Following the return of spontaneous circulation (ROSC) after a cardiac arrest, understanding the precipitating cause allows clinicians to provide definitive treatment for the patient (Nolan et al, 2017). As the majority of out-of-hospital cardiac arrests (OHCAs) are thought to be precipitated by coronary pathology, advanced life support guidelines suggest that a post-ROSC electrocardiogram (ECG) is obtained and interpreted.
A common post-ROSC ECG finding is ST elevation. This often indicates an underlying acute myocardial infarction (MI), which has been shown to benefit from immediate revascularisation via percutaneous coronary intervention (PCI) (Nolan et al, 2017).
Short contact-to-balloon times are associated with improved morbidity and mortality outcomes (Foo et al, 2018), which leads to pressure on prehospital clinicians to decide at an early stage on successful resuscitation whether the patient should be conveyed directly to the catheterisation laboratory to receive PCI or whether the patient should be conveyed to the emergency department (ED) for stabilisation, assessment and ongoing management (Nolan et al, 2017).
While ST elevation is considered to represent a coronary occlusion that would benefit from coronary revascularisation (Thygesen et al, 2012), the post-ROSC ECG has been shown to be poorly predictive of the precipitating cause. Of the eight studies to date describing the predictive value of ST elevation on the post-ROSC ECG (Anyfantakis et al, 2009; Dumas et al, 2010; Sideris et al, 2011; Zanuttini et al, 2013; Garcia-Tejada et al, 2014; Lee et al, 2015; Salam et al, 2015; Stær-Jensen et al, 2015), the sensitivity of the post-ROSC ECG (ST elevation correctly identifying those with an underlying coronary lesion) ranged from 42% to 88% and the specificity (the ability of the post-ROSC ECG to identify patients without causative coronary lesions) ranged from 65% to 95% (Table 1).
| First author | Year | Positive predictive value | Negative predictive value | Sensitivity | Specificity | Prediction |
|---|---|---|---|---|---|---|
| Anyfantakis | 2009 | 83% | 84% | 70% | 91% | Acute myocardial infarction |
| Dumas | 2010 | 96% | 42% | 42% | 95% | Significant coronary lesion |
| Sideris | 2011 | 76% | 92% | 88% | 84% | Acute myocardial infarction |
| Stær-Jensen | 2015 | 63% | 76% | 64% | 75% | Acute coronary lesion |
| Zanuttini | 2013 | 85% | 67% | 67% | 85% | Acute coronary artery occlusion |
| Garcia-Tejada | 2014 | 86% | 80% | 86% | 80% | Culprit lesion |
| Lee | 2015 | 76% | 61% | 42% | 87% | Acute lesion |
| Salam | 2015 | 65% | 73% | 74% | 65% | Acute myocardial Infarction |
Most notably, the Parisian Region Out of hospital Cardiac ArresT (PROCAT) registry demonstrated that 58% of patients who did not have ST elevation on their post-ROSC ECG had at least one significant coronary lesion (Dumas et al, 2010), throwing doubt on the reliability of the post-ROSC ECG. These results have been replicated in numerous other studies, with coronary lesions being found in 26% (Spaulding et al, 1997), 17% (Cronier et al, 2011), 46% (Garcia et al, 2016) and 10% (Sideris et al, 2014) of patients without ST elevation on their post-ROSC ECG. While the number of patients with false-negative post-ROSC ECGs varies between studies, what can be consistently seen is that the absence of ST elevation on the post-ROSC ECG cannot be reliably used to exclude the need for PCI.
Based on this, many authors have suggested that every patient without an obvious extra-cardiac cause should receive post-ROSC PCI, regardless of their post-ROSC ECG (Khan et al, 2017). However, this is not economically viable in many systems and not without risks, which include coronary perforation and dissection, distal embolisation, spontaneous intracoronary thrombus formation, acute stent thrombosis and aortic dissection (Godino and Colombo, 2015). Furthermore, patients who have had a cardiac arrest secondary to an extra-coronary aetiology such as pulmonary emboli will not only fail to benefit from PCI (Nolan et al, 2017) but also experience a delay in the diagnosis and management of the underlying cause of their arrest, which may prove fatal.
A further complication to the provision of PCI to all post-ROSC patients lies in the geography of centres. PCI facilities in many UK hospitals are found at sites without co-located EDs and the majority of EDs in the UK do not have access to on-site PCI facilities. Therefore, it is essential that factors that may alter the accuracy of the post-ROSC ECG are considered to ensure the patient is conveyed to the correct facility (Nolan et al, 2017).
False positive results are also possible. There are many causes of ST elevation in addition to coronary occlusions, including pericarditis, subarachnoid haemorrhage, Brugada syndrome and drug-induced coronary vasospasm (Gu et al, 2008; Thygesen et al, 2012). However, the clinical history surrounding the arrest can raise clinical suspicion to rule in or out many of these causes.
Defibrillation has also been shown to cause ST elevation through an unknown mechanism, with Kok et al (2000) demonstrating ST elevation in 15.4% of patients following transthoracic cardioversion of ventricular tachycardia at 200 J.
Furthermore, reperfusion injury upon successful resuscitation and myocardial hypoperfusion in shocked post-ROSC patients has also been shown to cause electrocardiographic artefact, resulting in an array of post-ROSC ECG changes including ST elevation and also non-specific changes such as global ST depression, a widened QRS complex in the absence of a bundle branch morphology, and arrhythmias such as atrial fibrillation (Adrie et al, 2002).
The suggestion that misleading post-ROSC ECG changes can be secondary to myocardial stunning, hypoperfusion and reperfusion injury implies that these changes may resolve over time as myocardial perfusion improves and the reperfusion injury subsides. However, to date, no studies have commented on the influence of time on the predictive value of the post-ROSC ECG, with the majority of studies describing the predictive value of the post-ROSC ECG commenting on ECGs obtained at hospital before PCI.
This pilot study aims to investigate the predictive value of the post-ROSC ECG immediately following resuscitation from an OHCA and the predictive value of the delayed post-ROSC ECG obtained at a later time while at hospital to assess the influence that time has on the predictive value of the post-ROSC ECG.
Methods
To investigate the influence of time on the predictive value of the post-ROSC ECG, patients over the age of 18 were sought who had experienced an OHCA, had been successfully resuscitated, had received coronary angiography to identify coronary pathology and who had a prehospital and a delayed hospital post-ROSC ECG (between 1–5 hours after ROSC but before coronary angiography) accessible for review.
To achieve this, electronic records from a large UK city-based tertiary ED between 28 February 2017 to 28 February 2018 were retrospectively searched to identify patients over the age of 18 brought in by the ambulance service who had been entered onto the ED patient booking system as a ‘cardiac arrest’, were seen by the cardiac arrest team or had ‘cardiac’ entered within the free text of their presenting complaint.
Electronic ED records were reviewed and patients who had not had a cardiac arrest or had experienced a cardiac arrest secondary to an obvious extra-cardiac aetiology were removed (e.g. patients presenting with ‘cardiac pain’ or traumatic cardiac arrests). Of the patients who had a confirmed cardiac arrest, ED records were reviewed in further detail to identify those who survived to be discharged from the ED (commonly to the intensive care unit).
Physical medical records for eligible patients were reviewed in full to identify those who had received coronary angiography with a documented outcome during their inpatient stay, such as the presence of a causative lesion. Of patients with a known coronary outcome, the earliest prehospital ECG available and a hospital ECG obtained between 1 and 5 hours after ROSC but before coronary intervention were obtained and interpreted. This was done using the third universal definition of myocardial infarction; it involved assessing for a ‘positive’ or a ‘negative’ outcome depending on the presence of ST elevation in two or more contiguous leads equal to or greater than 0.1 mV in all leads other than V2 or V3, where ST elevation must have been equal to or greater than 0.2 mV for men over 40 years old, 0.15 mV for women of all ages and 0.25 mV for men under the age of 40 years (Thygesen et al, 2012).
Prehospital and delayed hospital post-ROSC ECGs were collected, interpreted and recorded alongside PCI outcome; this allow the sensitivity, specificity, positive and negative predictive values of the prehospital and delayed hospital post-ROSC ECGs to be calculated.
This study met the Medical Research Council definition of a service evaluation (Twycross and Shorten, 2014) and ethics review was not required.
Results
Between 28 February 2017 and 28 February 2018, 191 patients attended the ED and were eligible for initial inclusion. The records showed that 30 of them died in the department and they were excluded. Of the 161 remaining patients, 87 were removed as they had not had a cardiac arrest because their cardiac arrest was secondary to trauma. This left 74 patients eligible for medical records review.
Upon analysis of medical records, 53 patients were excluded. This because ECG data was insufficient (n=49), they did not undergo angiography (n=3) or the final diagnosis concluded they did not have a loss of output but a vasovagal event (n=1). This left 21 patients eligible to be included within the pilot study.
The prehospital post-ROSC ECG had a sensitivity of 25%, specificity of 60%, positive predictive value of 66% and a negative predictive value of 20% for predicting a clinically significant coronary occlusion which would benefit from PCI, with an overall accuracy of 33%. In comparison, the delayed hospital ECG had a sensitivity of 69%, specificity of 100%, positive predictive value of 100%, a negative predictive value of 50% and an accuracy of 76%.
Classifying each ECG as either ‘correct’ or ‘incorrect’ in relation to the prediction of the coronary outcome allowed for a chi-squared value to be calculated of 7.78 (P=0.0053), demonstrating a statistically significant difference between the prehospital and delayed hospital post-ROSC ECG, significant at P<0.05. This (in combination with the aforementioned sensitivities and specificities) suggests the delayed post-ROSC ECG is significantly more accurate than the prehospital ECG.
Discussion
This study aimed to assess the predictive values of prehospital and delayed hospital post-ROSC ECGs and found that, while the predictive value of the delayed hospital post-ROSC ECG was in keeping with previously published values (Table 1) (Anyfantakis et al, 2009; Dumas et al, 2010; Sideris et al, 2011; Zanuttini et al, 2013; Garcia-Tejada et al, 2014; Lee et al, 2015; Salam et al, 2015; Stær-Jensen et al, 2015), the prehospital post-ROSC ECG was poorly predictive of an underlying coronary occlusion, most notably with a low sensitivity of 25%, a low negative predictive value of 20% and an accuracy of only 33%. Extrapolating these figures suggest that the majority of acute coronary occlusions are missed by the initial prehospital post-ROSC ECG with false-negative results seen.
This study also demonstrated that the delayed post-ROSC ECG was statistically significantly more accurate in predicting a causative coronary occlusion than the initial post-ROSC ECG, suggesting that time does influence the reliability of the post-ROSC ECG, with the predictive value of the post-ROSC ECG increasing over time.
There are limitations to the results produced by this small retrospective, observational pilot study.
First, there was significant variation in the time the hospital ECG was obtained, often owing to the natural clinical course. However, despite this, the delayed ECG was clearly more predictive of underlying coronary occlusions. In addition to this, post-ROSC ECGs were interpreted by a single assessor not blinded to the coronary outcome.
As data were collected from a single, tertiary, city-based ED with co-located PCI services, it is possible that the prevalence of cardiac arrests of a cardiac aetiology is higher than other centres as patients may have been selectively conveyed to the centre from outside the ED's normal catchment area on a case-by-case basis at the discretion of advanced clinicians (such as prehospital doctors or local critical care paramedics), who had high levels of clinical suspicion that the cause of the cardiac arrest was secondary to a coronary aetiology without post-ROSC ECG evidence. In addition to this, post-ROSC patients triaged under standard post-ROSC pathways to PCI owing to ST elevation on their post-ROSC ECG attend the ED first for assessment and anaesthetic support before transfer to receive coronary intervention, so could be included in this study.
Based on this, patients recruited to this study will not only include not only local post-ROSC patients who would have attended the ED regardless as it was their local department but also patients who had ST elevation on their post-ROSC ECG and were conveyed to the hospital for its co-located PCI ability from further afield. Because of a potentially skewed prevalence, positive and negative predictive values need to be interpreted with care.
However, despite these limitations, this study highlights an important association not yet described between time and the predictive value of the post-ROSC ECG, with numerous important clinical implications. These results suggest that instead of making triage decisions on scene based on the initial post-ROSC ECG, clinicians may consider allowing time for post-ROSC ECG artefact to settle before forming a diagnosis of the precipitating event. However, given that quicker contact-to-balloon time is associated with decreased morbidity and mortality (Foo et al, 2018), to suggest that prehospital providers wait on scene for a more reliable post-ROSC ECG would be catastrophic to patient care.
Therefore, the results suggest that it may not be possible to triage the prehospital post-ROSC patient to either ED or PCI but that a compromise needs to be achieved where the patient is conveyed to an ED with co-located PCI facilities on the basis that at this initial stage it is truly unknown what definitive care is required. In practice, this may mean bypassing local EDs and even PCI facilities without acute care services to arrive at regional ‘cardiac arrest centres’, as suggested by the 2015 International Liaison Committee on Resuscitation recommendations (Nolan et al, 2017).
These centres, akin to major trauma centres within a regional trauma network, have all the necessary facilities present within one hospital (McCullough et al, 2014). In addition to having all facilities available at one centre, the staff are well rehearsed in the management of this category of patients; the importance of cannot be overestimated, with Davenport et al (2010) describing these centres as ‘specialty hospitals, not a hospital of specialties’.
The results following the UK development of regional centres for major trauma, acute myocardial infarction and stroke have been excellent, with reductions shown in both morbidity and mortality (Antman et al, 2004; MacKenzie et al, 2006; Xian et al, 2011; Morris et al, 2014). These results have been replicated in cardiac arrest centres across Denmark (Søholm et al, 2013; 2015), the United States (Stub et al, 2015) and Australia (Nehme et al, 2015), which have demonstrated improved survival to hospital discharge.
Upon arrival at a regional cardiac arrest centre, the post-ROSC patient can be supported with anaesthesia and airway management, therapeutic hypothermia, central venous access and inotropic support alongside invasive monitoring, all while being continually reassessed with investigations such as a head computed topography (CT), CT pulmonary angiogram, point-of-care ultrasound scanning and arterial blood gas sampling to further rule out underlying causes of the arrest (Nolan et al, 2017).
With slick, well-rehearsed protocols, adequate pre-alert time and preparation, these investigations can be performed concurrently in a pit-stop manner while the patient is assessed and stabilised by the various critical care teams.
This allows common extra-coronary causes of cardiac arrest such as an intracranial bleed, pulmonary emboli, pneumothoracies, cardiac tamponade, abdominal aortic aneurysm rupture or electrolyte abnormalities to be considered promptly (Nolan et al, 2017). The combination of these investigations alongside a delayed post-ROSC ECG will allow a clearer understanding of the precipitating cause of the arrest and whether the patient should proceed to coronary intervention, which would reduce unnecessary PCI.
Furthermore, while the prehospital post-ROSC ECG has been shown to be poorly sensitive in the detection of an underlying coronary occlusion, this study demonstrated a prehospital positive predictive value of 66%; this suggests that, when ST elevation is seen, the majority of patients will benefit from coronary intervention. Therefore, a pathway could be developed where a patient with a positive prehospital post-ROSC ECG is fast-tracked through the ED to PCI in the same way that a code-red trauma patient is fast-tracked to theatre in a major trauma centre (Davenport and Khan, 2011). If it later transpires that the precipitating cause of the arrest was extra-coronary with false positive ST elevation (e.g subarachnoid haemorrhage), the patient is still in the best centre for the management of their condition.
Some may argue that by creating a cardiac arrest network, unstable post-ROSC patients are being transferred longer distances to reach definitive care, which may be detrimental. However, Spaite et al (2009) found no correlation between longer transport times and lower survival in a population of 15 559 patients post-ROSC or in cardiac arrest, concluding that bypassing local EDs to reach specialist centres is feasible, safe and effective.
Furthermore, the development of advanced prehospital services such as doctor-led critical care teams and critical care paramedics within local ambulance services means patients are able to receive advanced clinical interventions in the prehospital setting such as sedation and paralysis, advanced airway management and intravenous vasopressor support (Walmsley and Turner, 2015), which allow them to be safely transported longer distances to reach more appropriate care.
Conclusion
In this small pilot study, we have shown that a delayed post-ROSC ECG is more accurate in predicting a causative coronary occlusion than a prehospital post-ROSC ECG.
While the use of the prehospital post-ROSC ECG to triage patients between their local ED and PCI services remains common practice, this study questions whether this is a reliable method of triage and whether the prehospital phase is still too early to know what definitive treatment a patient may require. This raises the question of whether all post-ROSC patients should be conveyed to an ED with PCI facilities, known as regional cardiac arrest centres. Further research is required to fully understand the impact that time has on the post-ROSC ECG and the management of this category of critically ill patient.