We performed a retrospective study including patients admitted within 24 hours of CA to the intensive care department (ICD) of Faro Hospital in southern Portugal, between 1 January 2014 and 31 December 2018. This department consists of a general ICU and a general intermediate care unit, served by a shared medical team. The ICD also provides a permanent in-hospital emergency program staffed by at least one doctor and one nurse, which responds promptly to in-hospital emergency situations (including CA) following direct telephone activation.

The ICU has a normothermia protocol to be implemented in CA patients who are comatose after ROSC, aiming for a target central (esophageal) temperature below 37°C (Supplementary Appendix). This protocol is contraindicated in some situations: patients with Glasgow coma scale (GCS) >10, severe hemodynamic instability, septic shock, cranial trauma, major bleeding or pregnancy. In accordance with the European Society of Intensive Care Medicine and the European Resuscitation Council,9,10 neurologic prognosis is assessed 72 hours after the index event, using the GCS and electroencephalogram (EEG), complemented by brain computed tomography (CT).

It should be noted that, in our hospital, patients who have suffered CA due to ST-elevation acute myocardial infarction or ventricular arrhythmias due to cardiological causes are generally admitted directly to the coronary intensive care unit in the cardiology department, and undergo emergent coronary angiography. Other patients are first admitted to the general ICU, and if the cause of CA is presumed to be myocardial infarction, in view of the patient's clinical course and if there is no other obvious cause, the case is discussed with the cardiology department and coronary angiography is performed whenever clinically indicated.

Withdrawal of life-sustaining treatment or ‘do not resuscitate’ decisions are determined after health care team discussion (including ICU doctors and nurses), according to neurologic prognosis and comorbidities. Some successfully resuscitated patients develop post-CA shock, which is defined as the need for continuous vasopressor support to maintain mean arterial pressure above 60 mmHg for more than six hours following ROSC, despite adequate fluid loading.3

Clinical information was extracted from electronic clinical records, using the clinical software SClinic® and on a running database using B-ICU.CARE®. An initial screening was performed by searching for the diagnosis “arrest”, applying the statistical functionality of B-ICU.CARE. All patients aged 17 years or more were included. Patients with respiratory arrest only, and those with cardiac arrest already admitted to the ICD for other causes, were excluded. Patients admitted 24 hours after CA were also excluded, mainly those transferred from the coronary intensive care unit or from other hospitals. Data were collected and organized following the Utstein Style guidelines11,12 whenever possible.

Demographic factors, CA characteristics, immediate and in-hospital approach (including electrocardiogram (ECG), brain CT, EEG, application of the normothermia protocol, vasopressors, inotropic agents and invasive mechanical ventilation [IMV]), CA recurrence, epileptic seizures, mortality and neurologic outcomes were analyzed. Myoclonus was included in the epileptic activity recorded since it is associated with a worse prognosis (especially if starting within 48 hours of ROSC), although some patients may survive with good outcome.10 Severity indices were recorded, based on the Simplified Acute Physiology Score (SAPS) II score and Sequential Organ Failure Assessment (SOFA) score in the first 24 hours and SOFA score at ICU discharge. Overall length of ICU stay was also reported for each patient. Regarding vasopressor support and IMV, the duration of support is presented in absolute values and also indexed to the total length of ICU stay, to minimize the effect of early mortality on the analysis. Our results are compared between subgroups of gender, location of arrest and clinical outcomes. Neurologic outcome was assessed by cerebral performance category (CPC) at hospital discharge and the best CPC and GCS recorded during ICU stay were reported. Survival outcome was assessed at hospital discharge, 28 days after discharge, at six months and at one year of follow-up.

Categorical variables were described as absolute (n) and relative frequencies (%). Medians and percentiles were used for continuous variables. Mann-Whitney nonparametric tests were used as appropriate to test hypotheses concerning continuous variables, considering normality assumptions and the number of groups compared. The chi-square test and Fisher's exact test were used, as appropriate, to test hypotheses concerning categorical variables.

Univariate and multivariate logistic regression modeling was used to elucidate the factors associated with mortality and neurologic outcomes. The significance level used was 0.05. Statistical analysis was performed using IBM SPSS version 24.0.

During the five-year period, 187 patients were admitted to the ICD within 24 hours of successful cardiopulmonary resuscitation. Table 1 describes the documented study population and CA characteristics. The median age was 67 years and 121 patients (65%) were male; 61% of the CAs occurred in an in-patient setting and 80% were witnessed. However, bystander BLS was only performed in 73% of all patients. The first monitored rhythm was non-shockable in 87% of patients and the median time between CA and ROSC (downtime) was 10 min. Presumed cardiac causes accounted for 31% of CAs (acute coronary syndrome in 33% and acute or decompensated heart failure in 32%).

Table 2 describes the in-hospital management of CA patients, 90% of whom were admitted to the ICU. The others rapidly recovered consciousness, no longer requiring IMV, and were admitted directly to the intermediate care unit. During hospitalization, the median duration of IMV was two days, and one day for vasopressor support (including both survivors and non-survivors). Norepinephrine was the main vasopressor drug used in these cases. The normothermia protocol was applied in 53% of patients, but 75 patients to whom it was not applied had some predefined contraindications, which gives a rate of 93% of application for eligible patients. Patients with suspected acute coronary syndrome or other cardiac cause that precipitated the CA (such as ventricular arrhythmia) were discussed with the cardiology department and transferred whenever this was considered to be the best strategy (10%). Nineteen percent of patients had epileptic activity and neurologic lesions were the main cause of post-CA dysfunction in 43%. Recurrence of CA with successful resuscitation (during transportation or in-hospital) occurred in 18%, with a maximum of six CAs in a single patient. The median length of ICU stay was three days, but it should be noted that extubated patients were transferred to the intermediate care unit to maintain monitored care. In-hospital mortality was 63% (118 patients), 45% of which was associated with withholding or withdrawal of life support. Fifty-five percent of the 69 patients who survived were classified as CPC 1.

Comparisons between the genders can be seen in Supplementary Table A. In our population, only SOFA score at admission reached a statistically significant difference, with higher median values in males (11 vs. 10; p=0.007). There were no statistically significant differences in mortality (62% vs. 65%, p=0.668) or neurologic outcomes (70% male patients with CPC 1-2 at discharge vs. 65% in females; p=0.203).

Supplementary Table B shows comparisons between OHCA and IHCA. Statistically significant differences were observed in all variables except for gender (p=0.810), downtime (p=0.069), duration of IMV (p=0.132) and of vasopressor support (p=0.295), epileptic activity (p=0.200), GCS at ICU discharge (p=0.874) and neurologic outcome (p=0.887). There was also no statistically significant difference in mortality (63% in both groups, p=0.984).

Regarding OHCA, a smaller proportion of these patients had immediate initiation of BLS (52% vs. 87%, p<0.001), with a lower GCS after ROSC (3.8 vs. 5.0; p=0.003), while the ward was the main location of arrest in IHCA (22%). These patients less often had shockable rhythms at initial assessment (6% vs. 25%, p=0.001), and hence less presumed cardiac causes of CA (22% vs. 44%, p=0.002). Regarding the clinical conditions that led to hospital admission, they also had higher severity indices at ICU admission (p<0.001) and shock was the main dysfunction after CA (40% vs. 23%, p=0.016).

Mortality is analyzed in Table 3. Of the 118 non-surviving patients (63%), 64% were male and 61% suffered IHCA. The normothermia protocol was applied in 87% of patients without contraindications, and 19% experienced recurrence of CA with successful resuscitation. Nineteen percent presented epileptic activity. Patients who died during hospitalization less often had bystander BLS (68% vs. 83%, p=0.027), and had more frequent non-shockable rhythm at initial assessment (93% vs. 77%, p=0.001) and longer median downtime (10 vs. 8 min; p=0.052). All severity indices were higher (p<0.001) and these patients had longer indexed durations of IMV (1.0 vs. 0.50; p<0.001) and vasopressor support (0.6 vs. 0.3; p<0.001) indexed to total length of ICU stay. Survivors had higher GCS after ROSC (5 vs. 3; p=0.003) and longer median hospital length of stay (five vs. two days; p<0.001).

After multivariate analysis, non-immediate initiation of BLS (odds ratio [OR] 6.040 [1.533-23.801], p=0.010), higher SAPS II score (OR 1.105 [1.048-1.165], p<0.001) and higher indexed duration of vasopressor support (OR 8.375 [1.722-40.726], p=0.008) were independent predictors of in-hospital mortality. Shockable rhythms (OR 0.129 [0.024-0.791], p=0.018) were independently associated with improved survival.

Median follow-up was seven days, with a mean of 235 days, since all patients were included and most of them died in-hospital. Mortality at 28 days and 12 months after discharge was 67% and 72%, respectively. Figure 1 reveals cumulative survival during follow-up, indicating higher mortality within six months of CA. Comparison of cumulative survival according to first monitored rhythm and initiation of BLS is shown in Figure 2.

Figure 1.

Kaplan-Meier curve revealing cumulative survival during follow-up.

(0.13MB).

Figure 2.

Cumulative survival comparing the first monitored rhythm of cardiac arrest with log rank test p=0.009 (left) and prompt initiation of basic life support (BLS) with log rank test p=0.020 (right).

(0.17MB).

Of the 69 survivors, 31% had significant neurologic dysfunction (CPC 3 or 4) at discharge. Comparing CPC 1 or 2 patients with CPC 3 or 4 patients (Table 4), statistically significant differences were found concerning IMV duration (two vs. eight days, p<0.001), GCS (15 vs. 10, p<0.001), SOFA score (3.5 vs. 5, p=0.008) and length of ICU stay (four vs. 12 days, p<0.001). Also, CPC 3 or 4 patients had a higher incidence of epileptic activity (38% vs. 9%, p=0.015) and lower rates of post-CA shock (95% vs. 34%, p=0.003).

On multivariate regression analysis using GCS at ICU discharge, length of ICU stay, and SOFA at ICU discharge as independent variables, GCS (OR=0.727 [0.569-0.927]) and length of ICU stay (OR 1.153 [1.027-1.295]) remained statistically significantly associated with CPC. Although post-CA shock was associated with a better outcome in univariate analysis, multivariate analysis showed that, after adjusting for epileptic activity and IMV, it was no longer statistically significantly associated with CPC (OR=3.843 [0.700-21.100], p=0.121).