In keeping with the international and local literature, the majority of patients who present to our trauma unit are males between the ages of 25 and 36 yrs [1,2,3, 13, 15]. As demonstrated by the results of this study, stab wounds predominate in South Africa. These findings are supported by several local studies. A recent retrospective review of 4697 trauma patients by Bhana and colleagues at Chris Hani Baragwanath Academic Hospital (CHBAH), a level 1 trauma centre in Johannesburg, also found that males between the ages of 29 and 40 yrs constituted the majority (92.1%) of trauma victims and that stabbings are the most common mechanism of injury (71.8%) [16, 17].

The current recommendation by the World Society of Emergency Surgery is that the use of CT scan, complemented by serial clinical examination, should be used to guide surgical decision-making in the SNOM of penetrating abdominal trauma [8]. Mandatory surgical exploration is no longer routine as it is associated with a 61% non-therapeutic laparotomy rate, and the prevalence of hollow viscus and mesenteric injuries is low (17%). Negative laparotomies are also associated with an 8–40% morbidity rate [18, 19]. In South Africa, where thousands of patients present to the emergency department every month, resources are severely constrained. The emergency department at CHBAH receives approximately 11,100 patients per month, of which 65% is related to trauma. Challenges faced include limited access to emergency theatre and CT scans. Therefore, it is imperative to identify patients who potentially require surgery early, as limited access to theatre further delays surgical intervention [16].

Although triple-contrast CT has been proven to be accurate in identifying HVI in penetrating trauma, IV contrast-only CT scans are routinely performed by the CMJAH Trauma Unit [12, 13, 20]. A major disadvantage of enteric contrast is the time delay necessary for enteric opacification, which is the rationale for using single-contrast CT, especially in our resource-constrained environment. The time delay can be up to 68 minutes, a delay further compounded by limited access to theatre [21]. Delayed surgical intervention can result in increased morbidity and mortality [4]. Mingoli [22] and colleagues demonstrated that a delay of > 6 hours in the treatment of a blunt HVI is an independent predictor of postoperative morbidity. Other concerns include the risk of aspiration, especially in patients who will proceed to laparotomy [4, 21]. In addition, recent studies found that triple-contrast CT has a sensitivity equal to that of single-contrast CT in detecting HVI and that the addition of enteric contrast did not provide any diagnostic benefit [4, 14, 15]. As a result, most trauma centres, including the unit at CMJAH, have adopted single-contrast CT as the standard modality [8].

This study found that single-contrast CT has a sensitivity of 95.1% and a specificity of 44.23%. In the largest review to date by Jawad and colleagues, a lower overall sensitivity of 88% and a higher specificity of 72% were reported [4]. The lower specificity demonstrated by our study could be that most of our patients (72%) sustained stab wounds. Jawad et al. reported that gunshot wounds were responsible for 77% of their injuries and found that the sensitivity and specificity for stab wounds were lower, at 80% and 69%, respectively. Likely because it is more difficult to delineate stab wound tracts on CT compared to gunshot wounds [4, 20]. Secondly, Jawad and colleagues did not clearly describe what features determined a positive CT scan. We included intra-abdominal free air and fluid, which are non-specific, indirect features of an HVI, as they could also be due to peritoneal violation. The presence of only these findings on CT can result in high false-positive rates [13]. Another contributing factor to the low specificity of our study is that the majority of patients in this study had negative CT scans and were observed successfully. As there were no operative findings as the reference standard, these patients were excluded from the statistical analysis.

The majority of patients, 314 (71.9%), in this study had negative CT scans and proceeded to complete SNOM successfully. Furthermore, another 27 patients, in which an HVI was clinically unlikely based on the wound trajectory or the absence of abdominal pain, completed SNOM successfully. These results emphasize the importance of clinical evaluation and serial abdominal examinations. Only three patients who had negative CT scans and were selected for SNOM failed observation, none of which had an HVI at laparotomy. As a result, we were unable to identify the most common injury resulting in failure of SNOM. Two of these patients had persistent abdominal pain for more than 24 hours. The findings at laparotomy confirmed the CT findings which were a grade 2 liver injury with haemoperitoneum and a grade 2 liver injury with a grade 2 diaphragm injury, respectively. The third patient developed tachycardia, tachypnoea, and temperature spikes. At laparotomy, a zone 2 retroperitoneal haematoma was found. Subsequently, the patient was diagnosed with pneumonia. No HVI was missed in these patients, and CT accurately detected associated injuries. These results support the findings of Dayananda and colleagues that in select patients, SNOM is a safe strategy in the South African setting [9].

Direct signs of HVI include bowel wall defects and metallic fragments in the lumen or bowel wall. Indirect signs include bowel wall thickening or abnormal enhancement, mesenteric stranding, extraluminal air, and free fluid without solid organ injury [4, 8]. We found that extraluminal air locules, fluid collections, and pneumoperitoneum are often a result of air locules tracking along the wound tract or associated diaphragmatic injuries. Due to the small number of patients who proceeded to laparotomy and the numerous combinations of features of HVI on CT, we were unable to identify the most sensitive and specific feature suggestive of an HVI. However, Saksobhavivat and colleagues found that gastrointestinal wall thickening was the most accurate (accuracy 91%) feature of HVI, and the most sensitive feature is a penetrating wound tract up to the bowel wall (sensitivity 81–98%) [12].

Research into clinical features and investigations that can aid the early diagnosis of HVI is mostly performed in blunt abdominal trauma. In a 2017 prospective study by Matsumoto and colleagues, they evaluated the use of intestinal fatty acid-binding protein (I-FABP) levels in aiding the early diagnosis of HVI in blunt abdominal trauma. They also evaluated other clinical parameters and found that the heart rate, systolic blood pressure, and lactate levels were not significantly different between the groups with and without an HVI. Only abdominal tenderness and an elevated I-FABP level were statistically significant. Several other studies showed similar results, confirming that clinical examination together with a CT scan of the abdomen plays a vital role in the early diagnosis of an HVI [23,24,25,26]. In keeping with these studies, none of the clinical features, location of injuries, or metabolic parameters we evaluated were statistically significant. Data from patients who suffered injuries at multiple sites and those who suffered isolated abdominal injuries were analysed together, which potentially influenced our results and is one of this study’s limitations.

Another limitation to our study is that statistical analysis was performed only for the patients who underwent surgical exploration, as a reference standard is required. The majority of patients were treated non-operatively, and the false-positive and true-negative cases may have been underestimated. Secondly, we may have missed patients who failed SNOM and presented after 24 hours or to other hospitals, underestimating the false-negative rate. Lastly, our retrospective study design, conducted at a single institution, may be open to incomplete reporting and potential treatment practice bias.

Other studies within our context have evaluated and proven the safety of clinical evaluation and CT scan in the SNOM in penetrating abdominal trauma [1, 9, 27, 28]. However, our unit is the first to evaluate the accuracy of CT scan for detecting HVI in penetrating abdominal trauma in our environment.

Further research opportunities would be to proceed to a prospective study to identify the most sensitive and specific features of HVI on single-contrast CT and to determine which section of the intestine is the most sensitive.