Postresuscitation care with mild therapeutic hypothermia and coronary intervention after out-of-hospital cardiopulmonary resuscitation: a prospective registry analysis

The current study includes data from adult patients with OHCA, which was defined as the absence of signs of circulation and concomitant appearance of unconsciousness, apnea or gasping and pulselessness in accordance with the Utstein-style template [20]. After successful CPR, all patients were admitted to a hospital.

Patients with definite signs of death, patients with do-not-attempt-resuscitation orders, and patients presenting with injuries that were obviously associated with no chance of survival were excluded. In addition, patients initially resuscitated by basic life-support teams who subsequently did not receive any treatment from the advanced cardiac life-support team because the emergency physician decided to stop CPR due to pre-existing illness, medical history or after interviewing close relatives concerning the patient's supposedly negative intention for resuscitation were also excluded.

We recently evaluated the establishment of the GRR to record both OHCA and in-hospital cardiac arrest [19]. The database has been proven congruent with the Utstein style, and control mechanisms have optimized data collection and data quality. The GRR is currently the largest resuscitation registry launched in Germany. The dataset was approved by the German Society of Cardiologists and Internal Medicine. The registry was also accepted and recommended both by the German Resuscitation Council and the German Society of Emergency Physicians, and in addition it represents the German database within the European Registry of Cardiac Arrest provided by the European Resuscitation Council.

The GRR is a prospective web-based database to register all emergency physician-related resuscitation efforts, as previously reported by our group [19].

The Preclinical care dataset originated from the Utstein-style templates, aiming at documentation of resuscitation efforts with 118 variables - in particular, prehospital logistic issues, presumed etiology, resuscitation therapy and patient's initial outcome.

The Postresuscitation care dataset also originated from the Utstein-style templates, aiming at documentation of postresuscitation efforts. The Postresuscitation care dataset includes demographic data, ECG, temperature management, cerebral performance category (CPC), hemodynamic variables, blood glucose level, circulatory support and diagnostic procedures (for example, chest X-ray scan, ultrasound, computer tomography, and survival at both 24 hours and at hospital discharge) [21]. In the present study, we focused on MTH (body temperature of 32 to 34°C) and on primary PCI performed within 24 hours after ROSC, although further details concerning MTH (for example, type of induction, type of cooling device, surface vs. intravascular, target temperature) and coronary intervention (for example, TIMI flow, type of stents, type of infarct, event-to-needle-time) were not registered within the GRR. Data from the Postresuscitation care dataset were reported to the resuscitation registry by the hospitals themselves. These data were also allocated to the respective Preclinical care dataset.

In accordance with the Utstein definition, initial resuscitation success with ROSC was defined as a palpable pulse for more than 20 seconds.

The postresuscitation outcome was defined as 24-hour survival and neurological outcome at hospital discharge, since both endpoints represent variables within the Utstein style [20, 22]. In the revised Utstein definitions from 2004 [20], 24-hour survival was downgraded from core to supplementary compared with the original 1991 version [22]. Nevertheless, both endpoints are still core variables within the GRR dataset.

Assessment of the neurological status was based on the CPC [23]. The performance categories are defined as follows: CPC 1, conscious and alert with normal function or only slight disability; CPC 2, conscious and alert with moderate disability; CPC 3, conscious with severe disability; CPC 4, comatose or in a persistent vegetative state; and CPC 5, certifiably brain dead or dead by traditional criteria. The best CPC score achieved at hospital discharge was used for calculation. A CPC score of 1 or 2 represents favorable functional neurological recovery because patients with these scores have sufficient cerebral function for independent activities of daily living, and was therefore defined as good neurological outcome. We state that most relevant endpoints are neurological outcome and survival status at hospital discharge. A CPC score of 3, 4, or 5 reflects unfavorable functional neurological recovery.

Out of 584 patients, 154 patients (26%) received PCI and 430 patients (74%) did not. In patients without PCI, MTH was associated with increased 24-hour survival (unadjusted OR 7.02 (3.7 to 13.3), P < 0.001) and good neurological outcome (unadjusted OR 2.21 (1.23 to 3.96), P < 0.01).

Binary logistic regression analysis confirmed that MTH (adjusted OR 8.24 (4.24 to 16.0), P < 0.001), bystander CPR (adjusted OR 3.25 (1.84 to 6.76), P < 0.001) and VF/pVT as first ECG rhythm (adjusted OR 1.96 (1.22 to 3.16), P < 0.01) were associated with improved 24-hour survival, whereas systemic thrombolysis was associated with worse chance of 24-hour survival (adjusted OR 0.52 (0.28 to 0.98), P < 0.05).

With respect to neurological outcome, regression analysis further revealed that MTH (adjusted OR 2.13 (1.17 to 3.90), P < 0.05), age <60 years (adjusted OR 2.25 (1.24 to 4.07), P < 0.01) and VF/pVT (adjusted OR 2.27 (1.26 to 4.09), P < 0.01) were independent factors for good neurological outcome at hospital discharge. Detailed results are presented in Table 2 and in Tables S1 and S2 in Additional file 1.

Out of 584 patients, 179 patients (31%) received MTH. In normothermic patients (n = 405; 69%), PCI was associated with increased 24-hour survival (unadjusted OR 5.06 (2.63 to 9.71), P < 0.001) and good neurological outcome (unadjusted OR 11.31 (6.25 to 20.47), P < 0.001).

Binary logistic regression analysis revealed that PCI (adjusted OR 4.46 (2.26 to 8.81), P < 0.001), bystander CPR (adjusted OR 2.50 (1.34 to 4.69), P < 0.01) and VF/pVT as first ECG rhythm (adjusted OR 2.15 (1.33 to 3.48), P < 0.01) were associated with improved 24-hour survival.

PCI (adjusted OR 10.81 (5.86 to 19.93), P < 0.001) and age <60 years (adjusted OR 2.04 (1.10 to 3.78, P < 0.05) were independent predictors of good neurological outcome. Detailed results are presented in Table 2 and in Tables S3 and S4 in Additional file 1.

To evaluate the combination of hypothermia and coronary intervention, we again performed unadjusted Fisher exact tests followed by adjusted regression analysis of the total group of 584 patients.

According to the Fisher exact test, MTH was associated with increased 24-hour survival (unadjusted OR 7.6 (4.32 to 13.37), P < 0.001) and good neurological outcome (unadjusted OR 1.83 (1.23 to 2.74), P < 0.01).

Following adjustment of these results by binary logistic regression, MTH (adjusted OR 7.50 (4.12 to 13.65), P < 0.001), PCI (adjusted OR 3.88 (2.11 to 7.13), P < 0.001), age <60 years (adjusted OR 1.79 (1.14 to 2.82), P < 0.05), bystander CPR (adjusted OR 2.27 (1.26 to 4.08), P < 0.01), and VF/pVT as first ECG rhythm (adjusted OR 1.81 (1.17 to 2.80), P < 0.01) were associated with improved 24-hour survival.

In terms of good neurological outcome at hospital discharge, PCI (adjusted OR 5.66 (3.54 to 9.03), P < 0.001), age <60 years (adjusted OR 2.87 (1.83 to 4.49), P < 0.001), witnessed OHCA (adjusted OR 1.83 (1.02 to 3.27), P < 0.05), and VF/pVT as first ECG rhythm (adjusted OR 1.61 (1.01 to 2.54), P < 0.05) were found to be independent predictors, whereas MTH (adjusted OR 1.27 (0.79 to 2.03), P = 0.33) did not improve outcome statistically significantly. Detailed results are presented in Table 2 and in Tables S5 and S6 in Additional file 1.

In OHCA patients with VF/pVT as the first ECG rhythm, we found an increased 24-hour survival and a better neurological outcome at hospital discharge. An initial shockable ECG rhythm thus had a substantial influence on patient outcome. This result is in agreement with other studies [24‐27]. The prevalence of VF/pVT as first rhythm has decreased in recent years, however, from 34% to 21% dependent on witnessing cardiac arrest and bystander CPR [28‐32], but plausible explanation has not yet been found for this observation.

Although young adults are a minority among patients suffering from OHCA, these victims suffer from this catastrophic event when they are in a very active phase of life with a long life expectancy. Our registry analysis confirmed that patients aged <60 years had a better outcome in terms of good neurological outcome. We assume that most of the younger patients do not suffer from significant co-morbidities, and that the motivation of the medical team may be highest in these younger adults to make greatest efforts on any therapeutic option within the postresuscitation care period [33].

MTH is currently a mainstay of postresuscitation care [6‐9, 34]. Most clinical investigations, however, mainly included patients with VF/pVT as the first ECG rhythm, reporting good neurological outcome in this subset of patients. In our database, MTH was associated with increased 24-hour survival. Interestingly, these favorable effects were observed irrespective of the initial ECG rhythm; the 24-hour survival rate was 92% in patients with VF/pVT and 90% in those with an initial nonshockable rhythm. More interestingly, 24-hour survival regarding hypothermia is rather questionable since the cooling therapy itself was still ongoing. Although 24-hour survival is mentioned as a core variable in the original Utstein style, it is still recommended as supplementary data in the Utstein update [20], and therefore should be reported as a clinical endpoint in these kinds of resuscitation registry analyses. Being aware of this limitation, we further analyzed survival and the proportion of patients with good neurological outcome at hospital discharge as the more relevant primary endpoint. MTH was associated with increased good neurological outcome at hospital discharge in patients without PCI.

For the treatment of noncardiac arrest patients with myocardial ischemia, PCI is currently considered the treatment of first choice. But acute myocardial ischemia subsequent to coronary artery occlusion is also a common pathological correlate in cardiac arrest patients [35]. PCI has also been suggested to result in an increased chance of hospital survival in cardiac arrest patients suffering from myocardial ischemia. Gorjup and colleagues reported that OHCA patients with myocardial infarction may benefit from primary PCI similarly to noncardiac arrest patients with otherwise nonlethal myocardial infarction [36]. We are not, however, aware of any prospective randomized trial investigating the effect of primary PCI performed immediately after hospital admission in OHCA patients with successful CPR. Some smaller studies, however, have demonstrated beneficial effects of PCI in cardiac arrest patients [14, 16, 37].

In our registry analysis, PCI was an independent predictor of an increased chance of 24-hour survival and of good neurological outcome at hospital discharge. Our results revealed that the proportion of patients with CPC 1 or CPC 2 at hospital discharge increased from 10% to 54% in the group of normothermic patients if PCI was performed within 24 hours after ROSC. Interestingly, PCI was associated with increased 24-hour survival from 56% (159 out of 286 patients without PCI) to 88% (45 out of 51 patients with PCI) even in the subgroup of patients with an initial nonshockable rhythm. Patients with poorer baseline conditions (initial nonshockable rhythm) may thus also benefit from coronary intervention.

Our data may therefore support the hypothesis that a standardized postresuscitation care bundle, potentially including a liberal decision for coronary intervention, should be offered to most OHCA patients with successful resuscitation and hospital admission [17]. In addition, it should be noted that a typical history of coronary artery disease or ECG changes typical for ST-elevation myocardial infarction may be absent in up to 57% of OHCA patients, where coronary angiography revealed pathological findings with therapeutic options [35, 38]. Further, clinical symptoms such as chest pain or risk factors often are lacking in the setting of OHCA. Comparable with severe trauma patients, therefore, prompt transfer after successful resuscitation to specialized hospitals/cardiac arrest centers may allow patients to benefit from this invasive therapeutic option [39]. This hypothesis is further supported by the findings of Dumas and colleagues, who recently demonstrated in a multivariable analysis of 435 prospectively registered patients that successful immediate coronary angioplasty was independently associated with improved hospital survival in patients with or without ST-segment elevation [18]. The high incidence of coronary lesions in the Parisian Region Out of Hospital Cardiac Arrest cohort study confirmed previous findings that link acute coronary syndrome and OHCA. Coronary plaque rupture or erosion, fragmentation, and embolization of thrombus were identified as factors able to trigger cardiac arrest. Similar rates have been noted in studies based on postmortem examination of patients with OHCA [40] or angiographic data [41].

Seventy-three patients received both MTH and PCI, irrespective of first ECG findings. Ninety-six per cent of these patients survived 24 hours and 49% were discharged with CPC 1 or CPC 2 compared with 54% and 11% of patients without any therapeutic procedure, respectively. The proportion of patients with good neurological outcome at hospital discharge was thus much higher in patients receiving both forms of treatment compared with normothermic patients without PCI. We therefore suggest that a therapeutic bundle of hypothermia and coronary intervention in addition to standard critical care may be beneficial in selected successfully resuscitated patients. We are not aware of any randomized controlled study investigating the therapeutic approach of a combination of hypothermia and coronary intervention. A few small clinical studies including historical control groups and case reports, however, have recently indicated that the combination may be feasible and may indeed be associated with benefits for the individual patient [14, 17, 42, 43].

Considering the combination of MTH and PCI, we performed binary logistic regression analysis including all patients (n = 584). Both MTH and PCI were independently associated with increased 24-hour survival (MTH adjusted OR 7.50 (4.12 to 13.65), and PCI adjusted OR 3.88 (2.11 to 7.13)). In terms of neurological outcome at hospital discharge, however, only PCI was independently associated with increased chance of good outcome (adjusted OR 5.66 (3.54 to 9.03)). Although MTH was significantly associated with good neurological outcome in 44% and 21% of patients with VF/pVT and non-VF/pVT in contrast to 26% and 15% of normothermic patients, respectively (unadjusted OR 1.83 (1.23 to 2.74), P < 0.05), statistical significance was not reached in the subsequent binary logistic regression analysis (adjusted OR 1.27 (0.79 to 2.03), P = 0.327). These data are in some agreement with most of the recent studies demonstrating either a trend or a significant benefit for MTH in patients with VF/pVT and non-VF [43]. Very importantly, most of the published data did not undergo adjustment for multiple independent predictors, thus interpretation and comparison with our results is difficult. Our results may thus have a considerable heuristic value, and therefore additional international resuscitation registries should be encouraged to consider the same question with their data.

The GRR is based on voluntary participation of emergency services and hospitals. The registry cannot provide a complete picture of the total Germany-wide incidents of sudden cardiac arrest and resuscitation attempts at all. There is thus some degree of uncertainty with regard to representativeness of the register, but the GRR still reflects current practice throughout the country in both rural areas and big cities with different emergency medical system patterns. Nevertheless, voluntary registration and documentation by 23 medical emergency systems providing data for both the Preclinical care and Postresuscitation care datasets is probably associated with the risk of inclusion bias in the present study. But this problem is related to most of the published registries. For instance, the National Registry of Cardiopulmonary Resuscitation was started in 2000 as an international database of in-hospital resuscitation events worldwide, but it covers much less than 10% of potential hospitals. Further, the recent Parisian Region Out of Hospital Cardiac Arrest registry involved 68% patients with VF as initial rhythm, suggesting that there was also a highly selected cohort studied and a reasonable inclusion bias [18].

A total of 584 patients could be included in the present study, which may look like a rather small group; the reason for this number, however, was strict limitation to patients with complete Preclinical care and Postresuscitation care datasets, which resulted in a huge number of excluded patients. One of the main limitations of the present study is the selection bias for patients subjected to coronary intervention and hypothermia. Choice of postresuscitation therapeutic management was based on individual in-hospital postresuscitation treatment algorithms, so a bias in the selection of patients receiving any therapeutic option is highly likely. In addition, a substantial number of in-hospital variables that could influence survival and neurological outcome were not available in the database. These include body temperature management (for example, type of cooling induction, type of cooling device, surface vs. intravascular, target temperature), laboratory test levels, medications used, and details of revascularization procedures (for example, 'Thrombolysis In Myocardial Infarction' (TIMI) flow, type of stents, type of infarct, event-to-needle time). In addition, the present study does not differentiate between primary patient transports by the emergency medical system to the participating hospital or secondary transfer from one hospital to a hospital providing 24-hour coronary intervention services. Finally, our registry analysis is obviously limited by the nonrandomized and observational design, which contained no control group.