Reading Ventilator Waveforms

Educational reference — a way to understand and name what a ventilator’s tracings are telling you. It is not a bedside directive or a substitute for a device’s operator’s manual or for clinical judgment. The reading procedure below applies most closely to idealized, model-generated waveforms; real displays are distorted by noise, circuit mechanics, control delays, condensation, and leaks, so not every landmark is identifiable at the bedside.

What the tracings are, and why reading them matters

Every modern ICU ventilator paints three scalar waveforms across its screen — pressure, flow, and volume against time. They carry a continuous stream of information about both the patient’s respiratory physiology and the moment-to-moment interaction between the patient and the machine [Mireles-Cabodevila 2022]. Yet reading them is usually taught the wrong way: as ad-hoc pattern recognition — you memorize a picture of “double triggering” or a “missed trigger” and match against it — rather than as a systematic process reasoned from first principles [Mireles-Cabodevila 2022].

The corpus’s governing analogy is the electrocardiogram [Mireles-Cabodevila 2022; Chatburn 2026 · Waveforms]. Clinicians do not learn to read an ECG by collecting a scrapbook of tracings; they learn the underlying electrophysiology, memorize the normal complex as a comparator, then read deviations from it in a fixed order. Ventilator waveforms deserve the same rigor — but, the authors point out, unlike the ECG there has never been a widely accepted, formal, systematic method for reading them [Mireles-Cabodevila 2022]. That gap has deep roots: early ventilators had no graphical display at all, the tracings are fleeting and hard to capture, and there has never been a widely accepted method for reading them the way clinicians read an ECG — exactly the gap the 2022 flagship paper and Chatburn’s 2026 review set out to close [Mireles-Cabodevila 2022; Chatburn 2026 · Waveforms].

Two barriers stand between a clinician and a tracing, and the method is built to demolish both. The first is the explosion of proprietary mode names — solved by the taxonomy. The second is a lack of the mathematical intuition needed to reason about mechanical loads — solved by the equation of motion. Chatburn is reassuring about the second: expert bedside reading needs little more than high-school mathematics plus mentored practice, a “flight school” apprenticeship rather than a graduate course [Chatburn 2026 · Waveforms].

The one insight: the screen is graphing the equation of motion

Here is the fact that turns three squiggles into a physics readout. Every ventilator plotting pressure, volume, and flow is graphing the equation of motion for the respiratory system [Mireles-Cabodevila 2022]. In its bedside working form,

Pvent + Pmus = E·V + R·V̇ + PEEPauto

the ventilator’s pressure (Pvent) plus the patient’s muscular effort (Pmus) must at every instant overcome the elastic load (E·V — the force to expand the stiff lung and chest wall) plus the resistive load (R·V̇ — the force to drive flow through the airways and tube), with trapped-gas PEEPauto acting as an extra elastic load [Mireles-Cabodevila 2022; Chatburn 2026 · Waveforms]. The equation is a first-order linear differential equation in three variables — pressure, volume, flow — and the single-compartment model reduces the whole respiratory system to just those three [Chatburn 2026 · Waveforms] [established].

The consequence you can see: at any instant the height of the pressure curve equals the scaled volume height plus the scaled flow height — Pvent = PE + PR, where PE is the elastic term and PR is the resistive term [Mireles-Cabodevila 2022] [established]. The screen is not three unrelated pictures; it is one equation drawn three ways. There is even a lead–lag order to it: for unassisted breathing the general rule is that flow leads pressure, and pressure leads volume, which is why waveforms are conventionally stacked pressure → flow → volume, with volume last because it “generally gives the least amount of information” [Chatburn 2026 · Waveforms] [established].

Because it is one equation, memorizing the idealized passive shapes (Pmus = 0) is the whole trick. Know what the tracing should look like for a given mode and a normal lung, and every clinically useful finding is a deviation from that comparator [Chatburn 2026 · Waveforms].

The systematic method: three steps, then act

flowchart TD
  S1["Step 1 — identify the mode<br/>as a taxonomy tag"] --> S2["Step 2 — read the load off the waveform<br/>OPPOSITE the control variable"]
  S2 --> VC["Volume control<br/>→ read PRESSURE"]
  S2 --> PC["Pressure control<br/>→ read FLOW and VOLUME"]
  VC --> S3["Step 3 — diagnose interaction by phase<br/>trigger → inspiration → cycle → expiration"]
  PC --> S3
  S3 --> ACT["Then intervene by goal"]

Figure 1 — the three-step reading method: identify the mode, read the load off the waveform opposite the control variable, then diagnose interaction by phase. Adapted from [Mireles-Cabodevila 2022] and [Chatburn 2026 · Waveforms].

Both papers frame reading as an ordered procedure. The 2022 paper states it as three steps and supplies a one-page tool (“Standardized Ventilator Waveform Analysis”) that walks through them and lets the entire state collapse into a single sentence — e.g., “The patient is on PC-CMVa, with a high elastic load and has early triggers” [Mireles-Cabodevila 2022]. Chatburn’s 2026 review restates the same three steps and works through the first two in mathematical detail; it explicitly covers steps 1 and 2 only, deferring the phase-by-phase step 3 to the 2022 companion [Chatburn 2026 · Waveforms].

Step 1 — Name the mode as a taxonomy tag

You cannot read a waveform until you know what the ventilator is trying to do, and the commercial mode name does not reliably tell you [Mireles-Cabodevila 2022]. So the first move is to classify the mode as a TAG (Taxonomic Attribute Grouping): control variable + breath sequence + targeting scheme [Mireles-Cabodevila 2022].

• Control variable — pressure control (PC) or volume control (VC).
• Breath sequence — CMV, IMV (subtyped), or CSV.
• Targeting scheme — set-point (s), dual (d), bio-variable (b), servo (r), adaptive (a), optimal (o), intelligent (i).

The full machinery of this step lives in The Taxonomy of Ventilator Modes; the point here is why it comes first. The classic trap is “pressure-regulated volume control” (PRVC; equivalents include VC+, volume guarantee, autoflow): the name says volume, but the mode is actually PC-CMVa — pressure control, continuous mandatory, adaptive targeting [Mireles-Cabodevila 2022] [proposed]. Get that wrong and you will look for the load in the wrong waveform, because — as Step 2 shows — which curve you read is determined entirely by the control variable. Chatburn’s 2026 review frames the same step as “recognize the mode from the waveform,” using the control-variable and targeting display signatures covered below [Chatburn 2026 · Waveforms].

Step 2 — Read the load off the waveform opposite the control variable

This is the method’s center of gravity, and it follows straight from the equation of motion. During inspiration the ventilator predetermines one variable; the other two are then no longer free — they fall out of the mechanics. So whatever the ventilator controls carries no information about the patient; look at what it does not control [Mireles-Cabodevila 2022; Chatburn 2026 · Waveforms] [established]:

• In volume control, volume and flow are preset, so any change in mechanics or any patient effort has nowhere to go but the pressure waveform. Read the patient off pressure.
• In pressure control, pressure is preset, so changes in mechanics and effort show up in the flow and volume waveforms — flow is the primary focus (because ventilators control pressure less perfectly, a little can also leak into the pressure trace).

The goal of Step 2 is to state which load dominates — resistive or elastic — or to recognize that Pmus is present, which is itself the key finding, because while the patient is actively pulling you generally cannot cleanly separate R from E [Mireles-Cabodevila 2022] [established].

Separating elastic from resistive load is where memorized reference shapes pay off [Mireles-Cabodevila 2022; Chatburn 2026 · Waveforms] [established]:

• VC, constant (square) flow. Flow steps up to its set value and holds. The resistive load appears as the immediate initial step-up in pressure (higher R → taller step); the elastic load appears as the slope of the pressure rise that follows (stiffer lung → steeper slope). Add an end-inspiratory hold and the resistive term vanishes (R·0), leaving the pure elastic plateau — so the peak-to-plateau fall (Ppeak − Pplat) reflects the resistive load and indicates R, while the slope up to the peak is proportional to elastance [Chatburn 2026 · Waveforms].
• VC, descending-ramp flow. If flow falls to zero by end-inspiration, the end-inspiratory pressure is essentially all elastic (≈ plateau), so you can read elastic vs. resistive pressures without a hold; if flow does not reach zero, that shortcut fails and you need the hold [Mireles-Cabodevila 2022].
• PC, square pressure. Pressure steps to its set value and holds, so flow and volume rise and fall exponentially, governed by the time constant τ = R·C. Peak flow = ΔP/R at the very start; high resistance lengthens τ (lower peak flow, slower return to zero); low compliance shortens τ (fast return to zero flow) [Mireles-Cabodevila 2022]. One τ is a 63% change; inspiration or expiration is ~95% complete after three time constants — a rule of thumb worth carrying [Chatburn 2026 · Waveforms] [established].

These are the reference shapes to memorize — the same idealized tracings a learner meets in SEVA-basic. Each is computed here directly from the equation of motion for one representative patient (C 50 mL/cmH₂O, R 10 cmH₂O/L/s, τ = 0.5 s), so the shapes are a property of the physics, not of any one tracing.

Figure 2 — VC-CMV (constant flow). The square flow waveform is preset, so the airway pressure carries the patient: a resistive step (R·V̇) at the onset, then an elastic ramp (V/C) to PIP, then — during the end-inspiratory pause — a fall to the purely elastic plateau (Pplat); expiration is passive and decays on τ. The peak-to-plateau fall reflects resistance; the ramp’s slope reflects elastance. Original diagram; shapes follow the equation of motion as taught in [Mireles-Cabodevila 2022] and [Chatburn 2026 · Waveforms].

Figure 3 — PC-CMV (constant pressure). Now pressure is the square wave, so flow and volume carry the patient: flow starts at ΔP/R and decelerates exponentially, volume fills exponentially toward ΔP·C — both governed by τ = R·C. Read the load off flow and volume, not pressure. Original diagram; shapes follow the equation of motion as taught in [Mireles-Cabodevila 2022] and [Chatburn 2026 · Waveforms].

Figure 4 — PC-CSV / PSV (spontaneous, flow-cycled). Inspiration has the same square-pressure / decelerating-flow shape as PC, but every breath is patient- triggered and the ventilator cycles when inspiratory flow falls to a set fraction of its peak (25% here, the dashed line) — the flow-cycle criterion that makes the breath patient-cycled. Original diagram; shapes follow the equation of motion as taught in [Mireles-Cabodevila 2022] and [Chatburn 2026 · Waveforms].

These three shapes are not static. Change the lung and the ventilator and

watch them respond in the interactive simulator — the fastest way to build the reference intuition this step depends on.

Chatburn’s 2026 review operationalizes Step 2 as an ordered walk of eight numbered waveform reference points, each reasoned from the equation of motion — Pinit, Ppeak-insp, Pplat, V̇peak-insp, V̇end-insp, V̇peak-exp, V̇end-exp, and VT — with a short verdict at each landmark for VC versus PC [Chatburn 2026 · Waveforms] [established]. Two of its guardrails are worth stating plainly: a non-zero end-expiratory flow signals gas trapping / PEEPauto, and Pmus occurring at random points in the breath can make some or all landmarks impossible to identify — the model assumes a passive patient [Chatburn 2026 · Waveforms].

One more landmark that is always readable: expiration is always pressure-controlled with set-point targeting (the target is PEEP), so in every mode you read expiratory physiology and interactions off the flow and volume curves. The expiratory tracing is the corpus’s tie-breaker — it confirms inspiratory findings, or takes over when Pmus has contaminated inspiration [Mireles-Cabodevila 2022] [established].

Step 3 — Diagnose the interaction, phase by phase

With the mode named and the load characterized, walk the four phases of a breath in order — trigger → inspiration → cycle → expiration — and classify each [Mireles-Cabodevila 2022] [proposed]. The 2022 paper holds this step in full; Chatburn’s 2026 review defers to it [Chatburn 2026 · Waveforms]. Trigger and cycle are judged for timing (synchrony) relative to the patient’s effort; inspiration and expiration are judged for work (who is doing it):

• Trigger (who starts the breath, and when): normal, late trigger (response after a clinically important delay), early trigger (a machine breath precedes the effort — the descriptive relabeling of “reverse trigger”), false trigger (a non-Pmus signal such as cardiac oscillation or a leak starts the breath), and failed trigger (a real effort fails to cross threshold — most worryingly from auto-PEEP).
• Inspiration (work): normal (passive) or work shifting, where part of the work has moved to the patient. If, in VC, airway pressure is pulled below baseline PEEP, that is severe work shifting — the patient is doing work on the ventilator — which the paper states is never clinically appropriate.
• Cycle (who ends the breath, and when): normal, early cycle (inspiration ends before the effort peaks — a common cause of multiple triggering), or late cycle (inspiration runs past the effort peak; the PAV “runaway” is the type case).
• Expiration (work): normal passive decay, or expiratory work, where an active expiratory effort deflects flow away from baseline.

Two patterns are handled apart from the phase walk because a single picture can have several causes: multiple trigger (≥2 breaths stacked without full exhalation between — the “double triggering” family, dangerous mainly for volume-overdosing in VC) and tidal-volume discrepancies (the “square root sign,” an inhaled-vs-exhaled VT mismatch from a leak, active exhalation, air trapping, or a flow-sensor fault) [Mireles-Cabodevila 2022] [proposed]. The full taxonomy of these phase categories — with the deprecated synonyms each one replaces — belongs to Patient-Ventilator Interaction and Discordance.

Then act — choose the intervention by goal

The 2022 tool adds a fourth block: decide what to do. Anchor every decision to a single primary goal — safety, comfort, or liberation (the paper insists there are only three goals of mechanical ventilation), then choose to adjust settings, change the mode, do nothing, or something else [Mireles-Cabodevila 2022] [proposed]. Crucially, not every interaction needs fixing: many are transient, harmless, and clinically irrelevant (mild work shifting), while others are dangerous if frequent (multiple trigger in VC). And when discordances are causally linked — early trigger → early cycle → work shifting — the tool reports only the first one, because fixing the root event resolves the chain [Mireles-Cabodevila 2022] [proposed].

The targeting-scheme display signatures

Step 1 asks you to name the targeting scheme, and the screen itself will often tell you which one is running [Chatburn 2026 · Waveforms] [proposed — the taxonomy is an advocated standard, not universal nomenclature]:

Scheme	What you see on the pressure trace
Set-point	The preset inspiratory pressure (PIP / Pinit) is fixed and independent of R and C — it does not budge as mechanics change.
Adaptive (e.g., PRVC)	Harder to spot, but recognizable because inspiratory pressure decreases as the patient’s effort increases — the ventilator lowers its pressure target when effort pushes delivered volume above the set VT.
Servo (e.g., PAV, NAVA)	Inspiratory pressure is proportional to the patient’s effort — the machine’s pressure rises and falls with the diaphragm.

These signatures are also why work shifting behaves differently across schemes: in VC set-point the total work is fixed and effort only redistributes it; in PC set-point the ventilator’s work-per-liter is constant so total work climbs as VT rises; in servo modes the ventilator’s work rises with the patient’s, holding the ratio controllable — the whole point of PAV/NAVA [Mireles-Cabodevila 2022] [established].

Worked reading patterns

Chatburn’s 2026 review tabulates the eight-point analysis for a set of canonical distortions, computed against a normal reference lung (R = 10 cm H₂O/L/s, C = 60 mL/cm H₂O, passive). They are the fastest way to internalize the “read the opposite waveform” rule [Chatburn 2026 · Waveforms] [established]:

• Increased elastic load (stiffer lung, C falls). In VC, Ppeak and Pplat both rise (the E·V term grows) while tidal volume is unchanged (it is preset); in PC, the plateau rises but delivered VT falls (VT ≈ ΔP·C, and C is now smaller).
• Increased resistive load (R rises). In VC, Pinit and Ppeak rise (the R·V̇ step grows) while Pplat is little changed and VT holds; in PC, peak inspiratory and expiratory flows fall and VT falls as the longer time constant slows filling.
• Increased inspiratory effort (Pmus appears). In VC, the effort distorts the pressure waveform (concave-up, toward baseline) while VT stays preset; in PC, the effort adds to Pvent so peak flow, plateau, and VT all increase — the load shows up, as always, in the waveform opposite the control variable.

The through-line: in VC the tell is always in pressure; in PC it is always in flow and volume. Once that reflex is automatic, the eight landmarks are just bookkeeping.

How this chapter connects

• Reading a tracing is an application of two foundations: the equation of motion (Step 2’s physics) and the taxonomy (Step 1’s naming). The same physics and vocabulary that describe a mode are used to read the traces that mode produces [Chatburn 2026 · Waveforms].
• Step 3 — the phase-by-phase diagnosis of trigger, inspiration, cycle, and expiration, and the discordance vocabulary it uses — is developed in Patient-Ventilator Interaction and Discordance.
• The interventions block turns a reading into a decision, anchored to the single goal of safety, comfort, or liberation [Mireles-Cabodevila 2022].

Loose ends and honest caveats

• This is a proposed standard, not settled nomenclature. There is no widely accepted formal method for reading ventilator waveforms; the three-step procedure and its descriptive vocabulary are the authors’ well-argued proposal, adopted in a growing body of work but not universal [Mireles-Cabodevila 2022; Chatburn 2026 · Waveforms].
• Idealized vs. real displays. The procedure is built on model-generated waveforms. Real screens are distorted by noise, circuit mechanics, control-system delays, condensation, and leaks, so not every reference point is identifiable in practice — and active Pmus at random times can erase the landmarks entirely [Chatburn 2026 · Waveforms].
• [open] — the supplementary solution figures. The explicit pressure/volume/flow algebra for volume control and pressure control, and the sinusoidal unassisted-breathing solutions, live in the 2026 review’s Supplementary Figures S1–S6, which are not in the frozen main-text PDF; their exact equations remain [open] (supplementary data needed) [Chatburn 2026 · Waveforms]. Likewise the specific figure images of the 2022 paper are described from captions and body text, not reproduced.

Sources

Every claim above is drawn from the frozen, human-reviewed primary literature in the machine layer (raw/literature/, one artifact per paper). The short keys used inline resolve to:

• [Mireles-Cabodevila 2022] — Mireles-Cabodevila E, Siuba MT, Chatburn RL. A Taxonomy for Patient-Ventilator Interactions and a Method to Read Ventilator Waveforms. Respir Care 2022;67(1):129–148. PMID 34470804.
• [Chatburn 2026 · Waveforms] — Chatburn RL. How to Interpret Ventilator Waveforms Using the Taxonomy for Modes of Mechanical Ventilation. Respir Care 2026;71(6):566–587. PMID 41631602.

This chapter is the human-layer synthesis of two one-concept-per-page records in the machine layer — Ventilator Waveforms and Method to Read Ventilator Waveforms — and the frozen raw/literature/ artifacts they cite (see the derived_from and sources fields).