Injury is the leading cause of death and disability in children. Among children who experience trauma, those who present in hemorrhagic shock are at particularly high risk of death.1 The age-adjusted shock index has been identified as a valid measure of shock in pediatric trauma patients on initial presentation; however, the ideal resuscitation practices for these patients once identified remains unknown.2,3 The 10th edition of Advanced Trauma Life Support acknowledges the move toward crystalloid restriction and balanced transfusion for children with evidence of hemorrhagic shock. It was unable to make firm recommendations, however, because of the absence of conclusive evidence in the pediatric population.4
Prior pediatric retrospective studies have found that, while high-volume crystalloid resuscitation was not associated with increased risk of complications such as acute respiratory distress syndrome (ARDS), it was associated with greater duration of mechanical ventilation, intensive care, and hospitalization.5,6 Similarly, Department of Defense data demonstrated similar outcomes in children who received greater than 40 mL/kg of blood if they had received high-volume crystalloid.7 Additional retrospective data on pediatric trauma resuscitation demonstrated that the odds of requiring transfusion are similar following the first bolus, suggesting that crystalloid response achieves a plateau with the second bolus.8
Despite retrospective evidence, there continues to be a lack of defined consensus and variability in practice for resuscitation of pediatric trauma patients. In addition, many studies of pediatric resuscitation are centered on a volume-based definition of massive transfusion or activation of a massive transfusion protocol.7,9 Patients who would have benefited from early transfusion but received a crystalloid-heavy resuscitation may be missed if they do not meet the inclusion criteria of a massive transfusion activation or minimum volume of transfused products. This study was designed to capture pediatric trauma patients presenting in shock at an earlier stage and to determine the relationship between timing and volume of crystalloid before blood products and outcomes.
PATIENTS AND METHODS
A multi-institutional prospective observational study of pediatric trauma patients younger than 18 years who presented with elevated age-adjusted shock index (SIPA) from the scene of injury was performed from April 2018 to September 2019 at 24 trauma centers. Participating centers included 17 with state or American College of Surgeons level I pediatric verification, three with level II pediatric verification, and four with level I adult verification. Institutional review board approval was obtained at each center. Patients were screened based on their first and second set of vital signs and included if they had an elevated SIPA on either set. The SIPA has previously been validated to identify severely injured children.2 The exclusion criteria were the following: (1) a greater than 20% total body surface area burn, (2) an isolated burn or inhalation injury, (3) asphyxiation injury, (4) transfer from another facility, and (5) transport by private vehicle or police.
Participating centers entered deidentified data prospectively into a standardized Research Electronic Data Capture (REDcap) (Vanderbilt University, Nashville, TN) database. Study data were collected and managed using REDCap electronic data capture tools hosted at Cincinnati Children's Hospital Medical Center. REDCap is a secure, web-based software platform designed to support data capture for research studies. Monthly conference calls were held to discuss data collection and progress.
Patient and injury characteristics, volume and timing of prehospital, emergency department (ED), and initial admission resuscitation up to 30 hours were assessed. Resuscitation included receipt of crystalloid boluses or blood products—packed red blood cells, plasma, platelets, or cryoprecipitate. Transfused patients refer to patients who received blood products in this resuscitation period. Outcomes including ED disposition, intensive care stay, ventilator days, hospital complications, total length of hospital stay, and hospital survival were recorded. Complications included reintubation, acute renal failure, acute lung injury/acute respiratory distress syndrome, catheter associated urinary tract infection, hospital acquired pneumonia, blood stream infection, sepsis, deep venous thrombosis/pulmonary embolism, transfusion-related lung injury/transfusion-associated circulatory overload, or other complication. These were abstracted from participating sites' National Trauma Data Standard trauma registries.10 Injury Severity Score (ISS) and Abbreviated Injury Scale were added retrospectively from participating centers' trauma registries. Crystalloid boluses were defined as 20 ± 10 mL/kg of 0.9% normal saline or lactated ringers. The primary outcome was survival to discharge. Follow-up time was censored at discharge. Secondary outcomes were extended intensive care stay, ventilator days, and hospital stay, which were defined as 90th percentile or greater for the entire cohort.
Categorical patient characteristics are presented using frequencies and percentages, while continuous measures were summarized using means with SDs or medians with interquartile range (IQR). Cox proportional hazards and logistic regression models were used to evaluate the relationship between number of crystalloid boluses received before transfusion and primary and secondary outcomes. The following patient characteristics and clinical information were considered for inclusion in the final model for each outcome: age, sex, race, injury type, mechanism of injury, receipt of any prehospital crystalloid, receipt of any prehospital blood, crystalloid boluses administered before any transfusion, thromboelastograph obtained, massive transfusion protocol activation, time to transfusion, ED to operating room transfer, ISS, and initial Glasgow Coma Scale. Random clinical center intercepts were included in each model.
Statistical models addressed missing data values by maximum likelihood, under the data missing at random (MAR) assumption. Sensitivity analyses using pattern-mixture models were performed to evaluate this assumption. A total of 50 imputed data sets were created for use in the multivariable Cox proportional hazards model. SAS Proc PHreg and Proc MiAnalyze (SAS Institute, Cary, NC) were used to generate all estimates from the multiply imputed data sets. Using this approach, imputed ISS values were adjusted by −5%, 0%, and +5% of what they would be if the data were MAR (ISS was selected for assessment because it was the most frequently missing values in the analysis at 10%). Hazard ratio values by ISS adjustment level were nearly identical, supporting the validity of the MAR assumption.
Statistical analyses were conducted using SAS version 9.4; all reported p values were two-sided and considered statistically significant when less than 0.05.
Of 1,471 patients with elevated SIPA at 24 centers, 712 (48.4%) met the inclusion criteria (Fig. 1). The most frequently met exclusion criteria was transfer from another center (41.7%). For included patients, the mean age was 7.6 years, median (IQR) ISS was 9 (2–20) years, and hospital mortality was 5.3% (n = 38). The majority (87.9%) presented with blunt injuries, most frequently from motor vehicle incidents (40.8%). Fifteen percent (14.9%, n = 101) went to the operating room, 28.5% (n = 193) to the intensive care unit, and 31.6% to a general care floor from the ED. Among survivors, 82.4% were discharged home. Complications were infrequent (6.9%; Table 1).
There were 311 patients (43.7%) who received at least one crystalloid bolus, and 146 (20.5%) received blood including 65 (9.6%) with massive transfusion activation. The transfusion rate increased with greater number of crystalloid boluses administered, with more than half (53.3%) of patients who received greater than one crystalloid bolus receiving a transfusion (Fig. 2). Of the 149 transfused patients, 41 (28.1%) received blood before any crystalloid boluses. These patients had shorter median time to transfusion (19.8 vs. 78.0 minutes, p = 0.005) and less total fluid volume (50.4 vs. 86.6 mL/kg, p = 0.033) than those who received crystalloid before blood despite similar injury severity (median ISS, 22 vs. 27, p = 0.40). Notably, patients with penetrating injuries were more likely to receive blood first (39.5% vs. 23.3%, p = 0.047).
Characteristics of Transfused Patients
Among transfused patients, the median (IQR) volume of red blood cells was 33.2 (23.9–52.5) mL/kg; plasma, 26.6 (23.9–31.7) mL/kg; and platelets, 16.3 (12.4–17.7) mL/kg. Transfused patients were older, more likely to have penetrating injuries, and more severely injured (ISS, 26 vs. 5, p < 0.001) with significantly greater head, abdomen, and lower extremity Abbreviated Injury Scale (Table 2). Use of viscoelastic monitoring was infrequent among transfused patients (14.8%). In-hospital mortality was 22.8% in transfused patients versus 0.7% in nontransfused (p < 0.001). Transfused patients also experienced significantly more complications while hospitalized (15.5% vs. 4.4%, p < 0.001).
Multivariable Analyses of Outcomes
Multivariable analysis showed no association of resuscitation characteristics with in-hospital mortality (p = 0.51). Compared with those who received less than one crystalloid bolus, each crystalloid bolus after the first was associated with an incrementally greater odds of extended ventilator, intensive care, and hospital days (each p < 0.05; Table 3). Time to blood transfusion was additional associated with extended ventilator duration (odds ratio, 1.11; p = 0.04). Only greater ISS was associated with all outcomes (each p < 0.05).
The greatest burden of morbidity and mortality of children is trauma; however, there is no consensus on the optimal resuscitation of injured children in suspected shock. Adult trauma literature and existing retrospective pediatric studies suggest that a crystalloid-sparing, early transfusion approach is beneficial in children. This inclusive multicenter prospective observational study incorporated the experience of a diverse cohort of trauma centers caring for injured children to determine the impact of resuscitation approach on mortality and other important outcomes. A clear association between increased crystalloid use and prolonged mechanical ventilation, intensive care, and hospital stay was demonstrated.
High-volume crystalloid resuscitation has been associated with adverse effects including coagulopathy and hypothermia in the laboratory.11,12 In the clinical setting, this has been repeatedly associated with a longer ventilator duration, intensive care stay, and overall hospital stay.5,7,13 Coons et al.6 also found high-volume crystalloid to be associated with longer duration of nil per os in a large single institution retrospective study. The most frequent definition of high volume crystalloid resuscitation in pediatric trauma literature has been greater than 60 mL/kg per day. The data reported here demonstrate increased odds of prolonged mechanical ventilation, intensive care, and hospital stay after the first 20 ± 10 mL/kg crystalloid bolus, further supporting limitation of crystalloid specifically to, at most, one bolus in injured children with suspected bleeding before conversion to blood.
Additional crystalloid boluses were not associated with decreased transfusion in this study, and, rather, transfusion was more likely as the number of crystalloid boluses increased. This is consistent with existing retrospective data.8 For children with suspected hemorrhage from trauma who require resuscitation beyond one crystalloid bolus, we recommend blood products instead of further crystalloid boluses to avoid the deleterious effects of additive crystalloid as described above and the association of longer time to transfusion with prolonged mechanical ventilation. As noted in current Advanced Trauma Life Support guidelines, initial resuscitation with blood products should also receive strong consideration based on the success of this approach in adult trauma resuscitation and the association with less overall fluid volume and earlier transfusion in this study.4
While all injured children benefit from avoiding high-volume crystalloid resuscitation, tools for identification of those who require early transfusion are needed, as time to transfusion was associated with negative outcome in this study. Only half of patients with elevated SIPA received crystalloid boluses, with an even smaller subset ultimately transfused, suggesting that a more specific early indicator of shock is needed. We hypothesize that SIPA may be falsely elevated in those children who are tachycardic for other reasons including anxiety and hypotension is a late finding associated with increased odds of death.14 While identification of pediatric trauma patients who benefit from upfront transfusion is ideal, elevated SIPA and nonresponse to one crystalloid bolus can serve as a surrogate marker for prompt transition to blood as well as consideration for massive transfusion protocol activation, if available.
Prospective clinical trials that assign interventions are needed to confirm results in this study and further clarify which injured children benefit from early transfusion, as this study is limited by potential variable management at participating centers. In addition, more than half of eligible patients were excluded based on being transferred from other facilities where their resuscitation began. Because these patients still arrived to participating trauma center with elevated age adjusted shock index, trauma centers that care for children must take interest in resuscitation practices at referring facilities. Lastly, other aspects of resuscitation, including ratio of blood products and viscoelastic monitoring, and outcomes, including individual complications, were unable to be evaluated due to low occurrence.
In summary, pediatric trauma patients in suspected hemorrhagic shock benefit from earlier transfusion and fewer crystalloid boluses in terms of decreased duration of ventilation, intensive care, and hospitalization. We propose either immediate transfusion or transfusion after nonresponse to one crystalloid bolus in children with clinically suspected bleeding. Prospective interventional studies are needed to confirm this and to clarify optimal transfusion approaches.
S.F.P. and R.A.F. were involved in study design, interpretation of data, and drafting of the article. S.M. was involved in study design, data acquisition, and drafting of the article. T.M.J. was involved in data analysis and drafting of the article. R.F.W., M.L.K., E.C.A., R.S.B., T.J.S., J.E.B., A.M., W.B.R., L.A.B., E.M.C., C.R., R.M.N., C.J.R., D.I.G., C.J.S., M.G., J.K.P., C.G., S.P., A.M.W., R.T.R., B.K.Y., J.M., J.P., M.T.S., T.M., D.B.K., S.D.S., T.T., A.M.V., M.C., C.B., J.R., R.G.S., A.R.J., B.J.F., D.P.M., B.K., M.S.D., A.G.-S., and J.S.R. were involved in data acquisition and critical revision of the article.
We thank Olivia Beale, Magdolna Pakocs, Sherrie Murphy, and Karen Herzing for their assistance with the study.
The authors declare no conflicts of interest.
1. Neff LP, Cannon JW, Morrison JJ, Edwards MJ, Spinella PC, Borgman MA. Clearly defining pediatric massive transfusion
: cutting through the fog and friction with combat data. J Trauma Acute Care Surg
. 2015;78(1):22–28; discussion 28-9.
2. Acker SN, Ross JT, Partrick DA, Tong S, Bensard DD. Pediatric specific shock index accurately identifies severely injured children. J Pediatr Surg
3. Nordin A, Coleman A, Shi J, Wheeler K, Xiang H, Acker S, Bensard D, Kenney B. Validation of the age-adjusted shock index using pediatric trauma
quality improvement program data. J Pediatr Surg
4. American College of Surgeons Committee on Trauma. Student Course Manual ATLS ® Advanced Trauma Life Support®
. 10th ed. 2018.
5. Acker SN, Ross JT, Partrick DA, DeWitt P, Bensard DD. Injured children are resistant to the adverse effects of early high volume crystalloid resuscitation
. J Pediatr Surg
6. Coons BE, Tam S, Rubsam J, Stylianos S, Duron V. High volume crystalloid resuscitation
adversely affects pediatric trauma
patients. J Pediatr Surg
7. Edwards MJ, Lustik MB, Clark ME, Creamer KM, Tuggle D. The effects of balanced blood component resuscitation
administration in pediatric trauma
patients requiring transfusion
in Afghanistan and Iraq 2002 to 2012. J Trauma Acute Care Surg
8. Polites SF, Nygaard RM, Reddy PN, et al. Multicenter study of crystalloid
boluses and transfusion
in pediatric trauma
—when to go to blood? J Trauma Acute Care Surg
10. American College of Surgeons. National Trauma Data Standard (NTDS). Available at: https://www.facs.org/quality-programs/trauma/tqp/center-programs/ntdb/ntds
. Accessed November 28, 2019.
11. Sheppard FR, Schaub LJ, Cap AP, Macko AR, Moore HB, Moore EE, Glaser CJJ. Whole blood mitigates the acute coagulopathy of trauma and avoids the coagulopathy of crystalloid resuscitation
. J Trauma Acute Care Surg
12. Ponschab M, Schöchl H, Keibl C, Fischer H, Redl H, Schlimp CJ. Preferential effects of low volume versus high volume replacement with crystalloid
fluid in a hemorrhagic shock
model in pigs. BMC Anesthesiol
13. Elkbuli A, Zajd S, Ehrhardt JD Jr., McKenney M, Boneva D. Aggressive crystalloid resuscitation
outcomes in low-severity pediatric trauma
. J Surg Res
14. Leeper CM, McKenna C, Gaines BA. Too little too late: hypotension and blood transfusion
in the trauma bay are independent predictors of death in injured children. J Trauma Acute Care Surg
Terri Elsbernd MS, RN: I would like to thank all involved for the opportunity to review this manuscript. I applaud the authors and I thank them for this work.
As the current President of the Pediatric Trauma Society, I am even more pleased to be able to review this manuscript. As Dr. Polites so clearly stated in her presentation, we lack scientific evidence for what we do every day in the trauma bay to take care of injured children. She specifically cited the latest edition of Advance Trauma Life Support and the fact that no firm recommendations on what constitutes a balanced pediatric fluid resuscitation can be made because there is no conclusive evidence. The stakeholders in pediatric trauma care acknowledge the knowledge gap and strive daily to overcome the gap. In keeping true to scientific principles, the best way to address the gap in evidence-based care is to conduct a multi-center study: to provide the evidence, to promote the best care, and to improve outcomes for injured children.
Another aspect of this study that I appreciate is that the participating hospitals are not all verified pediatric level 1 trauma centers, stand-alone children’s hospitals, or even level 2 pediatric trauma centers. Instead, many of the participating facilities are adult trauma centers taking into account what we in the pediatric trauma world know: many children across the country are not cared for in pediatric trauma hospitals but instead in adult trauma hospitals or community hospitals. The fact the investigators involved a variety of participating institutions adds to the validity of the study and to advance the science of taking care of injured children. The ultimate goal for all pediatric trauma stakeholders is to be able to base our pediatric injury interventions on sound, evidence-based guidelines and this study is definitely a step in that direction.
This study also elicits more questions. Children who were transferred were not included in the study population, a criterion that eliminated as many as 50% of patients. The inclusion of this subset in a future study might yield results to further decrease our existing knowledge gap.