3,065 patients were admitted over the study period and 49 died (mortality rate=1.6%). Patients that died were primarily male (73%) with severe injuries (ISS 32 (19)) caused by a blunt (85%) mechanism. Fifteen deaths (31%) were the result of non-accidental trauma (NAT). Characteristics of study patients are presented in Table 1. Autopsy reports were available in 27 (55%) of cases. Based on autopsy findings, 15 of 27 (55%) patients required revision of both ISS and TRISS, resulting in a significant increase in ISS and decrease in TRISS (Table 2).

Seven patients had an ISS of 75 indicating non-survivable injuries. After calculating TRISS on the remaining 42 patients, 12 patients had a >50% probability of survival and 30 patients had a <50% probability of survival. There were no differences in age, gender, mechanism, ISS, head abbreviated injury scale (AIS) or rate of NAT between the two groups. All of the preventable deaths (n=3) occurred in patients with TRISS >50% (Table 3).

After multi-disciplinary review, three deaths were classified as preventable resulting in a PDR of 6%. The cause of death in two cases was failure to control hemorrhage and subsequent exsanguination. The cause of death in the third case was failure to secure an airway leading to respiratory arrest followed by cardiac arrest. Two of the PDs occurred in the pediatric intensive care unit (PICU) and one occurred in the operating room (OR). Table 4 provides a summary of the PDs.

Our data indicates that the overall mortality rate for pediatric trauma patients admitted to a mature (adult and pediatric) trauma center is low at 1.6% and the majority (57%) of patients that die have severe traumatic brain injuries (head AIS >4). The most important findings of the current report are that the PDR for pediatric trauma patients was 6%. Furthermore, a TRISS >50% correlates significantly with PD. Autopsy reports, when available, assist in the accurate calculation of ISS and TRISS and may clarify the actual cause of death in some cases.

Overall mortality rate is not an adequate reflection of pediatric trauma center performance and/or quality of care as the overall mortality rates in children are generally low. Several studies have attempted to determine which types of centers provide the best care to injured children (pediatric trauma centers v. adult trauma centers v. mixed trauma centers), however these studies use overall in-hospital mortality rates as a basis for comparison [5-9]. We believe that the PDR is a more appropriate area of focus as it quantifies the number of patients who died as a result of deficiencies in care. In-hospital PD has not been well characterized in pediatric trauma patients and a baseline PDR has not been established for comparative purposes between centers. Our process of identifying PD included individual case review of all pediatric deaths in trauma M&M conference as well as a subsequent secondary review by a multi-disciplinary panel that included the adult and pediatric trauma medical directors. As stated previously, a third level of review was performed by site reviewers from the ACS of 2014 and 2016 deaths.

Through this intensive review process, three PD’s were identified over the study period. Two deaths were a result of uncontrolled hemorrhage. One of these patients was transferred to our facility after initial presentation to a non-trauma center. She was a restrained backseat passenger in a motor vehicle collision with a Glasgow Coma Score of 15 at the scene complaining of abdominal pain. At the initial hospital she was taken to the operating room (OR) for the management of spleen and liver injuries. Surgeons were unable to control hemorrhage and she was transferred directly from the OR at the initial facility to the OR at our trauma center. The total time from her accident to transfer to a trauma center was seven hours. On arrival to the trauma center she was acidotic, hypothermic, coagulopathic and expired in the OR. From the operative notes, these appeared to be survivable injuries with a failure to control hemorrhage and subsequent exsanguination. The second patient was ejected from a motor vehicle and sustained traumatic liver lacerations with active extravasation, a right renal laceration with adrenal hematoma and left lung lacerations. The patient was taken to interventional radiology for angioembolization of the liver and admitted to the pediatric ICU (PICU). Approximately 7 hours after admission, the patient had a drop in hemoglobin to 6, became hypotensive and bradycardic, progressed to PEA arrest and expired in the PICU. On review, this was considered a potentially PD related to ongoing, unrecognized hemorrhage.

The third PD was classified as failure to secure an airway in a patient initially admitted after being “found down” at home and unresponsive with injuries concerning for non-accidental trauma. He had a traumatic subdural hemorrhage, required craniectomy and was admitted to the PICU. On day 19 of his hospital stay he became bradycardic and desaturated and the determination was made that his tracheostomy was occluded. A bedside tracheostomy change was attempted however the patient progressed to cardiac arrest during this procedure and expired. On review, this was considered a PD related to inability to secure an airway.

The PDs identified in this study represent opportunities for trauma performance improvement (PI). According to the American College of Surgeons Committee on Trauma, an effective PI process requires monitoring clinical processes, outcomes and systems issues and identifying sentinel events. The process must also result in the development of corrective actions for these events to prevent future recurrence [10]. In the case of the patient initially brought to a nontrauma center, extensive education and outreach was conducted with pre-hospital providers as well as surrounding hospitals (non-trauma centers) to emphasize the importance of bringing the “right patient to the right place the first time”. To mitigate systems issues and decrease delays, a transfer center was subsequently established at our hospital to ensure maximum efficiency for patient transfers. In addition our institution committed to becoming verified as a pediatric trauma center.

The in-hospital PD from exsanguination required individual provider education and remediation by the trauma medical director. In the case of the PD from a failure to establish an airway, a multidisciplinary group established an emergent airway protocol for the institution that clearly outlined the escalation pathway for assistance in difficult airway cases. All trauma providers were oriented to the protocol and adherence to the protocol is prospectively monitored.

The use of autopsy reports to calculate accurate ISS and TRISS was an important part of this retrospective review. In 1981, West JG advocated the use of autopsy reports as an inexpensive and valuable method to evaluate trauma care [11]. Ramenofsky ML et al. also used available autopsy reports to determine survivability in pediatric trauma patients [3]. In our review, there were several patients that died shortly after presentation without the benefit of CT scans to identify injuries. These patients had erroneously low ISS and high TRISS. Exhaustive attempts were made to collect autopsy reports from the office of the medical examiner and all available reports were reviewed by the trauma medical director and incorporated into the secondary review process. The trauma registry was updated to reflect the re-calculated ISS.

TRISS was also calculated specifically to determine whether or not higher TRISS were associated with PD. In 1991, Kaufmann CR et al. evaluated the validity of applying TRISS to injured children. They determined that despite the physiologic differences between adults and children, TRISS was applicable to children and a valuable tool to evaluate pediatric trauma care [12]. In 2006, Aaland et al also advocated the use of risk-adjusted mortality based on TRISS to critically evaluate pediatric trauma deaths [13]. Stelfox HT et al. systematically reviewed quality indicators for evaluating pediatric trauma care and suggested that “TRISS analysis may be a useful tool for identifying children with unexpected deaths whose care should be further reviewed” [14]. The results of our review support this statement as we found that all PD’s occurred in patients with a TRISS >50%. As a result, we currently calculate TRISS for all pediatric trauma deaths and apply the primary and secondary review process outlined previously those with a TRISS >50. Patients with a TRISS <50 are reviewed once during trauma M&M.

In conclusion, the overall PDR for pediatric trauma patients is 6% at a mature trauma center. A TRISS >50% correlates significantly with PD. The causes of PD in the pediatric population were failure to control hemorrhage and failure to secure an airway.