Surgical management of esophageal perforation

The recognition and management of esophageal perforation remain a problem. Diagnostic and treatment delays are common, and controversy continues regarding approaches to surgical intervention. Overall survival has increased with improved adjunctive modalities; however, morbidity and mortality remain high. A total of 115 consecutive cases of nonmalignant esophageal perforation were reviewed. There were 69 thoracic, 27 cervical, and 19 abdominal perforations. Etiology of the perforations was iatrogenic in 65 patients, traumatic in 28, and spontaneous perforation in 22. Symptoms included pain (71%), fever (51%), dyspnea (24%), and crepitus (22%). Contrast roentgenography was used in 78 patients and demonstrated the perforation in all but two patients. All but 20 patients had operations. In the last decade, the survival rate was 11.4 per cent for patients treated within 24 hours of perforation. Survival significantly improved in the last 10 years because of hyperalimentation, cardiopulmonary monitoring, and better antibiotic coverage. Treatment of choice is primary closure with drainage, regardless of the duration of the perforation. In selected patients who have cervical esophageal perforation, nonoperative management has a role.     

Esophageal perforation in adults: aggressive, conservative treatment lowers morbidity and mortality

OBJECTIVE: To evaluate the outcome of aggressive conservative therapy in patients with esophageal perforation. SUMMARY BACKGROUND DATA: The treatment of esophageal perforation remains controversial with a bias toward early primary repair, resection, and/or proximal diversion. This review evaluates an alternate approach with a bias toward aggressive drainage of fluid collections and frequent CT and gastographin UGI examinations to evaluate progress. METHODS: From 1992 to 2004, 47 patients with esophageal perforation (10 proximal, 37 thoracic) were treated (18 patients early [<24 hours], 29 late). There were 31 male and 16 females (ages 18-90 years). The etiology was iatrogenic (25), spontaneous (14), trauma (3), dissecting thoracic aneurysm (3), and 1 each following a Stretta procedure and Blakemore tube placement. RESULTS: Six of 10 cervical perforations underwent surgery (3 primary repair, 3 abscess drainage). Nine of 10 perforations healed at discharge. In 37 thoracic perforations, 2 underwent primary repair (1 iatrogenic, 1 spontaneous) and 4 underwent limited thoracotomy. Thirty-4 patients (4 cervical, 28 thoracic) underwent nonoperative treatment. Thirteen of the 14 patients with spontaneous perforation (thoracic) underwent initial nonoperative care. Overall mortality was 4.2% (2 of 47 patients). These deaths represent 2 of 37 thoracic perforations (5.4%). There were no deaths in the 34 patients treated nonoperatively. Esophageal healing occurred in 43 of 45 surviving patients (96%). Subsequent operations included colon interposition in 2, esophagectomy for malignancy in 3, and esophagectomy for benign stricture in 2. CONCLUSIONS: Aggressive treatment of sepsis and control of esophageal leaks leak lowers mortality and morbidity, allow esophageal healing, and avoid major surgery in most patients.     

Esophageal perforations. The need for an individualized approach

Since 1971 we have treated 33 patients with esophageal perforation caused by instrumentation in 21 patients, trauma in six, and spontaneous perforation in six. Chest pain, fever, mediastinal air, and an abnormal esophagogram were frequent but not invariable findings. Surgical therapy, consisting of primary repair and drainage in 12 patients, drainage alone in five, esophageal diversion in two, and esophagogastrectomy in one, was initiated within 24 hours in 14 patients, all of whom survived. A delay of more than 24 hours in six patients resulted in 33% mortality. Nine patients with small instrumental perforations were treated successfully with antibiotics alone, while three other patients with late traumatic (n = 2) and spontaneous (n = 1) perforations were treated nonoperatively; all three died. Overall mortality for the series was 15.5%. Except for small contained instrumental injuries, esophageal perforations demand prompt exploration, with primary repair and drainage as the procedure of choice.     

Esophageal perforation

Sixty-nine patients with perforation of the esophagus were treated at the University of California, San Francisco, from 1977 to 1988. The perforation was iatrogenic in 33 (48%) of the patients, spontaneous in 8 (12%), and a result of external trauma in 23 (33%). Clinical findings included chest pain in 36 (52%) of 69 patients, subcutaneous emphysema in 22 (32%) of 59 patients, and pneumomediastinum in 21 (36%) of 59 patients. Esophagograms demonstrated the perforation in 40 (93%) of 43 patients. Treatment delays of more than 24 hours occurred in about half of spontaneous and iatrogenic perforations, but when the perforation was due to external trauma, treatment was delayed infrequently. Operative therapy in 59 (86%) of the patients included primary closure in 44 patients, drainage alone in 9 patients, and Celestin tube placement in 1 patient. Four patients with benign strictures had esophagectomy, and 4 patients with achalasia had Heller myotomy in addition to closure of the perforation. Eight (12%) of the patients were treated nonoperatively. For thoracic perforations, nonoperative treatment was reserved for patients who were diagnosed late but who had minimal evidence of sepsis. Seven (10%) of the patients died. Factors that influenced outcome included cause of perforation, anatomic location, and patient age. Our study shows that a high index of suspicion, aggressive use of esophagography, and individualized treatment are necessary for the best results when treating esophageal perforation.     

Operative and nonoperative management of esophageal perforations.

During a 21-year period, 72 patients were treated for esophageal perforations; the diagnosis was made only at postmortem examination in 13 other patients. Fifty-eight of 85 patients (68%) sustained iatrogenic perforations, 11 patients (13%) had "spontaneous" perforation, nine patients (11%) had foreign body related perforation, and seven patients (8%) had perforation caused by external trauma. Eleven cervical perforations, contained between the cervical paravertebral structures, plus eight thoracic perforations, contained in the mediastinum, were treated with antibiotics, intravenous hydration, and nasogastric drainage. The mortality rate after this nonoperative approach was 16% (3/19 patients). Indications for operative treatment in 53 patients were hydropneumothorax with mediastinal emphysema, sepsis, shock and respiratory failure. The operative mortality rate in these instances was 17% (9/53 patients). Six of the nine patients who died had been operated on more than 24 hours after the onset of symptoms. For cervical perforations the best results were obtained by drainage plus repair of the perforation (mortality rate: 0%; 0/10 patients) and for thoracic perforations by suturing supported by a pedicled pleural flap (mortality rate: 11%; 1/9 patients). Simple drainage of thoracic perforation was followed by a mortality rate of 43% (3/7 patients).     

Nonoperative management of esophageal perforations. Is it justified?

OBJECTIVE: Experiences obtained with nonoperative treatment (NOT), i.e. total prohibition of per oral food intake for a minimum of 7 days, administration of combinations of broad-spectrum antibiotics, and parenteral hyperalimentation, are described in the management of esophageal perforations. SUMMARY BACKGROUND DATA: The place, value, and indication of NOT in the management of esophageal perforation has not yet been unequivocally defined. As a result, contradictory data have been published regarding the outcome of NOT. METHODS: During the past 15 years (1979 to 1994), 20 of 86 patients (23.3%) with esophageal perforation have been treated nonoperatively from the outset. In this group, perforations were located to the upper, middle, and lower third of the esophagus in 50%, 30%, and 20%, respectively. In the operative management group (OT)--in which conservative (drainage, endeprothesis), reconstructive (suture, reinforced suture), and radical (resection) surgical methods were applied--lesions were preponderantly located in the lower one third of the esophagus (56.1%--37/66). As to the interval between the perforation and the onset of treatment, 14 patients had been diagnosed within 24 hours, whereas in 6 cases treatment had been begun beyond 24 hours. RESULTS: NOT could be successfully carried out in 16 patients; the decision to use NOT had to be revised in 4 other cases (Table 1). Two patients were lost; the mortality rate was 10% (2 of 20). The rate of complications was lower in the NOT group (20%, or 4 of 20) than in the OT group (50%, or 33 of 66). CONCLUSIONS: NOT can be suggested for the treatment of intramural perforations. In the case of transmural perforation, this approach should be taken into consideration if the esophageal lesion is circumscribed, is not in neoplastic tissue, is not in the abdominal cavity, and is not accompanied by simultaneous obstructive esophageal disease; in addition, symptoms and signs of septicemia should be absent.     

Personal management of 57 consecutive patients with esophageal perforation

BACKGROUND: Esophageal perforation is a surgical emergency associated with high morbidity and mortality. There is no consensus regarding the appropriate management of this life-threatening condition. METHODS: A retrospective review was made of 57 patients with esophageal perforations treated at the Department of Surgery, Postgraduate Institute of Medical Education and Research, Chandigarh, India, between September 1986 and December 2001. RESULTS: Forty-four (77%) perforations were due to iatrogenic causes, spontaneous perforations occurred in 6 patients (11%). Foreign body ingestion caused perforation in 4 (7%), followed by blunt trauma in 2 (4%) and caustic injury in 1 patient. A total of 6 (11%) patients had cervical injury, 49 (86%) patients had thoracic, and 1 patient had abdominal esophageal injury. Thirty-three (58%) patients underwent emergency esophagectomy, 4 (7%) patients underwent primary repair, and 4 patients (7%) underwent drainage alone, whereas 16 (28%) patients were managed by nonoperative treatment. Using these treatment principles, we achieved 86% survival rate for all patients. Eight (14%) patients died. Spontaneous perforation had the highest mortality (67%). CONCLUSIONS: Esophageal perforation needs aggressive treatment. The treatment depends mainly on two factors: perforation in a healthy esophagus, and perforation with a preexisting underlying intrinsic esophageal disease causing distal obstruction. Esophageal perforation associated with stenotic lesions (benign or malignant) needs esophageal extirpation. Perforation in a healthy esophagus should be treated by primary closure if encountered early. Nonoperative conservative treatment is appropriate when esophageal perforation is encountered late.     

Esophageal perforation: a therapeutic challenge

The records of 64 patients with esophageal perforation treated since 1958 were reviewed. There were 19 cervical perforations, 44 thoracic perforations, and one abdominal perforation. Thirty-one perforations (48%) were due to injury from intraluminal causes. Twenty (31%) resulted from extraluminal causes: penetrating wounds, 11; blunt trauma, 3; and paraesophageal operations, 6. Eleven (17%) were spontaneous perforations, and two (3%) were caused by perforation of an esophageal malignancy. Ten (91%) of 11 patients with cervical perforations treated less than 24 hours after injury survived compared with 6 (75%) of 8 patients treated more than 24 hours after injury; hence 16 (84%) of the 19 patients in the cervical group survived. In the thoracic group, 19 patients were treated within 24 hours with 16 survivors (84%) compared with 25 patients treated beyond 24 hours with 12 survivors (48%); hence 28 (64%) of the 44 patients in the thoracic group survived. The patient with an abdominal perforation survived. Thirty patients underwent primary suture closure of the perforation, and 25 (83%) lived. Seventeen patients had drainage, and 10 (59%) lived. Total esophagectomy was performed in 9 patients, 7 (78%) of whom survived. Exclusion-diversion procedures were performed in 5 patients, and 1 (20%) survived.     

Esophageal perforation: which factors affect the prognosis? Results of a 10-year experience

Oesophageal perforations are a catastrophic event with a 10-40% mortality rate. The decisive prognostic factor is the time from the event to the diagnosis, while there is no agreement as to the therapeutic options. The aim of this study was to present our ten-year experience in the treatment of oesophageal perforations together with an evaluation of the prognostic factors. From January 1995 to January 2005, 18 patients (11 M, 7 F), mean age 49.3 years (range: 22-79), with oesophageal perforations were treated in our department. They were classified according to the cause and localization of the perforation and the time elapsing since the event. The perforation was localised in the cervical oesophagus in 4 patients (22.2%), in the abdominal oesophagus in 4 patients (22.2%) and in the thoracic oesophagus in 10 patients (55.5%). It was spontaneous in 4 patients (22.2%), traumatic in 4 (22.2%) and iatrogenic in the remaining 10 (55.5%). In 7 patients (38.9%), the treatment was started during the first 24 hours from the event, while the remaining 11 (61.1 %) were referred to us more than 24 hours after the perforation occurred. The overall mortality was 27.8% (5 patients). The only decisive prognostic factor was the time of observation: only 1 patient (14.3%) died in the group observed in the first 24 hours, while the remaining 4 who died (36.4%) were in the group treated more than 24 hours after the event (p < 0.05). Our series confirms that the time elapsing from the event to the diagnosis is the only decisive prognostic factor in the treatment of oesophageal perforations. There is no therapeutic option of choice since there is no significant influence of either cause or localisation of the perforation on outcome.     

Thoracic esophageal perforations

The authors review their experience with thoracic esophageal perforation at Inova Fairfax Hospital, June 1, 1988, to March 1, 2009. With the exception of 6 patients with occult perforation, all of whom survived with nonoperative therapy, aggressive surgical intervention was the standard approach. Among patients treated aggressively with surgery within 24 hours of perforation, hospital survival was 97 per cent versus 89 per cent for patients treated aggressively surgically after 24 hours. In the absence of phlegmon, implacable obstruction, or delay, primary repair resulted in 100 per cent survival. Where phlegmon or resolute obstruction existed, resection and reconstruction resulted in 96 per cent survival. Even when patients were deemed too ill to undergo surgery, cervical diversion was 100 per cent effective in eradicating continuing leak and achieved 89 per cent survival. Endoesophageal stenting was applied as primary treatment or secondarily such as where leak complicated primary repair. When stenting was used as the initial and primary treatment modality, survival was 88 per cent. Targeted drainage was helpful on occasion as an adjunct to initial therapies. Comfort measures alone were appropriate when clinical circumstances merited no effort at resuscitation. Finally, survivors were asked to self-categorize their ability to swallow; 95 per cent responded good to excellent.     

Diagnosis and management of esophageal perforations.

Esophageal perforation remains a difficult diagnostic and management problem. Recommendations regarding treatment remain controversial. A 15-year experience with perforation of the esophagus from all causes was reviewed at Louisiana State University and Veterans Administration, Medical Centers (Shreveport, LA). The majority of the injuries involved the thoracic esophagus (28 or 54%), followed by the cervical (21 or 40%), and the intraabdominal esophagus (3 or 6%). Iatrogenic causes constituted most of the injuries (52%), followed by external trauma (23%), barogenic rupture (15%), and ingested foreign bodies (10%). Diatrizoate methylglucamine 66 per cent, sodium diatrizoate 10 per cent (Gastrografin; Squibb, Princeton, NJ) contrast studies and flexible esophagoscopy were performed in 44 and 22 patients, respectively. In the cervical esophagus, contrast studies were more sensitive and specific than endoscopy (P less than .01), but both studies were equally effective as diagnostic methods in thoracic perforations. Cervical perforations were treated with either drainage alone (7 patients) or primary repair with drainage (14 patients) with an operative mortality of 4.8 per cent. Several procedures were used in thoracic perforations, which carried a mortality of 36 per cent and were more lethal than cervical tears (P less than 0.2). Any thoracic esophageal perforation treated more than 24 hours after the onset of symptoms, irrespective of what procedure was used, was associated with a significantly higher mortality than if operated on earlier (P less than .001). Five patients with perforated carcinomas were treated by esophageal resection with no mortality. Significantly higher mortality was seen with a delay in diagnosis, thoracic perforations, and Boerhaave's Syndrome. A subset of patients with perforated carcinomas may benefit from esophageal resection with delayed reconstruction.     

Colonoscopic perforations: incidence, management, and outcomes

Fiberoptic colonoscopy provides superior diagnostic and therapeutic capabilities in the treatment of lower gastrointestinal disease processes. A well-recognized, but uncommon, complication during the procedure is perforation. The purpose of this study was to determine the incidence of colonoscopic perforation, define risk factors, assess the management of these complications, and evaluate outcomes. From January 1997 through December 2003, 43,609 colonoscopies were performed in our medical center. There were 14 (0.032%) perforations (1 in 3115 procedures); 7 from diagnostic and 7 from therapeutic procedures. General surgeons performed 1243 procedures (2.9%), and their rate of perforation was 0.080 per cent compared with 0.031 per cent for gastroenterologists during the same period. Half of the perforations occurred in the rectosigmoid, and the most common mechanism was mechanical (n = 6). Perforation was identified immediately during endoscopy in 50 per cent of the patients. Thirteen of 14 perforations were treated within 24 hours; 1 was delayed 48 hours. Initial surgical management was undertaken in 11/14 patients. Initial nonoperative treatment was attempted in three and was successful in only one patient. The mean length of stay following perforation was 11.2 days (range, 4-36 days). Three patients (21.4%) had 7 postoperative complications. Colonoscopic perforations are uncommon but can be recognized early and managed surgically with acceptable morbidity and postoperative length of stay.     

Thoracic esophageal perforations: a decade of experience

BACKGROUND: Perforation of the thoracic esophagus is a formidable challenge. Treatment and outcome are largely determined by the time to presentation. We reviewed our experience with esophageal perforations to determine the overall mortality and whether the time to presentation should influence management strategy. METHODS: A retrospective chart review was performed on all patients treated for perforation of the thoracic esophagus from 1990 to 2001. There were 26 patients (14 men and 12 women; median age, 62 years; range, 36 to 89 years). Fourteen patients presented within 24 hours (early), and 12 patients presented after 24 hours (delayed). Nine of the 12 patients in the delayed group presented after 72 hours. The causes of the perforations were as follows: instrumentation (19 patients), Boerhaave's syndrome (2 patients), intraoperative injury (1 patient), and other (4 patients). In the early group, 3 patients were treated conservatively, 10 patients underwent primary repair, and 1 patient required esophagectomy for carcinoma. In the delayed group, 3 patients were treated conservatively, 6 underwent successful repair of the perforation, 1 had a T-tube placement through the perforation and eventually required an esophagectomy, and 2 had an esophagectomy as primary surgical treatment. RESULTS: Hospital mortality was 3.8% (1 of 26) and morbidity was 38% (10 of 26). Persistent leaks occurred in 3 patients, 2 after primary repair and 1 after T-tube drainage. All patients selected for conservative management successfully healed their perforation. CONCLUSIONS: Primary repair can be carried out in most cases of thoracic esophageal perforation regardless of time to presentation, with a low mortality rate. A small but carefully selected group of patients may be treated successfully without operation. Esophagectomy should be reserved for patients with carcinoma or extensive necrosis of the esophagus.     

The management of nonmalignant intrathoracic esophageal perforations

Eight patients with nonmalignant intrathoracic esophageal perforations recognized more than 48 hours (48 hours to 14 days) after rupture were treated at Toronto General Hospital between 1973 and 1978. Perforation was due to postemetic rupture in 7 patients and to instrumentation in 1. The patients were seen with pain (8), vomiting (7), fever (7), shock (4), respiratory insufficiency (5), pleural effusion (7), pulmonary infiltrates (7), and leukocytosis (6). All patients were managed with thoracotomy. Direct suture closure of the perforation was carried out in 4 patients with midesophageal perforations. Postoperative localized leaks developed in 2 of these patients but healed with conservative management. Cervical esophagostomy and esophageal diversion were used in 1 patient in whom a severe empyema developed in the postoperative period. Direct suture closure, reinforced with a gastric patch, was used to close three lower esophageal perforations. None of these patients had a postoperative leak but all developed subsequent reflux esophagitis. All 8 patients survived. In patients with delayed recognition of a nonmalignant intrathoracic esophageal perforation, elimination of continued chemical and bacterial contamination can be achieved by a clear definition and closure of the esophageal mucosal margins. The obliteration of potential pleural spaces by good tube drainage, lung decortication, and the elective use of mechanical ventilation with positive end-expiratory pressure decreases the incidence of uncontrolled intrapleural sepsis.     

Management of Esophageal Perforation

Despite recent advances in thoracic surgery, the management of esophageal perforation remains problematical and controversial. Thirty-one patients were treated for an esophageal perforation between 1986 and 1998. The esophageal perforation was iatrogenic in 25 cases, spontaneous in 2, traumatic in 2, and caused by a tumor and tuberculous lymphadenitis in 2 patients. There were 10 cervical, 19 thoracic, and 2 abdominal perforations. The interval from perforation to operation was less than 24 h in 12 patients and more than 24 h in 19 patients. The surgical procedures included a primary repair in 12 patients, a resection in 8, and conservative treatment with minor surgical approaches in 11. The mortality rate was 20% (4/20 patients) in the surgical treatment group and 45.5% (5/11 patients) in the conservative treatment with minor surgery group. The overall mortality was 29% (9/31 patients). The prognosis is thus concluded to depend on the cause and location of the perforation, the presence of underlying esophageal diseases, and the surgical procedure chosen. Received: October 12, 1999 / Accepted: May 30, 2000     

Experience with the Grillo pleural wrap procedure in 18 patients with perforation of the thoracic esophagus.

Perforation of the thoracic esophagus may be fatal unless diagnosed promptly and treated with an effective operation. The wide mortality range in different reports reflects the importance of these two factors. This range spans from as low as 11%, if operation is within 24 hours, to greater than 50% after two to three days. The high mortality with delayed treatment is principally due to inability to surgically close the perforation. Eighteen patients (aged from 31 to 78 years) were treated four hours to 14 days after thoracic esophageal perforation (less than 24 hours: 7 patients; 24 to 72 hours: 7 patients; greater than 72 hours: 4 patients). In 14 patients the perforation was sutured, after which the suture line was buttressed with a circumferential wrap of parietal pleura, originally described by Grillo. Underlying esophageal pathology was corrected and wide mediastinal drainage was instituted. All 14 patients recovered and were discharged from the hospital after a median stay of 20 days. Two patients had minor leaks at the suture line that soon closed. Four patients had perforations too extensive to close. Of these, one was resected, the Urschel procedure was used in two, and the Abbott T-tube drainage was used in one. Three of the four patients died. It was quite significant that the pleural wrap was equally effective with both early (6 patients) and delayed perforations (8 patients). These data indicate that the pleural wrap should be used routinely. Extensive perforations that cannot be closed should probably be treated by resection and drainage, followed by esophageal reconstruction at a later time.     

Mortality from iatrogenic esophageal perforations is high: experience of 54 treated cases

OBJECTIVE: To underline the severity of instrumental esophageal perforations and to discuss their management. PATIENTS AND METHODS: Data from patients treated for instrumental esophageal perforation were collected retrospectively from 1980 to 1995 then prospectively since 1995 to 2000. RESULTS: Fifty-four patients were treated for instrumental perforations. Perforation occurred after exploratory endoscopy (n = 24), endoscopic dilation (n = 13), attempted tracheal intubation (n = 5), foreign body extraction (n = 5), treatment of esophageal varices (n = 4), trans-esophageal echocardiography (n = 2), and duodenal prosthesis implantation (n = 1). Clinical manifestations were immediate in 18 cases and delayed in all others, with an interval before treatment ranging from 2 hours to 45 days (mean = 70 hours). All patients were operated after large spectrum antibiotherapy and intensive care, except 3 who were treated medically due to their poor general condition. Fourteen (26%) patients died, including the 3 non-operated ones. CONCLUSION: Instrumental esophageal perforations are associated with a high mortality, probably due to the poor general condition of the patients. Diagnosis of these perforations is often delayed. A good experience of endoscopic maneuveurs and adequate post-endoscopic monitoring could allow earlier surgical treatment with lower mortality.     

Traumatic perforations of the esophagus

Over the past 15 years nine patients with traumatic mechanical perforations of the esophagus have been treated. Seven perforations were iatrogenic, two were accidental. One patient treated conservatively did well. Two patients were operated on without delay. Their hospitalization was short and they had no complications. Six patients were referred to surgery after a delay ranging from 5 days to 17 days from the time of perforation. Their hospitalization ranged from 9 to 113 days, averaging 62.7 days. Complications were common and two patients died. In order to assure survival of patients with esophageal perforation, early aggressive treatment is essential in nearly all instances. In an occasional patient with a small and clean perforation at the esophageal inlet, conservative treatment may be justified.     

Options in the management of perforations of the esophagus

A study of 90 cases of esophageal perforation in the antibiotic era emphasizes individualized treatment and options of therapy based on a fundamental understanding of modifying pathophysiologic factors. If the patient is seen during the first 24 hours, surgical repair and irrigating tube drainage continue to be the treatment of choice in the thoracic and abdominal regions, with certain exceptions. The exceptions include small perforations proved by a thin media esophagram or esophagoscopy without pleural involvement or constitutional symptoms. Such patients may be treated nonoperatively, with gastric drainage, antibiotics, and parenteral alimentation. However, for large perforations with extensive contamination of the mediastinum and pleura, an esophageal exclusion operation may be life saving. In the cervical region, irrigating tube drainage may be just as effectual as repair and drainage. In patients seen after 24 hours, size of the perforation and the amount of mediastinopleural infection, rather than the time that has elapsed, dictate optimal treatment.     

Outcome after failed initial therapy for rupture of the esophagus or intrathoracic stomach

Survival after rupture of the esophagus or intrathoracic stomach is improving, but continued leakage after initial therapy remains a problem. We retrospectively reviewed patients with rupture of the esophagus or intrathoracic stomach to determine the prevalence of continued leakage after initial therapy and how this complication affects outcome. Our review included 58 patients, 38 (66%) of whom had preexisting esophageal disease. The etiology of perforation was spontaneous rupture in 17, penetrating trauma in four, and iatrogenic injury in 35; two patients had perforation from other causes. Initial therapy consisted of drainage in eight, primary repair in 24, resection in 18, bypass in two, and observation in six. The overall mortality rate was 12% (7 of 58 patients) and continuing leaks were identified in 21% (12 of 58 patients). These leaks were unrelated to patient age, existence of prior disease, or delay in therapy but were more common after initial treatment by primary repair with or without pleural flap coverage compared to other management strategies (6 of 9 vs. 6 of 49; P < 0.001). Salvage therapy with survival was possible in 10 (83%) of 12 patients by means of esophagectomy in four, exclusion in one, drainage in two, or observation in three. Continuing leaks can be avoided by providing soft tissue coverage other than pleura over a primary repair and by not leaving an intrathoracic esophageal stump. Aggressive management of continuing leaks results in survival in more than 80% of patients.