Higher Survival Rates Among Younger Patients After Pediatric Intensive Care Unit Cardiac Arrests
BACKGROUND. Age is an important determinant of outcome from adult cardiac arrests but has not been identified previously as an important factor in pediatric cardiac arrests except among premature infants. Chest compressions can result in more effective blood flow during cardiac arrest in an infant than an older child or adult because of increased chest wall compliance. We, therefore, hypothesized that survival from cardiac arrest would be better among infants than older children.
METHODS. We evaluated 464 pediatric ICU arrests from the National Registry of Cardiopulmonary Resuscitation from 2000 to 2002. NICU cardiac arrests were excluded. Data from each arrest include >200 variables describing facility, patient, prearrest, arrest intervention, outcome, and quality improvement data. Age was categorized as newborn (<1 month; N = 62), infant (1 month to <1 year; N = 105), younger child (1 year to <8 years; N = 90), and older child (8 years to <21 years; N = 207). Multivariable logistic regression was performed to examine the association between age and survival.
RESULTS. Overall survival was 22%, with 27% of newborns, 36% of infants, 19% of younger children and 16% of older children surviving to hospital discharge. Newborns and infants demonstrated double and triple the odds of surviving to hospital discharge from a cardiac arrest in an intensive care setting when compared with older children. When potential confounders were controlled, newborns increased their advantage to almost fivefold, while infants maintained their survival advantage to older children.
CONCLUSIONS. Survival from pediatric ICU cardiac arrest is age dependent. Newborns and infants have better survival rates even after adjusting for potential confounding variables.
- cardiac arrest
- cardiopulmonary resuscitation
- heart arrest
- intensive care
Age is an important determinant of outcome from adult cardiac arrests. Moreover, survival from cardiac arrests seems to be especially poor among extremely premature infants.1–4 Yet, there are no published studies supporting the effect of age on outcome from pediatric cardiac arrests among older infants and children.
Importantly, outcomes from cardiac arrest are substantially dependent on effective coronary, cerebral, and systemic perfusion during cardiopulmonary resuscitation (CPR).5,6 Animal studies indicate that coronary, cerebral, and systemic perfusion during CPR is markedly superior in young animals with compliant chest walls.7,8 However, several studies of pediatric in-hospital cardiac arrest suggest that age does not affect outcome from pediatric cardiac arrests after the neonatal period.9–12 Perhaps the small number of arrests evaluated in each of these studies limited their power to assess this issue.
Cardiac arrests occur in ∼1% to 4% of children admitted to PICUs.9,10,12,13 The frequency of these pediatric ICU cardiac arrests are ∼100-fold more than the frequency of out-of-hospital pediatric cardiac arrests.14 The limited data regarding these arrests are typically from a small series at a single institution.9–12,15 In addition, these studies have been limited by lack of consistent definitions and data collection. The American Heart Association National Registry of Cardiopulmonary Resuscitation (NRCPR) is a large multicenter database of in-hospital cardiac arrests with explicit standardized definitions and data reporting.16 In this report, we evaluate the effect of age on outcome from the first 464 consecutively reported index pediatric ICU cardiac arrests from the NRCPR. Because of studies suggesting that chest compressions may be more effective in an infant with a compliant chest wall than an older child or adult, we hypothesized that the outcomes would be better among infants than older children.
The Utstein-style data reporting guidelines for cardiac arrests and cardiopulmonary resuscitation have recently been reviewed and updated.17–19 The Utstein definitions and database elements are used consistently in the NRCPR registry database.
The NRCPR is a prospective, multicenter observational registry of in-hospital cardiac arrest and resuscitation. The current analysis reports on patients from 253 medical/surgical hospitals that provided ≥6 months of data from January 1, 2000, through December 31, 2002. Participating hospitals join the registry voluntarily and pay an annual fee for data support and report generation. On enrollment in NRCPR and annually, hospitals complete a form characterizing their facilities, staff, patients, and resuscitation services. Because the primary purpose of the NRCPR is quality improvement and the data are deidentified in compliance with Health Insurance Portability and Accountability Act, participating hospitals are not required to obtain institutional review board approval or individual informed consent. Nevertheless, this study was approved by the institutional review boards at the University of Arizona and the Children's Hospital of Philadelphia.
Specially trained NRCPR-certified research coordinators at each institution enter information for each cardiac arrest abstracted from hospital medical charts (including the patient's chart, cardiac arrest forms, and hospital paging system records) into a computer database that contains precisely defined variables. Data abstractors are required to successfully complete a certification examination consisting of multiple-choice questions and a mock scenario covering operational definitions and inclusion/exclusion criteria.
Case study methodology is used to evaluate data abstraction, entry accuracy, and operational definition compliance before acceptance of data transmission. Data are collected in 6 major categories of variables: (1) facility data, (2) patient demographic data, (3) pre-event data, (4) event data, (5) outcome data, and (6) quality improvement data (www.nrcpr.org). Explicit operational definitions have been generated for every data element. “Hospital type” is defined by the participating institution based on the primary patient population. “First recorded event rhythm” was defined as the first electrocardiogram rhythm documented at the time the patient required chest compressions and for those patients with unwitnessed/unmonitored arrests represents the first rhythm documented at the time a monitor arrives and is applied. “Index” events are defined as the patient's first cardiac arrest event during this hospitalization. Each patient is assigned a unique code, and no specific patient identifiers are transmitted to the central database repository, in compliance with Health Insurance Portability and Accountability Act regulations. Hospitals submit data on diskette or via encrypted, secure Internet transmission. A central data repository (Digital Innovation, Inc, Forest Hill, MD) facilitates data management and provides sites with quarterly reports summarizing their data and comparisons with grouped data. The American Heart Association provides oversight for the entire process of data collection, integrity, analysis, and reporting through staff, a science advisory board and an executive database steering committee.
Inclusion and Exclusion Criteria
All of the pediatric (<21 years of age) patients within an ICU setting who experienced pulseless cardiac arrest resuscitation were eligible for inclusion. A resuscitation event was defined as: a pulseless cardiopulmonary arrest requiring chest compressions and/or defibrillation that elicited a unit-based resuscitation response by ICU personnel and resulted in a resuscitation record. Pulseless cardiac arrest was defined as cessation of cardiac mechanical activity, determined by the absence of a palpable central pulse, unresponsiveness, and apnea. Events were excluded if the cardiac arrest began out of hospital, began out of the intensive care unit, involved a newborn in a delivery room or NICU, or was limited to a shock by an implanted cardioverter-defibrillator.
Age was categorized as newborn (<1 month; N = 62), infant (1 month to <1 year; N = 105), younger child (1 year to <8 years; N = 90), and older child (8 years to <21 years; N = 207), consistent with age groupings in the American Heart Association Pediatric Advanced Life Support Guidelines.20 Age <21 years was used to define the pediatric age as per the National Institutes of Health policy regarding research on pediatric subjects.21
The prospectively selected primary outcome measure was survival to hospital discharge.17–19 Consistent with the Utstein registry guidelines, only the first in-hospital “index” cardiac arrest and resuscitation were described and analyzed for patients with multiple arrests. Secondary survival measures included any return of sustained circulation (ROSC) >20 minutes, 24-hour survival, and survival to discharge with good neurologic outcome. Neurologic outcome was determined using adult cerebral performance category (CPC; for ages 18 to <21 years) and pediatric cerebral performance category (PCPC; for ages 0 to <18 years) scales, respectively.17,22,23 The CPC 1 is normal or mild cerebral disability, CPC 2 is moderate cerebral disability, category 3 is severe cerebral disability, category 4 is coma/vegetative state, and category 5 is brain death. The PCPC 1 is normal age-appropriate neurodevelopmental functioning, PCPC 2 is mild cerebral disability, category 3 is moderate cerebral disability, category 4 is severe disability, category 5 is coma/vegetative state, and category 6 is brain death. The pre-CPR neurologic categorization was based on historical data and chart review. Categorization at the time of discharge was determined by the discharge examination documentation. A single binomial variable was created to describe subject's neurologic status on discharge, independent of adult or pediatric classification. Good neurologic outcome was prospectively defined as CPC 1 or 2 for adults, the comparable PCPC of 1, 2, or 3 for children on hospital discharge, or no change from baseline CPC or PCPC.
The primary hypothesis that infants would have better survival to hospital discharge after an ICU cardiac arrest than older children was tested using χ2 with adjusted odds ratios (ORs). All of the reported P values are 2-tailed. Ninety-five percent confidence intervals (CIs) were calculated for the absolute difference in survival rates after cardiac arrest.
The 4 populations were compared by univariate analysis with regard to: patient demographics and hospital characteristics, preexisting conditions, interventions in place at time of arrest, immediate cause of arrest, first documented arrest rhythm, medication and nondrug interventions during event, and duration of CPR. Differences between children and adults for other data were analyzed with the Wilcoxon rank sum testing for ordinal variables, Fisher's exact test, χ2 test, or Student's t test, as appropriate.
Univariate analyses were conducted on all of the index cardiac arrests using Wilcoxon rank sum testing for continuous variables and χ2 analysis for dichotomous variables (Statistical Applications Software Version 9.1, SAS Institute, Cary, NC). Multivariable logistic regression analysis was performed on factors associated with survival in the univariate analysis (P < .05) to control for patient and event variables that may confound the relationship between age category and survival. Adjustment for multiple testing (eg, Bonferroni) was not used, because we were examining one relationship, age to survival to hospital discharge. ORs for survival and the 95% CIs were determined for factors that were independently associated with survival.
Data were checked for fidelity using a detailed periodic reabstraction process. NRCPR participants submitted randomly selected records each quarter. A random sampling of event records and corresponding NRCPR data sheets were reabstracted and reviewed for errors by NRCPR Scientific Advisory Board members. Mean error rates for all of the data were 2.4% ± 2.7%. Software data checks for out-of-range entries and a Web-based remediation program were developed to continuously remediate and support data integrity. Enrollment of new hospitals involves certification by testing accuracy of data abstraction before allowing data submission into the central database.
From January 1, 2000, through December 31, 2002, 487 pediatric index pulseless ICU cardiac arrest events were documented in 79 hospitals. Fourteen patients were eliminated, because the age data were inadequate (Fig 1). Nine patients (6 older children and 3 younger children) were also eliminated from further analysis because of inadequate postarrest data. Of the 464 remaining patients, 62 (13%) were newborns (<1 month), 105 (23%) were infants (1 month to <1 year), 90 (19%) were younger children (1 year to <8 years), and 207 (45%) were older children (8–21 years). Patient demographics did not differ significantly among age groups with respect to race, gender, or ICU size (Table 1). A total of 233 children (50%) survived the event, 174 children (37%) survived to 24 hours, 105 children (22%) survived to discharge, and 67 children (14%) survived with good neurologic outcome.
A total of 219 events occurred in 8 pediatric facilities, 31 in 15 adult facilities, and 214 in 56 mixed pediatric-adult facilities. Older children were more likely (P < .0001) to be admitted to mixed-type hospitals (59%) than any other age group (37% of newborns, 36% of infants, and 36% of younger children). The median number of total ICU beds of each participating hospital was 38 (range: 8–137). The geographic distribution of NRCPR facilities includes 35 states and the District of Columbia (Alaska, Arkansas, Arizona, California, Colorado, District of Columbia, Delaware, Florida, Georgia, Iowa, Illinois, Indiana, Kansas, Kentucky, Louisiana, Maryland, Minnesota, Missouri, North Carolina, Nebraska, New Hampshire, New York, Ohio, Oklahoma, Oregon, Pennsylvania, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Washington, Wisconsin, and West Virginia) and all US census regions.
Newborns were substantially more likely to be a cardiac surgical patient and more likely to have: active cardiac failure, a history of arrhythmia, an arterial catheter before cardiac arrest, and vasoactive infusions before cardiac arrest (Table 1). They were also more likely to have an event precipitated by hypotension and to be treated during the arrest with intravenous calcium, open-chest CPR, and extracorporeal life support (Table 2). Infants and younger children were more likely to have respiratory failure and pneumonia. However, there were no differences in respiratory support among age groups. Older children were more likely to have major trauma, a history of malignancy, and intravenous atropine treatment during the arrest.
This cohort consisted of patients who were pulseless at onset of resuscitation (n = 350) and patients who became pulseless while receiving chest compressions for bradycardia and/or hypotension (n = 114). At the onset of chest compressions, ventricular fibrillation was the first documented event rhythm in 39 patients (8% of patients who were initially pulseless or became pulseless during the event) and did not differ significantly among age groups. The proportion of pulseless ventricular tachycardia ranged from 0% in newborns to 6% in older children, with an overall incidence of 5% (Table 2). The median duration of CPR was shortest in children 8 to 21 years.
Survival to hospital discharge after cardiac arrest was higher in newborns and infants than younger and older children: 27% of newborns, 36% of infants, 19% of younger children, and 16% of older children (Table 3). Of the 128 children who survived the event but died before discharge, 79 children (60%) did not have withdrawal of technologic support, a “do-not-resuscitate” order, or brain death. Twenty-seven children had technological support withdrawn, 41 had a do-not-resuscitate order written, and 21 fulfilled brain death criteria. There were no significant differences among age groups with respect to withdrawal of technological support, do-not-resuscitate orders, or diagnosis of brain death.
When compared with older children, the crude OR for survival to hospital discharge by logistic regression for newborns is 2.0 (95% CI: 1.0–3.9), infants is 3.0 (95% CI: 1.7- 5.2), and younger children is 1.2 (95% CI: 0.6–2.3; Table 4). After adjustment by multivariable logistic regression analysis, the survival advantage for newborns increased to 4.9 (95% CI: 1.6–15.2) and for infants remained 3.1 (95% CI: 1.4–7.0) in comparison with older children; the younger children maintained a nonsignificant trend for survival in comparison with older children (Table 4). Effect of initial rhythm on survival was controlled using variables including initial pulseless rhythm, pulseless ventricular tachycardia, ventricular fibrillation, and asystole. The final regression of age on survival to discharge used 34 variables to control for potential confounding (Fig 2). When these same variables are used to control confounding on secondary end points, both newborns and infants continue to have survival advantages for ROSC, whereas only infants maintain survival advantage at 24 hours (Table 4). Children 1 to 8 years of age do not differ from their older counterparts at any end point. There were no significant differences in rates of survival to discharge with good neurologic outcome among age groups (Table 4).
These data from this large, multicenter in-hospital cardiac arrest registry establish that survival to discharge after a pediatric cardiac arrest in an ICU setting is age dependent. Newborns and infants were much more likely to survive after an ICU cardiac arrest than older children. This survival advantage was maintained even after adjusting for factors associated with better outcome (eg, arrests precipitated by ventricular arrhythmia or acute airway obstruction, extracorporeal life support during CPR, and short CPR duration), as well as those associated with worse outcome (eg, history of trauma, sepsis, and malignancy).24
The cohort of pediatric ICU arrests in this study was substantially larger than those in previous studies, and the data were prospectively collected using uniform Utstein-style definitions. These factors presumably improved the ability to identify a survival difference between age groups. In addition, the rigorous and uniform data reporting in this NRCPR database were intended to minimize information bias, observer bias, and surveillance bias.
Why did neonates and infants have better survival? This study cannot definitively answer the question “why.” Nevertheless, these data establish that age is an independent factor associated with outcome and, therefore, stimulates hypotheses for future study. Our initial hypothesis was that neonates and infants would have better survival because of more effective cerebral, coronary, and systemic perfusion during CPR compared with older children with less compliant chest walls. However, this database does not include hemodynamic data during CPR. Therefore, the relationship between these better outcomes and improved hemodynamics during CPR must remain speculative.
Our data also demonstrate that the children in these 4 age groups differed with regard to many other characteristics that could be associated with differential outcome. For example, the newborns were much more likely to have prearrest factors of cardiac surgery, cardiac failure, vasoactive infusions, and arterial catheters. Newborn cardiac arrest was more commonly precipitated by hypotension and more commonly treated with intravenous calcium, open-chest CPR, and extracorporeal life support. Clearly the younger age groups differed in many ways from the older groups: anatomically, physiologically, pathophysiologically, etiologically, and with respect to intra-arrest therapies. Although the age-related differences in outcome remained after controlling for potentially confounding factors, younger age may be a surrogate for other factors that confer a survival advantage.
In the original clinical investigation by Kouwenhoven8 describing closed-chest cardiac massage, the first patients to receive this novel therapy were children, because the author thought it would only be effective in small children with compliant chest walls.25,26 Furthermore, Dean et al demonstrated better CPR hemodynamics and chest wall recoil in immature animals with compliant chest walls compared with older animals with less compliant chest walls.7 The improved CPR hemodynamics are presumably related to better transmission of the external chest compressions to the cardiac chambers resulting in better stroke volumes and perhaps also to improved venous return associated with better chest wall recoil during the relaxation phase. The net result is superior cerebral, myocardial, and systemic blood flow in younger animals during CPR.7
Previous single-institution studies of pediatric ICU cardiac arrests were small and perhaps underpowered to evaluate associations of age with outcome (Table 5). The multicentered study by Slonim et al10 had a different definition of cardiac arrest (chest compression for >2 minutes), different age group definitions (0 to <1 month, 1 month to <1 year, 1 to <12 years, and ≥12 years), and less newborns (n = 25; 12%) in their cohort.10 They did not include patients with CPR in progress at the time of ICU admission who did not survive >2 hours or documented pulseless cardiac arrests with chest compression duration <2 minutes. In their study, survival to hospital discharge was 20% for patients <1 month, 13.5% for patients 1 month to 1 year, 15% for patients 1 to <12 years, and 5% for patients ≥12 years. Neither mean ages (survivors 30.8 ± 8.8 months; nonsurvivors 48.6 ± 4.8 months) nor distribution of ages were significantly associated with survival. The only factors related to survival included diagnoses (trauma and “other”), resuscitation duration, and underlying physiologic dysfunction. The physiologic dysfunction profile (categorized by Pediatric Risk of Mortality Score) was most highly associated with outcome. Unfortunately, direct comparison of outcomes from the Slonim et al10 cohort and the NRCPR cohort is limited, because the Slonim et al10 study did not use the Utstein style for data reporting, and the NRCPR database does not include physiologic data and Pediatric Risk of Mortality Scores.
The rate of initial ROSC and 24-hour survival in the present study is remarkably similar to previous in-hospital pediatric cardiac arrest data over the last 2 decades.9–12,27,28 Nevertheless, the survival-to-discharge rate is substantially superior to most previous studies (Table 5). Perhaps recent improvements in postresuscitation care have resulted in better outcomes.
We chose survival to discharge as our primary outcome measure, because it is the standard outcome measure recommended by consensus for in-hospital and out-of-hospital cardiac arrest studies.10–12,27–29 The analysis of our secondary outcome measure, “survival with good neurologic outcome,” was unable to achieve statistical significance, presumably because of the smaller numbers (ie, limited power). Nevertheless, the survival-with- good-neurologic-outcome data parallel the survival-to-discharge data in previous analyses.
As with all multicenter registries, analysis of the data may be limited by data integrity and validation issues at multiple sites. The rigorous abstractor certification process, uniform data collection, consistent definitions, scientific advisory board reabstraction process, and large sample size were intended to minimize these sources of bias. Another limitation is a potential sampling bias; the data represent consecutive, sequential data submission from volunteer centers. These centers compose more than one sixth of all hospitals with >500 beds in the United States but are a voluntary, convenience sample of hospitals. Therefore, the quality of care and outcomes may be different from other institutions. In addition, the neurologic outcome was determined at hospital discharge with no long-term neurocognitive follow-up. However, previous studies indicate that neurologic status at discharge is not substantially different from status at 6 months and 1 year postarrest.9,30,31 Finally, this study intentionally excluded newborns from NICUs and delivery rooms, thereby excluding most premature neonates.
What are the “take home” points from these NRCPR age-related data for clinicians resuscitating children in an ICU cardiac arrest setting? First, pediatric ICU cardiac arrests in all of the pediatric age groups are commonly associated with hypotensive shock, progressive respiratory failure, and/or ventricular tachyarrhythmias (ventricular fibrillation or pulseless ventricular tachycardia). Second, primary cardiac failure is relatively more common among the infants and neonates, progressive respiratory failure is relatively more common among children 1 month to 8 years old, and trauma and malignancy are relatively more common among the older children. Finally, newborns and infants <1 year of age, in an intensive care setting, have a better prognosis, independent of facility, patient, prearrest, and arrest interventions.
Survival after pediatric ICU pulseless cardiac arrest is age dependent. Age <1 year is associated with substantially better ROSC and survival to discharge. This survival advantage is sustained even after controlling for potentially important confounding patient and facility characteristics, prearrest conditions, cardiac arrest interventions, and process of care variables.
Endowed Chair of Pediatric Critical Care Medicine, Children's Hospital of Philadelphia, and the American Heart Association Emergency Cardiovascular Care Committee provided funding for the completion of this study but did not influence the design, conduct, management, analysis, or interpretation of the data or preparation of the article.
The American Heart Association National Registry of Cardiopulmonary Resuscitation investigators include the authors and Mary E. Mancini, RN, PhD; Mary Ann Peberdy, MD; Graham Nichol, MD; Tanya Lane-Truitt, RN; Jerry Potts, PhD; Brian Eigel, PhD; John F. Kutcher; and Joseph P. Ornato, MD.
We thank the American Heart Association and Michael C. Bell for their unwavering support of the National Registry of Cardiopulmonary Resuscitation; Kathryn Roberts, as well as innumerable staff and data abstractors from National Registry of Cardiopulmonary Rescucitation hospitals, for their time and effort; Yuling Hong from the Executive Database Committee of the American Heart Association for scientific review; Scott Carey, Tri-Analytics software; Sidney Atwood, Harvard School of Public Health; and Simeon Shaykavich, Harvard School of Public Health.
- Accepted August 8, 2006.
- Address correspondence to Peter A. Meaney, MD, MPH, Department of Anesthesia and Critical Care Medicine, 7th Floor, Room 7c03, Children's Hospital of Philadelphia, 34th Street and Civic Center Boulevard, Philadelphia, PA 19104. E-mail:
Financial Disclosure: Dr Berg received grant support to study swine defibrillation from Medtronic Emergency Response Systems. Dr Nadkarni served as an uncompensated education consultant for Laerdal Medical Corporation and Medical Education Technologies, Inc.
Drs Meaney and Nadkarni primarily designed the study, with analytic approach by Drs Meaney, Cook, and Testa; the data were collated and analyzed by Drs Meaney, Nadkarni, and Berg, all of whom vouch for the data and analyses; and the entire writing group wrote the article.
- ↵Sood S, Giacoia GP. Cardiopulmonary resuscitation in very low birthweight infants. Am J Perinatol. Mar.1992;9 :130– 133
- ↵Dean JM, Koehler RC, Schleien CL, et al. Age-related changes in chest geometry during cardiopulmonary resuscitation. J Appl Physiol.1987;62 :2212– 2219
- ↵Reis AG, Nadkarni V, Perondi MB, Grisi S, Berg RA. A prospective investigation into the epidemiology of in-hospital pediatric cardiopulmonary resuscitation using the international Utstein reporting style. Pediatrics.2002;109 :200– 209
- ↵Jacobs I, Nadkarni V, Bahr J, et al. Cardiac arrest and cardiopulmonary resuscitation outcome reports: Update and simplification of the Utstein templates for resuscitation registries: A statement for healthcare professionals from a task force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian Resuscitation Council, New Zealand Resuscitation Council, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Councils of Southern Africa). Circulation.2004;110 :3385– 3397
- Zaritsky A, Nadkarni V, Hazinski MF, et al. Recommended guidelines for uniform reporting of pediatric advanced life support: The Pediatric Utstein Style. A statement for healthcare professionals from a task force of the American Academy of Pediatrics, the American Heart Association, and the European Resuscitation Council. Resuscitation.1995;30 :95– 115
- ↵Cummins RO, Chamberlain D, Hazinski MF, et al. Recommended guidelines for reviewing, reporting, and conducting research on in-hospital resuscitation: the in-hospital ‘Utstein style’. American Heart Association. Circulation.1997;95 :2213– 2239
- ↵Department of Health and Human Services, National Institutes of Health. NIH Policy and Guidelines on the Inclusion of Children as Participants in Research Involving Human Subjects. Bethesda, MD: National Institutes of Health; 1998
- ↵Nadkarni VM, Larkin GL, Peberdy MA, et al. First documented rhythm and clinical outcome from in-hospital cardiac arrest among children and adults. JAMA.2006;1 :50– 57
- Copyright © 2006 by the American Academy of Pediatrics