Table 1 shows the patient characteristics and the results of univariate survival analysis.

The mean age was 47.5 months (range, 1-207 months) and the mean body weight, 16.2kg (range, 2-93kg). There were no differences in mortality related to age or sex. The mortality rate was lower among children weighing less than 10kg than among those whose weight was 10kg or over.

In all, 79.5% of the patients had underlying diseases. The mortality rate was lower in children with a history of heart disease and was higher in those with oncological diseases.

In 76.5% of the patients, the available data included functional and neurological status prior to CA, assessed using the pediatric overall performance category and PCPC scores, and 94.8% of them had good neurological status (PCPC score of 1 or 2). Mortality in the 8 patients with severe neurological disorders (previous PCPC greater than 2) was 100%, significantly higher than that of the remainder of the patient population.

In all, 11% of the patients had experienced CA previously. The mortality rate among these children was somewhat higher than that among the patients who had not, but this difference was not significant.

At the time of CA, 94.3% of the patients were being monitored, 92.9% required mechanical ventilation, and 64.6% were receiving vasoactive drugs. The mortality rate was significantly higher among the children receiving vasoactive agents.

Table 2 summarizes the univariate analysis of the relationship between the characteristics of CA and CPR and mortality.

Cardiac and respiratory diseases were the main causes of CA. However, the mortality rate was significantly higher among the patients with neurological disorders and sepsis.

The majority of CA episodes took place in the pediatric intensive care unit. The mortality rate associated with CA that occurred in the emergency department was higher than that recorded for any other area of the hospital, although the differences were not statistically significant.

There were no differences in mortality between episodes that began as respiratory arrests and those presenting as CA. Bradycardia was the most common electrocardiographic rhythm. Only 6.8% of the patients had shockable rhythms (ventricular fibrillation or pulseless ventricular tachycardia). The mortality rate among children with shockable rhythms was lower (36.4%), but the differences were not statistically significant. In 89.7% of the patient population, CPR was initiated less than 1min after the diagnosis of CA.

Mortality was significantly higher among the patients who required adrenaline, bicarbonate, and volume expansion (at least 10mL/kg during CPR). The mortality rate increased in relation to the duration of CA.

The multivariate logistic regression analysis (Table 3) revealed that hematologic/oncologic diseases, administration of continuous infusion of vasoactive drugs prior to the arrest, the need for bicarbonate administration, volume expansion during CPR, and CPR duration greater than 20min were risk factors associated with mortality.

The study performed previously in Spain involved 283 patients with CA and/or respiratory arrest occurring both in hospital and out of hospital.3 One hundred thirty-five children had in-hospital CA (71.1% of them in the pediatric intensive care unit). Survival was 25.9%, with a statistically significant difference (P<.001) between both studies. Duration of CPR greater than 20min was the factor most strongly associated with mortality.3

The mortality rate was lower among the children weighing less than 10kg compared with that in those weighing more. Other authors have found that the survival rate was higher among younger children than in older patients.10

Previous neurological status was normal in most patients. Although mortality was 100% among the children with a previous neurological disorder (PCPC greater than 2), the small number of patients makes their inclusion in statistical comparisons unviable.

In addition, the mortality among patients who received vasoactive drugs was significantly higher. This circumstance probably only reflects the fact that they were more seriously ill, an observation reported in the previous Spanish study3 and by other authors.8,11

The mortality rate was higher among patients with hematologic/oncologic diseases, a finding that had been reported in other studies.9,11

The most common causes of CA were respiratory and cardiac diseases. However, as observed in other studies, the causes most strongly associated with mortality were sepsis and neurological disorders. This finding highlights the importance of establishing measures for the early diagnosis and treatment of sepsis.21

Most of the episodes of CA occurred in the pediatric intensive care unit. The mortality associated with CA that took place in the emergency department was higher than the rates corresponding to all other areas of the hospital, although the differences were not statistically significant. For this reason, it is important that a program be developed for the prevention and treatment of CA throughout the entire hospital.24 A number of studies have demonstrated the usefulness of creating rapid response teams to improve the prevention and treatment of CA in children.25,26

Bradycardia was the most frequently recorded electrocardiographic rhythm (82.2%), coinciding with the findings reported in other studies.7,12,27 This is one of the most important differences with respect to the previous Spanish study, in which bradycardia was the initial rhythm in only 38% of patients with CA, whereas asystole was the most frequently detected rhythm.3 This finding is very important and is probably due to the fact that CA is diagnosed earlier and, thus, CPR is initiated sooner, a measure that increases the possibilities of recovery.

The percentage of children with shockable rhythms was low, a finding reported in other studies in children. For this reason, although early defibrillation improves survival, in cases of CA in children, the algorithms and management strategies should take into account mainly the treatment of nonshockable rhythms.

In out-of-hospital CA, the time to initiation of CPR is one of the most important prognostic factors. In contrast, in in-hospital CA, especially that occurring in the pediatric intensive care unit, CPR is begun immediately. Thus, this factor is not a highly important prognostic variable, a finding observed in our study.

The mortality rate was higher among patients who required adrenaline, bicarbonate and volume expansion, as reported in other studies.1–3,10,11,28 In recent years, some authors have pointed out that agents such as adrenaline and bicarbonate may not be useful and may even increase mortality among CA patients.29 However, there is insufficient evidence to support this hypothesis. It is unsurprising that the mortality rates among patients who require a greater number of doses of adrenaline or bicarbonate, which are only administered in situations of prolonged or refractory CA, are higher than in those who recover rapidly with massage and ventilation.

Comparative studies carried out in adults with out-of-hospital CA have found that adrenaline initially increases survival, although it has no effect on survival to hospital discharge.30,31

With respect to bicarbonate, some studies have reported a slight improvement in survival, whereas others have found no difference.32,33

The most consistent prognostic factor in our studies and in those of other authors is the relationship between the duration of CA and mortality,3 a finding that indicates that any maneuvers and treatments should be undertaken as soon as possible.

In all, 7 patients with refractory CA were resuscitated by means of extracorporeal membrane oxygenation. Other studies have demonstrated that this technique can be an effective rescue treatment in children with in-hospital CA refractory to CPR, but that the availability of a multidisciplinary team, specifically trained and with the necessary material prepared for immediate use, is indispensable. This framework exists in few hospitals, mainly in those with a considerable volume of cardiac surgery.34

Our study has several limitations. Although it included a large number of hospitals from nearly every region of Spain, randomized sampling was not employed. Thus, our data cannot be extrapolated to the Spanish population as a whole. The comparison with the previous Spanish study should be analyzed with caution because there may be uncontrolled factors that could have influenced the results.