Capnography

In depth

The one-page summary above is what the PDF prints. These sections go further.

CO₂ production, transport and elimination

A resting adult produces about 200 mL of CO₂ a minute (VCO₂) while consuming about 250 mL of O₂, a respiratory quotient near 0.8. The quotient is 1.0 when burning carbohydrate and 0.7 for fat, so overfeeding with carbohydrate raises VCO₂ and the ventilation needed to clear it. Fever raises metabolic rate by roughly 10–13% per °C; shivering, seizures, sepsis, burns and malignant hyperthermia raise it further, while hypothermia, sedation and paralysis lower it.

CO₂ is about 20 times more soluble than O₂. Of the CO₂ carried from tissue to lung, roughly 70% travels as bicarbonate (carbonic anhydrase in red cells, with the chloride shift), 20–25% as carbamino compounds on hemoglobin and 5–10% dissolved. Deoxygenated hemoglobin binds more CO₂ (the Haldane effect), so oxygenation in the lung helps unload it. Mixed venous PCO₂ is about 46 mm Hg against an arterial 40.

Elimination follows the alveolar ventilation equation: PaCO₂ = 0.863 × VCO₂ ÷ alveolar ventilation, where alveolar ventilation = (tidal volume − dead space) × rate. Halve alveolar ventilation and PaCO₂ doubles. The body holds large CO₂ stores, so after sudden apnea PaCO₂ is commonly quoted as rising about 10–15 mm Hg in the first minute and 3–5 mm Hg per minute after that, while SpO₂ on supplemental oxygen can stay normal for minutes.

How capnographs measure

Infrared capnographs pass light at about 4.26 µm, the CO₂ absorption peak, through the gas; absorption is proportional to the CO₂ partial pressure. Capnometry gives only the number; capnography adds the waveform, which is what shows obstruction, rebreathing and artifact. Values appear in mm Hg, kPa (1 kPa ≈ 7.5 mm Hg) or percent (5% ≈ 38 mm Hg at sea level). Monitors show a fast sweep for shape and a slow trend for minutes to hours.

Colorimetric detectors use pH-sensitive paper that changes color with CO₂. They confirm that CO₂ is present but give no waveform and no number, they cannot monitor continuously, and they lose accuracy when wet. They remain useful where waveform capnography is not available and, in some protocols, to detect a small-bore feeding tube entering the airway during insertion, which still needs radiographic confirmation.

The normal capnogram: phases, angles and slope

Phase I is the start of exhalation: gas from the airways and equipment, which held no gas exchange, so CO₂ stays at zero. Phase II is the steep rise as alveolar gas mixes with dead-space gas. Phase III is the alveolar plateau, which rises slightly because alveoli with low ventilation-to-perfusion ratios and higher CO₂ empty last. EtCO₂ is the value at the end of phase III. Inspiration (phase 0, or IV in some texts) brings fresh gas and a sharp fall to zero.

Shape tells you about the airway; height tells you about ventilation, perfusion and production. A tall wide waveform is slow breathing with CO₂ retention. Short narrow waves are fast or shallow breaths. A sloping upstroke without a plateau means alveoli are emptying unevenly through narrowed airways, so the number at the end of that slope underestimates alveolar CO₂.

PaCO₂–EtCO₂ gradient, dead space and volumetric capnography

Physiologic dead space is anatomic dead space (about 2 mL/kg, roughly 150 mL in an adult) plus alveolar dead space, alveoli that are ventilated but not perfused. Gas from unperfused alveoli contains almost no CO₂ and dilutes the gas from perfused alveoli, so EtCO₂ falls below PaCO₂. The normal gap is about 2–5 mm Hg; it widens with low cardiac output, hypovolemia, PE, emphysema, high PEEP that overdistends alveoli, and age. It can be zero or negative in pregnancy and with large tidal volumes at slow rates, where the late-emptying alveoli carry CO₂ above the arterial value.

The Bohr equation gives the dead-space fraction: VD/VT = (PACO₂ − PĒCO₂) ÷ PACO₂, where PĒCO₂ is mixed expired CO₂ collected over a minute or measured by volumetric capnography. Enghoff substituted PaCO₂ for the alveolar value, VD/VT = (PaCO₂ − PĒCO₂) ÷ PaCO₂, which is what bedside calculation uses. Because PaCO₂ also carries the effect of shunt and V/Q mismatch, Enghoff dead space is larger than true Bohr dead space. Normal VD/VT is about 0.3 breathing spontaneously; equipment adds to it on the ventilator.

In early ARDS, Nuckton 2002 (NEJM, 179 patients) found a mean dead-space fraction of 0.58, and each 0.05 increase raised the odds of death by about 45%, independent of oxygenation. A simpler bedside surrogate is the alveolar dead-space fraction, (PaCO₂ − EtCO₂) ÷ PaCO₂. Some PE studies combined a normal value with a negative D-dimer to exclude PE, but no guideline uses it on its own.

Volumetric capnography plots CO₂ against exhaled volume rather than time, using a mainstream sensor with a flow sensor. Phase I volume is anatomic plus equipment dead space (Fowler's method measures it at the midpoint of phase II). The area under the curve is the CO₂ eliminated per breath; summed over a minute it gives VCO₂. From these the monitor reports VD/VT, alveolar tidal volume and VCO₂ breath by breath.

• PEEP titration: dead space is lowest near the PEEP with best compliance; rising VD/VT as PEEP climbs suggests overdistension.
• Weaning: VCO₂ that climbs during a spontaneous breathing trial means rising work of breathing.
• Energy expenditure: VCO₂ (mL/min) × 8.19 estimates kcal/day when indirect calorimetry is unavailable (Stapel 2015).
• Perfusion: a sudden fall in VCO₂ at constant ventilation means less pulmonary blood flow.

Abnormal waveforms, pattern by pattern

Separate a fall to zero from a fall to a low value. A drop to zero in one breath is mechanical (no gas reaching the sensor) until proven otherwise. A fall over several breaths to a low but present value, with an unchanged ventilator, is perfusion. A sustained rise despite more minute ventilation is production, and in an anesthetized patient with rising temperature, tachycardia and rigidity that is malignant hyperthermia until proven otherwise.

Tube confirmation, cardiac arrest and prognosis

Tracheal gas carries CO₂ breath after breath as long as there is any pulmonary blood flow. Esophageal gas can contain CO₂ after bag-mask ventilation or carbonated drinks, but it washes out over a few breaths, which is why sustained CO₂ over at least 7 breaths is the test (PUMA 2022). Capnography confirms that the tube ventilates perfused lung; it does not detect a mainstem intubation, and a tube whose tip sits just above the cords can still show a waveform. Depth, bilateral chest rise and a chest film still matter.

During CPR, cardiac output is roughly a quarter to a third of normal and EtCO₂ follows it closely at a fixed ventilation rate. Overventilation lowers EtCO₂ and a slow rate raises it, so the number only reflects compressions when ventilation is steady at 1 breath every 6 seconds (AHA 2025). Sodium bicarbonate releases CO₂ (50 mEq can generate up to about 1 L), giving a transient rise that is not ROSC. Epinephrine can lower EtCO₂ for a few minutes despite raising coronary perfusion pressure.

Levine 1997 (NEJM, 150 out-of-hospital arrests with PEA) found that no patient with EtCO₂ of 10 mm Hg or less after 20 minutes of ALS survived to admission, which is the origin of the termination factor in the AHA guideline. Later cohorts have reported survivors below that value, and the number is confounded by ventilation rate, drugs, airway position and the cause of arrest. Asphyxial arrests start with high EtCO₂ as stored CO₂ washes out, so early values say little about prognosis.

Sedation, opioids, nasal capnography and transport

Pulse oximetry measures oxygenation; capnography measures ventilation. With supplemental oxygen, SpO₂ can stay above 95% for minutes while PaCO₂ climbs, so desaturation is a late sign of respiratory depression. The Joint Commission (Sentinel Event Alert 49, 2012) called for assessing both oxygenation and ventilation in patients on opioids, and capnography is the practical way to watch ventilation on the ward and during procedural sedation.

Nasal and oral-nasal sampling cannulas draw exhaled gas from the nares and mouth. Room air and oxygen flow dilute the sample, so the number runs low and is best used as a trend with the waveform and respiratory rate. Capnography gives a more reliable respiratory rate than chest impedance and detects apnea within seconds; set the no-breath alarm to the local policy, usually 20–30 seconds. Mouth breathers need an oral scoop, and high-flow oxygen can wash out the sample entirely.

• Early opioid effect: rate falls, waveform widens and EtCO₂ rises before SpO₂ changes.
• Sedative-hypnotics: shallow breaths, small waveform and a normal or low EtCO₂ while PaCO₂ rises.
• Partial obstruction: smaller, irregular waves; snoring; chest moving. Reposition, jaw thrust or airway adjunct.
• Apnea: no waveform, no effort. Stimulate, ventilate, consider naloxone or reversal.

In transport, disconnection and dislodgement are most likely during moves between bed, stretcher and aircraft, so keep the waveform visible and recheck after every move. Cold and humidity changes cause condensation in the line or on the sensor window. Monitors that report percent CO₂ depend on barometric pressure; most transport monitors report mm Hg with automatic compensation, but confirm this for flight. For head injury, target normocapnia and confirm with an ABG, because the gap widens with hypovolemia and chest trauma (Brain Trauma Foundation 2016 advises against prophylactic hyperventilation to PaCO₂ of 25 or less).

Ventilated patients, PE, sepsis and perfusion

On the ventilator, continuous EtCO₂ gives early warning of disconnection, tube migration, obstruction, rising dead space and changes in metabolism, but it replaces an ABG only after the gap is known and stable. A widening gap at constant settings points to a fall in cardiac output, a PE, or overdistension from PEEP or prone positioning. EtCO₂ that rises during a spontaneous breathing trial, with a faster rate, suggests the patient is tiring.

In PE, clot creates alveolar dead space: EtCO₂ falls and the PaCO₂–EtCO₂ gap widens while minute ventilation is unchanged or higher. Massive PE can cause an abrupt fall that mimics a disconnection but leaves a waveform. EtCO₂ rising after thrombolysis suggests reperfusion. Capnography cannot diagnose PE on its own; it prompts imaging.

In sepsis and other metabolic acidosis, compensatory hyperventilation lowers PaCO₂ and poor perfusion adds dead space, so EtCO₂ falls as lactate rises. Hunter 2013 (Am J Emerg Med) found EtCO₂ inversely related to lactate and mortality in suspected sepsis, and some EMS sepsis screens use 25 mm Hg or less as a trigger. EtCO₂ does not replace a lactate level. The same mechanism explains low EtCO₂ in DKA.

Under constant ventilation, a rise in EtCO₂ during passive leg raising reflects a rise in cardiac output. Monnet 2013 (Intensive Care Med) found a rise of 5% or more predicted fluid responsiveness with high specificity but modest sensitivity. Spontaneous breathing and changing ventilator settings invalidate the test.

Troubleshooting

• Keep the sampling port above the tube so condensate drains away from it, and follow the maker's guidance on where the port sits relative to the HME filter.
• Use the same cannula type for trending; switching from nasal to oral-nasal or from sidestream to mainstream changes the baseline.
• High oxygen or nitrous oxide concentrations can bias infrared readings; most current monitors compensate automatically, but some ask for the gas mix.
• Small infants on sidestream monitors can show falsely low values because the sample flow approaches their expiratory flow.

Exam pearls

• EtCO₂ depends on production, pulmonary blood flow and alveolar ventilation; at constant ventilation it tracks cardiac output.
• Normal EtCO₂ 35–45 mm Hg with a PaCO₂–EtCO₂ gap of about 2–5; a wider gap means more dead space.
• Shark fin with a widened alpha angle is bronchospasm; baseline above zero is rebreathing.
• Sudden zero is a disconnected or dislodged tube or apnea; an exponential fall over a minute or two is lost perfusion.
• In CPR, under 10 mm Hg means fix compressions; an abrupt sustained rise suggests ROSC; bicarbonate gives a transient rise.
• Capnography detects hypoventilation before SpO₂ falls, especially on supplemental oxygen.
• Waveform capnography confirms the tube is in the airway but not its depth; it misses mainstem intubation.
• Curare cleft is a notch in the plateau from returning diaphragm effort.
• Gradual rise despite more ventilation in the OR suggests malignant hyperthermia.
• In head injury, confirm EtCO₂ against an ABG before using it to guide ventilation.