The Journal — January 2017
Tex Med. 2017;113(1);e1.
By Michelle Buehner, MD; Jay Aden, PhD; Mathew Borgman, MD; Preston Love; Brandi Wright; and Mary Edwards, MD
Drs Buehner, Aden, Borgman, and Edwards, San Antonio Military Medical Center, Fort Sam Houston, Texas. Mr Love and Ms Wright, Southwest Texas Regional Advisory Council for Trauma. Send correspondence to Michelle Buehner, MD, Department of Surgery, San Antonio Military Medical Center, 3551 Roger Brooke Dr, Fort Sam Houston, TX 78234; email: firstname.lastname@example.org.
The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army, Department of the Air Force, or the Department of Defense.
The purpose of this study was to define the demographics of pediatric traumatic injuries and to understand the predictive value of injury type, prehospital, and emergency department (ED) data regarding the mortality of pediatric trauma patients (<14 years of age) in South Central Texas. We report a retrospective review of pediatric trauma patients presenting to Trauma Service Area P in South Central Texas during 2004–2013. The primary outcome was mortality; secondary outcomes were ventilator days, hospital days, and intensive care unit stay. Demographics, Abbreviated Injury Score (AIS) codes, ICD-9 codes, transport times, Emergency Medical Services (EMS) vital signs en route and on arrival to the emergency department (ED), and outcomes were evaluated. A total of 8004 traumatically injured children presented to EDs in South Central Texas during the study period; 4109 of these presented via EMS. Most patients were Hispanic and male. Distribution was even across age groups. Overall mortality was 2%, and the mortality of those arriving by EMS was 3.7%. Abnormal vital signs and Glasgow Coma Score upon presentation to both EMS and the ED were strongly associated with mortality. Increased Injury Severity Score, the need for transfusion in the ED, and increased maximal AIS were also strongly associated with mortality. African American race was associated with increased mortality, although transport time and age were not.
Most injuries overall were caused by motor vehicle collisions; however, burns and falls were most common in infants. The most lethal injuries were caused by firearms (mostly seen in preteens) and assaults (mostly seen in infants). This analysis of injured children in Southwest Texas offers insight into areas of needed quality improvement in the trauma system and potential areas to focus prevention efforts.
To achieve better outcomes and facilitate quality improvement and prevention efforts, regionalization of trauma care has become standard in both adult and pediatric trauma systems in the United States over the past several decades.1,2 Significant attention has been focused on the optimal care of pediatric trauma patients, with recent evidence that both young children and adolescents are best managed at pediatric trauma centers.3 However, tempering enthusiasm for universal rapid transfer of such patients to high levels of care are the known costs of transfer, particularly by air.4 As a result, rapid and accurate triage of patients is essential for appropriate distribution of resources.
In an attempt to study prehospital care in a rural, civilian trauma setting, the Remote Trauma Outcomes Research Network (RemTORN) was developed. Its purpose was to investigate civilian trauma patients with prolonged transport so as to assess demographics, timing, distance, injury characteristics, and potential interventions to help develop relevant clinical applications in combat casualty care.5 Ultimately, this project was enhanced by the evolution of the Southwest Texas Regional Advisory Council for Trauma (STRAC) Regional Trauma Registry. Sources of data for the initiative include a hospital registry (Collector, Digital Innovations, 2011) and another for Emergency Medical Services (EMS) data (Tablet PCR).5 This combined system allows provision of data from point of injury to discharge from the hospital.5 The system has been live for several years now and has been used in descriptive publications of adult trauma patients.6 At the time of data collection, Texas' Trauma Service Area P had a catchment area serving 2.4 million people in Southwest Texas. This region had 23 verified Texas State Level IV trauma centers, 3 Level III trauma centers, and 2 level I trauma centers, 1 with a pediatric designation (The University of Texas Health Science Center San Antonio), and more than 30 EMS providers.5 This region is largely rural, with 24-hour-a-day emergency services but very limited acute surgical availability.
The purpose of this study was to define the demographics and mortality risk factors of pediatric trauma patients in Texas Trauma Service Area P (South Central Texas) as the epidemiology of this population remains unknown. We sought to assess prehospital predictors of mortality and the impact of transfer time to a trauma center. We studied the data capture of the system for quality improvement. Also, we investigated the relationships among demographics, injury type, prehospital vital and emergency department (ED) signs, and transport times on outcomes.
After approval from the Brooke Army Medical Center Institutional Review Board, we conducted a retrospective analysis of pediatric trauma patients sustaining either blunt or penetrating injuries. Overseen by STRAC, the STRAC Regional Hospital Trauma Registry was queried from 2004 through 2013 for all pediatric trauma patients (aged ≤14 years) presenting to a treatment facility. The Trauma Service Area P includes a population of 2.4 million people from 22 counties. Patients aged 14 years or more and non-traumatic mechanisms of injury (eg, drowning) were excluded. In addition, patients arriving to the hospital by EMS were analyzed separately to ascertain the effect of transport time on mortality.
We collected demographics to include age, race, gender, height, weight, ethnicity, mechanism of injury, Injury Severity Score (ISS), and Abbreviated Injury Scale (AIS). Admission variables included both vital signs en route to a medical treatment facility and ED, time of transport, need for transfusion in the ED, and Glasgow Coma Score (GCS). Outcome data included length of stay, days in the intensive care unit (ICU), clinical outcome, cause of death if available, and days on ventilator. Categorical variables were summarized using percentages and chi-squared tests or the Fisher exact test as appropriate. Means and standard deviations or medians and the interquartile range were used as summary statistics for continuous variables; these were analyzed by using the t test and ANOVA or Wilcoxon's test, whichever was most suitable. Significance for results will be established when P values are less than .05. JMP v10.0 (Cary, NC) was used to perform all statistical analysis.
A total of 8004 children met inclusion criteria. The mean age of the children was 5.5 ± 4.2 years. Most were Hispanic (67.5%), male (62%), and suffered from blunt mechanism of injury (71%). ISS was 7 ± 8.6; overall mortality was 2%.
A comparison of demographics and injury severity between survivors and nonsurvivors is summarized in Table 1. Not surprisingly, the ISS and maximum AIS were significantly higher for the children who died; these scores were 4 [1,9] vs 30 [25,43] and 2 [1,3] vs 5 [5,5], P<.0001, respectively. Gender and age did not differ significantly between survivors and nonsurvivors.
Emergency department data collected on all children is listed in Table 2. Abnormal vital signs and lower GCS were associated with mortality (P<.0001). Compared with survivors, nonsurvivors had significantly longer stays in the ICU (1 [0,2] vs 0 [0,1], P=.02) and days on the ventilator (1 [1,2] vs 0 [0,0], P<.0001); survivors had longer hospital stays (2 [1,3] vs 1 [1,2], P<.0001).
Of the 8004 children, 7977 had a recorded mechanism of injury (Table 3). The most common causes of traumatic injury to children in Southwest Texas were motor vehicle injuries and burns (29% and 22%, respectively) with death mostly attributed to motor vehicle collisions. The most lethal injuries were due to firearms and assaults. The overall mortality of this cohort was 2%.
Of the 8004 patients, 4109 presented to care via EMS (Table 4). In general, the gender, age and racial characteristics of this group were similar to those of the entire study cohort. Mortality in this group was higher at 3.7%, P<.0001. As with the overall cohort, African American children carried a higher likelihood of death due to traumatic injury than did other children. Interestingly, transport time was not associated with mortality. As with ED vital signs, abnormal EMS vital signs and decreased GCS were more frequent in the group of nonsurvivors (P<.0001).
No significant variation was seen in the outcomes of mortality, hospital days, ICU days, or ventilator days; however, as a factor of age, significant differences did appear in mechanisms and patterns of injury. The distribution of patients was similar in the four age groups (age ≤1 year [24%], ages 2-4 years [25%], ages 5-8 years [23%], and ages 9-13 years [28%]). Infants were more likely to have a serious injury (AIS ≥3) to body region 1 (head and cervical spine) and older children (ages 9-13 years) to region 5 (extremity and bony pelvis). The distribution of injury mechanism within age groups and the proportion of age groups with each injury mechanism are displayed in Figures 1 and 2. Infants (≤1 year) were most likely to suffer from burns and falls in addition to assaults. Toddlers aged 2-4 years were most likely to suffer from a bite or sting, and children aged 2-8 years were most likely to suffer from a burn, a fall, or a motor vehicle injury. Older children aged 9-13 years most commonly sustained a motor vehicle injury, bicycle injury, or a firearm injury.
As seen in Table 5, children arriving via EMS had a lower GCS, systolic blood pressure (SBP), and heart rate (P<.05) compared with those transported to the hospital by personal transport. In addition, these pediatric trauma patients were more likely to be more severely injured with a higher ISS (P<.0001). Of note, despite their more severe injuries and depressed vital signs, no significant difference was seen in temperature between the two groups. On logistic regression (Table 6), SBP, GCS, African American race, age, and ISS were independently associated with mortality. After adjusting for GCS, SBP, and ISS, a significant predictor of mortality was race, which showed African American with a 3.6-fold increase in the risk of death over Caucasians. Age was also a significant predictor of mortality, with children younger than 2 years most at risk.
The first study of pediatric trauma patients in South Central Texas, our research revealed significant associations among age, race, injury mechanism, vital signs on presentation, and outcomes. We found no significant difference in mortality between females and males, unlike that of national data.7 African American children, those suffering from firearm injuries, and those who were victims of assault had the highest mortality. Increased maximum AIS, ISS, and need for transfusion in the ED were associated with mortality on univariate analyses, as were abnormal vital signs and decreased GCS as recorded in the ED and by EMS. Transport time did not vary significantly among survivors and nonsurvivors, although the cohort brought in by EMS had a higher overall mortality (3.7% vs 2%, P<.0001). Assaults were most likely to injure infants and firearms to injure older children. The most common injury mechanism was a motor vehicle collision, followed by burns and falls. Not surprisingly, traumatically injured children in this group who presented to medical care via EMS had a higher mortality given their higher ISS and depressed vital signs. Note, also, that African American children had inferior outcomes compared with other races, in a catchment area that is mostly Hispanic. Race was an independent predictor of mortality (P=.04); however, mechanism of injury could not be controlled for in our model in addition to the other variables. Overall mortality in this group was comparable to that published from pediatric trauma centers elsewhere.7
Triage is critical to ensuring appropriate rapid transfer in an area with limited surgical resources. In this limited analysis, transport time did not appear to be linked to outcome, which suggests that delays in transfer did not affect mortality. An interesting finding showed no difference in temperature between the EMS arrival group and those transported by personal vehicle, despite their higher injury severity.
Information in this study pertaining to mechanism is helpful in identifying areas to focus on in this community regarding prevention initiatives. Most assaults in this cohort involved infants, and this group of injuries carried the highest mortality rate (16%). Age was also an independent predictor of mortality with those younger than 2 years at the highest risk of death (P=.0129). As these injuries are higher risk for intentional injury, consideration should be given to focusing on community prevention and reporting of nonaccidental trauma in this area. Literature regarding prevention efforts in nonaccidental trauma is scarce, suggesting the need for future research into this area. Most thermal injuries and falls occurred also in the youngest children, suggesting a need for prevention efforts in the home. Fortunately, these injuries did not carry a high mortality; however, morbidity is not captured here. Motor vehicle collisions were the most common mechanisms of injury in all remaining age groups.
In spite of varying injury mechanisms with age, age in and of itself was not associated with significant changes in mortality. Infants had a propensity for serious head injury, while older children were more likely to sustain serious extremity injury. ISS, maximum AIS, African American race, abnormal vital signs, and decreased GCS ― both in the ambulance and in the emergency room ― were all much more common in nonsurvivors.
This analysis has several limitations. Vital signs and GCS were documented in the hospital trauma registry in less than 50% of patients transported by EMS. Database capture of transport time over time was similarly not well recorded in the hospital trauma registry over the study period reviewed and did not change over time. The predictive value of EMS vital signs has been well established in other studies and is a validated method of triage.4,8 Failure to capture vital signs in pediatric patients has also been noted during early development of pediatric trauma systems and is often used as a marker for quality improvement.9
The STRAC Regional Trauma Registry has been ongoing for several years, and thousands of records have been combined.6 Issues with data duplication, dys-synchrony, and human error are inherent in a retrospective database analysis and limit the ability to perform advance statistical analysis such as multivariate logistic regression to investigate independent associations with mortality. The issues with limited reporting of data points introduce the possibility of reporting bias, which might significantly affect the observation that EMS vital signs are predictive of mortality. However, given that this observation was also seen with ED vital signs (which were well recorded), it stands to reason that there is some validity to this. Given that this a retrospective review, the impact of survivor bias and selection bias also deserve mention, although neither of these should impact the analysis of demographics or mechanism of injury.
This study of approximately 8000 pediatric trauma patients in Southwest Texas identifies several potential areas of quality improvement for the pediatric trauma system and provides important insight on potential areas to target prevention efforts. Continued documentation of transport times is needed to identify potential gaps in triage and allow for ongoing quality improvement efforts in addition to potential interventions that might improve survival. Community safety initiatives should consider targeting preventing burns, falls, and nonaccidental trauma in children younger than 1 year. In addition, investigation into prevention measures for nonaccidental trauma and its potential effects on outcomes need to be evaluated. For our particular region, further research into the racial disparity of survival also needs attention.