Clinical management was led by an interprofessional team (an experienced endoscopist, an upper-gastrointestinal (GI) surgeon, an intensive care professional, a thoracic surgeon, and specialized nursing staff) with experience in all treatment options. After the diagnosis was confirmed, primary endoscopic treatment (stent or EVT), surgery (fundoplication or esophagectomy), or a combined approach was performed based on clinical parameters, including the perforation's location and size, mediastinal contamination, the duration between the injury and diagnosis, and the patient's overall condition. The baseline medical treatment included the avoidance of all oral intake, volume replacement, broad-spectrum antibiotic treatment, and parenteral or enteral (Jejunofix) feeding.

Our nursing staff is routinely trained once a year on the management of patients with complex esophageal diseases. This particularly includes the expected clinical courses and possible aberrations. When clinical warning symptoms (fever, pulmonary distress, shock, etc.) occur, drainage secretions are conspicuous, or the EndoVAC systems do not work, the surgeon on call or upper-GI surgeon is contacted.

The endoscopic approaches for managing esophageal perforation include the placement of fully covered esophageal stents and EVT. OTSCs were not routinely used in this patient cohort.

Briefly, stent application was carried out either in the endoscopy unit or in the intensive care unit (ICU). Fully covered self-expanding metal stents (FC-SEMS) with a shaft diameter of 23–25 mm were used. Prior to SEMS application, endoscopic cleansing of the leakage cavity was performed, if possible. After inserting a wire into the stomach, the SEMS was introduced. The positioning of the proximal edge of the SEMS and the release of the SEMS were carried out under endoscopic control and confirmed by x-ray.

The EVT procedures were performed under propofol sedation or general anesthesia (9). The sponge used was an Eso SPONGE (B. Braun Melsungen AG, Melsungen, Germany), with a gastric feeding tube and absorbent wound dressing (Suprasorb CNP Drainage Film; Lohmann & Rauscher International, Rengsdorf, Germany). First, the perforation and (if applicable) any extraluminal cavity were cleaned. Patients with defects large enough for endoscope passage and large extraluminal cavities received intracavitary EVT and patients with small defects received intraluminal therapy. In some cases, intracavitary and intraluminal therapies were combined.

The sponge was adequately positioned under endoscopic visualization before a negative pressure of −125 mmHg was applied. The frequency of sponge exchanges usually varied from 3 to 6 days, depending on the defect and clinical appearance. The treatment strategy for esophageal and/or upper-GI EVT treatment has been published elsewhere (9).

All the surgeries were performed using a minimally invasive approach. After establishing the capnoperitoneum and port placement, removal of contaminated fluid from the abdomen and mediastinum was performed. Under clean conditions, the hiatus was explored and mediastinal adhesiolysis with simultaneous intraoperative endoscopy was performed. This was followed by laparoscopic suturing of the esophageal defect (full-thickness) and coverage using Nissen or Dor fundoplication. Finally, a transhiatal drain was placed. In some cases, combined treatment with preemptive EVT was performed.

Primary salvage esophagectomy was performed in patients with large defects and gross contamination when endoscopic or local surgical treatment was considered insufficient. In these cases, debridement of the pleural cavity and mediastinum, along with esophagectomy with cervical esophagostomy and feeding gastrostomy, was performed. Secondary salvage esophagectomy was performed upon interdisciplinary discussion when endoscopic or local surgical options did not lead to sufficient healing of the defect. After esophagectomy, secondary reconstruction was performed in eligible patients after 12–26 weeks.

All statistical analyses were performed using IBM SPSS Statistics 29 (International Business Machines Corporation, Armonk, NY, USA). Descriptive data are reported as means with standard deviations, unless otherwise stated. Comparisons between the analyzed cohorts were performed using the chi-square test, Fisher's exact test, or Mann–Whitney U-test. For further risk analysis, a multivariate linear regression analysis was performed. The level of statistical significance was <0.05 (two-sided).

Overall, 32 patients were identified during the study period, with 15 diagnosed with BS and 17 with OEP (Table 1).

The diagnosis was established by endoscopy, a CT scan, or a combination of both. A combination of the two was carried out more often in patients with BS (66.7 vs. 35.3%, p = .077). These patients also had larger perforations at the initial endoscopy (22.50 mm vs. 15.00 mm, p = .05). Details on the location and reason for the OEP are provided in Table 1.

Overall, the majority (19/32, 59.4%) of the patients were secondarily referred to our tertiary center. This was more frequent in the BS group compared to the OEP group [12/15 (80.0%) vs. 7/17 (41.2%), p = .026]. Of the secondarily referred patients, 10/19 (52.6%) initially received antibiotics, while endoscopic or surgical treatment was performed in 3/19 (15.8%), with no differences between the groups. Sepsis at admission was present in 16/32 (34.4%) overall (Table 2).

One patient [1/32 (3.1%)] received conservative primary therapy with a gastric tube and antibiotics, but received local surgical treatment with hiatal exploration, defect suturing, and fundoplication within days due to clinical deterioration.

Endoscopic primary therapy was performed in 11/32 (34.4%) patients. Of these, 8/11 (72.7%) received EVT, with the remaining three patients receiving endoscopic stent placement. Closure using an OTSC was not tried due to the poor tissue conditions.

A total of 4/11 (36.4%) patients needed early escalation of their therapeutic regimen in this group, with two receiving fundoplication and two receiving salvage esophagectomy as a secondary treatment.

Surgery as the primary treatment was performed in 12/32 (37.8%) cases; 8/12 (66.7%) received defect closure and fundoplication. Defect closure and fundoplication were performed less often in patients with BS compared to those with OEP (42.9% vs. 100.0%, p = .038). Salvage esophagectomy as the primary treatment was performed in 4/12 (33.3%) patients. The patients with BS had a significantly higher risk for primary or secondary salvage esophagectomy than the patients with OEP (57.1% vs. 0.0%, p = .038).

Endoscopic treatment and fundoplication, as a combined primary therapy, were performed in 8/32 (25.0%) cases. Five of the patients who received primary combined therapy had EVT + fundoplication [5/8 (62.5%)], and three in the primary combined therapy group had a combination of endoscopic stent placement + fundoplication [3/8 (37.5%)]. Escalation to salvage esophagectomy as a secondary treatment was needed in 2/8 (25%) of the cases with a combined primary treatment regimen, with a significantly higher risk for patients with BS vs. those with OEP (66.7% vs. 0.0%, p = .035).

Additional therapy for pleural contamination was needed more frequently in the patients with BS [thoracic drain: 86.7% vs. 17.6%, p < .001, video-assisted thoracoscopy (VATS): 33.3% vs. 0.0%, p = .010] (Table 3).

Overall mortality was 3/32 (9.4%), with a strong trend toward higher mortality in the patients diagnosed with BS (20.0% vs. 0.0%, p = .053). Patients died due to septic multiorgan failure, pulmonary complications, or cerebral bleeding (Table 3).

Overall, the mean CCI score was 50.05 (20.9 −100.0), with higher CCI scores in the patients diagnosed with BS [61.8 (20.9–100.0) vs. OEP 45.6 (20.9–80.7), p = .076].

The patients with BS required a longer ICU stay (20.0 vs. 2.0 days, p = .022). Consistently, the duration of ventilation was substantially longer for the patients with BS (7.0 vs. 1.0 days, p < .001).

New-onset sepsis was more frequent in the patients with BS (40.0 vs. 0.0%, p = .004). Details are shown in Table 3.

A total of 8/32 (25.0%) patients needed either primary or secondary esophagectomy, with a higher risk in the patients with BS (46.7% vs. 5.9%, p = .008). All three deceased patients suffered from BS. Of the patients who survived after salvage esophagectomy, 3/5 (60%) received secondary reconstruction (retrosternal colon interposition or gastric pull-up) with cervical anastomosis (Supplementary Table S2). Of the non-reconstructed patients, one died of advanced pancreatic cancer and one had a too-low performance status for secondary reconstruction (Supplementary Table S2).

Diagnoses of BS and symptoms of sepsis showed a significant linear regression with CCI in the univariate linear regression analysis. In the multivariate analysis, sepsis at admission was identified as an independent risk factor for a higher CCI score (β = 28.387, p = .012). A diagnosis of BS showed a strong association with higher CCI scores in the multivariate analysis (β = 22.441, p = .078) (Table 4).