RESP (Respiratory ECMO Survival Prediction) Score
The Respiratory ECMO Survival Prediction (RESP) Score is a validated, pre-ECMO prognostic tool designed to estimate the probability of in-hospital survival for adult patients with severe acute respiratory failure who require veno-venous extracorporeal membrane oxygenation (VV-ECMO). Derived from the Extracorporeal Life Support Organization (ELSO) international registry, the RESP Score integrates twelve clinical variables collected at the time of ECMO initiation into a single numerical score that stratifies patients into five risk classes with corresponding predicted survival probabilities.
Introduced in 2014 through research by Schmidt and colleagues published in the American Journal of Respiratory and Critical Care Medicine, the RESP Score was developed in direct response to the absence of any validated, objective tool to guide clinical decision-making and patient selection for respiratory ECMO. By providing a reproducible, quantitative survival estimate before ECMO cannulation, the score supports clinicians in patient selection, family counseling, institutional benchmarking, and resource allocation for one of the most resource-intensive and high-stakes interventions in critical care medicine.
Extracorporeal Membrane Oxygenation in Severe Respiratory Failure
Veno-venous ECMO is an advanced cardiorespiratory support modality that provides temporary external gas exchange for patients with life-threatening respiratory failure refractory to conventional mechanical ventilation strategies. Deoxygenated blood is drained from the central venous circulation, circulated through an external membrane oxygenator that adds oxygen and removes carbon dioxide, and returned to the venous circulation, thereby bypassing the dysfunctional lungs entirely while they are allowed to rest and recover.
Indications for VV-ECMO
VV-ECMO is considered in adult patients with severe acute respiratory failure, most commonly caused by:
- Severe acute respiratory distress syndrome (ARDS), defined by PaO2/FiO2 ratio below 80 mmHg despite optimized conventional ventilation
- Viral pneumonia (influenza, SARS-CoV-2, and other viral pneumonitides)
- Severe bacterial pneumonia with lobar or multilobar consolidation
- Status asthmaticus refractory to maximal bronchodilator therapy
- Aspiration pneumonitis with severe hypoxemia
- Inhalation injury and burn-associated respiratory failure
- Primary graft dysfunction after lung transplantation
- Severe air leak syndromes (bronchopleural fistula)
The physiological rationale for ECMO in these settings is to provide oxygenation and carbon dioxide removal while allowing the application of ultraprotective mechanical ventilation strategies (very low tidal volumes, low driving pressures, low respiratory rates), minimizing ventilator-induced lung injury (VILI) and allowing the injured lung parenchyma time to recover.
The Resource and Risk Context of ECMO
VV-ECMO is among the most resource-intensive interventions in critical care. It requires continuous management by a highly trained team of intensivists, perfusionists, and specialized nurses. ECMO circuits are expensive, cannulation carries procedural risk (vascular injury, hemorrhage, air embolism), and maintenance of the circuit is associated with ongoing risks including systemic anticoagulation-related bleeding, circuit thrombosis, hemolysis, and nosocomial infection. The typical duration of VV-ECMO support ranges from days to weeks, with associated costs in the range of tens of thousands of dollars per patient course.
Given these resource implications and the profound ethical weight of initiating an intervention that may prolong dying rather than facilitate recovery in patients with no meaningful chance of survival, the availability of an objective, pre-ECMO survival prediction tool has significant clinical, ethical, and health system implications. The RESP Score was developed to fill precisely this gap.
Development and Derivation of the RESP Score
The RESP Score was derived from a retrospective analysis of 2,355 adult patients (age 16 years or older) from the ELSO registry who received VV-ECMO for acute respiratory failure between 2000 and 2012, across 355 centers in 36 countries. The ELSO registry is the world's most comprehensive database of ECMO outcomes, making it an ideal source for a multinational, broadly applicable prediction model.
Statistical Methodology
The derivation cohort underwent multivariate logistic regression analysis to identify variables independently associated with in-hospital survival after controlling for confounders. Variables were selected from 52 candidate predictors available in the ELSO registry. The final model was simplified from a continuous logistic regression model into an integer point-based scoring system to maximize clinical usability at the bedside, accepting a small sacrifice in mathematical precision in exchange for practical utility.
The model was internally validated using bootstrap resampling and demonstrated good discrimination, with an area under the receiver operating characteristic curve (AUC-ROC) of 0.74 in the derivation cohort. Subsequent external validation studies have confirmed acceptable performance across diverse ECMO centers in Europe, North America, and Asia-Pacific, with AUC-ROC values ranging from 0.60 to 0.78 depending on case mix and era of practice.
The Twelve RESP Score Variables
The RESP Score is composed of twelve preoperative or pre-ECMO clinical variables, each assigned a weighted integer point value derived from its regression coefficient in the multivariate model. The variables are collected from information available at the time of ECMO initiation, making the score practical for real-time bedside use.
1. Age
Age is one of the most consistent predictors of ECMO survival, reflecting differences in physiological reserve, comorbidity burden, immune function, and capacity for lung recovery.
| Age (Years) | Points |
|---|---|
| 18 – 49 | +3 |
| 50 – 59 | +1 |
| ≥ 60 | 0 |
Younger patients carry substantially better ECMO survival outcomes, consistent with their greater physiological reserve, higher likelihood of complete lung recovery, and lower competing mortality from non-respiratory comorbidities during the prolonged ECMO course.
2. Immunocompromised Status
Immunocompromise is defined as the presence of any of the following: hematologic malignancy, solid organ malignancy, solid organ transplantation, bone marrow transplantation, HIV infection, or liver cirrhosis.
| Immunocompromised Status | Points |
|---|---|
| Yes | −2 |
| No | 0 |
Immunocompromised patients have significantly worse ECMO survival, driven by impaired capacity to clear precipitating infections, heightened susceptibility to nosocomial superinfections during the ECMO course, and the competing mortality risk of the underlying immunosuppressive condition. Anticoagulation management is also more complex in patients with thrombocytopenia from hematologic disease or marrow transplantation.
3. Duration of Mechanical Ventilation Prior to ECMO Initiation
The duration of invasive mechanical ventilation before ECMO initiation is a surrogate for the extent of pre-existing ventilator-induced lung injury and the acuity of the respiratory decompensation.
| Duration of Mechanical Ventilation Before ECMO | Points |
|---|---|
| < 48 hours | +3 |
| 48 hours to 7 days | +1 |
| > 7 days | 0 |
Early ECMO initiation (within 48 hours of mechanical ventilation) is associated with the best survival outcomes. This likely reflects that earlier ECMO allows intervention before cumulative ventilator-induced lung injury has caused irreversible structural damage (fibroproliferative ARDS), and before the multiorgan dysfunction that accompanies prolonged hypoxemia and hemodynamic compromise has become established. Patients ventilated for more than 7 days before ECMO are at significantly higher risk of ARDS fibroproliferation, barotrauma sequelae, ventilator-associated pneumonia, and established organ dysfunction that limits recovery potential.
4. Acute Respiratory Diagnosis
The underlying diagnosis causing acute respiratory failure is the most heavily weighted individual variable in the RESP Score, reflecting the profound heterogeneity in lung recovery potential across different causes of respiratory failure.
| Diagnosis | Points |
|---|---|
| Viral pneumonia | +3 |
| Bacterial pneumonia | +3 |
| Asthma | +11 |
| Trauma and burn | +3 |
| Aspiration pneumonitis | +5 |
| Other acute respiratory diagnoses | +1 |
| Non-respiratory or chronic respiratory diagnoses | 0 |
Asthma receives the highest point value (+11) by a large margin, reflecting the uniquely favorable biology of status asthmaticus-related respiratory failure in this model. In status asthmaticus, respiratory failure is caused by dynamic hyperinflation and bronchospasm rather than intrinsic lung parenchymal destruction. Once the bronchospasm resolves, either spontaneously or with pharmacotherapy facilitated by the respiratory rest provided by ECMO, the underlying lung architecture is typically normal and recovery is rapid and complete. ECMO in status asthmaticus essentially functions as a bridge to bronchospasm resolution, with historical survival rates exceeding 80–90% in specialized centers.
Aspiration pneumonitis (+5) also carries a favorable prognosis because aspiration-induced lung injury, while severe, may resolve completely with supportive care once the acute inflammatory response subsides, provided secondary bacterial pneumonia is managed effectively.
Viral and bacterial pneumonia (+3 each) and trauma/burn-associated respiratory failure (+3) represent intermediate prognosis categories, where recovery is feasible but depends substantially on the patient's immune competence, the extent of lung destruction, and the management of extrapulmonary complications.
Non-respiratory or chronic respiratory diagnoses (0 points) carry the worst prognosis, as patients in this category often have fixed, non-reversible lung disease (end-stage pulmonary fibrosis, severe emphysema, chronic pulmonary hypertension) in which ECMO cannot provide meaningful lung rest because there is no acute, reversible injury to recover from. For these patients, VV-ECMO may be considered only as a bridge to lung transplantation rather than as a bridge to recovery.
5. Central Nervous System Dysfunction
Central nervous system dysfunction is defined as the presence of any of the following: neurotrauma, stroke, encephalopathy, cerebral embolism, or intracranial hemorrhage at the time of ECMO initiation.
| CNS Dysfunction | Points |
|---|---|
| Yes | −7 |
| No | 0 |
CNS dysfunction carries the largest single negative weight in the RESP Score (−7), reflecting the critical importance of neurological integrity for ECMO survival and meaningful recovery. Patients with significant CNS injury who are also in severe respiratory failure face a dual-organ insult with compounded mortality risk. Furthermore, systemic anticoagulation required to maintain the ECMO circuit substantially elevates the risk of hemorrhagic expansion of intracranial hemorrhage or hemorrhagic conversion of ischemic stroke. The presence of CNS dysfunction also raises profound questions about the goals and expected benefits of ECMO support, as survival to hospital discharge in a persistent vegetative state or with severe neurological disability is a fundamentally different outcome than the intact survival that ECMO is intended to facilitate.
6. Acute Associated Non-Pulmonary Infection
This variable captures the presence of any concurrent infection outside the lungs at the time of ECMO initiation, including bacteremia, urinary tract infection, abdominal sepsis, or catheter-related bloodstream infection.
| Acute Non-Pulmonary Infection | Points |
|---|---|
| Yes | −3 |
| No | 0 |
Concurrent non-pulmonary infection at ECMO initiation reflects a systemic septic burden that reduces survival probability through multiple mechanisms: ongoing organ dysfunction from sepsis physiology, greater antibiotic exposure with attendant risks of Clostridioides difficile colitis and antimicrobial resistance, and the immunological stress of multifocal infection limiting lung recovery capacity. The presence of concomitant infection also increases the risk of ECMO circuit colonization and secondary nosocomial complications during the ECMO run.
7. Neuromuscular Blockade Agent Use Prior to ECMO
| Neuromuscular Blockade Prior to ECMO | Points |
|---|---|
| Yes | +1 |
| No | 0 |
The use of neuromuscular blocking agents (NMBAs) before ECMO carries a modest positive weight, likely reflecting that NMBA use is a marker of the severity of respiratory dyssynchrony and the degree of ventilator optimization that was attempted before ECMO. Patients in whom NMBAs were used were being managed with guideline-concordant lung-protective strategies, suggesting a higher level of pre-ECMO institutional expertise in ARDS management. This is a marker of process quality rather than a direct physiological predictor.
8. Nitric Oxide Use Prior to ECMO
| Inhaled Nitric Oxide Prior to ECMO | Points |
|---|---|
| Yes | −1 |
| No | 0 |
Inhaled nitric oxide (iNO) is a selective pulmonary vasodilator used as a rescue therapy for severe hypoxemia in ARDS, acting on well-ventilated lung regions to improve ventilation-perfusion matching. Its negative weight in the RESP Score reflects its role as a marker of severity: patients who required iNO before ECMO had already failed conventional lung-protective ventilation and were in the most refractory category of respiratory failure, associated with worse underlying lung injury and less potential for recovery.
9. Bicarbonate Infusion Prior to ECMO
| Bicarbonate Infusion Prior to ECMO | Points |
|---|---|
| Yes | −2 |
| No | 0 |
Bicarbonate infusion prior to ECMO is typically administered to manage severe life-threatening acidemia, most commonly hypercapnic respiratory acidosis in patients with severe ventilation-perfusion mismatch and insufficient alveolar ventilation despite maximal conventional support. Its presence as a negative predictor reflects the severity of the physiological derangement it is attempting to correct: patients requiring bicarbonate infusions before ECMO have profound acidemia indicating critical failure of gas exchange that may be associated with irreversible end-organ injury.
10. Cardiac Arrest Prior to ECMO
| Cardiac Arrest Prior to ECMO | Points |
|---|---|
| Yes | −2 |
| No | 0 |
Pre-ECMO cardiac arrest carries a negative weight, reflecting the significant anoxic insult to the brain and other organs associated with the period of circulatory arrest and resuscitation. Post-arrest hypoxic-ischemic encephalopathy is a major competing cause of death during the ECMO course and reduces the meaningful recovery potential of survivors. Additionally, cardiac arrest is often itself a marker of the most extreme physiological decompensation at ECMO initiation, representing the final common pathway of refractory hypoxemia, severe right ventricular failure from hypoxic pulmonary vasoconstriction, or dysrhythmia from severe acidemia.
11. PaCO2 ≥ 75 mmHg Prior to ECMO
| PaCO2 ≥ 75 mmHg | Points |
|---|---|
| Yes | −1 |
| No | 0 |
Severe hypercapnia (PaCO2 ≥ 75 mmHg) indicates profound failure of alveolar ventilation that cannot be corrected by conventional mechanical ventilation optimization. At this level of hypercapnia, pH is typically below 7.20 even with bicarbonate compensation, producing severe respiratory acidemia with significant adverse effects on myocardial contractility, pulmonary vasomotor tone, and cellular metabolism. This degree of ventilatory failure is a marker of the severity and reversibility of the underlying respiratory pathophysiology.
12. Peak Inspiratory Pressure ≥ 42 cm H2O Prior to ECMO
| Peak Inspiratory Pressure ≥ 42 cm H2O | Points |
|---|---|
| Yes | −1 |
| No | 0 |
Peak inspiratory pressures of 42 cm H2O or greater indicate either profoundly reduced respiratory system compliance (stiff, consolidated, or edematous lungs requiring very high pressures to achieve even modest tidal volumes), very high airflow resistance (as in status asthmaticus or severe mucus plugging), or both. In the context of ARDS, this degree of mechanical abnormality suggests extensive alveolar flooding and consolidation, associated with worse lung injury severity and less recovery potential. Critically, high peak inspiratory pressures also indicate that significant ventilator-induced lung injury is being accrued during pre-ECMO management, further justifying the urgency of ECMO initiation to enable pressure de-escalation.
Calculating the RESP Score
The RESP Score is calculated by summing the integer point values assigned to each of the twelve variables based on the patient's clinical status at the time of ECMO initiation. The theoretical range of the score extends from approximately −22 (worst possible scenario across all negative variables) to +28 (most favorable profile). In clinical practice, scores typically range from −10 to +15.
RESP Score = Sum of points for all 12 variables
Theoretical range: −22 to +28
The score can be computed in minutes at the bedside using clinical variables that are routinely available in the ICU setting at the time ECMO is being considered. No additional laboratory tests or investigations beyond those already obtained in the standard pre-ECMO workup are required.
Risk Classes and Predicted Survival
The summed RESP Score maps to one of five risk classes, each associated with a predicted in-hospital survival probability derived from the ELSO registry derivation cohort:
| Risk Class | Score Range | Predicted In-Hospital Survival | Clinical Interpretation |
|---|---|---|---|
| Class I | ≥ 6 | ~92% | Very high survival probability. Strong candidate for VV-ECMO if meeting clinical indications. Favorable patient and disease characteristics. |
| Class II | 3 – 5 | ~76% | Good survival probability. ECMO is generally appropriate when clinical indications are met. Most patients in this class benefit from ECMO support. |
| Class III | 0 – 2 | ~57% | Moderate survival probability. ECMO may be appropriate but warrants careful discussion of goals, expected course, and institutional resource capacity. Shared decision-making with patient representatives is essential. |
| Class IV | −2 to −1 | ~33% | Low survival probability. Careful, individualized consideration of ECMO appropriateness is required. Goals-of-care discussion, input from palliative care, and explicit discussion of what constitutes an acceptable outcome for this patient are important. Consider whether ECMO bridges to a realistic recovery or bridges to prolonged dying. |
| Class V | ≤ −3 | ~18% | Very low survival probability. ECMO carries a very high probability of non-survival. This risk class should trigger a mandatory, comprehensive goals-of-care conversation. Palliative or comfort-focused care may be the most appropriate and compassionate option for many patients in this class. |
These survival estimates are derived from the original ELSO registry cohort (2000–2012) and reflect population-level probabilities, not individual deterministic outcomes. A patient with a Class V score retains an approximately 1 in 5 chance of survival, while a Class I patient retains approximately 1 in 12 risk of in-hospital death. The score quantifies risk; it does not dictate outcome.
Clinical Applications of the RESP Score
Patient Selection for VV-ECMO
The most direct application of the RESP Score is informing the patient selection decision. In the setting of severe ARDS or refractory respiratory failure where ECMO is technically feasible, the question of whether to proceed is among the most consequential in critical care. ECMO commits patients to days or weeks of intensive intervention with significant procedural risks, mobilizes enormous institutional resources, and creates complex ethical obligations regarding withdrawal.
The RESP Score provides an objective survival probability estimate that complements clinical judgment. While there is no universally mandated score threshold for or against ECMO, many centers use the RESP framework to structure the decision:
- Class I and II patients (scores ≥ 3): ECMO is generally supported when clinical indications are met and contraindications are absent.
- Class III patients (scores 0–2): ECMO may be appropriate but warrants explicit institutional review, multi-disciplinary discussion, and careful informed consent with patient families or surrogates regarding the approximately 43% mortality risk.
- Class IV and V patients (scores ≤ −1): ECMO should trigger mandatory goals-of-care discussion. The score does not prohibit ECMO but creates an ethical obligation to ensure that the intervention is aligned with the patient's previously expressed or reasonably inferred values, and that the goals of the intervention (bridge to recovery, bridge to transplantation, or time for family to gather) are explicitly defined.
Family and Surrogate Counseling
Families or surrogate decision-makers of patients with severe respiratory failure facing ECMO decisions are in an acutely distressed state, making complex probabilistic information difficult to process under optimal circumstances. The RESP Score provides a concrete, quantitative survival probability that can be communicated compassionately alongside a clear explanation of what ECMO entails, the duration of expected support, and the criteria by which the team would reassess the trajectory of recovery.
Framing the RESP class in the context of what a given survival probability means practically, for example explaining that a Class III score means roughly 4 in 10 patients with this profile do not survive to hospital discharge, allows families to make more genuinely informed decisions about whether to proceed, and to establish clear expectations about the possibility of a non-survival outcome. This reduces the risk of subsequently unexpected or traumatic outcomes and supports family psychological adjustment regardless of whether the patient survives.
ECMO Program Benchmarking and Quality Assessment
Institutional ECMO programs are appropriately assessed by their outcomes, but raw survival rates without risk adjustment are fundamentally misleading. An ECMO program that accepts primarily Class I and II patients will appear to have dramatically better survival rates than a program that accepts all risk classes, including the most complex Class IV and V cases referred from tertiary centers. Without controlling for pre-ECMO severity as quantified by the RESP Score, inter-institutional outcome comparisons are not meaningful.
The RESP Score provides the basis for risk-adjusted benchmarking: comparing observed survival against the expected survival predicted by the case mix's RESP score distribution allows fair, valid comparison of ECMO program quality independent of patient selection philosophy. Programs that significantly outperform their risk-adjusted expected survival are demonstrating genuine clinical excellence; programs that underperform their risk-adjusted baseline may identify areas for quality improvement.
Clinical Research and Trial Design
In randomized controlled trials of ECMO versus conventional treatment (such as the EOLIA trial comparing VV-ECMO to conventional mechanical ventilation in severe ARDS), the RESP Score provides a validated stratification variable to ensure balanced allocation of high- and low-risk patients across treatment arms. It also enables pre-specified subgroup analyses evaluating whether the treatment effect of ECMO varies across survival probability strata, a question of direct clinical relevance: patients with very high RESP scores may derive less absolute benefit from ECMO (since they are likely to survive with conventional treatment) while patients with intermediate scores may show the greatest absolute risk reduction.
Pathophysiology of ARDS and ECMO's Therapeutic Rationale
Understanding the RESP Score in its full clinical context requires a working understanding of ARDS pathophysiology and the mechanisms by which ECMO alters its natural history.
ARDS Pathophysiology
ARDS is defined by the Berlin Definition (2012) as acute-onset bilateral pulmonary infiltrates on chest imaging with hypoxemia (PaO2/FiO2 ratio below 300 mmHg with PEEP ≥ 5 cm H2O) not fully explained by cardiac failure or fluid overload. The underlying pathophysiology involves:
- Phase 1 (Exudative phase, days 1–7): Diffuse alveolar damage with disruption of the alveolar-capillary barrier, flooding of alveoli with protein-rich edema fluid, neutrophil influx and activation, surfactant dysfunction causing alveolar collapse, and formation of hyaline membranes. Gas exchange is severely impaired by alveolar flooding, atelectasis, and intrapulmonary shunting.
- Phase 2 (Proliferative phase, days 7–21): Type II pneumocyte proliferation, fibroblast migration and collagen deposition, and resolution of alveolar edema. In patients who survive and recover, this phase marks the beginning of lung repair. In those who do not recover, it may evolve into fibroproliferative ARDS with permanent structural lung damage.
- Phase 3 (Fibrotic phase): In survivors, remodeling and near-complete functional recovery may occur over months. In non-survivors or patients with irreversible injury, progressive pulmonary fibrosis with architectural distortion and fixed restriction of lung compliance prevents meaningful gas exchange recovery.
How VV-ECMO Modifies ARDS Progression
VV-ECMO intervenes in ARDS pathophysiology through several mechanisms:
- Ultra-protective ventilation: By providing external gas exchange, ECMO allows reduction of mechanical ventilation to minimal settings (tidal volumes of 1–2 mL/kg predicted body weight, driving pressures below 10–12 cm H2O, respiratory rates of 5–10 breaths per minute), dramatically reducing the mechanical stress on injured alveoli. This “lung rest” strategy minimizes ongoing VILI and may prevent the transition from exudative to fibroproliferative ARDS.
- Correction of life-threatening hypoxemia: ECMO provides oxygenation independent of lung function, breaking the cycle of hypoxia-driven pulmonary vasoconstriction, right ventricular failure, and hemodynamic instability that is the most common immediate cause of death in severe ARDS.
- Carbon dioxide removal: ECMO is highly efficient at CO2 removal, correcting hypercapnic acidemia rapidly even in patients with near-complete alveolar collapse, allowing normalization of pH and reversal of acidemia-related organ dysfunction.
- Time for recovery: ECMO buys days to weeks of time for spontaneous lung recovery, clearance of precipitating infection with antibiotics, resolution of pulmonary edema with diuresis, and treatment of reversible causes (such as bronchospasm in status asthmaticus).
RESP Score Compared with Other ECMO Prognostic Tools
Several other prognostic tools have been developed for ECMO patients, and understanding their relationship to the RESP Score helps clinicians select the appropriate instrument for different clinical questions.
PRESERVE Score
The PRESERVE Score (Predicting Death for Severe ARDS on VV-ECMO) was developed from a European cohort of ARDS patients treated with VV-ECMO and predicts 6-month mortality rather than in-hospital mortality. It incorporates age, body mass index, immunosuppression, SOFA score, plateau pressure, and duration of mechanical ventilation. PRESERVE focuses on longer-term outcome and is complementary to RESP, which predicts in-hospital survival. The PRESERVE score may be more informative for patients who survive to hospital discharge but remain at risk of late death from ongoing organ dysfunction or underlying disease.
SAVE Score
The SAVE (Survival After Veno-Arterial ECMO) Score is specifically designed for veno-arterial ECMO (VA-ECMO), used in cardiogenic shock and cardiac arrest, rather than VV-ECMO for respiratory failure. It incorporates different variables (including intracardiac diagnosis, pre-ECMO cardiac arrest, acute myocardial infarction, and peak lactate) that are physiologically relevant to cardiogenic shock but not directly applicable to the respiratory ECMO context. The RESP Score should not be applied to VA-ECMO patients, and the SAVE Score should not be applied to VV-ECMO patients; they evaluate fundamentally different patient populations and disease processes.
ELSO Recommended Indications and Contraindications
The ELSO guidelines provide qualitative criteria for ECMO indications and contraindications, including absolute contraindications (irreversible lung disease without transplant candidacy, non-survivable neurological injury, advanced multi-organ failure with no recovery potential) and relative contraindications (prolonged high-pressure mechanical ventilation, immunocompromised state, age extremes). These guidelines are complementary to, not a replacement for, the RESP Score's quantitative risk estimation. The RESP Score operationalizes several of the qualitative risk factors that ELSO guidelines describe, translating them into a quantitative survival probability.
Severity of Illness Scores in ARDS (APACHE II, SOFA)
General critical illness severity scores such as APACHE II and SOFA are not designed specifically for ECMO patient selection and perform significantly worse than the RESP Score for predicting ECMO survival (AUC-ROC of APACHE II approximately 0.60–0.64 versus 0.74 for RESP in the original validation). This is because APACHE II and SOFA capture general physiological derangement but do not incorporate ECMO-specific predictors such as diagnosis type, duration of mechanical ventilation before ECMO, or ECMO-relevant interventions (nitric oxide, bicarbonate, neuromuscular blockade).
Special Clinical Contexts and Nuanced Interpretation
COVID-19-Associated ARDS
The COVID-19 pandemic dramatically expanded the use of VV-ECMO globally and generated substantial data on ECMO outcomes in a specific viral ARDS phenotype. COVID-19 ARDS is classified as “viral pneumonia” in the RESP Score diagnosis category (+3 points), but several characteristics of severe COVID-19 modify its prognostic landscape relative to other viral pneumonias:
- COVID-19 ARDS frequently features a prolonged, biphasic course with an initial inflammatory phase followed by a fibroproliferative phase, extending the required ECMO duration beyond that typical of influenza ARDS.
- Obesity, a highly prevalent comorbidity in severe COVID-19, increases cannulation complexity, circuit flow requirements, and prone positioning challenges during ECMO.
- Hypercoagulability associated with severe COVID-19 increases the risk of circuit thrombosis, pulmonary embolism, and systemic thromboembolic events during the ECMO run.
- The RESP Score was derived from pre-COVID era data; its calibration in COVID-19 populations has been evaluated in several studies with generally acceptable but somewhat attenuated discrimination, reflecting these COVID-specific factors.
Bridge to Lung Transplantation
For patients with irreversible respiratory failure (end-stage pulmonary fibrosis, severe emphysema, primary graft dysfunction after lung transplantation) who are listed or potentially listable for lung transplantation, VV-ECMO may be appropriate as a bridge strategy rather than a bridge to native lung recovery. In this context, the RESP Score's negative weighting for non-respiratory/chronic respiratory diagnoses is appropriate for predicting bridge-to-recovery outcomes, but may not reflect the very different prognosis of bridge-to-transplantation ECMO, where survival depends on organ availability and recipient-donor matching rather than native lung recovery. The RESP Score should be interpreted cautiously in the bridge-to-transplant context.
Prone Positioning During ECMO
The combination of prone positioning (which redistributes perfusion and ventilation, improves oxygenation, and reduces VILI) and VV-ECMO is increasingly practiced at experienced ECMO centers. Prone ECMO requires careful cannula management to avoid positional displacement during turning, but has been associated with improved outcomes in some case series. The RESP Score does not account for the planned use of prone positioning during ECMO and therefore may be less accurate in institutions that routinely combine these modalities.
Awake ECMO
Awake ECMO, the maintenance of patients on VV-ECMO without invasive mechanical ventilation, supported by high-flow nasal oxygen or non-invasive ventilation with the patient awake and spontaneously breathing, is an emerging strategy at specialized centers. Awake ECMO avoids the complications of mechanical ventilation (ventilator-associated pneumonia, sedation-related delirium, muscle wasting) and allows patients to participate actively in rehabilitation. The RESP Score's variables are designed for patients who are mechanically ventilated before ECMO, and its application to awake ECMO candidates requires careful consideration.
Ethical Dimensions of ECMO Prognostication
The availability of a validated survival prediction score for ECMO does not resolve the profound ethical complexity of ECMO decision-making; rather, it sharpens the terms in which these ethical questions must be addressed.
Futility and the Obligation to Offer ECMO
A RESP Class V score (~18% predicted survival) does not constitute physiological futility in the sense of a survival probability of zero. An 18% survival probability represents a meaningful chance of life for many patients. Whether this probability justifies the resource commitment, procedural risks, and prolonged intensive care required by ECMO depends on patient values, family preferences, institutional capacity, and the broader healthcare system context. The RESP Score informs this discussion but cannot resolve it; that resolution requires the irreplaceable work of thoughtful, compassionate clinical ethics.
Transparency and Trust
Sharing RESP Score predictions explicitly with families, rather than translating them into vague prognostic language, serves the goal of genuine informed consent and maintains trust in the therapeutic relationship. Families who understand from the outset that a patient has a 1-in-5 chance of surviving are better positioned to maintain realistic expectations, engage meaningfully in goals-of-care conversations during the ECMO course, and accept a non-survival outcome without experiencing it as a failure or deception.
Withdrawal of ECMO Support
The RESP Score is a pre-ECMO tool. Once ECMO has been initiated, decisions about continuation or withdrawal should be based on the trajectory of recovery observed during the ECMO course rather than on the baseline pre-ECMO score. Patients who fail to show evidence of lung recovery (improving radiographic appearance, improving lung compliance, reducing fraction of inspired oxygen requirements) after an appropriate trial of ECMO support (typically 7–14 days in most centers) face a clinical and ethical reassessment that incorporates post-ECMO variables well beyond the pre-ECMO RESP framework.
Limitations of the RESP Score
Derivation Era and Evolving ECMO Practice
The RESP Score was derived from ELSO registry data spanning 2000 to 2012, a period during which ECMO technology, cannula design, circuit management, anticoagulation protocols, and intensive care medicine more broadly underwent substantial evolution. Modern ECMO practice features centrifugal pump technology, polymethylpentene hollow-fiber membrane oxygenators with longer circuit longevity, heparin-coated circuits, and increasingly sophisticated hemodynamic and gas exchange monitoring that was not available to the original cohort. The generalizability of historical survival probability estimates to current practice requires critical appraisal.
Institutional Volume and Experience
ECMO outcomes are highly volume-dependent. High-volume ECMO centers (defined variably as centers performing more than 20–30 ECMO runs per year) consistently demonstrate better risk-adjusted survival than low-volume centers, reflecting accumulated clinical experience, dedicated multidisciplinary teams, institutional protocols, and the compounding benefits of iterative quality improvement. The RESP Score does not incorporate institutional ECMO volume or experience as a variable, potentially generating miscalibrated survival estimates when applied at low-volume centers.
Post-ECMO Functional Outcomes
The RESP Score predicts in-hospital survival, not functional recovery, quality of life, or long-term survival. Patients who survive to hospital discharge following VV-ECMO may have prolonged physical and psychological rehabilitation needs. Post-intensive care syndrome (PICS), encompassing cognitive impairment, psychiatric sequelae (post-traumatic stress disorder, depression, anxiety), and physical deconditioning, is common among ECMO survivors and is not captured by the binary outcome of in-hospital survival. The RESP Score therefore represents only one dimension of a comprehensive ECMO outcome assessment.
ELSO Registry Data Quality
The ELSO registry relies on voluntary center participation and center-level data entry, with variable data completeness and quality across contributing institutions. Reporting biases (centers more likely to report favorable outcomes, or variation in how complications are classified and recorded) may influence the survival probability estimates derived from registry data. These limitations apply to the derivation dataset of the RESP Score and to any tool derived from voluntary clinical registries.
Does Not Capture Intraoperative or Circuit-Related Variables
Outcomes during ECMO depend significantly on factors not captured by any preoperative score: cannulation complications, circuit thrombosis events, hemorrhagic complications related to anticoagulation management, nosocomial infections acquired during the ECMO run, and the skill of the managing team in recognizing and addressing complications. These within-ECMO factors may dominate outcomes in individual patients, limiting the prognostic precision of any pre-ECMO tool.