Pulmonary Physiology and Respiratory Failure


Normal minute ventilation is from 4-7 L/min. Normal deadspace ventilation makes up 30% of that. The shape of the oxygen/hemoglobin curve starts to flatten at SpO2 of 90% – for this reason, PaO2 is a poor indicator of oxygen content; however, the A-a gradient is an excellent indicator of gas exchange.

PAO2 = FiO2 (760 – PH2O) – PaCO2/0.8
PAO2 = FiO2 (713) – PaCO2/0.8


Shunting immediately affects PaO2, but PaCO2 does not begin to rise until the shunt reaches ~ 50% [Med Clin North Am 67: 557, 1983]. In high shunt conditions (ex. ARDS, pneumonia, pulmonary edema), increases in oxygen do not necessarily lead to increases in PaO2, thus FiO2 can often be reduced to non-toxic levels. In fact, if the shunt is 50% or more, PaO2 is completely independent of FiO2 [Med Clin North Am 67: 557, 1983]. PaO2/FiO2 is a rough estimate of shunting – if PaO2/FiO2 < 200, shunt fraction > 20%, whereas if PaO2/FiO2 > 200, shunt fraction < 20% [Crit Care Med 11: 646, 1983]

Arterial Blood Gas Variability

Average variability can be as high as 13 mm Hg for PaO2 and 2.5 mm Hg for PaCO2, thus ABG should not be used for routine monitoring of patients unless a clinical change has occurred – otherwise changes in PaO2 and PaCO2, if found, are not necessarily abnormal [J Clin Monitor 8: 111, 1992; Chest 106: 187, 1994]


There are three main classes of etiologies for hypoxemia (hypoventilation, D/V mismatch, V/Q abnormality). The A-a PO2 gradient and PvO2 can help distinguish them – PvO2 is particularly important in high-shunt patients because it can be a major determinant of PaO2. Note that while the A-a PO2 gradient may be useful, the a/A PO2 fraction may be more useful as it is independent of FiO2 (the A-a PO2 gradient increases with increasing FiO2 because the lung doesn’t shunt from as well).

1) Calculate the A-a PO2 gradient but be sure and adjust it for age (at 20 years, normal A-a PO2 is 4 – 17 mm Hg, at 50 it’s 14 – 27 and at 80 it’s 25 – 38) and FiO2 (normal A-acorrected  0.57 x FiO2 + 4.6, which is ~ 16 at 20% and 38 at 60%). If this is normal, measure peak inspiratory pressure (central vs. drug induced hypoventilation).

2) When the A-a PO2 gradient is elevated, measure O2 from the SVC or a distal port in the PA catheter (PvO2). If this value is low, suspect anemia, low cardiac output, or hypermetabolic state. Otherwise, suspect V/Q abnormality.

Shunt Workup: Summary

  • A-a Gradient: excellent measure of shunt, but not completely FiO2-independent. Normal = age/4 + 4
  • a/A Ratio: FiO2-independent measure of shunt. Normal a/A is > 0.75, a/A > 0.25 is predictive of successful weaning
  • P/F Ratio: rough estimate of shunting – if PaO2/FiO2 < 200, shunt fraction > 20%. Normally > 350 (80/0.21). Used to diagnose ARDS (< 200) and ALI (< 300)
  • SvO2: should be checked to ensure that low SvO2 (anemia, low CO) is not the cause of an increased A-a gradient


Always check pH first because this could be a normal response to metabolic alkalosis. If pH is normal, the three major categories are hypoventilation, increased dead space, and excess production (rare because CO2 will normal enhance the respiratory drive, thus elevated pCO2 usually represents a lung abnormality).

1) Calculate the A-a PO2 gradient but adjust it for age and FiO2.

2) VCO2: normally 90 – 130 L/min/m2.

3) PImax: normally > 80 cm H2O, may lead to CO2 retention if < 32 cm H2O. Opiates (and benzodiazepine in the elderly) are a common cause of central hypoventilation. Critical illness polyneuropathy may be the cause of a neuromuscular insufficiency.

Algorithm for Hypoxemia AND Hypercapnia

Hypoxemia in Neurosurgical Patients

Two main classes of hypoxemia in neurosurgical patients: V/Q Mismatch: can include virtually anything – pneumonia, ARDS, COPD, pulmonary edema, and pulmonary embolism, etc. Can be corrected by 100% oxygen. Shunt: pneumonia, atelectasis, pulmonary embolism, etc. Cannot be corrected by 100% oxygen.


Ordinarily, methemoglobin and HbCO make up less than 5% of the total hemoglobin concentration, however in smoke inhalation (CO) or high-dose nitroglycerin (methHb), this may not be valid and a two channel oximeter will be wrong. Otherwise, at SpO2 > 70%, pulse oximetry differs from co-oximetry by < 3%. Pulse oximetry is also accurate down to blood pressures of 30 mm Hg, [Hg] of 2 – 3 g/dL, and through bilirubinemia. It can be altered by very dark skin, blue or back nail polish, onychomycosis, or methylene blue injections (treatment for methemoglobinemia). Pulse oximetry is generally more accurate than arterial blood gases [Anesth Intensive Care 21: 72, 1993] for measuring saturation and is superior to ABGs at detecting hypoxia [Crit Care Clin 11: 199, 1995] and has been shown to be superior to serial ABGs for assessing hypoxemia [Crit Care Clin 11: 199, 1995]. The one drawback of oximetry is that in the normal operative range of the lungs, large changes in PAO2 go undetected because the sigmoid curve is flat. An interesting new development is forehead sensors, which detect changes in SaO2 much more rapidly than fingertip oximeters [Crit Care Med 29: A115, 2001]. SpO2 is an accurate marker for hypoventilation in patients on room air but not in patients on supplemental oxygen [Chest 126: 1552, 2004]

SvO2 (mixed venous oximetry) can be measured in specialized PA catheters and is equal to [CO x Hb x SaO2] / VO2, thus it is a good screening method for these four variables but in and of itself tells you nothing.

Mixed venous saturation (SvO2 , must be measured in the PA) below 70-75% suggests inadequate O2 delivery either due to decreased DO2 (low cardiac output, anemia, hypoxemia) or increased VO2 (hypermetabolism) – continuous SvO2 monitoring by PA catheters is accurate to within 1.5% of SvO2 measured by clinical labs [Intensive Care Med 20: 484, 1994] i.e., it is highly accurate.

Central venous saturation (ScvO2) is measured in the SVC and tends to be slightly lower than SvO2 (brain is metabolically active), and this difference is exaggerated in shock states [Curr Opin Crit Care 7: 204, 2001] – single measurements can differ by 10%, but multiple measurements are within 5% [Anesthesiology 103: 249, 2005], thus ScvO2 may be better suited for trending and not absolute values. Still, because of the increased morbidity associated with PA catheters, therapeutic guidelines for sepsis now recommend maintaining a ScvO2 > 70% [Crit Care Med 32: 858, 2004] as a therapeutic endpoint.

The most accurate parameter is likely to be (SaO2 – Sc/vO2), which takes into account oxygen delivery – normal range for extraction is 20 – 30%, whereas values 50 – 60% or higher suggest impending dysoxia and a risk for anaerobiasis.

Capnography (CO2 Detection – Colormetric And Infrared Methods)

Breath sounds are unreliable when attempting to locate ET tubes post-placement [J Clin Mon 7: 232, 1991]. Colorimetric CO2 monitors (purple  yellow if > 5% CO2) should be used in addition to the stethoscope, but may also be inaccurate as a means of detecting ET tube placement (ex. following cardiac arrest, CPR, or carbonated beverage ingestion), thus 4 breaths are prudent before measuring for CO2. Keep in mind that the colorimetric monitor, while otherwise 99-100% sensitive/specific, is only 23% specific for esophageal intubation in cardiac arrest cases [Ann Emerg Med 21: 518, 1992], mostly because absent cardiac output leads to minimal CO2 production.

Colorimetry has recently been used to evaluate feeding tube placement, and if these recent positive results [Crit Care Med 30: 2255, 2002] are corroborated, this may replace the chest X-ray.

Alternatively, infrared capnography can be used following intubation In a normal person, PaCO2 – PETCO2 (end-tidal CO2) is < 3 mm Hg (the first requires an ABG to measure, the second a CO2 monitor i.e., capnography). When gas exchange in the lung is impaired (virtually any pulmonary or cardiac disorder), the PaCO2 – PETCO2 gradient increases. Any change in gas exchange (ex. ventilator settings) will alter this value, so the gas must be redrawn and the number recalculated if it is to be followed. PETCO2 can now be calculated even in non-ventilated patients (using a modified nasal cannula [J Clin Mon 7: 249, 1991]), and has several uses:

Cardiac Output Monitoring (during volume or cardiopulmonary resuscitation): as cardiac output increases, PETCO2 should increase [r = 0.87, J Clin Monit 8: 175, 1992] Ventilation: drop in PETCO2 can signify a disconnected ventilator or migration of the ET tube into a mainstem bronchus. Nosocomial Disorders: drop in PETCO2 accompanied by an increase in the PaCO2 – PETCO2 gradient can signify PE, atelectasis, overdistension of alveoli (ex. excessive PEEP), low CO, pneumonia, or pulmonary edema. The PaCO2 – PETCO2 gradient is becoming more popular for ruling out PE in combination with the d-dimer assay, but its use is not established in the ICU where many patients have such a gradient at baseline [Chest 120: 115, 2001] Ventilator Weaning: can act as a surrogate for PaCO2. Increased PETCO2 can signify increased work of breathing, whereas decreased PETCO2 with an increase in the PaCO2 – PETCO2 gradient respiratory weakness and shallow breathing. Controlled Hyperventilation: can act as a surrogate for PaCO2.

A recent development is the earlobe CO2 monitor, which appears to be accurate over a range from 25 to 100 mm Hg [Crit Care Med 33: 2203, 2005].