Leakage of the repair of a duodenal lesion with or without fistula formation is one of the most feared complications in the surgical treatment of duodenal trauma, with a median of 6.3% in the published series[1, 19, 12, 13, 20–27, 14, 15, 28, 29, 6, 30, 17, 31, 32, 8, 18, 9, 33]. It has been associated with a higher risk of intra-abdominal infection[15, 8], the need for support[15, 29, 8], prolonged stay[15, 29, 8], and a higher death risk[1, 13, 20, 21, 24, 25, 27, 15, 29, 8, 9].
In the PTS cohort, we identified leakages in 17.4% of the cases, which showed association with a higher risk of intraabdominal abscess, sepsis, ICU admission, and ventilatory support. ICU and hospital stay were longer.
The multivariate analysis of the sepsis risk factors revealed that DL contributes independently of trauma severity, shock, massive transfusions, and the technique used to repair the duodenal injury.
The probability of death was 1.8 times higher in the subjects with leakage. This difference did not reach statistical significance. Except for Levison´s study[23], which reported a slightly lower mortality rate in the group of the patients who leaked, the authors who analyzed this association found a higher risk of death in the leak subjects, with a median of 2.8[1, 13, 20, 21, 24, 25, 27, 15, 29, 8, 9]. The intriguing Levinson's finding may be the consequence of survival bias. The author did not exclude the early deaths. Eight of the 17 patients who died did it intraoperatively by exsanguination. They did not have a chance to leak despite the severity of their trauma, modifying the result falsely.
The risk factors for DL have not been appropriately studied. Previous publications examined all duodenal complications, performed univariate analyses, or had low statistical power. In 1999 Timaran and coworkers studied 152 patients, 27 of them with duodenal complications. In a multivariate analysis, they found shock, ATI > 25, and the coexistence of colonic, pancreatic, or superior mesenteric vessels injury as independent risk factors[15]. In 2008, Fraga et al., in univariate analysis of duodenal and non-duodenal complications, occurring in 47 of 77 patients, identified association with altered RTS, ATI > 25, ISS > 25, and procedures different to primary repair[17]. In 2016, Schroeppel et al. compared subjects who leaked with individuals who did not. They did not identify significant differences in the compared variables[8]. In 2019 Weale published a similar comparison reporting a lower arterial PH, a higher lactic acid, and more frequent damage control surgeries in the patients who developed a duodenal leak[9].
Our study collected patients from 11 trauma centers from North, Central, and South America. It included an adequate number of subjects and outcomes to perform the statistical analysis required to identify the variables associated with the leak of the duodenal repair. We confirm the role of shock and trauma severity as risk factors of DL and evidence the risk associated with the more complex repairs, independently of the presence or the magnitude of the other factors.
Complex procedures were devised, to decompress the duodenum or to deviate the intestinal content from the repair, to prevent the fistula formation or to ameliorate its impact. Some of them, such as diverticulization, proved to be excessively aggressive or morbid. The merits of others, such as pyloric exclusion or duodenal decompression, are still debated.
Pyloric exclusion with gastro-jejunostomy, as described by Vaughan[12], or without it as proposed by Ginzburg[34] and Ferrada[35], has been the preferred method to treat duodenal injuries judged as severe.
One of the main difficulties in selecting candidates for a PE is the definition of severe duodenal trauma. Ben Taub Hospital[22, 12] and Denver Hospital[36] surgeons reported using PE in severe duodenal or pancreatoduodenal injuries without clearly defining severe trauma. Both groups reported PE in 41% of their cases. Nassoura et al., on the other hand, proposed ATI > 40 or duodenal injury score ≥ 4 as severity criteria. They performed PE in 3 out of 66 patients[14]. Additionally, the reports describing the surgical treatment according to trauma severity showed PE was used among severity grades 2 to 5, giving evidence of inconsistencies in the indication[36, 27, 37, 18].
The technique was created to reduce the risk of complications, which has not been proven. The publications from Houston containing the technique's description showed leakages only in the group treated by PE[12, 22].
Some studies have evaluated the impact of PE. Seamon and coworkers studied patients with penetrating DI OIS ≥ 2, who survived > 48 hours. They compared 14 subjects with PE with 15 managed with PR. PE patients had a higher proportion of grade 4 injuries (21% vs. 0), suffered complications more frequently (71% vs. 33%), and had a more extended hospital stay (24.3 ± 19.7 vs. 13.5 ± 7.7 days). None of the differences reached statistical significance[6].
Velmahos et al. included 50 patients with OIS > 2 DI, 16 with PE. The proportion of cases with injuries in D1 and D2 and subjects with pancreatic trauma were higher in the PE group (79% vs. 42%, p = 0.02) and (63% vs. 24%, p = 0.02), respectively. DL, intraabdominal infections, and systemic complications occurred with similar frequencies[31].
Dubose and coworkers analyzed patients from the National Trauma Data Bank with DI grades 2 to 5 who survived more than 24 hours. They compared 119 subjects primarily repaired with 28 patients treated with PE. The proportions of patients with ISS > 20, abdominal AIS > 3, and DI > 3 were higher in the PE group, without statistical significance. Adjusted morbidity, mortality, ICU stay, hospital stay, and hospital charges were similar[30].
Our data showed a fourfold increase in the risk of leakage when a PE was used. It cannot be attributed to the trauma severity. The association persisted after adjustment by the other identified risk factors.
Duodenal decompression with tubes comprises a heterogeneous set of intraluminal lines, including gastrostomy, duodenostomy, and proximal and distal jejunostomy. It was proposed by Stone et al. as an adjunct to reduce the pressure within the duodenal lumen without opening or resecting the stomach[38]. Original Stone's publication reported zero duodenal complications in 18 patients treated with this method[38]. Corley and coworkers informed 15% of duodenal complications in decompressed patients, compared with 26% in not decompressed subjects[1]. Stone and Fabian reported 302 cases of DI. Decompression was used in 78%. Duodenal complications occurred in 0.4% of the patients treated with decompression and in 19% of the cases treated without it[13].
Other authors reported a high frequency of use of decompression, without similar results. Snyder et al. complemented the duodenal repair with decompression techniques in 53% of their cases. Duodenal morbidity was more frequent among decompressed patients, 12% vs. 8%[21]. Schroeppel and coworkers informed using decompression in 50% of their cases. Duodenal leakage happened in 10% when decompression was used and 2% when it was not[8].
In our report DL was three times more frequent in the repair + duodenostomy. The association persisted and its strength increased after the multivariate analysis. It confirms the independent contribution and suggest a role in increasing the risk of DL.
Nassoura et al. proposed primary repair as the management technique for most penetrating DI. Duodenal fistula developed in 4% of the PR patients[14]. Some authors have documented an increase in PR use without a parallel increase in the complications[39, 40]. In most contemporary reports, Talving and Weale informed PR in 87% and 97% of their cases, respectively, with a low leakage rate[29, 9].
The available literature and our results identify trauma severity (systemic and local) as the main determinant of leakage after the repair of a duodenal injury[15, 28, 17, 9]. Complex procedures including diverticulization, pyloric exclusion, and tube duodenostomy have failed to reduce the risk of duodenal complications. In fact, as our analysis shows, they can contribute to increase the risk. Resecting, practicing incisions, and anastomoses or inserting tubes for decompression sum to the traumatic burden and the operation's length, which can increase the risk of infectious complications. There is enough evidence of the biological and clinical impact of the trauma from the injury and the surgery[41–44] and the additional risk derived from unnecessary procedures[45–48]. Our findings can be considered part of this evidence.