Capnography & ETCO₂: Visualizing Ventilation

The Waveform, the Number, and What Each Phase Actually Tells You

Respiratory Monitoring · CO₂ Transport · Waveform Analysis · Airway Safety

TL;DR

In practice, ETCO₂ is best understood as a real-time window into three coupled physiologic systems: metabolism, circulation, and ventilation, all read through the imperfect lens of a single sampled gas. The number is a first approximation; the waveform is where the actual diagnosis lives.

Background:

Capnography is the continuous, breath-by-breath measurement of carbon dioxide concentration in expired gas. It is plotted as CO₂ partial pressure against time. The peak value at the end of each exhalation (end-tidal CO₂ ETCO₂) is a single number anesthesia providers use as a real time measurement for both ventilation and pulmonary perfusion.

The clinical waveform was first described by Smalhout and Kalenda in the 1970s, and capnography is now the ASA-mandated standard for confirming endotracheal intubation and monitoring ventilation under any anesthetic.

Because CO₂ is the metabolic product of every cell in the body, the gas leaving the airway at the end of exhalation has spent the longest time equilibrating with mixed alveolar gas, and therefore most closely reflects what’s happening at the alveolar–capillary interface. The waveform itself adds a second layer of information that the number alone misses: shape, slope, and baseline tell you how the patient is ventilating, not just how much.

Important Terms

Capnometry — the numeric display alone
Capnography — the full waveform plus the number, allowing pattern recognition.
ETCO₂ — the specific peak value at the end of expiration, normally 35–45 mmHg in a healthy adult.

Physiology:

To understand why ETCO₂ is useful, it helps to trace a single CO₂ molecule from where it’s made to where the sampling line picks it up. CO₂ is generated in the mitochondria as a byproduct of aerobic metabolism.

It then diffuses out of the cell into venous blood, is carried (mostly as bicarbonate) to the right side of the heart, then it is pumped through the pulmonary capillaries, diffuses across the alveolar membrane down its partial-pressure gradient, mixes with alveolar gas, and finally exits the airway during exhalation.

From this we can see that ETCO₂ is downstream of three independent physiologic processes — metabolism, circulation, and ventilation — and a change in any one of them will move the number.

In a healthy patient, ETCO₂ runs about 2–5 mmHg lower than arterial PaCO₂. This gradient is the result of physiologic dead space; alveoli that are ventilated but not perfectly perfused dilute the CO₂-rich gas from well-perfused alveoli on its way out.

The standardized normal values:

PaCO₂: 35–45 mmHg (arterial blood gas)
ETCO₂: 35–45 mmHg (end of exhalation)

The clinical takeaway: a falling ETCO₂ doesn’t automatically mean hyperventilation. It could be a drop in cardiac output (less CO₂ delivered to the lungs per minute), a sudden increase in dead space (pulmonary embolism), a circuit disconnect, or most ominously in the OR, cardiac arrest, where ETCO₂ collapses within seconds because there’s no longer a pulmonary circulation to deliver CO₂ to the alveoli at all.

The Waveform:

The capnogram is not a single endpoint but a sequence of clinically distinct phases, each indexing a different stage of the breath cycle.

Phase I — Inspiratory baseline (0 mmHg): the flat segment during inhalation and the very start of exhalation, when only CO₂-free anatomic dead space gas is leaving the airway.
Phase II — Expiratory upstroke (rapid rise to ~35 mmHg): alveolar gas arrives at the sampling line and CO₂ rises sharply. A sluggish upstroke is a hallmark of obstruction, bronchospasm, kinked tube, or partial circuit obstruction.
Phase III — Alveolar plateau (35–45 mmHg, nearly flat): pure mixed alveolar gas, ending in the ETCO₂ value at the very last point of exhalation. A steeply upsloping plateau (“shark-fin”) indicates uneven alveolar emptying, classically seen in asthma and COPD.
Phase 0 — Inspiratory downstroke (rapid fall back to 0 mmHg): fresh gas displaces alveolar gas at the sensor as the next breath begins. A slow downstroke or failure to reach zero again points to rebreathing or a leak around the cuff.

Moreover, the shape of the capnogram is not arbitrary, it is a physical record of how gas mixes and empties from the lung. Phase II is steep because alveolar gas all reaches the sensor at roughly the same time in a healthy lung. Phase III is flat because, in well-matched ventilation and perfusion, all alveoli are emptying gas of nearly the same CO₂ concentration. When that synchrony breaks down, the plateau tilts upward and the curve loses its rectangular shoulder. The waveform is, in effect, a real-time readout drawn one breath at a time.

Abnormal capnogram patterns including shark-fin, curare cleft, rebreathing, and cardiac arrest

A handful of abnormal patterns are worth recognizing on sight:

Shark-fin: (loss of plateau, sloping Phase III), bronchospasm, or other obstruction.
Curare cleft: (a notch in Phase III) occurs sometimes when a recovering paralyzed patient takes a small spontaneous breath against the ventilator
Sudden flatline: in an intubated patient means one of three things: dislodged tube, circuit disconnect, or no cardiac output. All three demand action within seconds.

In Practice:

ETCO₂ is not a fixed property of the patient — it is a moving target shaped by metabolism, circulation, and ventilation acting in parallel.

Cardiac output is the modifier with the most dramatic and time-sensitive signal. Because CO₂ has to be physically delivered to the lungs by pulmonary blood flow, a sudden drop in cardiac output produces an immediate fall in ETCO₂ . This is often before blood pressure or pulse oximetry reflect the change. During CPR, ETCO₂ becomes a real-time quality metric: a rise above 10–15 mmHg indicates effective compressions, and a sudden surge to near-normal values is one of the earliest signs of return of spontaneous circulation (ROSC). This is why ETCO₂ is now part of every ACLS algorithm.

Ventilation moves ETCO₂ in the direction you’d expect: hyperventilation drops it (CO₂ blown off faster than it’s produced), hypoventilation raises it. A rough rule is that doubling minute ventilation roughly halves ETCO₂, and vice versa, in a metabolically stable patient.

Metabolic rate shifts the production side of the equation. Fever, sepsis, shivering, and most dramatically malignant hyperthermia all push CO₂ production upward. A rapidly rising ETCO₂ in the OR with unchanged ventilator settings is in fact one of the earliest and most sensitive signs of MH, often preceding temperature changes by 15–30 minutes. Hypothermia, neuromuscular blockade, and deep anesthesia all push ETCO₂ down.

Dead space changes widen the a–ET gradient without necessarily changing ventilation. Pulmonary embolism is a textbook example: a clot blocks perfusion to a lung region, those alveoli become dead space, ETCO₂ falls (fewer well-perfused alveoli contributing), but PaCO₂ may rise — the gradient widens from a normal 2–5 mmHg to 10+ mmHg.

Other notable modifiers include airway leaks (↓ falsely, gas escaping around the cuff), bicarbonate administration (transient ↑ as bicarb is converted), and tourniquet release (transient ↑ as pooled CO₂-rich blood returns to circulation).

Limitations

ETCO₂ is a secondary measurement of arterial CO₂. The a–ET gradient assumed in the OR (2–5 mmHg) is a healthy-adult assumption that breaks down in patients with significant changes in perfusion: COPD, ARDS, pulmonary embolism, or single-lung ventilation. In situations like this, the gradient can exceed 10–15 mmHg and the number on the monitor systematically underestimates true PaCO₂. In those patients, capnography tracks trends reliably but cannot replace an arterial blood gas for absolute values.

Capnography also doesn’t tell you about oxygenation — a patient can have a perfect waveform and a normal ETCO₂ while desaturating, which is why pulse oximetry and capnography are complementary rather than redundant. And like BIS, the number is only as good as the signal: a kinked sample line, a wet sensor, or a leak around the cuff can produce a misleading waveform that looks worse than the patient.