We appreciate the interest of and insightful comments by Dr Joffe et al concerning the relationship between extracorporeal cardiopulmonary resuscitation (E-CPR) and in-hospital pediatric cardiac arrest outcomes. During the 10-year study period, only 227 of the 3419 patients received E-CPR, with very few patients in several illness categories. 1 Although some E-CPR patients received shorter durations of CPR, 168 received E-CPR for >35 minutes. Of those who received >35 minutes of CPR, survival

more »

... hospital discharge was as follows: overall, 56 of 168 (33.3%); general surgical, 0 of 7 (0%); surgical cardiac, 40 of 104 (38.5%); medical cardiac, 15 of 34 (44.1%); general medical, 1 of 23 (4.4%); and trauma, 0 of 0. Survival to hospital discharge after >35 minutes of CPR without E-CPR was as follows: overall, 102 of 819 (12.5%); general surgical, 9 of 63 (14.3%); surgical cardiac, 26 of 156 (16.7%); medical cardiac, 22 of 145 (15.2%); general medical, 40 of 390 (10.3%); and trauma, 5 of 65 (7.7%). With the use of a 2-sided Fisher exact test (Stata 12.1) without adjustment for other potential confounding factors to compare those who received E-CPR and those who did not, only surgical cardiac (P<0.0001), medical cardiac (P=0.001), and all patients (P<0.0001) were statistically significant. Therefore, increased survival outcomes for medical and surgical cardiac illness category patients at longer CPR durations may be partially explained by E-CPR. Although previous studies in adults have also shown most impressive survival rates with E-CPR for illness categories of medical or surgical cardiac disease, 2,3 the low number of patients receiving E-CPR for several of the illness categories in our study precludes adequate assessment of this issue. Neurological outcome among survivors was not statistically different with or without E-CPR for CPR durations >35 minutes in any of the illness categories. Of the survivors who received E-CPR and >35 minutes of CPR, favorable neurological outcome was as follows: overall, 35 of 56 (62.5%); general surgical, 0 of 0; surgical cardiac, 28 of 40 (70.0%); medical cardiac, 7 of 15 (46.7%); general medical, 0 of 1 (0%); and trauma, 0 of 0. For the survivors who did not receive E-CPR and received >35 minutes of CPR, favorable neurological outcome was as follows: overall, 60 of 102 (58.8%); general surgical, 4 of 9 (44.4%); surgical cardiac, 17 of 26 (65.4%); medical cardiac, 16 of 22 (72.7%); general medical, 20 of 40 (50.0%); and trauma, 3 of 5 (60.0%). Both with and without E-CPR, the likelihood of favorable neurological outcome was impressive for survivors of >35 minutes of chest compressions. We also appreciate the interest of Dr Xue et al in our article and that limitations inherent to registry studies include the inability to completely adjust for severity of illness. Although the Get With The Guidelines-Resuscitation (GWTG-R) registry includes >400 variables and remains the most comprehensive database for CPR studies to date, it does lack physiological variables. We recognize this limitation, but we do not think that this changes the importance of the findings. Regardless of whether the patients who received longer CPR were sicker or had some cause of arrest the providers felt was reversible, we still find it notable that these patients reported to the registry with prolonged CPR survived with favorable neurological function. These patients would not have survived if their providers had stopped CPR at 15 to 20 minutes. The intent of our statement that these "children would presumably die without CPR" may have been misinterpreted. To clarify, the GWTG-R registry does exclude patients with preexisting do-not-resuscitate orders. The patients identified with do-not-resuscitate orders who were included in the study were children who had full resuscitation for their initial in-hospital cardiac arrest but whose family created an advanced directive subsequent to their successful resuscitation from their initial arrest. Only 111 such patients were included in this study. We also agree that quality of CPR is vitally important and was not quantitatively measured and reported in the registry. We presume that higher-quality CPR might result in an even greater proportion of patients surviving after prolonged CPR. Finally, we agree that for now CPR duration must be established on a case-by-case basis and that no simple solution for when to discontinue CPR currently exists. Therefore, we want providers to recognize that CPR duration alone does not dictate futility. The registry tells us that there are selected resuscitation circumstances, most likely personalized to the reason for the arrest and resources of the hospital, that can result in relatively good outcomes after prolonged in-hospital CPR events.