MIRACLE2 Risk Score for Early Prediction of Neurologic Outcome in Out-of-Hospital Cardiac Arrest

Background Information: Out-of-hospital cardiac arrest (OHCA) remains a diagnostic challenge to providers and a significant burden on healthcare systems globally. Despite the advancement of invasive medical therapies such as percutaneous coronary intervention (PCI) and extracorporeal membranous oxygenation (ECMO) at designated cardiac catherization centers, the majority of these patients sustain poor outcomes due to hypoxic brain injury. Clinical features of neurologic injury are typically delayed until 72 hours after admission. As a result, many neuro-prognostication tools have been developed to assist with clinical decision making as well as reduce expensive futile interventions.1 Some of these neuroprognostication tools include the Cardiac Arrest Hospital Prognosis (CAHP), OHCA and Targeted Temperature Management (TTM) risk tools. Unfortunately, these are complex and time consuming, thus limiting their use in the emergency department (ED). The authors of the following study sought out to develop and validate a point-based risk score to support clinical decision making and predict neurologic outcomes using the cerebral performance category (CPC) scale (Figure 1)

Paper: Pareek N, et al. A practical risk score for early prediction of neurological outcome after out-of-hospital cardiac arrest: MIRACLE2. Eur Heart J. Jul 2020. PMID: 32731260

Figure 1: Breakdown of the Cerebral Performance Category (CPC) Scale

Clinical Question:

  • Can the development of a practical point-based risk score applied to OHCA patients arriving to a specialized cardiac catherization center accurately reflect long-term prognosis and support clinical decision making?

What They Did:

  • Retrospective chart review from a single large quaternary center in South East London’s OHCA database (KOCAR) between May 2012 and December 2017
  • Authors also included prehospital data from London Ambulance Service, such as:
    • Zero-flow time: Duration from the cardiac arrest to commencement of CPR
    • Low-flow time: Duration of CPR until ROSC
    • Initial Rhythm 
    • Use of bystander CPR
  • Baseline cardiovascular investigations such as ECG, Echocardiography and coronary angiography were also collected
  • Patients who did not follow-up at 6-months and were known to be alive were approached by telephone interview. Those who had CPC scores of 3 – 4, had a nominated consultee or designated person perform the phone interview for them
  • Variables for development of the risk score included those with the strongest evidence of clinical relevance, statistical association and practical availability for application at point of arrival
  • Seven predictor variables resulting in a final score ranging from 0 to 10 were used in the final model and it was named MIRACLE2 (See Figure 2)
  • Variables exhibiting significant association in the multivariable analysis were assigned 1 point with the exception of age and administration of epinephrine as these had particularly large effect sizes. Age was further divided into three categories each with designated cut-off points
  • Three risk groups were created depending on the patient’s MIRACLE2 score:
    • 0 – 2 = LOW risk of poor neurologic outcome
    • 3 – 4 = INTERMEDIATE risk of poor neurologic outcome
    • > 5 = HIGH risk of poor neurologic outcome
  • Following development of the risk score, an external validation was performed in two cohorts from two separate designated PCI centers across two separate countries
  • Using the same data, the MIRACLE2 Risk Score was also compared to the following other prediction scores: The OHCA Score, Cardiac Arrest Hospital Prognosis (CAHP), and Target Temperature Management (TTM) risk tool


Figure 2:
Breakdown of the MIRACLE2 Risk Score

Inclusion Criteria:

  • All OHCA patients over the age of 18 years with ROSC prior to hospital arrival
  • All patients with ST-elevation on ECG and those without ST-elevation if there was an absence of non-cardiac etiology

Exclusion Criteria:

  • Under the age of 18 years old
  • Died prior to hospital arrival
  • Evidence of an obvious non-cardiac cause of arrest (ie. suicide, trauma, drowning, overdose)
  • Confirmed intracerebral bleeding
  • Prior neurological disability (CPC 3 or 4)
  • Any survival limiting disease (comorbidity leading to <6 month life expectancy)
  • Patients lost to follow-up (1 patient)
  • Any patient with intact conscious state defined as a GCS of 15

Outcomes:

Primary

  • Poor neurologic outcome defined as CPC of 3 – 5 (severe disability to death) at 6-month follow-up

Secondary

  • None mentioned

Results:


Development Cohort

  • Of the 1055 OHCA patients, 291 failed to regain ROSC and 129 died prior to arrival
  • Out of the 635 who reached the hospital, 232 had non-cardiac causes for their arrest and following exclusion of those who had a GCS of 15, 373 were left for final analysis
  • The patients were then divided further into Good Outcome (CPC 1-2) and Poor Outcome (CPC 3-5)
  • The area under the curve (AUC) for the MIRACLE2 Score in this cohort was: 0.90 (95% CI 0.865 to 0.928)

Validation Cohorts

  • 325 patients were recruited at the PCI center in Ljubljana, Slovenia
  • 148 patients were recruited at the Royal Free Hospital, another PCI center in London, UK
  • At these two sites, the MIRACLE2 score was compared to the OHCA and CAHP scores
  • In the Ljubljana cohort, the MIRACLE2 AUC was 0.84 and in the RFH cohort it was 0.91

Critical Results:



  • Primary endpoint occurred in 5.6% of those with low-risk, 55.4% in the intermediate risk, and 92.3% in the high-risk groups
  • With a MIRACLE2 score >2 (low-risk), the sensitivity was 97.9%. The specificity increased to 90.8% when the score was > 5 (high-risk)
  • When compared to other predictive scoring tools at their end-point range, the MIRACLE2 score was the only tool that had both a high sensitivity (70.6%) and a high specificity (90.8%)
  • At low risk, MIRACLE2 had a NPV of 94.4% (95% CI: 86.4-98.5) in the development cohort and 81.8% (95% CI: 72.2-89.2) and 91.5% (95% CI: 82.5-96.8) in the two validation cohorts
  • At high-risk, MIRACLE2 had a PPV of 92.3% (95 CI: 86.7-96.1) in the development cohort and 92.8% (95% CI: 84.9-97.3) and 89.6% (95% CI: 77.3-96.5) in the two validation cohorts
  • The AUC values were matched to sensitivity and specificity for the various risk tools across the two cohorts (Figure 3)

Figure 3: AUC values of the multiple risk tools in both the development and validation cohorts

Strengths:

  • Primary outcome was a blinded analysis which decreases bias in this result
  • Minimized overfitting with further model refinement selecting variables with the strongest evidence of clinical relevance, statistical association, and practical availability for application
  • Attempted to create an easy-to-use risk tool in the emergency setting to help with clinical decision making that also helps to reduce healthcare costs, specifically in under-resourced medical centers
  • Clinically relevant patient-oriented primary outcome using the common and easy to understand CPC score
  • Authors formed a data collaboration with London Ambulance Service to ensure accurate collection of pre-hospital data
  • Created their own built-in external validation cohorts at two different PCI centers in two different countries with no overlap of any patients between any of them
  • Compared their own developed risk tool with already existing cardiac arrest outcome prediction scores using the same set of data
  • Had a contingency plan in the form of a telephone interview for patients known to be alive who did not show up to their 6-month follow-up appointment
  • Justified points of their risk score by using strongest evidence of clinical relevance and practical availability
  • Performance of MIRACLE2 in two external validation cohorts with differing rates of predictors provides a more robust assurance of its validity

Limitations:

  • Specificity of pupillary reflexes after ROSC for neurological outcome after OHCA is not great and can easily be confounded by ambient light and drug administration
  • Predictive accuracy of the MIRACLE2 score may not transfer to community hospitals and institutions that don’t have 24-hour access to coronary angiography
  • External validity is limited to countries with more widespread CPR education and robust pre-hospital EMS systems
  • Although zero- and low-flow times are well established markers of poor outcome, they are often unknown or inaccurately recorded which can add challenges in predicting neurologic outcome on arrival
  • ABGs results available within 30 minutes of patient arrival which is fast, given how many other things need to occur to improve patient oriented outcomes
  • Proportion of pupil reactivity and ABGs in the Slovenia cohort were recorded upon ICU admission which may have affected the score’s performance
  • The MIRACLE2 score was derived and validated in retrospective cohorts as opposed to prospective cohorts
  • The Royal Free Hospital in the validation cohort is less than 10 miles from King’s College Hospital in the development cohort indicating that the patient populations were likely very similar when compared to the Slovenian cohort

Discussion:

  • The authors made conscious efforts to develop a very useful and quick scoring system that can specifically be used in the emergency department. This is in contrast to the already existing outcome prediction risk tools which are complex and too time-consuming to perform in this particular setting
  • It is important to note that hypoxic brain injury sustained prior to arrival is the main driver of morbidity and mortality in OHCA survivors
  • Current recommendations suggest that favorable cardiac arrest circumstances should be present before consideration for angiography. This, however, is an ambiguous recommendation with limited clinical benefit in deciding which patients should undergo invasive investigation
  • Despite their heterogeneity, the authors justify specifically including OHCA patients who were being considered for coronary angiography because this group requires objective guidelines to support their decision making
  • The authors demonstrate the potential impact on clinical decision making that the MIRACLE2 score may have in the clinical setting. Early angiography was performed in 285 patients from the KOCAR registry. A total of 90.1% of patients treated with an early invasive strategy and a MIRACLE2 score of > 5 had poor neurological outcome. Further studies using this score have the potential to answer a potential healthcare saving benefit in procedures that are not performed due to futility.
  • This study’s findings are system dependent. As seen from the results of their database (including Slovenia’s data), certain variables such as the zero-flow time and the rate of bystander CPR are nowhere near as high in the United States.2 Countries who have a more robust EMS system (quick response, able to provide zero and low flow data time) and widespread CPR education are more likely to successfully replicate the findings in this study
  • Although not incorporated into the MIRACLE2 Score, 71.8% of the total patients in the KOCAR registry received bystander CPR. Additionally, 70.2% of the total patients had shockable rhythms. These factors may have biased the study towards better outcomes
  • This study had ABG results within 30 minutes of patient arrival. This is very impressive and not always available at most institutions
  • The authors are applauded in highlighting that the score is NOT intended to replace clinical assessment but to provide a practical means for objective neurological evaluation prior to delivery of expensive and invasive therapies

Author’s Conclusions:

  • The MIRACLE2 is a practical risk score for early accurate prediction of poor neurological outcome after OOHCA, which has been developed for simplicity of use on admission

Our Conclusion:

  • Although simple and quick to use, the MIRACLE2 Risk Score is currently reserved for resource-rich academic institutions which is not the majority of healthcare systems around the world. This risk score also requires further validation in larger cohorts and across different healthcare systems as certain difficult-to-replicate variables in this study may have biased results towards favorable outcomes

Clinical Bottom Line:

  • The MIRACLE2 Risk Score is a promising clinical decision-making tool for predicting neurologic outcomes in OHCA patients presenting to resource-rich academic emergency departments. However, additional validation studies in larger and more diverse cohorts are required before it can be regularly used, especially in community settings.

REFERENCES:

  1. Pareek N, et al. A practical risk score for early prediction of neurological outcome after out-of-hospital cardiac arrest: MIRACLE2. Eur Heart J. Jul 2020. PMID: 32731260
  2. BYSTANDER CPR: Girotra S, et al; CARES Surveillance Group and the HeartRescue Project. Regional Variation in Out-of-Hospital Cardiac Arrest Survival in the United States. Circulation. May 2016. PMID: 27081119

 Post Peer Reviewed By: Salim R. Rezaie, MD (Twitter: @srreziae)

Cite this article as: Mark Ramzy, "MIRACLE2 Risk Score for Early Prediction of Neurologic Outcome in Out-of-Hospital Cardiac Arrest", REBEL EM blog, January 14, 2021. Available at: https://rebelem.com/miracle2-risk-score-for-early-prediction-of-neurologic-outcome-in-out-of-hospital-cardiac-arrest/.

Like this article?

Share on Facebook
Share on Twitter
Share on Linkdin
Share via Email

Want to support rebelem?

Sponsored