all necessary data, it is
calculated for each of these four
points shunt and fA2 according to
previous equations.
ο§ fitting parameter model to data =
finding the one parameter value that will
subsequently describe patientβs data
with utmost precision
ο§ one parameter models show dependence on inspired O2 fraction
ο§ they cannot appropriately describe gas-exchange
36. Beginnings of Two Parameter Models
The PIO2 vs. SpO2 diagram: A non-invasive measure of pulmonary oxygen
exchange, EUROPEAN JOURNAL OF ANAESTHESIOLOGY 1995, Sapsford and
Jones - Cambridge
ο§ The two parameters are shunt and V/Q mismatch +
PACO2/R effect measured as % and as P I O2-PcO2
(kPa) respectively
ο§ Mass balance for O2 in blood and air, ODC equation,
computer algorithm based on fitting the model
parameters to P I O2/SaO2 data pairs
37. Beginnings of Two Parameter Models
A noninvasive method for evaluating the effect of thoracotomy on shunt and
perfusion inequality, ANAESTHESIA 1997, Gray and Jones
38. Beginnings of Two Parameter Models
- course of family of curves -
Noninvasive assessment of shunt and ventilation/perfusion ratio in neonates with
pulmonary failure, Arch Dis Child Fetal Neonatology Ed. 2001, J G Jones et al
39. We need the numbers
ο§ In A there is dependency of PIO2 vs SaO2 on aVDO2.
Given that aVDO2 is dependent on Q, we infer Q dependency.
Simply eyeballing might not be enough. We need the numbers.
ο§ In B there is dependency on Hb. Hb is nonetheless more
stable.
The PIO2 vs. SpO2 diagram: A non-invasive measure of pulmonary oxygen
exchange, EUROPEAN JOURNAL OF ANAESTHESIOLOGY 1995, Sapsford and
Jones - Cambridge
40. Reverse avDO2 dependency
- monitoring cardiac output -
Cardiac output estimation using pulmonary mechanics in mechanically
ventilated patients, Biomedical Engineering Online 2010, Sundaresan et al
41. Refinement of the two parameter models
β’Rdiff(βPO2)Shunt
β’AlveolarDS(βPO2)Shunt
β’V/Q mismatch(βPO2)Shunt
Mathematical models of pulmonary gas exchange - validation and application to
postoperative hypoxaemia , Aalborg Hospital, Denmark, Soren Kjaergaard
43. Fitting two parameter models to data
ο§ All three models are equivalent in assessing
shunt
ο§ All three models are equivalent in assessing
βPO2
ο§ VDalv inferred from an O2 based 2 parameter
model is NOT equivalent to the one determined
from a CO2 based model
ο§ Rdiff is not supported by MIGET as an
important constituent of gasexchange
disturbances
ο§ VDalv O2 based has no meaning in day to day
clinical practice
44. V/Q mismatch and Shunt Model
- shunt and fA2 impact on ODC -
SHUNT V/Q or fA2
46. Predicting risk of hypoxemia
DISCRIMINATING BETWEEN THE EFFECT OF SHUNT AND REDUCED VA/Q ON ARTERIAL OXYGEN
SATURATION IS PARTICULARLY USEFUL IN CLINICAL PRACTICE, J Clin Monit and Comp 2000, Jones et al
50. PaO2/FIO2
ο§ Risk indicator as in Berlin ARDS definition
ο§ Global gas - exchange parameter
ο§ Non independent behavior with respect to shunt,
avDO2, PaCO2, RQ, Hb
ο§ Non independent parameter when FIO2 is varied
51. PaO2/FIO2 FIO2
dependency according to shunt
ο§ avD02 is constant, that is constant metabolism
ο§ Three shunt values
ο§ At each shunt value, PaO2/FIO2 shows FIO2 dependence
52. PaO2/FIO2 FIO2
dependency according to avDO2
ο§ avD02 varies, that is changing CO for a constant VO2
ο§ Same shunt value
ο§ At each avDo2 value, PaO2/FIO2 shows FIO2 dependence
53. PaO2/FIO2 FIO2
dependency according to shunt
ο§ Shunt varies from 0% to 30%
ο§ Thick lines stand for clinically important SaO2 (92%-98%)
ο§ At each shunt value, PaO2/FIO2 shows FIO2 dependence
54. PaO2/FIO2 FIO2
dependency according to V/Q
ο§ βPO2 ( image of V/Q ) varies from 0 kPa to 30 kPa
ο§ Thick lines stand for clinically important SaO2 (92%-98%)
ο§ At each βPO2 value, PaO2/FIO2 shows FIO2 dependence
55. PaO2/FIO2 FIO2
dependency β switching risk groups
ο§ Six pacients, graphs with SaO2/FIO2 and PaO2/FIO2 FIO2
dependency, two models are used β shunt and shunt+V/Q,
thick lines pertain to SaO2 = 92%-98%, dashed line is shunt model whereas solid line is the other
ο§ PaO2/FIO2 FIO2
dependency brings about different risk groups even though shunt or V/Q do not really change.
57. PaO2/FIO2 FIO2
dependency β switching risk groups
ο§ risk group βswitchingββ is 50% for shunt model and 38% for two
parameter model
ο§ by β FiO2 (SpO2=92-98%)
- shunt model ALI 14β40
- shunt model ARDS 18β38
- two parameter model ALI 23β31
- two parameter model ARDS 18β24
ο§ The shunt model has a poor fit to the data
ο§ PaO2/FiO2 is FIO2 dependent (use the same FIO2 when tracking
evolution)
ο§ PaO2/FiO2 is a poor gas exchange tracker
58. βPerhaps more appropriate would be to replace the PaO2/FiO2 ratio
with two parameters, a parameter to describe the oxygenation
problems due to V/Q mismatch and one to describe oxygenation
problems due to shunt.β
Kjaergaard and Rees, Critical Care 2007