Artificial intelligence may help decision-making for withdrawal of life supporting treatment among the injured elderly

Abstract

W e read with interest the article Diagnostic accuracy of computed tomography findings for hollow viscus injuries following thoracoabdominal gunshot wounds by Lian et al. 1 Computed tomography (CT) has been found to diagnose solid organ injuries with a good predictive value. 2 The concern for missing a hollow viscus injury (HVI); however, remains the biggest hurdle in selective nonoperative management (SNOM) of penetrating injuries.The authors have reported a reassuring sensitivity of more than 70% and specificity of almost 90% for several CT findings. Similar signs of HVI on CT have been previously described for blunt abdominal trauma as well. 3,4 These results are not surprising as the findings of a perforated viscus would remain similar irrespective of the mechanism of injury.An important distinction of penetrating injuries from blunt injury of the abdomen is the introduction of air into the peritoneal cavity through the wound track. Therefore, presence of free air, which is a definitive sign of HVI in blunt trauma, may not be so in penetrating trauma. The diagnostic dilemma is more when the amount of air in the peritoneal cavity is small. We conducted a randomized trial among patients with anterior abdominal stab wounds comparing SNOM and diagnostic laparoscopy (DL), wherein we faced a similar situation. 5 Of the 54 patients randomized to the SNOM arm, seven patients had a few specks of air and were operated. Surgery was nontherapeutic in two of these patients, in whom the presence of air was attributed to the stab wound. Other authors have reported this dilemma as well. 6 A comment on such CT findings and the ultimate management of such patients would have added value to the study as they represent the group that often causes a diagnostic dilemma.