Failure of nonoperative management of pediatric blunt liver and spleen injuries: A prospective Arizona-Texas-Oklahoma-Memphis-Arkansas Consortium study

Linnaus, Maria E. MD; Langlais, Crystal S. MPH; Garcia, Nilda M. MD; Alder, Adam C. MD; Eubanks, James W. III MD; Maxson, R. Todd MD; Letton, Robert W. MD; Ponsky, Todd A. MD; St. Peter, Shawn D. MD; Leys, Charles MD; Bhatia, Amina MD; Ostlie, Daniel J. MD; Tuggle, David W. MD; Lawson, Karla A. PhD; Raines, Alexander R. MD; Notrica, David M. MD

Journal of Trauma and Acute Care Surgery: April 2017 - Volume 82 - Issue 4 - p 672–679
doi: 10.1097/TA.0000000000001375
AAST Plenary Papers
Editor's Choice

BACKGROUND: Nonoperative management (NOM) is standard of care for most pediatric blunt liver and spleen injuries (BLSI); only 5% of patients fail NOM in retrospective reports. No prospective studies examine failure of NOM of BLSI in children. The aim of this study was to determine the frequency and clinical characteristics of failure of NOM in pediatric BLSI patients.

METHODS: A prospective observational study was conducted on patients 18 years or younger presenting to any of 10 Level I pediatric trauma centers April 2013 and January 2016 with BLSI on computed tomography. Management of BLSI was based on the Arizona-Texas-Oklahoma-Memphis-Arkansas Consortium pediatric guideline. Failure of NOM was defined as needing laparoscopy or laparotomy.

RESULTS: A total of 1008 patients met inclusion; 499 (50%) had liver injury, 410 (41%) spleen injury, and 99 (10%) had both. Most patients were male (n = 624; 62%) with a median age of 10.3 years (interquartile range, 5.9, 14.2). A total of 69 (7%) underwent laparotomy or laparoscopy, but only 34 (3%) underwent surgery for spleen or liver bleeding. Other (nonexclusive) operations were for 21 intestinal injuries; 15 hematoma evacuations, washouts, or drain placements; 9 pancreatic injuries; 5 mesenteric injuries; 3 diaphragm injuries; and 2 bladder injuries. Patients who failed were more likely to receive blood (52 of 69 vs. 162 of 939; p < 0.001) and median time from injury to first blood transfusion was 2.3 hours for those who failed versus 5.9 hours for those who did not (p = 0.002). Overall mortality rate was 24% (8 of 34) in those who failed NOM due to bleeding.

CONCLUSION: NOM fails in 7% of children with BLSI, but only 3% of patients failed for bleeding due to liver or spleen injury. For children failing NOM due to bleeding, the mortality was 24%.

LEVEL OF EVIDENCE: Therapeutic study, level II.

From the Level I Pediatric Trauma Center, Phoenix Children's Hospital (M.E.L., C.S.L., D.M.N.), Phoenix, AZ; Pediatric Trauma Center, Dell Children’s Medical Center (N.M.G., K.A.L., D.W.T.), Austin, TX; Pediatric Surgery Department, Children's Medical Center (A.C.A.), Dallas, TX; Department of Pediatric Surgery, Le Bonheur Children's Hospital (J.W.E.), Memphis, TN; Department of Pediatric Surgery, Arkansas Children's Hospital (R.T.M.), Little Rock, AR; Department of Surgery, The Children’s Hospital at OU Medical Center (R.W.L., A.R.R.), Oklahoma City, OK; Department of Pediatric Surgery, Akron Children’s Hospital (T.A.P.), Akron OH; Pediatric Surgery, Mercy Children’s Hospital (S.D.S.), Kansas City, MO; Pediatric Surgery, American Family Children’s Hospital (C.L.), Madison WI; Department of Pediatric Surgery, Children’s Healthcare of Atlanta (A.B.), Atlanta GA; and Department of Surgery, Phoenix Children’s Hospital, Phoenix AZ & American Family Children's Hospital (D.J.O.), Madison WI.

Submitted: September 1, 2016, Revised: November 20, 2016, Accepted: December 19, 2016, Published online: January 17, 2017.

This study was presented at the 76th annual meeting of the American Association for the Surgery of Trauma, September 13–16, 2016, in Waikoloa, Hawaii.

Address for reprints: David M. Notrica, MD, Division of Pediatric Surgery and Level I Pediatric Trauma Center, Phoenix Children's Hospital, 1919 E Thomas Rd, Phoenix, AZ 85016; email:

Article Outline

Over the last four decades, the management of blunt abdominal trauma in adult and pediatric patients has evolved from operative management with high morbidity and mortality to successful nonoperative management (NOM) as the mainstay of therapy for patients with solid organ injuries.1,2,3 Since its introduction in the late 1960s, NOM has been increasingly accepted in the management of patients with blunt liver and spleen injuries (BLSI) and more recently, NOM of even high-grade injuries has been successful and more readily attempted.4 Surgeons have simultaneously decreased the time patients require bed rest, shortened the length of hospitalization, and decreased unnecessary health care resources dedicated to diagnosis and treatment of these injuries.1,5–9 This movement has only been possible with the recognition that children injured by blunt abdominal trauma infrequently fail conservative therapy regardless of grade of injury.10 Limited literature is available for those children failing NOM, and no prospective studies currently exist to investigate the failure of NOM in pediatric trauma. Therefore, we sought to identify the frequency and characteristics of failure of NOM in pediatric patients using a multi-institutional prospective observational cohort study of centers using the Arizona-Texas-Oklahoma-Memphis-Arkansas Consortium (ATOMAC) management guideline for NOM of BLSI.3

Back to Top | Article Outline


We conducted a prospective observational study on patients 18 years or younger presenting to one of 10 Level I pediatric trauma centers (PTC): Phoenix Children’s Hospital (Phoenix, AZ), Dell Children’s Medical Center (Austin, TX), Child’s Medical Center, part of Children’s Health (Dallas, TX), Arkansas Children’s Hospital (Little Rock, AR), The Children’s Hospital at OU Medical Center (Oklahoma City, OK), Le Bonheur Children’s Hospital (Memphis, TN), American Family Children’s Hospital (Madison, WI), Akron Children’s Hospital (Akron, OH), Children’s Mercy Hospital (Kansas City, MO), and Children’s Healthcare of Atlanta (Atlanta, GA). Patients were included if they presented with a BLSI identified on computed tomography (CT) between April 2013 and January 2016. Institutional review board approval was obtained at each institution.

All institutions had previously adopted the ATOMAC protocol for NOM of BLSI.3 Data collected included demographics; injury characteristics; vitals and scores for injury severity; hemoglobin within 48 hours of injury; all radiology findings; procedures; and initial, short- (2 weeks post-injury) and long- (60 days post-injury) term outcomes. Failure of NOM was defined as a patient undergoing a surgical procedure in their abdominal cavity and included exploratory laparotomy or laparoscopy for splenectomy, splenorrhaphy, hepatectomy, hepatorrhaphy, abdominal packing, intestinal resections/repairs, mesenteric repairs, pancreatectomy, drain placement, and diagnostic surgical procedures. Angiography and embolization were not considered failures of NOM. A subset of failure for bleeding was defined as patients undergoing splenectomy, splenorrhaphy, hepatectomy, hepatorrhaphy, or abdominal packing. For patients that did not fail for bleeding, primary cause of failure was further classified. For grade-related summary analysis on patients with both liver and spleen injury, the higher grade of injury was used.

To gain a better understanding of all children who fail NOM, trauma registry data at all 10 sites were queried for any additional patients with BLSI who could not be enrolled in the NOM study due to lack of CT scanning before death or operative exploration.

Back to Top | Article Outline

Data Analysis

Descriptive statistics were used to summarize patient characteristics, injuries, and outcomes. Univariate comparisons were made using chi-square and Mann-Whitney for categorical and continuous variables, respectively. Statistical significance was assessed at the 0.05 level with all tests being two-tailed. All analyses were done using Stata 13.1(College Station, TX).

Back to Top | Article Outline


During the 34-month study period, 1,008 patients were enrolled at the ten sites. Men accounted for 624 (62%) of the patients; median age was 10.3 years (interquartile range [IQR], 5.9–14.2). Patients presented to the PTC a median of 3.6 hours (IQR, 1.1–6.3) after injury, most commonly from a motor vehicle collision (n = 340, 34%), fall (n = 138, 14%), or being struck by a car as a pedestrian or cyclist (n = 111, 11%). The liver was the most commonly injured organ at 499 (50%), whereas 410 (41%) had a spleen injury, and an additional 99 (10%) had both a liver and spleen injury. Patients presented to the trauma center with a median shock index (maximum heart rate/minimum systolic blood pressure) of 1.7 (IQR, 1.4–2.3); 1.3 (IQR, 1.1–1.6), 1.1 (IQR, 0.9–1.4), 1.0 (IQR, 0.8–1.2), and 0.9 (IQR, 0.8–1.0) for patients younger than 4 years, 4 to younger than 7 years, 7 to younger than 13 years, 13 to younger than 17 years, and 17 years or older, respectively. Of the 1,008 patients, 214 (21%) patients required transfusion. Median time from injury to blood transfusion was 5.0 hours (IQR, 2.0–17.8), a median of 3.0 hours (IQR, 0.7–14.4) after arrival for the 171 patients with known time of arrival to ED. Overall, 939 (93%) of enrolled patients were successfully managed nonoperatively (NOM) and 69 (7%) patients underwent an intra-abdominal surgery during their hospital stay (failed NOM). Table 1 compares patient and clinical demographics between failed and successful NOM.

Back to Top | Article Outline


The overall failure rate for enrolled patients was 7% (69 patients). The failure rate for bleeding was 3.4% (34 patients), but an additional 35 patients (3.5%) underwent laparotomy or laparoscopy at their index admission. Other (nonexclusive) reasons for surgery were bowel injury (n = 21); evacuation of hematoma, abdominal washout, or drain placement (n = 15); pancreatic injury (n = 9); mesenteric injury (n = 6); nontherapeutic laparoscopy (n = 3); diaphragm injury (n = 3); or bladder repair (n = 2). A completed list of procedures is listed in Table 2.

The failure rate for bleeding in children with a combined liver and spleen injury was 10 (10.1%) of 99. When including the combined injuries, the total failure rates by organ were 44 (7.4%) of 598 for liver injury (with 52% for bleeding), and 42 (8.3%) of 509 for spleen injury (54% for bleeding). Failure rates for patient with an isolated liver or isolated spleen injuries (with no additional significant injuries of any kind) were 6 (3.8%) of 156 and 0 (0%) of 173, respectively.

Back to Top | Article Outline

Timing of Failure

Sixty-three (91%) of the patients who failed NOM had a known time of injury. The median time to failure for any reason was 5.0 (IQR, 3.0–21.0) hours after injury and 3.2 (IQR, 2.0–5.0) hours for hemorrhage. Failure frequencies and timing of failure for all failures and specifically for hemorrhage are presented by grade of injury in Table 3. Time to first operation is shown in Figure 1.

Back to Top | Article Outline

Angiography and Embolization

Overall, 32 (3%) patients underwent an angiography, and 19 of these patients underwent an embolization. Seven (37%) of the 19 patients with angioembolization failed NOM despite embolization.

Back to Top | Article Outline


Overall mortality rate was 2.5% (n = 25). The rate was higher in the patients that failed NOM compared with patients that did not (8 of 69 vs 17 of 939; p < 0.001). All eight of the patients who died after failure of NOM failed due to bleeding; the median time to initial operation for those patients was 2.9 (IQR, 2.0–3.7) hours from injury, and 1.0 (IQR, 0.5–2.6) hours from arrival to the PTC. Overall, median abdominal Abbreviated Injury Scale score of the cohort was 3 (IQR, 2–4), but was higher in those who failed versus those who did not (4 [IQR, 3–5] vs 3 [IQR, 2–3]; p < 0.001). Overall median pediatric intensive care unit and hospital length of stay were 48 hours (IQR, 24–127) and 57 hours (IQR, 36–127), respectively (excluding deaths).

Complications during the hospital stay were documented in 121 (12%) patients including: acute respiratory distress syndrome (n = 24), urinary tract infection (n = 15), unplanned return to the operating room (n = 9), unplanned transfer to the pediatric intensive care unit (n = 8), deep vein thrombosis (n = 8), iatrogenic complications (n = 5), pseudoaneurysm (n = 4), and biloma (n = 2). There were no delayed intra-abdominal bleeds, but one persistent liver bleed did not stop, and another patient returned at 25 days postinjury with an upper gastrointestinal bleed found to be due to hemobilia. Most of the late surgeries were predominantly for abdominal washouts and drainage procedures. The one case of continued slow bleeding occurred in a patient with a grade V liver, grade III spleen, and grade 4IV kidney injury. The patient received a total of 2,050 mL (56.9 mL/kg) of packed red blood cells and was taken to the operating room 187 hours after injury. The patient was found to have slow, persistent liver parenchymal bleeding, felt to be due to the portal system. The only delayed bleeding presented at 25 days postinjury as an upper gastrointestinal bleed. At the initial presentation, the patient had a grade V liver injury and a grade II kidney injury. The patient was managed by the NOM algorithm, never requiring blood transfusion, embolization, or operative intervention. At 25 days postinjury, the patient returned to the hospital with bloody emesis. Reimaging showed a pseudoaneurysm of a right hepatic artery branch. The patient required rehospitalization and underwent angioembolization. The patient was doing well at 60-day follow-up appointment.

Follow-up was available for 724 (72%) patients. Of these, 87 (12%) were seen in the ED at least once after discharge. The majority of these ED visits were unrelated to the original BLSI injury (i.e., new trauma, neurologic or orthopedic issues, illness); however, 20 (2.7%) patients were seen in the ED for abdominal pain. Five of these patients (0.7%) were rehospitalized. One patient was rehospitalized in another state (at a nonstudy site) for a fluid collection after pancreatic surgery and underwent additional procedures related to the pancreatic injury. In addition to the study patients, the 10 sites cared for 10 patients who were not formally enrolled and died before CT scanning and without surgery. Another 14 underwent operative exploration without CT and without an attempt at NOM. Considering the additional 14 patients that failed NOM, the total failure rate in this population is 8% (83 of 1,022) and probably closer to 4.7% for bleeding (34 study patients + 14 nonstudy patients).

Back to Top | Article Outline


This study represents the first prospective multicenter study to show that the frequency of failure for NOM in pediatric BLSI is low, but higher than previously reported. Although the overall rate in this study is 7%, nearly half of failures are attributable to injuries other than hemorrhage from BLSI. Hollow viscus injury and isolated pancreatic injury have been previously demonstrated as accounting for a significant portion of failures of NOM, particularly in pediatric patients.10 Similarly, in this cohort, nearly a third of failures were on patients with intestinal injury; an additional subset of patients underwent delayed surgery for abdominal washout or drainage of retained bile or blood rather than ongoing hemorrhage. This technique of laparoscopic washout of peritonitis was performed at seven of the ten centers and has been previously described in adults, but has not been reported in the management of pediatric BLSI.11 In many cases, the bleeding was successfully managed without surgery, but the patient was not.

The organ-specific failure rates were 8.3% for spleen and 7.4% for liver in this study, but only 4.5% and 3.8% for hemorrhage, respectively, and 3.9% and 0% for isolated injury (without any significant additional injuries). These rates are similar to the 3% failure of NOM for isolated liver reported by Holmes et al.,10 but significantly lower than the 4% reported for isolated spleen injury. Over half (12 of 23) of the splenic injury patients that failed for bleeding also had a liver injury and were therefore counted in both groups.

Several patient characteristics were identified as increasing the risk of failure. Those patients with multiple intraabdominal injuries and those with a higher grade of injury documented on CT were more likely to fail in this cohort. Conversely, children with isolated injuries have a decreased risk of failure, and no child whose only injury was a splenic injury failed NOM. Sex was not associated with risk of failure, nor was age. There was no difference in the risk of failure of NOM for liver compared with splenic injury, but children with combined liver and spleen injuries were more likely to fail NOM, as were children who also sustained a pancreatic injury. Children who were transferred to the PTC were less likely to fail NOM than those who came directly to the PTC. Similarly, the shorter times from injury to arrival at the PTC were associated with an increased risk of failure.

Although higher heart rate is an expected risk of failure, the association of low blood pressure and failure has been somewhat controversial in pediatric trauma, where several authors have questioned the strength of association of hypotension and bleeding in children with BLSI.12 These vital sign findings also suggest shock index or the recently described pediatric-adjusted shock index may have merit.13–15

In contrast to other pediatric studies, children with contrast extravasation (CE) on CT were more likely to fail NOM than those without.1,16 This is the first large study to document an association between failure of NOM and CE in children.

When failure of NOM occurred, the mortality was fairly high at 12% in this cohort. Overall, 25 patients died; eight after abdominal surgery for bleeding. Of those eight patients, the primary cause of death was bleeding in only 2 (25%). Other causes of death were traumatic brain injury (TBI) with or without hypoxia,4 multisystem trauma with respiratory failure,1 and 1 undetermined. For the 17 patients who died without abdominal surgery, 12 deaths were due solely to TBI, 2 TBI with anoxia, 2 due to TBI with bleeding, and 1 due to atlanto-occipital dislocation with anoxia. Although other studies have shown that when operative management of BLSI is required, mortality is high, the data from this report show that it is actually much lower than previously reported.10,17 An algorithmic approach to management defining failure of the NOM algorithm at 40 mL/kg of blood, or four units, potentially may have accounted for the lower mortality either through higher success by utilization of a defined end point, or via selection bias.3,18

Among the 14 nonenrolled patients who went straight to the OR without imaging, all were hemodynamically unstable except one who presented several days after injury with peritonitis. Several had witnessed cardiac arrest and ED thoracotomies, and many had a Glasgow Coma Score of 3. At least eight of the patients died, and several of the deaths were in the operating room.

Back to Top | Article Outline


Ten patients with BLSI died without undergoing CT or laparotomy and were not enrolled in the study. This would have increased the mortality rate from 2.5% to 3.4%. An additional 14 patients failed NOM immediately without CT scan or any attempt at NOM, which would increase the total NOM failure frequency to 8%. Additionally, although the ATOMAC algorithm for management is used at all 10 centers, surgeon adherence to the algorithm was not analyzed. Nonadherence to the algorithm may have impacted the failure rate of NOM. Routine reimaging was not performed in these patients, consistent with the literature.18,19 However, during follow-up, 17% of the patients required some form of radiographic imaging for symptoms. Despite good follow-up for 72% of patients, the study may have missed patients who returned to other hospitals with a complication secondary to their BLSI. Although most of the literature supports no further follow-up imaging, some authors suggest there may be a role for imaging in high-grade injuries looking for pseudoaneurysms, and these may have been missed if presenting with symptoms after the 60-day follow-up period.3,20,21

Back to Top | Article Outline


Overall, failure of NOM for pediatric BLSI is low, but higher than previously reported. Half of the failures are for reasons other than ongoing hemorrhage. When failure occurs for bleeding, it typically occurs at a median of 5 hours after injury, and 2.5 hours of arrival to PTC. Failure of NOM of isolated liver and spleen injuries is uncommon at 3.8% and 0%, respectively, but much higher for multiply injured patients. CE, early transfusion, and injury of multiple intraabdominal organs are associated with failure of NOM. When angioembolization was done for continued bleeding, 37% failed despite embolization. The overall mortality after BLSI is 2.5%, but most deaths are attributable to reasons other than hemorrhage.

Back to Top | Article Outline


M.E.L. contributed in the interpretation of data, performance of the literature search, wrote the initial draft of the article, and did critical revision of the article. C.S.L. contributed to data acquisition, data analysis, integrity of the data, drafting of the article, critical revision of the article. N.M.G contributed to interpretation of the data, critical revision of the article, and study supervision. A.C.A contributed to interpretation of the data, critical revision of the article, and study supervision. J.W.E contributed to interpretation of the data, critical revision of the article, and study supervision. R.T.M. contributed to interpretation of the data, critical revision of the article, and study supervision. R.W.L. contributed to interpretation of the data, critical revision of the article, and study supervision. T.A.P. contributed to interpretation of the data, critical revision of the article, and study supervision. S.D.S. contributed to interpretation of the data, critical revision of the article, and study supervision. C.L contributed to interpretation of the data, critical revision of the article, and study supervision. A.B. contributed to interpretation of the data, critical revision of the article, and study supervision. D.J.O. contributed to interpretation of the data, critical revision of the article, and study supervision. D.W.T. contributed to interpretation of the data, critical revision of the article, and study supervision. K.A.L. contributed to the interpretation of data and critical revision of the article. A.R.R. contributed to the interpretation of data and critical revision of the article. D.M.N. conceived and supervised this study and the initial study design, led the study group, wrote sections of the article, did critical revision of the entire article, and provided study supervision.

Back to Top | Article Outline


The authors declare no conflicts of interest.

Back to Top | Article Outline


1. Bansal S, Karrer FM, Hansen K, Partrick DA. Contrast blush in pediatric blunt splenic trauma does not warrant the routine use of angiography and embolization. Am J Surg. 2015;210(2):345–350.
2. Bass BL, Eichelberger MR, Schisgall R, Randolph JG. Hazards of nonoperative therapy of hepatic injury in children. J Trauma. 1984;24(11):978–982.
3. Notrica DM, Eubanks JW 3rd, Tuggle DW, Maxson RT, Letton RW, Garcia NM, Alder AC, Lawson KA, St Peter SD, Megison S, et al. Nonoperative management of blunt liver and spleen injury in children: evaluation of the ATOMAC guideline using GRADE. J Trauma Acute Care Surg. 2015;79(4):683–693.
4. Notrica DM. Pediatric blunt abdominal trauma: current management. Curr Opin Crit Care. 2015.
5. St. Peter SD, Sharp SW, Snyder CL, Sharp RJ, Andrews WS, Murphy JP, Islam S, Holcomb Iii GW, Ostlie DJ. Prospective validation of an abbreviated bedrest protocol in the management of blunt spleen and liver injury in children. J Pediatr Surg. 2011;46(1):173–177.
6. Golden J, Mitchell I, Kuzniewski S, Lipskar A, Prince JM, Bank M, Stylianos S, Rosen NG. Reducing scheduled phlebotomy in stable pediatric patients with blunt liver or spleen injury. J Pediatr Surg. 2014;49(5):759–762.
7. McVay M, Kokoska E, Jackson R, Smith S. Throwing out the “grade” book: management of isolated spleen and liver injury based on hemodynamic status. J Pediatr Surg. 2008;43(6):1072–1076.
8. Fremgen HE, Bratton SL, Metzger RR, Barnhart DC. Pediatric liver lacerations and intensive care: evaluation of ICU triage strategies. Pediatr Crit Care Med. 2014;15(4):e183–e191.
9. Dodgion CM, Gosain A, Rogers A, St Peter SD, Nichol PF, Ostlie DJ. National trends in pediatric blunt spleen and liver injury management and potential benefits of an abbreviated bed rest protocol. J Pediatr Surg. 2014;49(6):1004–1008.
10. Holmes JH, Wiebe DJ, Tataria M, Mattix KD, Mooney DP, Scaife ER, Brown RL, Groner JI, Brundage SI, Tres Scherer LR 3rd, et al. The failure of nonoperative management in pediatric solid organ injury: a multi-institutional experience. J Trauma. 2005;59(6):1309–1313.
11. Franklin GA, Richardson JD, Brown AL, Christmas AB, Miller FB, Harbrecht BG, Carrillo EH. Prevention of bile peritonitis by laparoscopic evacuation and lavage after nonoperative treatment of liver injuries. Am Surg. 2007;73(6):611–617.
12. Partrick DA, Bensard DD, Janik JS, Karrer FM. Is hypotension a reliable indicator of blood loss from traumatic injury in children? Am J Surg. 2002;184(6):555–559; discussion 9–60.
13. Acker SN, Ross JT, Partrick DA, Tong S, Bensard DD. Pediatric specific shock index accurately identifies severely injured children. J Pediatr Surg. 2015;50(2):331–334.
14. Linnaus ME, Notrica DM, Langlais CS, St Peter SD, Leys CM, Ostlie DJ, Maxson RT, Ponsky T, Tuggle DW, Eubanks JW 3rd, et al. Prospective validation of the shock index pediatric-adjusted (SIPA) in blunt liver and spleen trauma: an ATOMAC+ study. J Pediatr Surg. 2016. pii: S0022-3468(16)30417-1.
15. Haider AA, Azim A, Rhee P, Kulvatunyou N, Ibraheem K, Tang A, O'Keeffe T, Iftikhar H, Vercruysse G, Joseph B. Substituting systolic blood pressure with shock index in the national trauma triage protocol. J Trauma Acute Care Surg. 2016;81(6):1136–1141.
16. Davies DA, Ein SH, Pearl R, Langer JC, Traubici J, Mikrogianakis A, Wales PW. What is the significance of contrast “blush” in pediatric blunt splenic trauma? J Pediatr Surg. 2010;45(5):916–920.
17. De Jong WJJ, Nellensteijn DR, ten Duis HJ, Albers MJIJ, El Moumni M, Hulscher JBF. Blunt splenic trauma in children: are we too careful? Eur J Pediatr Surg. 2011;21(4):234–237.
18. Notrica DM. Pediatric blunt solid organ injury: beyond the APSA guidelines. Curr Surg Rep. 2015;3(4):1–6.
19. Stylianos S. Evidence-based guidelines for resource utilization in children with isolated spleen or liver injury. The APSA Trauma Committee. J Pediatr Surg. 2000;35(2):164–167; discussion 7–9.
20. Safavi A, Beaudry P, Jamieson D, Murphy JJ. Traumatic pseudoaneurysms of the liver and spleen in children: is routine screening warranted? J Pediatr Surg. 2011;46(5):938–941.
21. Navarro O, Babyn P, Pearl RH. The value of routine follow-up imaging in pediatric blunt liver trauma. Pediatr Radiol. 2000;30(8):546–550.
Back to Top | Article Outline


Dr. Barbara A. Gaines (Pittsburgh, Pennsylvania): It gives me great pleasure to discuss this excellently-presented and well-written paper exploring the failures of nonoperative management of liver and spleen injuries in the pediatric population.

Nonoperative management of blunt injuries to the liver and spleen was first described in 1977, a millennium ago. In 1981, Wesson published a case series of 63 children initially managed nonoperatively at the Hospital for Sick Children in Toronto.

The 71% success rate demonstrated that the majority of these injuries do not require operative intervention. This publication was truly revolutionary. And by the late 1990s the paradigm had completely shifted to where operative intervention for these injuries was distinctly unusual.

In 2000, Stylianos, the chair of the American Pediatric Surgical Association Committee on Trauma, published guidelines on nonoperative management, based primarily on grade of injury.

A second publication in 2002 validated the guidelines. So, by 2002, nonoperative management of blunt liver and spleen injuries in children was the norm, with expected success rates of 90% to 95%.

Saint Peter in 2008 and later the ATOMAC group revised the APSA guidelines, basing intervention such as hospital length-of-stay and ICU admission on physiologic criteria instead of purely anatomic grade.

The rate of splenectomy in pediatric trauma patients is now an important quality indicator, particularly for pediatric trauma centers.

Peds-TQIP measures rates of operative intervention. And recent reports demonstrate an all-comers rate of 5.1%, with only 1.8% of patients with an isolated splenic injury undergoing an operation.

Thus, at this time, it is well established that most children with blunt injury to the liver or spleen do not require an operation; but most is not all. Who are the patients who do require an operation? When they do they fail nonoperative management?

Previous studies have tried to shed some light on these issues. For example, in a 2009 multi-institutional study the distinction was made between those who required immediate operation—defined as within three hours of hospital arrival—versus those who failed nonoperative management.

The immediate operation group was hemodynamically labile; whereas, there was a high rate of pancreatic injuries in the failure group.

Failure was generally within 12 hours of injury. Fortunately, in this study, failure was not associated with adverse outcomes, including mortality and blood transfusion.

The current study adds to this body of literature. It is a multi-center study conducted by a group of institutions that follow similar protocols. A large number of children were enrolled in the study and the 72% follow-up rate is commendable.

Significant results include an overall failure rate of 7% with half of those failing secondary to bleeding and half for other injuries.

Time to failure was five hours from injury and three hours for the time of injury for bleeding patients. Interestingly, there was no incidence of delayed or late bleeding.

Patients with both liver and spleen injuries had a higher rate of failure, as did those with injuries to the pancreas and the kidney.

Other results include a 25% requirement for operative intervention with evidence of a contrast blush on CT and a 37% rate of failure of angiography.

I have several questions for the authors.

First is in regard to study methodology. The title of the study states that this is a prospective study. Were individual subjects enrolled in this study? Did they sign informed consent? Or was it really that their data was prospectively entered into a database and essentially retrospectively reviewed?

Along those lines, aside from the basic atomic protocol, were there standardized study protocols? Did a study coordinator arrange for the follow-up appointments? Was standard information collected at those visits? What was done to contact the 28% of subjects lost to follow-up?

I was also interested in your definition of “failure of nonoperative management.” In previous work the definition of “ nonoperative management” was greater than three hours of observation in the hospital.

It appears that in this study bleeding patients were in the OR within three hours from the time of injury. How long were these patients really in the emergency department? In other words, were these patients really failures of nonoperative management or actually successes in operative intervention?

Three. You demonstrated a 25% rate of failure for patients with a blush on CT and a 37% failure of angioembolization to prevent operative intervention.

In light of these findings what would you recommend we do for a ten-year old, hemodynamically-stable child with a splenic injury and a contrast blush on CT?

Should we observe the patient—in other words, nonoperatively management them? Or perform angiography and possible angioembolization? Or take them to the OR immediately?

And, finally, what should we do with this information? How do your findings change our management of patients? Alter our existing protocols? Change recommendations for follow-up?

Again, I would like to commend the authors for their contribution to the field and to the Association for the privilege of discussing this important paper.

Dr. Ronald J. Simon (New York, New York): A well-presented paper. Interesting stuff. Just to amplify a question by Dr. Gaines, we’re currently having a little bit of a debate about angiography in young children.

And I was wondering whether or not there is an age cut-off in which you would take a child with a splenic blush and not do angiography or take them to the operating room? So, basically, how young are you going that you are doing angiography?

Thank you.

Dr. Martin R. Eichelberger (Washington, D.C.): Congratulations. Excellent combination of patients and well done by this whole group. I have two questions.

I always have a concern when we include adolescents into the database for “children,” because I think anything less than 14 years of age has a different type of physiologic response and their mechanism of injuries are different than the drivers that are teenagers. Could you tease out the less than 14 year old age group to analyze if there was any difference in mortality or in complication rate?

The second question is what was the criteria for hypotension and shock in the children; what was the actual guidelines for initiation of transfusion for shock? How many ccs per kilo and when did you infuse? It would help all of us understand the definition of shock and when to infuse blood in this cohort of children.

Thank you very much. It was excellent.

DR. GREGORY J. “JERRY” JURKOVICH (Sacramento, California): What percent of patients were transferred into these institutions versus presented there primarily? One could argue that the transfers in with enough delays already have demonstrated that they are going to survive a nonoperative approach.

Also, was there any way to measure the intent of the admitting team to be a nonoperative patient? Or is this just a matter of the time delay? I struggle with the concept, like Dr. Gaines said, of calling this nonoperative failure.

To be a nonoperative failure it implies you had the intent to manage these patients nonoperatively. And I don’t think the data is clear on whether that was the intent on these patients or not. Thank you.

Dr. R. Stephen Smith (Gainesville, Florida): Nicely presented paper. I’d just ask for a little bit of clarification.

In your abstract you quoted an overall mortality rate of 21% for the patients who failed nonoperative management. I was trying to do some quick math, adding up your various mortality rates, and I just wanted to make sure that was the correct number. It seems somewhat high.

And I am assuming that these, at least two deaths, that resulted from ongoing hemorrhage were categorized as unanticipated mortalities at the various centers.

And what’s the PI process for these? I mean, quite honestly, I don’t think kids should bleed to death from splenic injuries or liver injuries, even if they’ve had an attempt at nonoperative management.

So, finally, what would be your strategy to prevent those occasional children who fail nonoperative management, who have ongoing hemorrhage, who require operation? How do you identify those earlier and get them to definitive therapy at an earlier state?

Dr. Saman Arbabi (Seattle, Washington): Nice presentation. I had two questions. You said the blush or active extravasation was associated with higher failure rate in nonoperative management. Did angioembolization change that? Was there a decrease in the rate of operative intervention in patients that had angioembolization?

The other thing you said is half of the patients had failure because they required bowel resection. Was this failure of management or failure of diagnosis? Was there anything in the CT scan that you could have used to diagnose bowel injury in these cases? Thank you.

Dr. Maria E. Linnaus (Phoenix, Arizona): So, first, thank you for the questions. Regarding Dr. Gaines’ first question, these patients were prospectively enrolled in this observational study once they were determined to meet inclusion criteria, mainly, having a blunt liver or spleen injury. Each patient was followed in a prospective manner and data was collected throughout the treatment course. Regarding consent, management of these study patients was based on an evidence-based algorithm that was the standard of care at all ten sites. Therefore, none of the 10 sites’ Institutional Review Boards required consent.

Next, the question was regarding standardized protocols. We do have standardized protocols for these patients at all of the ten ATOMAC sites. The management protocol for these patients was previously published by the ATOMAC group in the Journal of Trauma and Acute Care Surgery as discussed earlier in the presentation. Additional follow-up care was also standardized across the sites. This included standardized follow-up at fourteen and sixty days.

All patients received the same explicit discharge instructions on symptoms that should prompt an emergency department visit, activity restriction, medications after injury and so forth. Low-grade injuries had a follow-up phone call at fourteen days and again at sixty. Patients with high-grade injuries followed up in the trauma surgeon’s office around fourteen days post-injury and had a 60-day follow-up phone call in addition. For those discharged after fourteen days, only a 60-day follow-up phone call was performed. At each of these encounters, a standardized set of data was collected for each individual patient. The 28% that were lost to follow-up were unreachable by phone.

Next, the question was regarding the semantics of failure of nonoperative management versus success of operative intervention. This terminology can invoke some negative emotions, however, it is important to understand that a failure of nonoperative management merely implies a failure of the patients to stop bleeding or development of additional problem that required surgery rather than a failure of the surgeon or managing services. In these instances, operative intervention is the correct course of action. Given the high rates of successful nonoperative management in the literature, we felt it important to include all cases undergoing abdominal operation to provide an accurate and complete picture of pediatric patients requiring surgical intervention. In the manuscript, an analysis of isolated solid organ injury is also included, which provides additional important benchmark data.

Regarding the importance of contrast extravasation: What do we recommend? This is actually going to be the topic of another paper that comes out of this study on angioembolization so we will further delineate that. Again, we do not consider angioembolization to be a failure of nonoperative management. Angioembolization may have a role but requires a focused data analysis which will be forthcoming.

And then what do we do with this information? I think it is important to know the failure rates and the characteristics of those children at risk of failing nonoperative management because, like Dr. Gaines said, benchmarking and quality indicators are the upcoming trend for how trauma centers are judged.

With this increasing focus, we need to adjust the failure rates and characteristics of failure appropriately so that trauma centers aren’t being compared unfairly to those centers who have or do not have the sicker patients.

Dr. Simon asked about angiography, how young can we go? Again, we will further elicit this in our angioembolization study that is coming out of this. However, one year of age was the youngest done in this study.

Dr. Eicheleberger asked about adolescents- I think that’s a great topic of discussion. We haven’t done that for this paper but I think it’s a great suggestion and we will look into doing that for another manuscript. I am unaware of any studies assessing they fare better.

Hypotension we defined as systolic blood pressure of 55 plus 2 times the age of the patient or less. In children, shock and hypotension are not the same thing. For shock, we had a data variable that asked if the patient was in shock per the opinion of the treating surgeon. In many instances, this was no different than the other parameters used in the adult centers, such as acidosis, hypotension not responsive to fluid resuscitation, etc. For transfusion, we give 20mLs per kg for each transfusion. However, many of the patients received almost 40mL per kg or more for transfusion when they came in, indicating that they were very sick.

Regarding Dr. Jurkovich’s questions, approximately 50% of the patients were transferred from an outside institution and yes, a large proportion was successfully managed nonoperatively. However, over 40% of those who were considered a failure of nonoperative management were indeed transferred from an outside hospital, but I do think that is a valid point. In regard to the intent of nonoperative management, these patients were all with a CT-documented blunt liver or spleen injury. Therefore, already all of these patients were stable enough for a scan. Additionally, given that nonoperative management has been shown to be successful in the literature, any patient who is hemodynamically stable without peritonitis deserves an attempt at nonoperative management, particularly for isolated splenic injuries.

However, if they are hemodynamically unstable or if they are in need of an operation, I think that you absolutely need to operate and it’s not a failure of the surgeon; like I said, it’s a failure of the patient to stop bleeding, or a need for surgery for an associated injury or complication of nonoperative management.

Thank you Dr. Smith for bringing that up. Once our finalized data was entered and the numbers were looked at again, we actually had a 12% mortality rate for those who failed and a 2.5% rate overall. We analyzed if those patients who died somehow may have been delayed in operative therapy, and we found that only 2 patients died as a result of bleeding. One of these patients presented with a horrible TBI following a motor vehicle collision. Although the patient was hemodynamically unstable from head injury, intraabdominal injury, and pelvic fracture, the parents elected for comfort care only due to the prognosis of the head injury. The patient died in the trauma bay without an attempt at operative management. The other patient was transferred from an outside hospital with an ISS of 22. This patient arrived to the trauma center 2 hours post injury and was in the OR within an hour of arrival. In both cases, we felt their other injuries were severe enough to cause death.

To Dr. Arbabi’s question on how to identify those children who are going to fail, we have described some of the characteristics in this presentation, but high transfusion requirements and active blush on CT are certainly characteristics to be wary of when treating pediatric patients. In addition, patients with combined liver and splenic injuries or pancreatic injuries should be considered very high-risk for needing an operation.

As previously discussed, we will look more closely at those who underwent angioembolization but those who required angioembolization were more likely to have significant injuries warranting an operation. Then to your other point, those requiring a bowel operation were considered a failure to be congruent with the other literature regarding this topic. I do not think they were failures of diagnosis as everyone knows, CT imaging is not sensitive for picking up bowel injury.

I think that’s all the questions, everybody. Thank you.


Pediatric trauma; blunt liver and spleen injury; nonoperative management; failure

© 2017 Lippincott Williams & Wilkins, Inc.