Evidence Database

Adults on peripheral femoral VA ECMO

Review of observational data; the pooled DPC effect comes from non-randomised studies with confounding by indication. No randomised comparison of prophylactic versus reactive DPC exists.

3, 12, 17

Limb ischaemia occurs in 10-30% of peripheral femoral VA ECMO patients (historically up to 70% without preventive technique). A summarised meta-analysis found DPC reduced limb ischaemia from 25.4% to 9.7%, absolute risk reduction 15.7%, risk ratio 0.41 (p<0.01); in one series 3.4% with DPC versus 21.4% without required intervention for critical limb ischaemia. Guidance is prophylactic DPC at ECMO initiation in all patients. Technique: short 6-8 Fr armoured cannula, target flow about 100 mL/min, antegrade femoral/superficial femoral or retrograde posterior tibial. Monitoring: NIRS above 50% and preferably 60% with bilateral difference under 20%; duplex ultrasound as adjunct; clinical examination limited in sedated patients.

No distal perfusion cannula

Distal perfusion cannula and other distal limb perfusion strategies

Decide about distal limb perfusion at the time of cannulation, not after ischaemia appears, and monitor the limb on every round.

DOI 10.1177/02676591241236650

Computational model of a patient-averaged right atrium and venae cavae; two commercial bicaval dual-lumen cannulae (Avalon Elite and MC3 Crescent) scaled to 27 Fr

In silico only. Patient-averaged anatomy, single cannula size, no patient outcomes. Does not capture patient movement, volume state or RV function.

3, 5, 9

Both designs achieved recirculation fraction below 7% at correct position - essentially zero at 2 and 4 L/min, rising to 3.0-6.2% at 6 L/min - with near-identical pressure-drop curves. Rotation to plus or minus 60 degrees changed recirculation little (2.4-3.1%). Short insertion depth raised recirculation to 31.2-44.6% at all flow rates; long insertion depth caused 24.0% recirculation only at maximum flow. Caval pressures of 16.2-23.9 mmHg at low flows; high-velocity jets above 5 m/s at 4 L/min with shear stress above 413 Pa.

Cannula designs against each other, and correct versus incorrect positioning

Computational fluid dynamics simulation at 2-6 L/min, with varied rotation and insertion depth

When a dual-lumen cannula underperforms, suspect insertion depth before device failure. Confirm depth echocardiographically before considering exchange.

DOI 10.1038/s41598-023-34655-1

Adults requiring configurations beyond standard VV or VA ECMO

Case series and anecdotal reports only. Confounding by indication is total: patients reach a hybrid configuration because a simpler one failed. Reported mortality describes the population, not the intervention. Mortality figures are taken from this review rather than the primary papers.

3, 16, 78

Taxonomy of hybrid configurations with the specific mismatch each addresses. VAV for differential hypoxaemia on VA or new circulatory failure on VV, with reported mortality of roughly 50-61% across small series; VVA for combined lung and cardiac failure with inadequate single-limb drainage; V-Pa (typically a 17 F cannula into the pulmonary artery) as a percutaneous RVAD that bypasses the RV; VVVA and VVAV anecdotal. Flow splitting between arterial and venous return limbs requires a partial-occlusion clamp and dedicated flow monitoring.

None - descriptive review of case series

Hybrid configurations: VAV, VVA, V-Pa, VVVA, VVAV, and LV unloading strategies including surgical vents, transaortic catheters, atrial septostomy, IABP, Impella and Tandem-Heart

Hybrid configurations solve specific physiological mismatches and should be a considered decision, not an escalation reflex; each added limb adds bleeding, infection, thrombosis and haemolysis risk on an anticoagulated patient.

DOI 10.21037/jtd.2018.03.84

Adults on venovenous ECMO

Modelled physiological relationships, not outcome-validated targets. No clinical endpoints.

2, 6, 7, 9

Expresses arterial and pulmonary arterial saturation as the flow-weighted average of circuit blood and blood bypassing the circuit. Practical anchor: an effective ECBF/CO ratio of about 0.6 with venous saturation 75% and a fully saturating oxygenator yields roughly 90% arterial saturation in complete native lung dysfunction; the ratio must rise as venous saturation falls. Effective rather than total circuit flow must be used when recirculation is present. Sweep gas flow is the analogue of native minute ventilation and the primary determinant of CO2 removal; CO2 removal rises logarithmically with blood flow and plateaus according to membrane surface area.

Not applicable

Not applicable - physiological synthesis and modelling

Gives a quantitative basis for interpreting arterial saturation on VV ECMO as a mixture, and for recognising that a falling venous saturation lowers arterial saturation with the circuit behaving normally.

DOI 10.1177/02676591241238156

18 critically ill adults in two tertiary ICUs: 9 septic, 9 non-septic, studied during discontinuation of life support

Only 18 patients. Measured during withdrawal of life support - a low and falling metabolic state that does not represent an actively treated ICU patient. Population estimate with wide standard deviations; individual thresholds vary with temperature, sedation and disease.

2, 25, 30, 46

Critical oxygen delivery identified from the biphasic DO2-VO2 relationship: 3.8 +/- 1.5 mL/min/kg in septic versus 4.5 +/- 1.3 mL/min/kg in non-septic patients, not significantly different. Critical and maximal oxygen extraction ratios also did not differ. Values considerably lower than previously reported.

Septic versus non-septic patients

None - observational measurement of the DO2/VO2 relationship as delivery fell

Supports a critical delivery threshold that is a floor to stay above rather than a target to maximise, and undercuts the rationale for driving DO2 to supranormal levels. Do not apply the numeric threshold to an individual patient as a target.

DOI 10.1001/jama.1993.03510140084034; PMID 8411504

Neonates, children and adults receiving ECMO, 2009-2022

Voluntary reporting; contributing centres are not a random sample; no comparator, so no causal inference about effect is possible. Case mix shifts over the period confound temporal trends.

1, 30

154,568 ECMO runs across 780 centres (557 reporting in 2022). Median annual adult runs per centre rose from 4 to 15; paediatric/neonatal fell from 12 to 7. Survival to hospital discharge 68.5% neonatal respiratory, 29.5% adult ECPR. By 2022 the Registry had enrolled its 200,000th patient and 100,000th patient discharged alive.

None - descriptive registry cohort

ECMO as delivered in reporting ELSO centres

Use for calibration, case-mix context and programme benchmarking only. Never cite registry survival as evidence that ECMO caused survival.

DOI 10.1097/MAT.0000000000002128

Not applicable - methodological paper on ELSO guideline production

Methodological and organisational description; carries no clinical recommendations.

1

Describes three ELSO guideline methodologies: narrative (expert synthesis plus expert judgment), consensus (formal voting to quantify agreement), and GRADE. States that no ELSO guideline currently uses GRADE methodology. The Guideline Subcommittee expanded from four to twelve members in 2023. Roadmap proposes new guidelines on ECMO during CPR, trauma and education, plus formal 3-5 year update cycles.

Not applicable

Not applicable

Sets the weight a reader should give any ELSO document: high-quality expert synthesis rather than graded evidence appraisal. This book labels ELSO recommendations accordingly.

DOI 10.1177/02676591261425548