| | More

August 2010

Critical Care

Reviewer Name:        Mark W. Hall, MD   


Article Title/Citation

Srinivasan V, Nadkarni VM, Helfaer MA, Carey SM, Berg RA.  Childhood Obesity and Survival After In-hospital Pediatric Cardiopulmonary Resuscitation.  Pediatrics 2010;125(3):e481-8.

Study objectives/purpose

This study was designed to test the hypothesis that childhood obesity would be independently associated with increased mortality after in-hospital, pediatric CPR.

Brief background

Obesity is a public health problem of epidemic proportions in children.  Obesity which might impact the effectiveness of CPR in children due to the physical challenges of airway management, performance of effective chest compressions, and placement of intravenous access.  Additionally, drug dosing in pediatric resuscitation algorithms is typically based on weight.  Depending on the lipid-solubility of the medication, this could result in marked over- or under-achievement of plasma drug concentrations in obese children. 
The relationships between obesity and outcomes following in-hospital CPR are poorly understood in children. 

Funding source of study

This study was funded by the Endowed Chair of Pediatric Critical Care Medicine, Children’s Hospital of Philadelphia and the American Heart Association Emergency Cardiovascular Care Committee.


Study design and methodology: Retrospective analysis of a prospectively-collected database of resuscitation data obtained through the American Heart Association’s National Registry of Cardiopulmonary Resuscitation (NRCPR).  The NRCPR is a voluntary, fee-based registry in which certified coordinators at member institutions input data on precisely defined variables from every child receiving CPR. 

Patient selection and enrollment

Data representing children < 18 years of age from 167 centers between January 1, 2000 and July 31, 2004 were analyzed.  Only first-time, in-hospital arrests involving children without limitations on their care were evaluated.  Children whose arrests were resolved via an implantable cardioverter-defibrillator, and children whose arrests began outside the hospital or in the delivery room were excluded.

Body mass index (BMI) data for subjects were estimated, given that the NRCPR collects weight data, but not height/length data.  This estimation was done using median height-for-age values adjusted for gender using World Health Organization software.  BMI-for-age (for children ≥ 2 years old) or weight-for-length (for children < 2 years old) were then calculated.  Children were then assigned to one of three categories (adjusted for gender):  obese (≥ 95th% for BMI-for-age or weight-for-length), underweight (< 5th% for BMI-for-age or weight-for-length), or normal-weight.

Outcomes/endpoints being measured

The primary outcome variable for this study was survival to hospital discharge.  Secondary outcome variables included survival of the arrest event (return of spontaneous circulation [ROSC] for > 20 minutes) and survival with a “favorable” neurologic outcome (defined as a Pediatric Cerebral Performance Category (PCPC) scale score of 1, 2, or 3 representing the spectrum from normal neurologic status through moderate disability). 

Statistical analysis
Comparisons between groups were made using ANOVA (continuous data) or chi-squared (categorical data).  Predictors of overweight and underweight status and for primary and secondary outcomes were analyzed by stepwise, multivariate, logistic regression.  Co-variables included in the models were epidemiologic factors (including primary illness category and co-morbidities) and other factors known to impact outcomes after cardiac arrest which also showed significant correlation in prior univariate analysis.


Enrollment & baseline characteristics

Arrest data from a total of 1477 children were evaluated.  Of those 209 subjects were excluded due to the presence of an incomplete data set with outcome characteristics similar to the included population.  Of the remaining 1268 patients, 213 (17%) were classified as obese, 484 (38%) were normal-weight, and 571 (45%) were underweight. 
Clinical factors associated with obesity by multivariable regression analysis included male gender, medical non-cardiac illness, and cancer.  Obese children were unlikely to have heart failure.  By contrast, being underweight was associated with young age, male gender, surgical cardiac illness, and prematurity.  Underweight children were unlikely to have cancer.

Immediately prior to arrest, underweight children were more likely to have acute respiratory insufficiency and obese children more likely to have acute pulmonary edema.  Most pre-arrest factors were similar between groups.

Obese children were more likely to have absent pulses throughout the arrest compared to normal-weight children and underweight children.  Obese children received more epinephrine than normal-weight or underweight children (4 [2 – 7] vs 3 [2 – 5] vs 3 [1 – 5] doses, p<0.005).  There were no differences between obese and normal-weight children in terms of fluid boluses administered or other PALS/ACLS code drug administration. Obese children were not different from normal-weight children in terms of arrest rhythm type.  Similarly, there were no significant differences between obese and normal-weight children in terms of time to initiation of CPR, duration of CPR, incidence of ventricular fibrillation/pulseless ventricular tachycardia, time to defibrillation, or time to first epinephrine dose.  The use of ECMO was rare and was similar between the obese and normal-weight groups.

Analysis of self-reporting from the arrest events revealed more deviations from PALS protocols in the obese patients along with a trend toward more problems with vascular access.

Summary of primary & secondary outcomes
Compared to normal-weight children, obesity was independently associated with lower survival even after adjusting for the aforementioned covariables by multivariable logistic regression (Odds Ratio [95% CI]:  0.62 [0.38 – 0.93]) with a survival rate of 23% in obese children vs. 34% in normal-weight children.  Obese children were less likely to have ROSC after the initial arrest event (OR: 0.58 [0.35 – 0.76]) compared to normal-weight children.  There was no difference between any of the groups in terms of survival with favorable neurologic outcome.  Interestingly, there was no difference in survival or ROSC between normal-weight and underweight children.   

Author’s Discussion and Conclusions  

Brief summary of author’s main discussion points

Dr. Srinivasan et al conclude that obesity represents an independent risk factor for mortality following in-hospital cardiac arrest in children.  They suggest that this difference be the subject of future research and should be considered in the development and evaluation of pediatric CPR guidelines. The authors hypothesize that the worsened outcomes in obese children could be due to inadequate force and depth of compressions, improper medication dosing or non-optimal dose of energy used for defibrillation.  Lastly, they note that team dynamics were more problematic in the obese children. 

The authors mention that there is a discrepancy between their findings relative to underweight subjects compared with similar studies in adults, in which patients with low BMI had higher mortality rates and worse functional status upon hospital discharge.  They point out that underweight children, many of whom were premature infants and/or were hospitalized for cardiac surgery, likely represent a very different patient population than underweight adults.  Malignancy, for example, was rare in the underweight pediatric population whereas it was more commonly associated with low BMI in adult studies.

Dr. Srinivasan et al highlight several important limitations of their study.  Chief among them is the fact that the NRCPR does not collect height/length data, resulting in the need to estimate BMI-for-age and weight-for-length data using normative tables.  They also point out the limitations and potential biases associated with the use of registry data.  They note, however, that uniform operational definitions, data collection, training and certification, periodic re-abstraction, and large sample size may mitigate these limitations.

Reviewer’s Discussion and Conclusions  

Study strengths

The relationships between obesity and CPR outcomes in children are largely unknown, making this study the largest of its kind to evaluate these relationships.  The authors are experienced investigators in the field of pediatric cardiopulmonary resuscitation research and they made use of a very robust, highly quality-controlled, pre-existing database from 167 hospitals across the U.S., which speaks to the generalizability of their findings.  The large number of children studied provided substantial power for their study and allowed for the performance of detailed, multivariable regression analyses. 

Study limits, weakness (potential for bias, etc)
As the authors note, a major weakness of this study lies with the fact that the most important variables entered into their analytical models (BMI-for-age, weight-for-length) were never directly measured.  Instead, they paired actual weight data with estimated height/length data.  While it is true that some heavy children could have been exceptionally tall and/or some thin children could have been unusually short, their large sample size reduces the risk of bias here somewhat. 

While the authors include a long list of pre-arrest conditions in their analyses, it is still not exactly clear why the children arrested.  Similarly, we are not given objective data by which to judge pre-arrest severity of illness (e.g. PRISM III score, PELOD score).  Perhaps these data are not captured in the NRCPR registry.  This study was not designed to evaluate cause-and-effect relationships between obesity and CPR outcomes, but the absence of these pre-arrest data make it impossible to know if the obese patients were simply sicker just prior to arrest compared to normal-weight children.  Such a finding could shift the focus of future investigations more toward the management of the pre-arrest state rather than to the CPR itself.  It is notable that, among patients who experienced ROSC, there was much less of a difference between weight groups in terms of survival to hospital discharge.  This raises the possibility that obesity may not play as great a role in contributing to late mortality compared to its pre- or intra-arrest effects.    

Applicability & impact on healthcare providers
There is no doubt that pediatric critical care medicine providers are faced with the challenge of caring for obese patients with increasing frequency.  Simply by being aware of the differences in CPR outcomes between obese and non-obese children, the pediatric ICU community could take several important steps to insure that the current PALS standard of care is applied to obese children.  First, centers can work to assure that appropriately sized equipment is readily available to resuscitate obese patients.  Second, the resuscitation of the obese patient could be added to simulation/training scenarios so that providers could familiarize themselves with the special challenges faced in the care of these children.  Third, centers could strive to achieve uniformity regarding drug dosing weights (ideal vs actual) for resuscitative medications.  Centers could also focus on improving vascular access algorithms for obese patients, perhaps through emphasis of intra-osseous needle placement or use of ultrasound-guided vascular access placement. 

Unfortunately the current study does not provide insight into the cause-and-effect relationships between obesity and adverse CPR outcomes.  Additional prospective research is needed to identify ways in which current resuscitative practice could be improved to better serve the obese child.

Conclusions and recommendations
This study has highlighted yet another challenging, and previously unappreciated, aspect of the obesity epidemic among America’s children.  The explanation for the increased CPR mortality rate seen in obese children is likely to be multi-factorial, and the current study does little to shed light on the specific reasons underlying these differences.  Since childhood obesity is likely to remain a substantial public health problem for years to come, however, it behooves the pediatric critical care community to address these important findings by being prepared to resuscitate obese children with appropriate equipment and training.  We should also strive to prospectively measure and hopefully improve the quality of CPR, the use of code drugs, and other aspects of resuscitation in this at-risk pediatric population.