Journal of Emergencies, Trauma, and Shock

: 2015  |  Volume : 8  |  Issue : 1  |  Page : 30--33

Preoperative evaluation of penetrating esophageal trauma in the current era: An analysis of the National Trauma Data Bank

Daniel Dante Yeh1, John O Hwabejire2, Marc de Moya1, David R King1, Peter Fagenholz1, Haytham MA Kaafarani2, Eric N Klein3,  
1 Massachusetts General Hospital, Department of Surgery, Division of Trauma, Emergency Surgery, and Surgical Critical Care; Harvard Medical School, Boston, MA, USA
2 Howard University Hospital, Washington, DC, USA
3 Department of Surgery, Hartford Hospital, Hartford, Connecticut, USA

Correspondence Address:
Daniel Dante Yeh
Massachusetts General Hospital, Department of Surgery, Division of Trauma, Emergency Surgery, and Surgical Critical Care; Harvard Medical School, Boston, MA


Background: Preoperative diagnostic evaluation (PDE) of penetrating esophageal injury (PeEsIn) can delay treatment and increase morbidity. We sought to study the relationship among PDE, delay in definitive treatment, and patient mortality in PeEsIn. Materials and Methods: The 2008-2010 National Trauma Data Banks were queried for PeEsIn. Exclusion criteria were death within 1 day of injury, and missing data about survival to discharge or operative intervention. Data extracted included demographics, vital signs, injury severity, diagnostic procedures (endoscopy, computed tomography, and fluoroscopy), time to procedures and/or operation, hospital-free days, and mortality. Results: Of 280 patients, 75 underwent PDE and 205 did not. There were no significant differences in baseline demographics, vital signs or injury severity between the two groups. The median time to the first operation was shorter in the nonPDE cohort compared to the PDE cohort (2 vs. 3 h; P = 0.018). Median hospital-free days at day 60 were significantly less in nonPDE (42 days, interquartile range ([IQR] = [28, 50]) versus PDE patients (47 days, IQR = [38, 51]) (P = 0.007). Mortality was not statistically different. Conclusions: PDE in PeEsIn slightly delays the time to operation without worsening mortality, and is a predictor of more hospital-free days

How to cite this article:
Yeh DD, Hwabejire JO, de Moya M, King DR, Fagenholz P, Kaafarani HM, Klein EN. Preoperative evaluation of penetrating esophageal trauma in the current era: An analysis of the National Trauma Data Bank.J Emerg Trauma Shock 2015;8:30-33

How to cite this URL:
Yeh DD, Hwabejire JO, de Moya M, King DR, Fagenholz P, Kaafarani HM, Klein EN. Preoperative evaluation of penetrating esophageal trauma in the current era: An analysis of the National Trauma Data Bank. J Emerg Trauma Shock [serial online] 2015 [cited 2021 Apr 19 ];8:30-33
Available from:

Full Text


Penetrating esophageal injuries (PeEsIns) are rare, accounting for <0.5% of all trauma admissions even in busy urban trauma centers. [1] Overall mortality is high (44%), but the vast majority of these deaths occur within the first 24 h and are likely not attributable to the esophageal injury, but rather to severe associated injuries. [2] Since the diagnosis of esophageal injury is not always straightforward, preoperative diagnostic evaluation (PDE) adjuncts such as endoscopy, computed tomography (CT), and fluoroscopy may be helpful. However, performing these studies may delay operative repair and potentially increase morbidity. [3],[4],[5],[6] Because of the rarity of this injury, previously published studies trying to address the pros and cons of PDE have been limited by small sample size and single center bias. [1],[6],[7]

In 2001, Asensio et al. reported in a multicenter American Association for the Surgery of Trauma (AAST) retrospective study that preoperative evaluation was associated with a significant delay from admission to operative repair (13 h vs. 1 h, P < 0.001) and with increased complications. [8] The study recommended that the definitive repair should be a priority and not be delayed unnecessarily. In the years since that landmark publication, the immediate availability of adjunct PDE modalities has improved in most centers, and thus, the time it takes to obtain these diagnostic tests might have decreased. We sought to re-examine the impact of PDE in the current era. We also hypothesized that promptly obtaining adjunctive diagnostic studies to aid in the diagnosis of PeEsIn is associated with improved outcomes.


The National Trauma Data Bank (NTDB) from 2008 to 2010 was queried for patients who suffered esophageal injuries based on International Classification of Diseases codes 862.22 (injury of esophagus closed) or 862.32 (injury of esophagus open). The query was limited to patients who suffered penetrating trauma based on the E-code classification of their mechanism of injury. The NTDB was created by the American College of Surgeons Committee on Trauma in 1997 in order to collect retrospective data about trauma patients. Data are submitted from a variety of trauma settings which captures the effects of differing practice patterns on patients with similar injuries. A research dataset is published each year which can facilitate outcomes studies on rare injuries like PeEsIns.

Patients who expired within 1 day were excluded from the study because their injuries were likely not survivable, and their death was unlikely to be attributable to the esophageal injury. Two data elements were critical in our analysis, an operative intervention with a documented time delay and whether or not the patient survived to hospital discharge. If either of these data were missing, the patient was excluded from the study.

Additional data about each patient were extracted from the NTDB. Demographic data included age and gender. The systolic blood pressure (SBP), heart rate (HR), respiratory rate (RR), and Glasgow Coma Scale (GCS) upon arrival at the hospital were captured. The modified shock index (HR/SBP) and the revised trauma score (RTS, a composite score based on GCS, SBP, and RR) were calculated for each patient. The patients' injury severity score (ISS) submitted by the trauma center was used in our analysis. Esophageal diagnostic procedures were defined as: CT, endoscopy, barium swallow, and other diagnostic procedures on the esophagus. We did not include esophageal diagnostic procedures that occurred more than 1 day after the operative repair, as it may represent routine postoperative care (e.g., postrepair swallow study) or was obtained for unrelated reasons (e.g., CT chest to assess for pneumonia). The time in hours of the first esophageal diagnostic procedure, the first surgical procedure and the first surgical procedure on the esophagus were queried from the database. The population were divided into two cohorts based on whether or not they underwent an esophageal diagnostic procedure prior to the conclusion of their first surgical procedure.

The following outcomes were extracted: Length of hospital stay (LOS) and whether the patient survived until discharge. The patients' hospital-free days at day 60 were calculated (60 - LOS). Readmission data are not available in the NTDB and could not be included in our calculation. Patients who did not survive to discharge were considered to have zero hospital-free days.

Statistical analysis was performed using R version 2.15.1 (R Foundation for Statistical Computing, Vienna, Austria). A P < 0.05 was considered statistically significant. Continuous variables were reported as means and standard deviations or medians and interquartile ranges (IQRs) as appropriate. Means were compared using a Welch two-tailed t-test. Medians were compared using the Wilcoxon rank sum test. Proportions between two groups were compared using Fisher's exact test. Linear regression was used to correlate continuous variables to other continuous and categorical variables; model coefficients of independent variables are reported with 95% confidence intervals. Analysis of variance with Tukey's posthoc test was used to correlate continuous variables to categorical variables, and results are reported as absolute differences with 95% confidence intervals.



Our query of the NTDB found 523 patients who suffered esophageal injuries as a result of penetrating trauma. Of these 523 patients, 165 (31.5%) were excluded because they expired within 1 day, 74 (14.1%) were excluded because no surgical procedure was documented, and 4 (0.8%) were excluded because survival to hospital discharge was not documented. Therefore, 280 patients remained for further analysis [Figure 1]. These 280 patients were divided into two cohorts based on whether they underwent a PDE to assess for esophageal injury prior to the conclusion of their first operation (PDE, n = 75, 26.8%) or whether they left the operating room without having undergone such a procedure (nonPDE, n = 205, 73.2%). The characteristics of the two cohorts are displayed in [Table 1]. When comparing the demographics of the two cohorts, there were no significant differences in baseline demographics, vital signs upon admission or calculated injury scores (ISS and RTS).{Figure 1}{Table 1}


The distribution of the first esophageal diagnostic procedures performed in the PDE cohort is listed in [Table 2]. The majority of these patients (66.7%) underwent endoscopy. The median time until the first operation was 1 h shorter in the nonPDE (2 h, IQR = [1, 5]) as compared to the PDE cohort (3 h IQR = [2, 6]) (P = 0.018).{Table 2}


[Table 3] lists the unadjusted outcomes of the two cohorts. Before controlling for ISS, mortality was significantly greater in the NO_DIAGNOSTIC group than the DIAGNOSTIC group, 5.4% and 0%, respectively (P = 0.04). Median hospital LOS of survivors was also significantly longer in the NO_DIAGNOSTIC group than the DIAGNOSTIC group, 18 days (IQR = {10, 29}) versus 13 days (IQR {9, 23}) (P = 0.030). Not surprisingly, median hospital-free days at day 60, which represents a composite endpoint of these two measures, was significantly less in the NO_DIAGNOSTIC group (42 days, IQR = {28, 50}) than the DIAGNOSTIC group (47 days, IQR = {38, 51}) (P = 0.007). When adjusting for ISS using a multiple regression model, PDE was no longer a statistically significant predictor for hospital LOS (P = 0.10) or mortality (P = 0.99) but remained a statistically significantly predictor of hospital-free days at day 60 (5 greater hospital-free days in the DIAGNOSTIC group than the NO_DIAGNOSTIC group, IQR = [0.6, 8.7], P = 0.025). The coefficient of ISS was −0.55 (0.55 less hospital-free days for each 1 point increase in ISS, IQR = [−0.71, −0.39], P < 0.00001). Overall adjusted R-squared was 0.16.{Table 3}


In this analysis of the NTDB, we report that obtaining PDE for the diagnostic workup of PeEsIn only slightly delays time to operation and is not associated with worse outcomes when performed in a prioritized and expeditious fashion. While there was no difference in mortality, obtaining a PDE was associated with more hospital-free days at day 60. Although the performance of such adjuncts resulted, on average, in a delay of definitive repair by approximately 1 h, the median time to esophageal repair (3 h) was still well below those reported in older studies. For example, Asensio et al. reported in 1997 an average of 16.7 h delay from admission to OR in the preoperative evaluation group. [1] In the AAST multicenter trial from 2001, Asensio et al. reported an average time delay of 13 h from admission to the OR in the preoperative evaluation group. [8] Our findings likely reflect a real shift in practice, with the recognition that delay of operation is associated with worse outcomes like esophageal-related infectious complications.

We initially believed that the nonPDE cohort would be more severely injured and would present with worse physiological parameters, thus justifying the decision to forgo adjunctive diagnostic procedures. However, this proved to be untrue as the groups were not significantly different in terms of physiologic derangement and of overall injury severity. It goes without saying that an unstable patient suffering penetrating injury should be taken for immediate operative exploration. However, in the absence of instability, performing an adjunctive procedure such as CT scan or intraoperative endoscopy, if associated with minimal time delay as was the case in our study, has low risk and may contribute to superior outcomes (increased hospital-free days).

In the AAST multicenter Asensio study, roughly half of the patients had undergone preoperative evaluation. This preoperative evaluation consisted of esophagoscopy in 73%, CT scan in 24%, and contrast studies in 59%. The mean ISS of 28 and the late death rate of 5.4% in that study were very similar to our findings. While we report similar rates of esophagoscopy (66.7%) and CT scan (29.3%), we found that contrast fluoroscopy was used much less frequently (4%). Fluoroscopic esophagography is considered the gold-standard in diagnosing esophageal injuries, but recent studies have found CT esophagograms to have acceptable diagnostic accuracy. CT esophagograms can be performed more expeditiously, at a lower expense, and without the direct supervision of a radiologist. [11],{12} An alternative would be to perform upper endoscopy once the patient has been stabilized in the operating room. [7],{13} Endoscopy has been shown to be highly sensitive for esophageal injury.

Our study has a few limitations. First, it was not possible to determine the reason why diagnostic procedures (e.g., CT neck) were delayed or omitted in many cases. Second, intraoperative details and operative reports are not captured. Third, the NTDB does not permit identification and analysis of specific relevant complications, so we exclusively analyzed hospital LOS, mortality, and hospital-free days.


If the time to definitive repair is not excessively delayed, obtaining preoperative evaluation for the diagnostic work-up of PeEsIn is associated with increased hospital-free days in patients with PeEsIn. Further studies are needed to better identify the patients who would benefit from prompt PDE prior to definitive surgical repair.


1Asensio JA, Berne J, Demetriades D, Murray J, Gomez H, Falabella A, et al. Penetrating esophageal injuries: Time interval of safety for preoperative evaluation - how long is safe? J Trauma 1997;43:319-24.
2Patel MS, Malinoski DJ, Zhou L, Neal ML, Hoyt DB. Penetrating oesophageal injury: A contemporary analysis of the National Trauma Data Bank. Injury 2013;44:48-55.
3Cornwell EE 3 rd , Kennedy F, Ayad IA, Berne TV, Velmahos G, Asensio J, et al. Transmediastinal gunshot wounds. A reconsideration of the role of aortography. Arch Surg 1996;131:949-52.
4Yap RG, Yap AG, Obeid FN, Horan DP. Traumatic esophageal injuries: 12-year experience at Henry Ford Hospital. J Trauma 1984;24:623-5.
5Glatterer MS Jr, Toon RS, Ellestad C, McFee AS, Rogers W, Mack JW, et al. Management of blunt and penetrating external esophageal trauma. J Trauma 1985;25:784-92.
6Smakman N, Nicol AJ, Walther G, Brooks A, Navsaria PH, Zellweger R. Factors affecting outcome in penetrating oesophageal trauma. Br J Surg 2004;91:1513-9.
7Ahmed N, Massier C, Tassie J, Whalen J, Chung R. Diagnosis of penetrating injuries of the pharynx and esophagus in the severely injured patient. J Trauma 2009;67:152-4.
8Asensio JA, Chahwan S, Forno W, MacKersie R, Wall M, Lake J, et al. Penetrating esophageal injuries: Multicenter study of the American Association for the Surgery of Trauma. J Trauma 2001;50:289-96.
9Fadoo F, Ruiz DE, Dawn SK, Webb WR, Gotway MB. Helical CT esophagography for the evaluation of suspected esophageal perforation or rupture. AJR Am J Roentgenol 2004;182:1177-9.
10Stassen NA, Lukan JK, Spain DA, Miller FB, Carrillo EH, Richardson JD, et al. Reevaluation of diagnostic procedures for transmediastinal gunshot wounds. J Trauma 2002;53:635-8.
11Horwitz B, Krevsky B, Buckman RF Jr, Fisher RS, Dabezies MA. Endoscopic evaluation of penetrating esophageal injuries. Am J Gastroenterol 1993;88:1249-53.