Cardiac Output Measurement

Definition

Cardiac output (CO) measurement methods fall into five classes: (1) Fick principle (direct O₂ and CO₂ rebreathing), (2) indicator dilution (thermodilution via PAC, transpulmonary thermodilution, lithium dilution, ultrasound dilution), (3) pulse contour analysis (PiCCO, PRAM, LiDCO/PulseCO, Vigileo/FloTrac, Modelflow), (4) echo-Doppler (transoesophageal, transthoracic), and (5) thoracic bioimpedance/bioreactance. The "holy grail" — a method that is accurate, precise, operator-independent, fast-responding, non-invasive, continuous, easy to use, cheap, and safe — does not exist. Every method involves trade-offs between accuracy, invasiveness, and practicality.

Key Concepts

Ideal Method Criteria and Statistical Evaluation

• Critchley-Critchley criterion: a new CO method is acceptable only if its 2 SD-precision does not exceed that of the reference method; when the reference (PAC thermodilution) has 2 SD-precision of ±20%, the Bland-Altman limits of agreement for the new method must be <√(20² + 20²) = ±28% — not the commonly used ±30% threshold (which is an oversimplification) sources/co-bjcp-2010 (high)
• Reference precision tiers (PAC thermodilution): 2 SD-precision 10% = 3 injections synchronized to ventilatory cycle; 20% = 3 random injections; 30% = single injection sources/co-bjcp-2010 (high)
• Linear regression/correlation coefficients are insufficient for method comparison; Bland-Altman plot (bias ± 2 SD) is the appropriate approach sources/co-bjcp-2010 (high)

Fick Principle

Direct Fick for Oxygen

• Formula: CO = VO₂ / (CaO₂ − CvO₂), where VO₂ = O₂ uptake (mL O₂ min⁻¹), CaO₂ and CvO₂ = arterial and mixed venous O₂ content (mL O₂ L⁻¹) sources/co-bjcp-2010 (high)
• VO₂ measured by spirometry (CO₂ absorber) or indirect calorimetry monitor; mixed venous blood (CvO₂) requires PAC for sampling sources/co-bjcp-2010 (high)
• Most accurate method — laboratory reference standard to which all other methods are compared; not practical for routine clinical monitoring sources/co-bjcp-2010 (high)
• Accuracy caveats: (i) many variables → compound permutation of errors; (ii) FiO₂ >60% decreases accuracy; (iii) haemodynamic stability required; (iv) labour-intensive
• Summary: invasiveness +++, accuracy high, precision moderate; requires PAC + spirometer/mechanical ventilator

Partial CO₂ Rebreathing (NICO)

• Applies Fick principle to CO₂: CO = ΔVCO₂ / (S × ΔEtCO₂), where ΔVCO₂ = change in CO₂ production, ΔEtCO₂ = change in end-tidal CO₂ during rebreathing, S = slope of CO₂ dissociation curve sources/co-bjcp-2010 (high)
• Disposable rebreathing loop with CO₂ infrared sensor + differential pressure transducer + pulse oximeter; CvCO₂ eliminated mathematically (CO₂ diffuses 22× faster than O₂ → CvCO₂ unchanged during brief rebreathing) sources/co-bjcp-2010 (high)
• Measures effective lung perfusion, not total CO — anatomic shunts and V/Q inequality cause underestimation; clinically acceptable only in mechanically ventilated patients with minor lung abnormalities sources/co-bjcp-2010 (high)
• Comparative performance (vs PAC thermodilution): bias −4.35%, limits of agreement ±35%; 2 SD-precision 29% (ref 20%) — cannot replace triplicate thermodilution sources/co-bjcp-2010 (high)
• Summary: invasiveness +, accuracy low, precision low; requires mechanical ventilation

Indicator Dilution

Mathematical Basis — Stewart-Hamilton Equation

• CO = amount of indicator / ∫c(t)dt, where ∫c(t)dt = area under the concentration-time curve; assumes complete mixing, no indicator loss, and constant blood flow between injection and detection sites sources/co-indicator-anesthanalg-2010 (high)
• Stewart 1897: CO = C₀V₀ / (C₁ × t); Hamilton 1928 modification: CO = C₀V₀ / ∫c(t)dt — accounts for laminar flow and variable path lengths
• Thermodilution substitutes temperature change for chemical indicator concentration: CO inversely proportional to blood temperature depression × transit duration sources/co-indicator-anesthanalg-2010 (high)

PAC Revival and Modern Features

• PAC declined after outcome studies in medical ICUs showed no survival benefit; however use in cardiac ICUs has been revisited for RV failure, PAH, LV failure, and mixed shock sources/co-cocc-2022 (medium)
• RV port: allows simultaneous display of PA and RV pressure waveforms; overlap of RV end-diastolic pressure over PA end-diastolic pressure = earliest indicator of RV decompensation sources/co-cocc-2022 (medium)
• Continuous CO Monitoring (CCOM) algorithm: temperature filament adds heat to blood flow; rapid response thermistor detects downstream temperature change; displays CO, SV, SvO₂, SVR, and RVEF every 20s without cold saline volume overload sources/co-cocc-2022 (medium)
• PAC preferred in cardiac surgical ICU for differentiating mixed shock; misinterpretation of PAC data is the leading cause of complications — training essential sources/co-cocc-2022 (medium)
• PAC CO measures RV output — not a valid surrogate for LV CO in intracardiac shunts or tricuspid valve abnormalities sources/co-cocc-2022 (medium)
• Wedging contraindicated in PAH and severe MR: PA rupture risk; in severe MR, wedged waveform resembles PA waveform and is prone to misinterpretation sources/co-cocc-2022 (medium)

Intermittent Bolus PA Thermodilution (IB-PATD)

• Cold saline (10 mL, iced or room-temperature) injected into right atrium via PAC; thermistor near tip in pulmonary artery records dilution curve sources/co-indicator-anesthanalg-2010 (high)
• Clinical gold standard for >40 years; measures RV output (= systemic CO except in intracardiac shunts); PAC additionally measures PA pressures and mixed venous O₂ saturation sources/co-indicator-anesthanalg-2010 (high)
• Best precision: 3 injections equally distributed over ventilatory cycle → 2 SD-precision 3.5%; 3 random injections → 10%; single injection → 15% sources/co-bjcp-2010 (high)
• Reproducibility threshold: 22% change required for statistical significance (single measurement); 13% for triplicate sources/co-indicator-anesthanalg-2010 (high)
• PAC complications: PA rupture, pulmonary embolism, catheter-related infection; 3 major RCTs showed no mortality benefit vs no PAC — may reflect interpretation failure rather than device failure sources/co-bjcp-2010 (high)

Nine sources of measurement error:

1. Pre-injection indicator loss — injectate warming (1°C rise → ~3% CO overestimate); discard first measurement in each series sources/co-indicator-anesthanalg-2010 (high)
2. Intra-catheter conductive warming — 9–17% loss → ~20% CO overestimate; corrected by catheter-specific computation constant K₂ sources/co-indicator-anesthanalg-2010 (high)
3. Post-injection conductive rewarming — worse in low-flow states and longer transit distances (especially TCPTD) sources/co-indicator-anesthanalg-2010 (high)
4. Injectate volume error — 9 mL assumed to be 10 mL → 11% CO overestimate sources/co-indicator-anesthanalg-2010 (high)
5. Recirculation — L→R intracardiac shunt → indicator detected multiple times → CO underestimate sources/co-indicator-anesthanalg-2010 (high)
6. Malpositioning — catheter in collapsed lung branch → prolonged curve → CO underestimate sources/co-indicator-anesthanalg-2010 (high)
7. Tricuspid regurgitation — conflicting direction of error (overestimate in low-flow, underestimate in high-flow); severity-dependent; unresolved sources/co-indicator-anesthanalg-2010 (high) — see contradiction below
8. Baseline temperature fluctuations — concurrent IV infusions, respiratory oscillations; requires stable baseline sources/co-indicator-anesthanalg-2010 (high)
9. Respiratory cycle variation — SV varies up to 50% across respiratory cycle; 3 synchronized injections standard but may be insufficient sources/co-indicator-anesthanalg-2010 (high)

Continuous PA Thermodilution (CPATD)

• Electric filament on PAC heats blood intermittently in SVC; thermistor at tip detects downstream response; stochastic cross-correlation reconstructs dilution curve sources/co-indicator-anesthanalg-2010 (high)
• Systems: Vigilance II (Edwards; pseudorandom 1–4s pulses) and Q2plus (Hospira; 20s pulses every 40s)
• Main limitation — response delay: 50% response at ~9 min; 80% response at ~12 min → unsuitable for detecting rapid haemodynamic changes; displayed value = averaged previous 1–6 min (up to 12 min in extreme conditions) sources/co-indicator-anesthanalg-2010 (high) sources/co-bjcp-2010 (high)
• In hypothermia (post-CPB, liver transplantation): IB-PATD more reliable until temperature normalises
• CCOM (2022): newer fast CCOM algorithm claims CO/SV/SvO₂/SVR/RVEF display every 20s via temperature filament — see contradiction below re: response lag sources/co-cocc-2022 (medium)

Transcardiopulmonary Thermodilution (TCPTD) — PiCCO

• Cold indicator injected centrally; femoral (or axillary/brachial) arterial thermistor records dilution curve after traversing entire cardiopulmonary circuit; no PAC required sources/co-indicator-anesthanalg-2010 (high)
• Correlation with IB-PATD: r >0.9, bias <10%; slight LV vs RV CO difference (cold-induced sinus slowing affects RV more) sources/co-indicator-anesthanalg-2010 (high)
• Unique additional variables: GEDV = ITTV − PTV (superior to PAWP/CVP as preload estimate); EVLW = ITTV − GEDV − intrathoracic blood volume (EVLW-guided management: shorter extubation, decreased ICU stay) sources/co-indicator-anesthanalg-2010 (high)
• Also provides pulse contour CO (see below); combined system = PiCCO device
• Requires femoral arterial catheter; cannot measure PA pressures or mixed SvO₂

Transpulmonary Ultrasound Dilution — COstatus (Paediatric/Shunt Detection)

• Isotonic saline at body temperature injected into venous port of extracorporeal AV loop (existing arterial + central venous lines); saline/blood ultrasound velocity difference (1,530 vs 1,560–1,585 m/s) produces measurable transient signal sources/co-indicator-anesthanalg-2010 (high)
• CO by Stewart-Hamilton; L→R shunts cause delayed descent of arterial dilution curve (recirculation) → Qp/Qs estimated from curve asymmetry sources/co-indicator-anesthanalg-2010 (high)
• Shunt detection (n=44, mean age 12 months, ASD/VSD): sensitivity 95.7%, specificity 97.6%, AUC 0.97; significantly underestimates Qp/Qs magnitude in moderate/small shunts (sources/shunt-nature-sr-2020 — medium)

Lithium Dilution (LiDCO)

• 1–2 mL isotonic LiCl (150–300 mmol) via venous bolus; ion-selective electrode in peripheral arterial line; CO = (Li dose × 60) / [(1 − PCV) × ∫Δc_li dt] — corrected for PCV because lithium distributes only in plasma sources/co-indicator-anesthanalg-2010 (high)
• Peripheral venous injection feasible — only requires arterial + peripheral venous access; 3 measurements needed for precision sources/co-bjcp-2010 (high)
• LiDCO-Plus combines lithium dilution calibration with continuous pulse contour CO
• Contraindications: active lithium therapy, high-dose neuromuscular blockers (electrode drift), weight <40 kg, first trimester pregnancy

Pulse Contour Analysis

General Principles

• CO estimated from arterial pressure waveform using a mathematical model — not a mass balance; deviations from model assumptions (rapid SVR change, valve disease, arrhythmias, poor signal quality) directly cause error sources/co-bjcp-2010 (high)
• Origin: Otto Frank's Windkessel model (1899) — aorta as compliant reservoir + peripheral resistance; peripheral arterial pressure substitutes for aortic pressure (backward filtering required)
• Calibrated systems (PiCCO, LiDCO): patient-specific calibration factor from independent CO; recalibration needed with SVR changes
• Uncalibrated systems (PRAM, Vigileo/FloTrac): rely on demographic data or pressure morphology alone; higher susceptibility to error in haemodynamic instability
• Key advantage: beat-to-beat CO, continuous, requires only arterial catheter ± central venous catheter

PiCCO (Pulsion Medical Systems)

• Modified Wesseling cZ algorithm; COpi = K × HR × ∫[P(t)/SVR + C(P) × dP/dt]dt; K = transpulmonary TD calibration factor sources/co-bjcp-2010 (high)
• Recalibration required at ≤1h intervals and after major SVR changes; radial or femoral artery usable
• Comparative performance (Table 1): bias 0.73%, limits ±32%, 2 SD-precision 30% (ref 20%) — replaces single but not triplicate thermodilution sources/co-bjcp-2010 (high)

PRAM (Pressure Recording Analytical Method, Vytech Health)

• SV = A[(P/t) × K(t)]; characteristic impedance Z derived from pressure curve morphology without external calibration; CO = mean of 12 beats sources/co-bjcp-2010 (high)
• Vulnerable to signal quality and valve disease (aortic stenosis, regurgitation alters waveform morphology); sparse validation (3 studies as of 2010; excluded from Table 1)

LiDCO/PulseCO (LiDCO, London)

• Remington-Noback non-linear arterial pressure-volume relationship; nominal SV converted to actual SV by calibration factor (from thermodilution or lithium dilution) sources/co-bjcp-2010 (high)
• Recalibration every 8h or with major haemodynamic changes; no PAC needed
• Comparative performance (Table 1): bias 0.91%, limits ±24%, 2 SD-precision 22% (ref 20%) — can replace averaged triplicate thermodilution sources/co-bjcp-2010 (high)

Vigileo/FloTrac (Edwards Lifesciences)

• No external calibration; SV = σAP × Khi; σAP = SD of arterial pressure over 20s; Khi = multivariate polynomial (HR, σAP, MAP, Langewouters compliance, BSA, waveform skewness/kurtosis) updated on rolling 60s average sources/co-bjcp-2010 (high)
• Worst performer in Table 1: bias 4.55%, limits ±41%, 2 SD-precision 40% (ref 20%) — only acceptable at single-thermodilution reference precision; note: reflects software version ≥1.07 (pre-2009); algorithm substantially revised since sources/co-bjcp-2010 (high)

Modelflow (FMS, Amsterdam)

• 3-element Windkessel: characteristic impedance + Windkessel compliance + peripheral resistance; aortic compliance is pressure- and demographics-dependent (Langewouters equation) sources/co-bjcp-2010 (high)
• Can be used calibrated (independent CO) or uncalibrated (demographics only)
• Best pulse contour performer calibrated (Table 1): bias 0.00%, limits ±17%, 2 SD-precision 16% (ref 20%) — can replace averaged triplicate thermodilution; uncalibrated: limits ±31% sources/co-bjcp-2010 (high)

Comparative Performance vs PAC Thermodilution (Table 1, Geerts 2010)

No method replaces 3-synchronized-injection thermodilution (2 SD-precision ≤10%). Impedance excluded (pre-existing meta-analysis: insufficient agreement). Ultrasound excluded (insufficient comparative data).

Echo-Doppler Ultrasound

• Transoesophageal Doppler (TOD): probe at 35–45 cm in oesophagus; V = (Fd × c) / (2 × Fo × cosθ); aortic CSA from nomogram (age, weight, height); minimal bias but limited agreement (meta-analysis Dark & Singer 2004); trend monitoring reliable if probe position unchanged; poorly tolerated in awake patients; operator-dependent; contraindicated with oesophageal disorders sources/co-bjcp-2010 (high)
• EDM: NICE guidelines support use in complex/high-risk surgical procedures to reduce complications, hospital stay, and central line use; limitations include 10–12% operator-dependent variability, alignment error >20° → poor accuracy sources/co-cocc-2022 (medium)
• Transthoracic Doppler (TTD): non-invasive; jugular notch probe; larger inter/intra-observer variability than TOD; measures aortic or pulmonary valve outflow; portable sources/co-bjcp-2010 (high)
• Summary: TOD — invasiveness +, accuracy high, precision low; TTE — non-invasive, accuracy moderate, precision low

Critical Care Echocardiography (CCE)

• CCE is both a CO monitor and a differential diagnostic tool — uniquely valuable in ICU hemodynamic assessment sources/co-cocc-2022 (medium)
• TTE most commonly used; TEE preferred in intubated/ventilated patients sources/co-cocc-2022 (medium)
• TTE vs PAC accuracy: percentage error 25% (within the ≤30% acceptable threshold) — accurate and precise for CO estimation in critically ill mechanically ventilated patients sources/co-cocc-2022 (medium)
• Severely ill pregnant women: TTE CO shows excellent agreement with PAC sources/co-cocc-2022 (medium)
• Focused Doppler/VTI training added to basic TTE training allows noncardiologist ICU physicians to achieve reproducible, accurate CO assessment in most mechanically ventilated patients sources/co-cocc-2022 (medium)
• CCE is complementary to PAC (not exclusive) in septic shock requiring advanced hemodynamic monitoring sources/co-cocc-2022 (medium)
• Proposed future role: given its noninvasive nature and acceptable agreement with PAC, TTE is proposed as a future reference method for validating other CO techniques sources/co-cocc-2022 (medium)
• Barriers: most countries lack formal CCE training programs and clearly defined competencies; ESICM published core critical care ultrasound competency recommendations; ASE offers CCE certification sources/co-cocc-2022 (medium)

Thoracic Electrical Bioimpedance / Bioreactance

• Alternating current applied across thorax; bioimpedance changes converted to CO by mathematical formula; inaccurate — motion artifacts, abnormal thoracic anatomy, valve disease, arrhythmias; meta-analysis (Raaijmakers): 3 decades of validation insufficient for clinical acceptance sources/co-bjcp-2010 (high)
• Meta-analysis (Sanders 2020): modest agreement and inadequate percentage error for bioimpedance — not interchangeable with bolus thermodilution; electrical interference and increased lung water render devices ineffective in ICU sources/co-cocc-2022 (medium)
• Bioreactance (Cheetah Medical): phase-shift based; less susceptible to thoracic fluid artifacts vs bioimpedance; awaits larger validation sources/co-bjcp-2010 (high)
• NICOM study (cardiogenic shock): bioreactance showed poor correlation with both Fick and PAC thermodilution; likely due to thoracic fluid overload and low-flow state in cardiogenic shock sources/co-cocc-2022 (medium)

Pulse Wave Transit Time (esCCO)

• esCCO (Nihon Kohden): noninvasive continuous CO using R-wave to pulse oximeter waveform rise-point interval; also incorporates ECG and arterial pressure sources/co-cocc-2022 (medium)
• Poor accuracy and precision vs invasive TPT CO monitoring in cardiac surgery and liver transplant patients sources/co-cocc-2022 (medium)
• Cannot track CO changes induced by preload increase or vasomotor tone variations in ICU patients sources/co-cocc-2022 (medium)
• Not suitable for clinical decision-making in settings where CO monitoring matters

Transthoracic Doppler (USCOM)

• USCOM (continuous-wave Doppler): suprasternal notch (trans-aortic, left heart CO) or left sternal edge (trans-pulmonary, right heart CO); SV = VTI × valve CSA (from height-indexed nomogram) sources/co-cocc-2022 (medium)
• Meta-analysis of 10 studies: pooled weighted percentage error 42.7% vs bolus thermodilution (threshold for acceptable precision: ≤30%) — does not achieve acceptable agreement sources/co-cocc-2022 (medium)
• Operator-dependent in ICU; observational data show strong USCOM-echocardiography SV correlation, suggesting CO estimates could be practitioner-dependent

Contradictions / Open Questions

• Tricuspid regurgitation and thermodilution accuracy: sources/co-indicator-anesthanalg-2010 reports significant TR makes IB-PATD generally unreliable with conflicting error direction (overestimate low-flow, underestimate high-flow); sources/rhc-hf-ehj-2025 states thermodilution "remains accurate even in significant TR — contrary to common belief." Contradiction unresolved; likely reflects TR severity studied and publication year gap.
• CCOM response latency vs stated 20s update interval: sources/co-indicator-anesthanalg-2010 (high) reports continuous PAC thermodilution has 50% response at ~9 min and 80% response at ~12 min — unsuitable for detecting rapid changes. sources/co-cocc-2022 (medium) states the modern CCOM algorithm displays CO/SV/SvO₂/SVR/RVEF "every 20s." This likely reflects a technology generation gap (older stochastic cross-correlation vs newer fast filament algorithm), but the 2022 source does not provide validation data for the 20s claim; the underlying response delay may persist.
• CO₂ rebreathing — effective lung perfusion vs total CO: underestimation magnitude from shunts is unpredictable without concurrent arterial shunt measurement; limits utility in the most common critical care patients (pulmonary disease)
• Direct Fick vs thermodilution — no large prospective comparison: direct Fick accepted as gold standard by convention and physiological logic, not a prospective RCT
• Optimal injection number for thermodilution: 3 injections at same respiratory phase is standard; averaging over full ventilatory cycle is theoretically superior but optimal asynchronous injection number is unknown sources/co-indicator-anesthanalg-2010 (high)
• EVLW by lithium dilution: promising animal data (Maddison 2008) not confirmed clinically — EVLW by LiDCO should not be used clinically sources/co-indicator-anesthanalg-2010 (high)
• Uncalibrated pulse contour in haemodynamic instability: all models assume stable pressure-flow relationship; threshold of instability that invalidates each system is undefined sources/co-bjcp-2010 (high)
• Pulse contour software version confound: major algorithm updates (especially Vigileo) make cross-version comparisons unreliable; ±41% for Vigileo reflects pre-2009 software and may not represent current performance sources/co-bjcp-2010 (high)
• PAC survival benefit: multiple RCTs show no mortality benefit in medical ICUs; sources/co-cocc-2022 (medium) argues this reflects misinterpretation of data rather than device failure, and PAC is regaining acceptance in cardiac ICUs. No RCT has specifically evaluated PAC in cardiac ICU populations.

Connections

• Related to concepts/Right-Heart-Catheterization — thermodilution as standard CO method; pitfalls in PAWP/CO measurement; TR accuracy contradiction
• Related to concepts/Invasive-Hemodynamic-Monitoring-CS — CO monitoring in cardiogenic shock; CPO = (MAP × CO)/451 requires accurate CO
• Related to concepts/Pulmonary-Artery-Pulsatility-Index — PAPi derived from PAC data; same catheter used for IB-PATD
• Related to concepts/Intracardiac-Shunts — COstatus exploits L→R shunt recirculation for Qp/Qs estimation; recirculation is also a thermodilution error source
• Related to entities/Heart-Failure — hemodynamic monitoring; EVLW in acute/critical HF
• Related to concepts/ECPELLA — CO monitoring in VA-ECMO + Impella; pulse contour unreliable on ECMO circuit

Sources

• sources/co-indicator-anesthanalg-2010
• sources/shunt-nature-sr-2020
• sources/co-bjcp-2010
• sources/co-cocc-2022