Stent-associated esophagorespiratory fistulas: Incidence and risk factors

Background: Esophageal self-expandable stents (SESs) effectively treat strictures and leaks but may be complicated by a stent-associated esophagorespiratory fistula (SERF). Little is known about SERFs. Objective: To determine the incidence, morbidity, mortality, and risk factors for SERF. Design: Retrospective case-control study. Setting: Single referral center. Patients: All adults undergoing esophageal SES placement during a 10-year period. Intervention: Stent placement. Main Outcome Measurements: Occurrence of SERF, morbidity, and mortality. Results: A total of 16 of 397 (4.0%) patients developed SERF at a median of 5 months after stent placement (range 0.4-53 months) including 6 of 94 (6%), 10 of 71 (14%), and 0 of 232 (0%) of those with lesions in the proximal, middle, and distal esophagus, respectively (overall P <.001). SERF occurred in 10% of those with proximal and mid-esophageal lesions, including 14% with benign strictures, 9% with malignant strictures, and none with other indications for SES placement (P =.27). The risk was highest (18%) in patients with benign anastomotic strictures. Risk factors for development of SERF included a higher Charlson comorbidity index score (odds ratio [OR] 1.47 for every 1-point increase; P =.04) and history of radiation therapy (OR 9.41; P =.03). Morbidity associated with SERF included need for lifelong feeding tubes in 11 of 22 (50%) and/or tracheostomy or mechanical ventilation in 5 of 22 (23%). Median survival after diagnosis was 4.5 months (range 0.35-67), and 7 patients survived less than 30 days. Limitations: Retrospective design, limited statistical power. Conclusion: SERF is a morbid complication of SES placement for strictures of the proximal and mid-esophagus. The dominant risk factors for development of SERF are prior radiation therapy and comorbidity score.