Overview of the RESCUE-IHCA Score

The RESCUE-IHCA (Resuscitation Using ECPR During In-Hospital Cardiac Arrest) score is a validated, six-variable clinical prediction model developed to estimate in-hospital mortality probability for adult patients receiving extracorporeal cardiopulmonary resuscitation (ECPR) during in-hospital cardiac arrest (IHCA). Published by Tonna et al. in JACC: Cardiovascular Interventions (2022), the score was derived from 1,075 patients across 219 centres in the American Heart Association Get With The Guidelines-Resuscitation (GWTG-R) registry and externally validated in 297 patients from the Extracorporeal Life Support Organization (ELSO) registry.

The score incorporates six pre- and intra-arrest variables: patient age, pre-existing renal insufficiency, time of day of the arrest, illness category (medical vs. surgical, cardiac vs. non-cardiac), presenting cardiac rhythm, and duration of cardiac arrest before ECMO initiation. These inputs generate a composite score that maps to an estimated in-hospital mortality probability ranging from approximately 22% (lowest-risk profiles) to greater than 99% (highest-risk profiles). Importantly, five of the six variables are non-modifiable patient or arrest characteristics, meaning the score can be calculated at or near the time of arrest decision-making with minimal additional data collection. Only arrest duration is potentially modifiable through rapid ECMO deployment protocols.

The RESCUE-IHCA score was designed not to replace clinical judgment or dictate patient selection for ECPR, but to provide a structured, reproducible framework for estimating mortality risk. This probability estimate can inform shared decision-making conversations with surrogate decision-makers, guide team-level discussions about resource allocation and goals of care, and support institutional quality improvement efforts in ECPR programs.

Background: In-Hospital Cardiac Arrest and the Role of ECPR

In-hospital cardiac arrest (IHCA) affects an estimated 290,000 adults annually in the United States alone, with survival to hospital discharge rates of approximately 25-27% across all patients. However, outcomes vary substantially based on the cardiac rhythm, arrest location, witnessed vs. unwitnessed status, time to initiation of cardiopulmonary resuscitation (CPR), and the underlying etiology of the arrest. A subset of patients fails to achieve sustained return of spontaneous circulation (ROSC) despite prolonged conventional resuscitation, a state termed refractory cardiac arrest.

Extracorporeal cardiopulmonary resuscitation (ECPR) refers to the initiation of venoarterial extracorporeal membrane oxygenation (VA-ECMO) during ongoing cardiac arrest to restore mechanical circulatory support and provide time for myocardial recovery, definitive treatment of a reversible underlying cause, or bridge to advanced cardiac therapies. By maintaining end-organ perfusion during a period of otherwise lethal circulatory failure, ECPR offers a potential pathway to survival for patients who would otherwise die from refractory arrest.

The potential benefit of ECPR is most compelling when cardiac arrest results from a reversible cause (acute coronary occlusion, massive pulmonary embolism, refractory ventricular fibrillation, post-cardiac surgery complications) that can be definitively treated once circulation is mechanically restored. In these scenarios, ECMO effectively replaces the heart as a pump while coronary revascularization, thrombolysis, surgical repair, or other targeted therapy addresses the underlying problem.

However, ECPR is a highly resource-intensive intervention requiring specialized cannulation skills, dedicated perfusionist support, a complex ICU care environment, and substantial institutional infrastructure. Overall survival in published ECPR series ranges from 20-40%, with significant variability across centres reflecting differences in patient selection, institutional experience, protocol maturity, and post-resuscitation care quality. This broad survival range underscores the critical importance of thoughtful patient selection: ECPR deployed indiscriminately carries high rates of futility, resource burden, and potential harm to patients who cannot benefit, while overly restrictive selection may deny a survival chance to patients who could recover.

The RESCUE-IHCA score was developed specifically to address this selection challenge by providing a data-driven, evidence-based estimate of mortality risk at the moment ECPR is being considered. By quantifying the degree to which individual patient and arrest characteristics predict in-hospital death, the score helps clinicians identify which patients are most and least likely to benefit from this high-acuity intervention.

Score Development: Registry Data and Methodology

The derivation and validation study by Tonna et al. used two large, independent datasets to develop and test the RESCUE-IHCA score:

Derivation Cohort: GWTG-R Registry

The American Heart Association Get With The Guidelines-Resuscitation (GWTG-R) registry is a prospective, multi-centre quality improvement database tracking in-hospital resuscitation events across hospitals in the United States. For the RESCUE-IHCA derivation, data from 1,075 adult patients (≥18 years) who sustained IHCA and were treated with ECPR across 219 participating centres between 2000 and 2018 were analysed. Patients with prior out-of-hospital cardiac arrest, non-index arrest events, and missing outcome data were excluded.

The primary outcome was in-hospital death. In the derivation cohort, 28% of patients survived to hospital discharge, meaning approximately 72% died before discharge. This overall mortality figure is consistent with pooled estimates from published ECPR series and underscores the high-acuity, high-mortality nature of the ECPR-treated IHCA population.

Candidate predictor variables were identified from the GWTG-R dataset that were consistently collected, clinically plausible, and available at or near the time of ECPR initiation. Variables requiring laboratory values (arterial pH, lactate, creatinine at arrest) were excluded because they were not systematically available in the GWTG-R registry, a design choice that simultaneously limits the score's completeness and enhances its practical real-time applicability.

Statistical model development used Bayesian model averaging after multiple imputation across 20 augmented datasets (using the MICE package for regression-based imputation of missing covariates). Bayesian averaging assigns probability weights to multiple candidate models rather than selecting a single best-fitting model, producing more stable coefficient estimates and reducing overfitting. Variables retained in the final model were those with consistently high posterior inclusion probability across averaging iterations.

The final six-variable model achieved an area under the receiver operating characteristic curve (AUC) of 0.719 (95% CI 0.680-0.757) in the derivation cohort, with acceptable calibration on Hosmer-Lemeshow testing (p = 0.079). AUC values above 0.70 are generally considered adequate for clinical prediction models in complex resuscitation populations.

External Validation Cohort: ELSO Registry

External validation was conducted in 297 adult patients from the Extracorporeal Life Support Organization (ELSO) Registry (2017) who received ECPR for IHCA. The ELSO registry is an international multi-centre database maintained by the Extracorporeal Life Support Organization, collecting data on all ECMO runs at member institutions worldwide. This provided a geographically and institutionally distinct validation cohort, strengthening confidence in the score's generalizability beyond the GWTG-R hospitals.

In the ELSO validation cohort, overall survival to hospital discharge was again 28%, closely matching the derivation cohort. The RESCUE-IHCA score achieved an AUC of 0.676 (95% CI 0.606-0.746) in this external validation, with excellent calibration (Hosmer-Lemeshow p = 0.66, indicating no significant departure from predicted vs. observed mortality rates). The modest reduction in AUC from derivation to validation (0.719 to 0.676) is expected in external validation and reflects the inherent uncertainty of any prediction model applied to a new population.

A subsequent independent external validation by Ho et al. (2024) in an Asian ECPR cohort reported an AUC of approximately 0.63, suggesting that score performance may vary in populations with different baseline characteristics, ECPR protocols, and healthcare system structures. This finding reinforces the importance of understanding the score's derivation context when applying it in different institutional settings.

Score Variables and Calculation

The RESCUE-IHCA score is calculated by summing six component point values. The score typically ranges from -15 to greater than 40, with higher values corresponding to higher predicted in-hospital mortality.

1. Age

Age is incorporated as a continuous variable converted to a point scale: 2 points for age ≤20 years, with an additional +2 points for each additional 10 years above age 20. For example, a 40-year-old patient scores 6 points (2 for ≤20 base, +2 for the 21-30 decade, +2 for the 31-40 decade), while a 70-year-old scores 12 points.

In the derivation cohort, older age was significantly associated with in-hospital mortality: survivors had a median age of 58 years compared to 61 years in non-survivors (p = 0.019). While the absolute age difference was modest, the continuous age contribution across a wide range of patients produces clinically meaningful point variation in the composite score.

Younger patients, particularly those under 40, accumulate relatively few age-based points, partially reflecting the greater physiologic reserve and likelihood of reversible arrest etiology in this population. Elderly patients above 70-80 years accumulate 14-18 age-based points alone, substantially elevating their baseline composite score before other factors are considered.

2. Pre-existing Renal Insufficiency (+8 points)

A history of pre-existing renal insufficiency documented prior to the cardiac arrest event adds 8 points to the score. This is the largest single point increment for any dichotomous (yes/no) variable in the model, reflecting the strong and independent association between renal dysfunction and post-arrest mortality.

In the GWTG-R derivation cohort, 80% of patients with documented renal insufficiency died before hospital discharge, compared to 69% of those without renal insufficiency (p = 0.001). Pre-existing renal insufficiency is a marker of multisystem comorbidity burden, reduced physiologic reserve, and vulnerability to the ischemia-reperfusion injury associated with cardiac arrest and ECMO support.

Renal insufficiency is defined as any pre-existing documented history of kidney disease or impaired renal function, consistent with the GWTG-R registry data fields. This includes patients with chronic kidney disease (CKD) at any stage, as well as those on dialysis. Acute kidney injury occurring as a consequence of the cardiac arrest itself is not included in this variable.

Beyond its direct contribution to the score, renal insufficiency in ECPR patients creates compounding clinical challenges: anticoagulation management on ECMO is more complex in the setting of uremia and platelet dysfunction, fluid management is constrained by reduced renal clearance, and recovery from multi-organ dysfunction is more difficult in patients with limited baseline renal reserve.

3. Time of Day of Arrest

The time of cardiac arrest is classified into three shifts, each associated with different in-hospital mortality rates:

Time Window	Points	Observed Survival (Derivation)
7:00 AM - 2:59 PM (Daytime)	0 (reference)	33%
3:00 PM - 10:59 PM (Afternoon/Evening)	+4	26%
11:00 PM - 6:59 AM (Nocturnal)	+13	15%

The time-of-day variable carries one of the highest point weights in the model, with nocturnal arrests (+13 points) conferring a risk burden comparable to pre-existing renal insufficiency (+8 points) combined with a decade of age. The difference in survival between daytime (33%) and nocturnal (15%) arrests is striking and statistically highly significant (p < 0.001).

Several mechanisms likely contribute to the adverse nocturnal effect. ECPR requires a specialized, multi-disciplinary team including intensivists, perfusionists, cardiac surgeons, and interventional cardiologists who can cannulate the patient within minutes of arrest. During night-time hours, many of these specialists operate on reduced staffing levels or on-call coverage, extending time to ECMO initiation. ECMO cannulation by less experienced overnight personnel may result in more cannulation complications and longer low-flow time. Additionally, post-ECMO management in overnight ICU settings may be less proactive, and access to interventional suites for coronary angiography or surgical procedures may be delayed.

The nocturnal variable highlights an important systems-level insight: institutional ECPR outcomes depend not only on patient characteristics but on the round-the-clock capacity to deliver the full spectrum of ECPR care. Centres considering ECPR program expansion should evaluate whether adequate nocturnal capacity exists before offering ECPR to all-hours arrests.

4. Illness Category

Patients are classified into one of four illness categories based on whether their admission is for a cardiac vs. non-cardiac condition and whether the care context is medical vs. surgical, as determined by CPT and ICD diagnostic codes at the time of the hospitalization. Medical non-cardiac patients serve as the reference category with the highest baseline predicted mortality (0 points). The three alternative categories all carry negative point adjustments reflecting lower predicted mortality:

Illness Category	Points	Observed Survival (Derivation)
Medical Non-Cardiac (reference)	0	~22%
Medical Cardiac	-2	~25%
Surgical Non-Cardiac	-6	~29%
Surgical Cardiac	-11	~34%

The substantial survival advantage of surgical cardiac patients (-11 points) is the largest single negative point adjustment in the model and reflects several interrelated factors. Patients admitted for cardiac surgery represent a selected population undergoing operations in centres with established cardiac surgical and perfusion infrastructure, where ECMO circuits and cannulation expertise are immediately available. Their cardiac arrests frequently stem from potentially reversible perioperative causes: bypass graft occlusion, cardiac tamponade, severe vasoplegic shock, or difficulty weaning from cardiopulmonary bypass. These conditions, while immediately life-threatening, are often definitively treatable if perfusion is rapidly restored.

Medical non-cardiac patients represent the highest-risk group in part because their cardiac arrests are more likely to occur in settings with less immediate ECMO access, and their underlying conditions (sepsis, multi-organ failure, respiratory failure, neurologic catastrophe) are less amenable to the targeted post-ECMO interventions that drive recovery in cardiac populations.

The illness category assignment should reflect the primary diagnosis driving the patient's hospitalization at the time of arrest, not the arrest etiology itself. A patient admitted for pneumonia who arrests from a hospital-acquired ventricular arrhythmia should still be categorized as Medical Non-Cardiac based on admission diagnosis.

5. Presenting Cardiac Rhythm

The initial cardiac rhythm documented at the time of cardiac arrest is a fundamental determinant of resuscitation outcomes, reflecting the underlying electrophysiologic mechanism and the likelihood of a reversible substrate. In the RESCUE-IHCA score, asystole serves as the reference category (0 points, worst prognosis), and shockable or more organized rhythms carry negative point adjustments:

Presenting Rhythm	Points	Observed Survival (Derivation)
Asystole (reference)	0	~24%
Pulseless Electrical Activity (PEA)	-1	~25%
Pulseless Ventricular Tachycardia (pVT)	-5	~34%
Palpable Pulse Initially	-5	~35%
Ventricular Fibrillation (VF)	-8	~40%

Ventricular fibrillation (VF) carries the largest survival advantage (-8 points), consistent with decades of resuscitation research demonstrating that shockable rhythms are associated with substantially better outcomes than non-shockable rhythms in both conventional and ECPR-augmented resuscitation. VF typically indicates a primary myocardial electrical instability (often from acute ischemia or channelopathy) with intact myocardial architecture that can recover function once the arrhythmic substrate is treated. In the derivation cohort, VF survival was 40% compared to 24% for asystole (p < 0.001).

Asystole and pulseless electrical activity (PEA) generally indicate either profound myocardial injury with irreversible electrical silence or a severely compromised metabolic state from hypoxia, acidosis, or electrolyte derangement. PEA carries a marginally better prognosis than asystole (-1 point), consistent with the knowledge that some PEA arrests have a reversible cause (tension pneumothorax, hypovolemia, massive PE, tamponade) that may respond to ECPR bridge to targeted therapy.

Patients with an initially palpable pulse who are cannulated for ECMO receive -5 points, reflecting that this group likely includes cardiogenic shock patients cannulated pre-arrest or peri-arrest who have not yet progressed to pulseless cardiac arrest. This population has inherently better hemodynamic reserve and a higher probability of recovery.

6. Duration of Cardiac Arrest (+2 per 10 minutes)

The total duration of resuscitative efforts before durable ROSC or decision to cannulate for ECMO adds 2 points for every 10-minute interval of arrest duration. This is the only variable in the RESCUE-IHCA score that is potentially modifiable through system-level interventions.

In the derivation cohort, survivors had a median arrest duration of 23 minutes (IQR 10-44 minutes) compared to 44 minutes (IQR 20-80 minutes) for non-survivors (p < 0.001). This striking difference reflects the cumulative ischemic injury inflicted upon the brain, heart, kidneys, liver, and gastrointestinal tract during periods of inadequate cardiac output, even with high-quality CPR. While CPR provides approximately 20-30% of normal cardiac output, it is insufficient to prevent progressive end-organ ischemia, acidosis, and cellular injury beyond a critical duration.

The linear scoring increment (+2 per 10 minutes) means that an arrest of 20 minutes adds 4 points, while an arrest of 60 minutes adds 12 points, representing a substantial shift in predicted mortality. This non-trivial point weight for arrest duration underscores the fundamental principle that time is myocardium and time is brain: the faster ECMO is initiated, the lower the cumulative ischemic burden and the higher the probability of meaningful recovery.

From an operational standpoint, the arrest duration variable provides the most actionable lever for ECPR program improvement. Institutions with dedicated ECPR activation protocols, pre-positioned ECMO circuits, and perfusionists with defined response time standards can meaningfully reduce time-to-cannulation. Every 10 minutes of low-flow time averted by faster cannulation effectively removes 2 points from the composite score, potentially shifting a patient from a high-risk to a moderate-risk tier.

Score Interpretation and Risk Stratification

The RESCUE-IHCA score maps to in-hospital mortality probability across a continuous spectrum. Based on the derivation and validation data, score ranges can be grouped into approximate risk tiers:

Score Range	Risk Tier	Approximate In-Hospital Mortality	Clinical Interpretation
≤ 10	Lower Risk	~22-45%	Favorable patient and arrest profile; favorable anatomy for ECPR benefit; strongest candidate for aggressive ECPR pursuit if no other contraindications
11-20	Moderate Risk	~46-70%	Intermediate mortality probability; benefit of ECPR possible but diminished; careful consideration of underlying reversibility and institutional capacity warranted
21-30	High Risk	~71-90%	High predicted mortality; ECPR may be considered if there is a clearly reversible etiology and patient/surrogate goals align with aggressive intervention; early goals-of-care discussion essential
> 30	Very High Risk	> 90%	Very high predicted mortality; ECPR in this range is likely futile in most patients; individual etiology-specific reversibility and surrogate goals must guide decision-making

These risk tiers are intended as a framework for clinical reasoning rather than rigid decision thresholds. No single score value has been validated as a cutoff above which ECPR should never be initiated or below which it should always be offered. The score quantifies probability, not certainty, and individual patients with high predicted mortality can and do survive with meaningful neurological outcomes.

Clinical Applications of the RESCUE-IHCA Score

Real-Time Resuscitation Decision Support

The primary intended application of the RESCUE-IHCA score is at the bedside during active resuscitation, when the ECPR team must make a rapid decision about whether to initiate ECMO cannulation. In this high-pressure, time-sensitive environment, the score provides a structured framework for synthesizing key prognostic information that might otherwise be processed intuitively and inconsistently.

A calculated score of 8 in a 45-year-old surgical cardiac patient with ventricular fibrillation arrested during daytime hours, without renal insufficiency and with a 20-minute arrest duration, provides a concrete mortality estimate that can anchor the resuscitation team's discussion. Conversely, a score of 35 in an 80-year-old medical non-cardiac patient with asystole, renal insufficiency, a nocturnal arrest, and 60 minutes of CPR contextualizes the futility concern in quantitative terms that can be communicated clearly to all team members.

Shared Decision-Making with Surrogate Decision-Makers

IHCA patients are by definition incapacitated and unable to participate in real-time decision-making about ECPR. Surrogate decision-makers (family members, healthcare proxies) are typically asked to provide substituted judgment about whether the patient would want aggressive resuscitation, often within minutes and with limited ability to process complex prognostic information.

The RESCUE-IHCA score provides a data-derived probability estimate that can be communicated in accessible terms: "Based on your loved one's age, medical history, and the circumstances of this cardiac arrest, our best estimate is that there is approximately a [X]% chance of dying in the hospital even if we place them on the heart-lung bypass machine." This framing gives surrogates a tangible probabilistic anchor for their decision, rather than relying solely on physician gestalt or vague statements about "doing everything."

It is essential that these conversations acknowledge the score's uncertainty, the specific limitations of the prediction model, and the irreplaceable role of the patient's known values and preferences. The score is one input into a complex human decision, not a verdict.

ECPR Program Development and Patient Selection Criteria

Institutions developing or refining ECPR programs often use prediction scores like RESCUE-IHCA to inform written patient selection criteria. Rather than relying on ad hoc case-by-case decisions, programs can establish prospective criteria that incorporate score thresholds alongside other clinical factors (underlying diagnosis reversibility, advance directive status, functional baseline, availability of post-ECMO care resources).

For example, a program might define ECPR as routinely offered to patients with a RESCUE-IHCA score below 20, considered with multidisciplinary input for scores 20-30, and generally not offered (absent exceptional circumstances) for scores above 30, in patients meeting other eligibility criteria. Such criteria provide internal consistency, reduce the influence of resuscitation team composition on individual decisions, and create a framework for prospective data collection and outcomes analysis.

Quality Improvement and Benchmarking

Because the RESCUE-IHCA score generates a continuous mortality probability estimate, it enables risk-adjusted outcome comparison across time periods, provider teams, and institutions. An ECPR program can calculate the expected mortality for its patient cohort based on case-mix-adjusted RESCUE-IHCA scores and compare this to observed mortality to assess whether outcomes are better or worse than predicted.

Systematic better-than-predicted performance may reflect institutional strengths in ECMO technique, post-resuscitation care, or patient selection. Worse-than-predicted performance may identify opportunities for protocol improvement, training interventions, or reconsideration of patient selection criteria. This type of risk-adjusted benchmarking is increasingly used in ECMO quality improvement programs and ELSO-affiliated institutional reviews.

Research and Clinical Trial Design

The RESCUE-IHCA score provides a validated risk-stratification tool for use in clinical research involving ECPR populations. Randomized or observational studies can use score-defined strata to ensure balanced enrollment across risk categories, apply the score as a covariate to adjust for case-mix differences between comparison groups, and define pre-specified subgroups for heterogeneity of treatment effect analyses (e.g., does ECPR benefit differ across RESCUE-IHCA score terciles?).

The ARREST trial (Yannopoulos et al., Lancet, 2020), a landmark randomized trial of ECPR for out-of-hospital refractory VF, and ongoing IHCA-focused ECPR trials have highlighted the need for risk stratification tools to ensure that trial populations are comparable and that results can be interpreted in the context of patient-level risk. The RESCUE-IHCA score fills this role for IHCA-specific ECPR research.

Comparison with Other Cardiac Arrest and ECMO Scoring Systems

The RESCUE-IHCA score occupies a specific niche in the landscape of cardiac arrest and ECMO prediction tools, distinguished by its development specifically for the ECPR-treated IHCA population. Understanding its relationship to other available scores clarifies appropriate application.

SAVE (Survival After Veno-Arterial ECMO) Score

The SAVE score (Schmidt et al., European Heart Journal, 2015) was developed to predict hospital survival in patients placed on VA-ECMO for refractory cardiogenic shock, a population that overlaps with but is distinct from the IHCA-ECPR population. The SAVE score incorporates 12 variables including diagnosis, cardiac arrest before ECMO (as a binary variable), weight, acute renal failure, chronic renal failure, serum bicarbonate, arterial pulse pressure, and duration of intubation before ECMO, among others.

While the SAVE score has been validated across large ELSO registry populations, its primary population is cardiogenic shock patients, not necessarily those in active cardiac arrest. Several SAVE score variables (bicarbonate, pulse pressure) are not readily available in the first minutes of resuscitation. The RESCUE-IHCA score's advantage is its derivation in a population defined by active cardiac arrest requiring ECPR and its exclusive use of variables available at or before the moment of cannulation.

APACHE and SOFA Scores

General critical care severity scores such as APACHE II, APACHE IV, and SOFA were not designed for real-time cardiac arrest decision-making and require laboratory values, physiologic parameters, and admission diagnoses that are either unavailable or unreliable during active resuscitation. They are appropriate for post-admission ICU prognosis but do not address the specific decision point of whether to initiate ECPR.

OHCA vs. IHCA Prediction Models

Several prediction models have been developed for out-of-hospital cardiac arrest (OHCA), including the OHCA score, CAHP score, and SCAI SHOCK classification. These tools incorporate bystander CPR, initial rhythm, time to first shock, and prehospital factors that are specific to the out-of-hospital context and not applicable to IHCA. The RESCUE-IHCA score was specifically derived from the IHCA-ECPR population and should not be applied to OHCA patients without caution. Some ECPR programs that treat both OHCA and IHCA may use different scoring tools for each population.

Targeted Temperature Management and Post-Resuscitation Care Tools

Post-ROSC prognostication tools (such as the TTM trial-derived neuroprognostication algorithms and ERC/ESICM post-resuscitation care guidelines) address a different and later decision point: neurological outcome prediction after successful resuscitation. The RESCUE-IHCA score operates at the pre- or peri-arrest decision point about whether to initiate ECPR, not post-resuscitation neurological prognosis. These tools are complementary, not competing.

The Nocturnal Arrest Phenomenon: A Deeper Look

The outsized weight of the nocturnal arrest variable (+13 points for 11 PM to 7 AM) in the RESCUE-IHCA score warrants additional clinical attention. The score designates nocturnal arrests as carrying a mortality-risk increment equivalent to more than six additional decades of age, or the combination of renal insufficiency and a decade of age together. This is a remarkable finding with important systems-level implications.

Multiple mechanisms have been proposed to explain the nocturnal survival disadvantage in ECPR-treated IHCA:

• Reduced in-hospital monitoring intensity during overnight hours may result in a longer interval between the physiologic onset of hemodynamic deterioration and recognition of cardiac arrest, increasing no-flow and low-flow time before both conventional CPR and ECMO initiation.
• Staffing differences: Overnight hospitals typically operate with reduced nurse-to-patient ratios, fewer senior physicians physically present, and on-call rather than in-house ECMO specialists. The quality of CPR provided by fewer responders may be lower, and time to assembling a full ECPR team is longer.
• Cannulation expertise: ECMO cannulation is a high-skill procedure. If overnight cannulators are less experienced (residents, fellows, or surgeons outside their routine call rotation), cannulation complications and time-to-flow may be higher than during daytime hours when primary ECMO specialists are present.
• Delayed access to definitive post-ECMO therapies: Coronary angiography, cardiac surgery, and other post-resuscitation interventions may face longer door-to-procedure times during overnight hours, reducing the benefit of mechanically supported circulation in patients whose survival depends on rapid revascularization or surgical correction.
• Circadian physiology: Some investigators have proposed that circadian rhythm variation in myocardial vulnerability, autonomic tone, and hormonal milieu may contribute to both the timing of arrests and the physiologic capacity for recovery from ischemia-reperfusion injury.

The practical implication of the nocturnal variable is that ECPR programs should explicitly evaluate whether their overnight staffing model supports outcomes comparable to daytime operations. Where nocturnal outcomes are significantly worse, targeted interventions (24/7 dedicated ECMO perfusionist in-house coverage, mandatory rapid response team activation protocols, night-time ECPR simulation training) may mitigate the nocturnal survival deficit.

Renal Insufficiency in ECPR: Mechanisms and Clinical Significance

The +8-point contribution of pre-existing renal insufficiency to the RESCUE-IHCA score reflects a multifaceted relationship between chronic kidney disease and outcomes after cardiac arrest and ECMO support. Understanding these mechanisms helps clinicians appreciate why renal insufficiency carries such a high point weight.

Baseline vulnerability to ischemia-reperfusion injury: Chronically diseased kidneys have reduced functional reserve and impaired tubular adaptation to ischemia. Even brief periods of no-flow and low-flow during cardiac arrest can precipitate acute-on-chronic kidney injury, accelerating progression to dialysis and extending ICU stay.

Anticoagulation challenges on ECMO: VA-ECMO requires systemic anticoagulation, typically with unfractionated heparin. Uremia impairs platelet function and coagulation factor activity, creating a paradoxically prothrombotic and hemorrhagic state simultaneously. Achieving target anticoagulation while minimizing both circuit thrombosis and bleeding complications is substantially more difficult in patients with pre-existing CKD.

Fluid management constraints: ECMO-supported patients often require large volumes of blood products, crystalloid resuscitation, and pharmacologic agents. In patients without renal insufficiency, excess fluid is mobilized and excreted during recovery. In patients with CKD or acute-on-chronic renal failure, fluid overload accumulates rapidly, increasing positive pressure ventilation requirements, impairing cardiac recovery, and prolonging ECMO duration.

Multi-organ dysfunction trajectory: Pre-existing renal insufficiency is a surrogate marker for more widespread multisystem comorbidity (hypertension, diabetes, cardiovascular disease, anemia, nutritional deficiency) that reduces overall physiologic resilience during the post-arrest recovery period.

Arrest Duration as a Modifiable Quality Target

Unlike the five non-modifiable variables in the RESCUE-IHCA score, arrest duration (+2 per 10 minutes) represents a systems-level quality target that ECPR programs can actively work to reduce. The relationship between arrest duration and outcome in ECPR-treated patients is well established and provides the clearest direct rationale for investing in rapid ECMO deployment infrastructure.

Strategies that may reduce arrest duration and thereby reduce RESCUE-IHCA score point accumulation include:

• Pre-positioned ECMO circuits: Having circuits primed and ready in high-risk locations (cardiac catheterization laboratory, cardiac ICU, cardiovascular operating rooms) reduces preparation time from 20-40 minutes to fewer than 5 minutes after cannulation decision.
• 24/7 in-house ECMO cannulation teams: Programs with cannulators physically present in the hospital rather than on-call from home achieve shorter time-to-cannulation, particularly during overnight hours.
• Mechanical CPR devices: Devices such as the LUCAS or AutoPulse provide consistent, fatigue-free CPR quality during the cannulation procedure, reducing interruptions and allowing the cannulating operator to work without competition from manual compressions.
• Protocolized ECPR activation criteria: Clear, pre-defined activation criteria (rhythm, arrest duration thresholds, location, diagnosis criteria) allow immediate team mobilization without requiring attending physician-level approval at each step, reducing decision-to-cannulation time.
• Simulation-based training: Regular high-fidelity ECPR simulation drills improve team coordination, reduce cannulation time, and identify procedural bottlenecks that can be addressed prospectively.
• ECPR activation scoring systems: Some institutions use real-time decision support tools during ongoing resuscitation that trigger ECPR activation when predefined criteria are met (e.g., specified arrest duration without ROSC in a shockable rhythm), removing the cognitive burden of activation decision from the resuscitation leader.

Neurologically Intact Survival: The Score's Outcome Gap

A critical limitation of the RESCUE-IHCA score is that it predicts in-hospital mortality, not neurologically intact survival. From both a patient-centered and societal resource-allocation perspective, the distinction between survival and meaningful neurological recovery is fundamental. A patient who survives to hospital discharge with severe anoxic brain injury requiring permanent institutional care represents a qualitatively different outcome from a patient discharged home with full neurological recovery.

In the RESCUE-IHCA derivation cohort, greater than 85% of survivors had good neurological outcomes recorded (Cerebral Performance Category 1 or 2), suggesting that most survivors of IHCA-ECPR do achieve meaningful neurological recovery. However, neurological outcome data had high rates of missingness in the GWTG-R registry, making definitive statements about the neurological outcome distribution across score tiers unreliable.

Future refinements of ECPR prediction scores should incorporate neurological outcome as the primary endpoint, and post-resuscitation neuroprognostication frameworks should be applied in ECPR survivors to guide ongoing care decisions after initial resuscitation success. The RESCUE-IHCA score's mortality prediction should be interpreted in conjunction with available post-arrest neurological assessment tools (EEG, somatosensory evoked potentials, brain MRI, clinical neurological examination) when making withdrawal-of-life-sustaining-treatment decisions in ECPR survivors.

The Role of Illness Category: Understanding the Surgical Cardiac Advantage

The -11-point adjustment for surgical cardiac illness category is the most powerful single categorical modifier in the RESCUE-IHCA score and deserves careful clinical analysis. Several structural and contextual factors explain why cardiac surgical patients treated with ECPR have substantially better survival than other illness categories.

Infrastructure proximity: Cardiac operating rooms and cardiac surgical ICUs maintain ECMO circuits and perfusionists immediately available as a core operational requirement. ECPR in this setting benefits from essentially zero circuit preparation delay and cannulation by personnel who perform cardiac and vascular surgery daily.

Reversible etiologies: Post-cardiac surgery arrest commonly results from bypass graft thrombosis, coronary spasm, cardiac tamponade, aortic dissection flap extension, or difficult weaning from intraoperative cardiopulmonary bypass. Many of these causes are precisely diagnosable and surgically correctable within the ECMO support window, allowing the intervention to serve its intended purpose as a bridge to definitive therapy.

Preserved myocardial architecture: Unlike patients with acute myocardial infarction complicated by cardiac arrest, post-surgical patients' myocardium is structurally intact (barring intraoperative injury) and may recover function once ischemic or obstructive causes are corrected.

Monitored setting: Cardiac surgical patients are continuously monitored with invasive arterial and pulmonary artery catheters, making both the recognition of hemodynamic deterioration and the physiologic assessment during resuscitation more informative and timely.

Clinicians should recognize that the "surgical cardiac" category captures a heterogeneous group and that not all surgical cardiac arrests are equally reversible. A patient arrested after cardiac transplantation for primary graft dysfunction in an anatomy-unfavorable configuration may have a poorer prognosis than the category average, while a patient who arrests from a technically correctable tamponade 12 hours after coronary bypass grafting may have a substantially better prognosis.

Practical Implementation: Common Scenarios and Score Calculation Examples

Scenario 1: Favorable Profile (Lower-Risk)

A 35-year-old (age points: 2 for ≤20 base + 2 for 21-30 decade + 2 for 31-35 = 6 points, using the 31-40 decade tier) previously healthy cardiac surgical patient (illness: surgical cardiac = -11 points) with no renal insufficiency (0 points) arrests at 10:00 AM (daytime = 0 points) with ventricular fibrillation (VF = -8 points) after coronary artery bypass grafting. CPR is ongoing for 15 minutes at the time of cannulation decision (duration = 2 points for first 10 minutes, rounded = +4 for 15 minutes, typically +2 per completed 10-minute interval, so 1 interval = +2 points).

Approximate score: 6 + 0 + 0 + (-11) + (-8) + 2 = -11 points. This is firmly in the lower-risk tier, corresponding to a predicted mortality of approximately 22-30%. ECPR is strongly supported by the score in this patient.

Scenario 2: High-Risk Profile

A 72-year-old (age points: 2 + [5 additional decades × 2] = 12 points) medical non-cardiac patient admitted for sepsis with known CKD stage 4 (renal insufficiency = +8 points) arrests at 3:30 AM (nocturnal = +13 points) with pulseless electrical activity (PEA = -1 points). CPR is ongoing for 50 minutes before the team discusses ECPR (duration: 5 complete 10-minute intervals = +10 points).

Approximate score: 12 + 8 + 13 + 0 + (-1) + 10 = 42 points. This falls in the very-high-risk tier with predicted mortality above 99%. The score quantifies the extreme futility risk in this case and should prompt a goals-of-care discussion with surrogate decision-makers before any cannulation attempt.

Scenario 3: Intermediate Profile Requiring Careful Deliberation

A 55-year-old (age points: 2 + 3 additional decades × 2 = 8 points) medical cardiac patient (illness: medical cardiac = -2 points) with no renal insufficiency (0 points) arrests at 7:00 PM (afternoon/evening = +4 points) with ventricular fibrillation (VF = -8 points) after an acute STEMI. ECPR team arrives at 35 minutes of CPR (+6 points for 3 complete 10-minute intervals, with 30 minutes = 3 × 2 = +6 points, but for 35 minutes use 3 completed intervals = +6).

Approximate score: 8 + 0 + 4 + (-2) + (-8) + 6 = 8 points. This falls at the lower end of the lower-risk tier, with favorable anatomy (shockable rhythm, cardiac etiology, no renal disease) tempered by the evening timing and moderate arrest duration. ECPR as a bridge to primary PCI is well-supported at this score level, provided immediate catheterization laboratory access is available.

Special Populations and Applicability Considerations

Pediatric Patients

The RESCUE-IHCA score was derived exclusively in adults (≥18 years) and has not been validated in pediatric cardiac arrest patients. Pediatric ECPR outcomes are governed by different age-specific physiology, distinct underlying etiologies (congenital heart disease, respiratory failure, myocarditis, metabolic disorders), and different normative values for physiologic parameters. Dedicated pediatric ECPR prediction tools (or institutional pediatric cardiac arrest registries) should be used for patients under 18 years.

Post-OHCA Patients Transferred to ECPR Centres

Patients who sustained out-of-hospital cardiac arrest (OHCA) and are transferred to ECPR-capable centres with ongoing refractory arrest represent a distinct population with additional prognostic factors related to prehospital resuscitation quality, transport time, and environmental conditions not captured in the RESCUE-IHCA score. The score was explicitly derived from IHCA patients and should not be applied to OHCA patients without recognition of this important derivation-population mismatch.

Post-Cardiac Surgery Arrest: Resternotomy Context

Many cardiac surgical arrests are managed with immediate resternotomy and open cardiac massage before ECPR consideration. When resternotomy is performed, it alters both the arrest physiology (open-chest massage provides better cardiac output than closed-chest CPR) and the cannulation approach. The RESCUE-IHCA score does not specifically account for whether open- vs. closed-chest CPR was employed, and its performance in the resternotomy-ECPR subset of the surgical cardiac population deserves specific study.

Patients with Advance Directives

Patients with documented do-not-resuscitate (DNR) or do-not-intubate (DNI) orders should not be evaluated with the RESCUE-IHCA score unless those orders have been explicitly suspended or overridden by the patient's prior stated wishes in the specific context of an operative or peri-procedural setting. The score is applicable to patients in whom full resuscitative measures have been authorized and ECPR is under active consideration.

Frequently Asked Questions

Can the RESCUE-IHCA score be used to decide whether to terminate resuscitation?

The score was developed to predict in-hospital mortality and can inform termination discussions by quantifying predicted futility risk. However, it was not designed or validated as a termination-of-resuscitation rule and should not be used as a standalone criterion to stop resuscitation. Termination decisions in the ECPR context require integration of arrest duration, CPR quality, clinical trajectory, reversibility of underlying cause, and surrogate decision-maker goals, alongside but not replaced by the score.

What if the illness category is unclear or the patient has mixed diagnoses?

Illness category classification follows the primary reason for hospital admission as coded by CPT and ICD codes at the time of the arrest-precipitating hospitalization. In ambiguous cases, use the primary admitting diagnosis or the most prominent diagnosis driving the current hospitalization. A patient admitted primarily for cardiac bypass surgery who develops pneumonia pre-arrest should still be classified as surgical cardiac. If genuine ambiguity exists, document the uncertainty and consider sensitivity analyses across the two most plausible categories.

Should the score be recalculated during resuscitation as arrest duration increases?

Yes. Because arrest duration is a continuous, real-time variable, the composite score changes as resuscitation continues. Recalculating the score at defined intervals (e.g., every 10 minutes of ongoing arrest) allows the team to track the trajectory of the mortality estimate and recognize when prolonged resuscitation has shifted the patient into a higher-risk tier. Some programs use this dynamic recalculation as a formal checkpoint mechanism for ongoing ECPR candidacy reassessment.

How should the score be interpreted for patients with very short arrest durations (<10 minutes)?

For patients who are cannulated within the first 10 minutes of arrest, the arrest duration component contributes 2 points (1 completed 10-minute interval). This reflects the minimum low-flow contribution. Even with very short arrest duration, the composite score is still substantially influenced by the other five variables, maintaining its prognostic utility across all arrest duration ranges.

Does a high RESCUE-IHCA score automatically preclude ECPR?

No. The score predicts mortality probability, not mandates clinical decisions. Patients with high predicted mortality who have a clearly reversible underlying cause (e.g., massive pulmonary embolism amenable to catheter-directed thrombolysis, acute coronary occlusion amenable to primary PCI) may still be offered ECPR if the etiology-specific reversibility and patient-surrogate goals justify the intervention. The score is a decision support tool, not a decision replacement tool.

Limitations and Important Caveats

• Observational registry derivation: The RESCUE-IHCA score was derived from retrospective registry data subject to selection bias, centre-level variation in ECPR protocols, missing data, and heterogeneous coding practices. The GWTG-R registry depends on voluntary hospital participation, which may over-represent academic and high-volume ECPR centres relative to community hospitals.
• Mortality-only outcome: The score predicts in-hospital death and does not specifically address neurologically intact survival, long-term functional outcomes, or quality of life. Most ECPR survivors do achieve meaningful neurological recovery, but this is not guaranteed and should not be assumed from the score alone.
• No laboratory variables: Potentially powerful predictors such as initial arterial pH, lactate, bicarbonate, and end-tidal CO₂ were unavailable in the GWTG-R dataset and are not incorporated. Future score revisions with laboratory data may improve predictive performance.
• External validation AUC attenuation: AUC declined from 0.719 in derivation to 0.676 in ELSO validation and approximately 0.63 in the Asian validation cohort. This suggests that score performance may be population-dependent and that institutional calibration efforts are warranted before applying the score in very different healthcare settings.
• Time-of-day as a systems proxy: The nocturnal variable captures the aggregate effect of reduced overnight staffing and resources on outcomes but does not directly measure ECMO deployment time, cannulation quality, or post-ECMO care intensity. Institutions with robust 24/7 ECPR infrastructure may observe better nocturnal outcomes than the registry-derived score suggests.
• Post-arrest care not incorporated: Variables that govern recovery after ECMO initiation (targeted temperature management, coronary revascularization, hemodynamic management quality, weaning strategy, ECMO circuit management) are not captured in a pre-arrest prediction score by definition. These factors substantially influence final outcomes and represent the major modifiable domain after ECMO is initiated.
• Intentional exclusion of race: Race was excluded from the final model to avoid encouraging racial bias in ECPR resource allocation decisions, though it was associated with survival in univariate analysis. Clinicians should be aware of and actively work against structural and systemic biases in ECPR patient selection that are not captured by the score.
• Evolving ECPR technology and protocols: The GWTG-R derivation cohort spans 2000-2018, a period during which ECMO technology, cannulation techniques, and post-resuscitation care practices evolved substantially. Score performance may differ in contemporary practice using current-generation ECMO circuits, cannulas, and management protocols.