Contemporaneous managementof esophageal perforationGhulam Abbas, MD, Matthew J. Schuchert, MD, Brian L. Pettiford, MD, Arjun Pennathur, MD, JamesLandreneau, Joshua Landreneau, James D. Luketich, MD, and Rodney J. Landreneau, MD,Pittsburgh, PA

Background. Esophageal perforation is an important therapeutic challenge. We hypothesized thatpatients with minimal mediastinal contamination at the time of diagnosis could be managedsuccessfully with nonoperative treatment modalities.Methods. We performed a retrospective review of 119 consecutive patients with esophageal perforationfrom 1998 to 2008. Demographics, cause of perforation, clinical presentation, diagnostic methods, andmanagement results were evaluated. The decision to operate was based on the extent of mediastinalcontamination and systemic sepsis rather than cause of perforation.Results. Median time to diagnosis among all patients was 12 hours (range, 1--120). Spontaneous(Boerhaave’s) perforation occurred in 44 (37%) patients. Iatrogenic perforations constituted theremaining patients (n = 75). After instrumental perforation, 9 patients (13%) required esophagectomy,48 patients were managed with repair and drainage, and the remaining 18 were managed nonopera-tively. All 34 patients undergoing operative therapy for spontaneous perforations were treated withesophageal repair. Overall mortality was 14%, with intrathoracic perforations having 18% mortality,cervical 8%, and gastroesophageal junction 3%. Patients undergoing nonoperative therapy had ashorter hospitalizations (13 vs 24 days), fewer complications (36% vs 62%), and less mortality (4% vs15%) compared with those undergoing operative intervention.Conclusion. An approach to esophageal perforation based on injury severity and the degree ofmediastinal and pleural contamination is of paramount importance. Although operative managementremains the standard in the majority of patients with esophageal perforation, nonoperative managementmay be successfully implemented in selected patients with a low morbidity and mortality if favorableradiographic and clinical characteristics are present. (Surgery 2009;146:749-56.)

From the University of Pittsburgh Medical Center, Pittsburgh, PA

ESOPHAGEAL PERFORATION remains a morbidcondition with grave outcomes. Escape of gastriccontents into the mediastinum produces mediasti-nitis and sepsis, associated with a high morbidityand mortality. Published mortality rates haveranged historically between 20% and 30% depend-ing on the etiology and the interval between theperforation and initiation of treatment.1,2 Im-proved technology has stimulated the expandeduse of a wide variety of esophageal and transeso-phageal diagnostic and therapeutic procedures,each of which carries an associated risk ofperforation.

Management of esophageal perforation remainscontroversial, and is influenced by surgeon

Accepted for publication June 30, 2009.

Reprint requests: Rodney J. Landreneau, MD, Shadyside Medi-cal Center, 5200 Centre Avenue, Suite 715, Pittsburgh, PA15232. E-mail: landreneaurj@upmc.edu.

0039-6060/$ - see front matter

� 2009 Mosby, Inc. All rights reserved.

doi:10.1016/j.surg.2009.06.058

experience and judgment. Most reports to daterepresent small retrospective case series reflectingspecific techniques in management, including pri-mary repair, repair over a drain, stenting, esopha-geal resection, or exclusion. Given the lack of level1 evidence to establish clear standards, a lack ofconsensus exists regarding various aspects of man-agement of this complex and life threatening prob-lem. In the current study, we evaluate the outcomesof 119 patients who presented with spontaneous orinstrumental perforation. We hypothesize that dis-crete clinical factors can be identified at the time ofpresentation that might impact optimal operativeapproach and resulting patient outcomes. Further-more, the initial clinical presentation may serve toidentify the subset of patients amenable to nonop-erative therapy.

PATIENTS AND METHODS

We conducted a retrospective review of 119consecutive patients with esophageal perforationat the University of Pittsburgh Medical Centerfrom 1998 to 2008. Approval for this retrospective

SURGERY 749

SurgeryOctober 2009

750 Abbas et al

analysis was obtained from the Institutional ReviewBoard of the University of Pittsburgh. Patients withanastomotic leak were excluded from analysis.

Clinical evaluation and classification. Thediagnosis of esophageal perforation was estab-lished on the basis of clinical presentation,computed tomography (CT), contrast esopha-gram, and endoscopic evaluation. The patientswere divided into 2 groups according to the causeof esophageal perforation (group 1, spontaneousand group 2, iatrogenic/traumatic, including thepatients with esophageal perforation caused byinstrumentation or foreign body ingestion). Thedecision to operate was based on the extent ofmediastinal contamination and systemic sepsisrather than cause of perforation. The extent ofperforation was classified as contained or non-contained. A noncontained perforation wasdefined by free extravasation of contrast in themediastinal or peritoneal cavities (Fig 1). Acontained perforation was defined as minimalextravasation of contrast at the perforation site orthe presence of pneumomediastinum or pneumo-peritoneum without apparent extravasation of con-trast (Fig 2). The interval between the time ofperforation and initiation of treatment wasconsidered in the management decision.

Management strategies. Repair was the preferredoperative approach. Primary repair was typicallyperformed in patients with limited mediastinalcontamination and necrosis irrespective of thetime of presentation. Repair over a drain (eg,T-tube) was preformed in the remaining patients.Gastrostomy and jejunostomy tubes were frequentlyplaced as an adjunct for drainage and enteralfeeding access, respectively. Minimally invasive re-pair included laparoscopic and/or thoracoscopicapproaches. Side of thoracotomy was selected onthe basis of known extravasation within the pleuralspace, taking into consideration extent of contam-ination and endoscopic findings. Laparotomy orlaparoscopy was performed for the repair of gastro-esophageal junction perforations into peritonealcavity. T-tube or Jackson Pratt drains were used whenthe perforation was repaired over a drain.

Esophagectomy was performed in the setting ofcancer or severe esophageal stricture. Esophagealexclusion was performed in 1 patient, and isconsidered to be an historical approach utilizedin unstable patients with severe contamination.Nonoperative management was preferred inpatients with contained perforation and no clinicalsigns of sepsis. Most patients were managed withno oral intake for 24--72 hours, intravenous hydra-tion and antibiotics. Covered esophageal stents

were placed in few patients to occlude theperforation.

Perforation severity score. A perforation severityscore (range, 0--18) was established using 10 clinical

Fig 1. Contained thoracic esophageal perforation withextravasation of oral contrast into the left pleural space, as-sociated with left pleural effusion, and pneumothorax.This patient was managed operatively with esophagealrepair.

Fig 2. Noncontained thoracic esophageal perforationwith minimal oral contrast in the periesophageal area.This patient was managed nonoperatively with intravenousantibiotic and was NPO for 3 days.

SurgeryVolume 146, Number 4

Abbas et al 751

variables that were deemed to be potentially impor-tant indicators of injury severity and patient out-come. Points (range, 1--3) were given to eachvariable, as documented at the time of presentation,according to the following scale: 1 = age >75,tachycardia (>100 bpm), leukocytosis (>10,000WBC/ml), pleural effusion (on chest x-ray, CT, orbarium swallow); 2 = fever (>38.5�C), noncontainedleak (on barium swallow or CT), respiratory com-promise (respiratory rate >30, increasing oxygenrequirement, or need for mechanical ventilation),time to diagnosis >24 hours; and 3 = presence ofcancer, hypotension.

Statistical analysis. Statistical analysis wasperformed utilizing the SPSS 16.0 statistical soft-ware package (SPSS, Inc., Chicago, IL). Statisticalcomparisons between groups were performedwith the t test and the Fisher exact test. Actuarialsurvival estimates were calculated utilizingthe Kaplan-Maier method, with significance beingassessed by the log-rank test. P < .05 was consideredsignificant.

RESULTS

Patient demographics, as well as etiology andlocation of perforation, are depicted in Table I.The average age of the entire cohort was 62.1years. Perforation was localized to the thoracicesophagus in 52% of patients. Overall mortalityfor the entire group was 14%. Thoracic esophagealperforations had the highest associated mortalityrate of 18%, followed by the cervical esophagus(8%; P = .33) and gastroesophageal junction(3%; P = .05). Median time to diagnosis amongall patients was 12 hours (range, 1--120). Time todiagnosis was >24 hours in 33 patients. Mortalitywas 25% in patients in whom the diagnosis ofesophageal perforation was established after 24hours, as compared with 8% in patients with the di-agnosis established within 24 hours (P = .04).

Twenty-eight (24%) patients were treated with anonoperative approach. There was 1 patient death,which was not related directly to the esophagealperforation. The mortality rate for this group was4%, as compared with 15% in the 91 patientstreated operatively (P = .19). Patients undergoingnonoperative therapy had median hospital stay of13 days as compared with 24 days in the operativegroup (P = .032). In addition, patients treated witha nonoperative approach had fewer complicationsthan those who were managed operatively (36% vs62%; P = .018; Table II).

Among 91 patients treated operatively, repair ofesophageal perforation was performed in 72 patients

(79%). Thirty-three patients underwent primary re-pair, with the remaining undergoing repair over adrain. The mortality rate was 3% (n = 1/33) amongpatients who underwent primary repair as comparedwith 18% (n = 7/39) among patients undergoing re-pair of esophageal perforation over a drain (P = .13).In the subset of patients with spontaneous perfora-tion who underwent repair over a drain, the mortalityrate was 38.5%. Nine patients (13%) required esoph-agectomy after instrumental perforation in the set-ting of obstructive pathology (5 cancers; 4 pepticstrictures). By comparison, all 34 patients undergo-ing surgery for spontaneous perforations were trea-ted with esophageal repair.

Table I. Demographics, etiology, and locationof esophageal perforation

Age 62.1 years

GenderMale 67Female 52

Esophageal pathologyNo prior pathology 56Stricture 11Cancer 7Diverticulum 6Others 39

Cause of perforationSpontaneous 44Iatrogenic 75

Location of perforationCervical 26Thoracic 61GEJ 32

Overall mortality 14.2%

Table II. Comparison of outcome between theoperative and nonoperative groups

Operative(n = 91)

Nonoperative(n = 28) P value

Age (yrs) 65.3 56.3 .014Instrumentation (n = 75) 57 (76 %) 18 (24%) 1.00Spontaneous/

Boerhaave’s (n = 44)34 (77%) 10 (23%)

Cervical (n = 26) 15 (58%) 11 (42%) .017Thoracic (n = 61) 49 (80%) 12 (20%) .388GE Junction (n = 32) 27 (84%) 5 (16%) .329Duration of stay (d) 24.0 13.0 .032Complication (%) 61.5 35.7 .018Mortality (%) 15.4 3.6 .189

SurgeryOctober 2009

752 Abbas et al

Table III compares the outcomes afterspontaneous versus iatrogenic esophageal perfora-tion. There was a significantly larger number ofmale patients within the spontaneous perforationgroup (77%) as compared with the iatrogenicgroup (41%). No significant differences werenoted in duration of stay, complications, ormortality when comparing these groups.

Perforation severity scores correlated withincreasing duration of stay, morbidity, and mortal-ity (Tables IV and V). All survivors had a clinicalscore of #3.3 and all patients who died had clini-cal score of $6.5 (P = .00004). All patients withclinical score of >9 died. Among the recordedclinical variables, the presence of a noncontainedleak (P = .002) and respiratory compromise/fail-ure (P = .047) were associated most strongly withthe use of an operative approach over nonopera-tive management. Among all patients who had acontained esophageal perforation with no respira-tory impairment, 31 were treated operatively and22 were managed nonoperatively. Although therewas no difference in mortality, the nonoperativegroup had a shorter duration of stay (7.5 vs 15days; P = .001) and lesser morbidity (32% vs72%; P = .006; Table VI).

Kaplan-Meier curves showed an overall survivalrate of 82%, with no difference in the survivalbetween nonoperative and operative approachesand spontaneous versus iatrogenic perforation(Figs 3--5).

DISCUSSION

The management of esophageal perforationremains a formidable challenge. Managementoptions include primary repair, repair over a drain,

Table III. Comparison of outcomes betweenspontaneous and iatrogenic esophagealperforation groups

Spontaneousesophagealperforation(n = 44)

Iatrogenic/traumaticesophagealperforation(n = 75)

Pvalue

Gender (M:F) 36:8 31:44 .0001Management 1.00

Operative 34 57Nonoperative 10 18

Duration ofstay (d)

24.1 20.3 0.59

Complications (%) 65.9 66.5 1.00Mortality (%) 11.4 13.3 1.00

esophageal exclusion, esophagectomy, and nonop-erative management in selected patients. Recently,there has been increasing interest in the nonop-erative management of esophageal perforation.The term ‘‘nonoperative management’’ is mislead-ing because most of these patients will undergo$1 so-called ‘‘palliative interventions,’’ whichinclude endoscopy, stent placement, drainage gas-trostomy, feeding jejunostomy, and/or tube thora-costomy. True nonoperative patients are treatedwith observation and or intravenous antibiotics.

There are no absolute guidelines that identifypatients suitable for nonoperative management.This decision is based on surgeon’s experience andjudgment. We created an esophageal perforationseverity score using 10 clinical variables availablecommonly at the time of presentation. The aver-age clinical score was much lower in the nonoper-ative group as compared with the operative group;and patients with a lesser clinical score had a betteroutcome with shorter hospital stays and a lessermorbidity rate (Tables IV and V). Patients with acontained leak and without respiratory compro-mise (low clinical score), who were managed oper-atively, had worse outcomes than the patientstreated nonoperatively with similar clinical scores(Table VI). It is clear that the extent of contamina-tion and signs of sepsis are the most importantclinical variables in the outcome of esophageal per-foration. Patients with contained leaks withoutsigns of sepsis may be probably managed safelywith a nonoperative approach. Operative interven-tion can be performed subsequently if clinicalsigns of sepsis arise. The outcome of these patientsrequiring operative therapy after initial failed non-operative management should not be worse thanthe similar patients who were treated operativelywithout a nonoperative trial. Although patientswith delayed diagnosis of esophageal perforationhad greater mortality in our series, patients witha failed initial course of nonoperative manage-ment can not be categorized in the same group be-cause their perforation is being treated with

Table IV. Perforation score (0--18)*

Clinical score #2 (n = 44) 3--5 (n = 49) >5 (n = 26)

Complications,n (%)

23 (53) 32 (65) 21 (81)

Mortality, n (%) 1 (2) 3 (6) 7 (27)Duration of

stay (d)10 16 28

*Cumulative score based on following variables: 1 = age >75, tachycardic,leukocytosis, effusion; 2 = fever, noncontained, >24 hours, respiratorycompromise; and 3 = cancer, hypotension.

SurgeryVolume 146, Number 4

Abbas et al 753

Table V.

Operative (n = 91) Nonoperative (n = 28)

Spontaneous (n = 34) Instrumental (n = 57) Spontaneous (n = 10) Instrumental (n = 18)

Clinical factors 2.2 2.2 1.4 1.4Clinical score 3.9 3.8 2.5 2.6Duration of stay 13 17.5 11 6Complications 22 (64.7%) 41 (71.9%) 5 (50%) 7 (38.9%)Mortality 4 (11.8%) 6 (10.5%) 0 (0%) 1 (5.6%)

hydration, intravenous antibiotics, and the other‘‘palliative’’ adjuncts as mentioned.

In our series, the nonoperative group hadbetter perioperative outcomes as compared withthe operative group. The only mortality in thenonoperative group was not directly related to theunderlying esophageal perforation. This elderlypatient with dementia had simultaneous laryngealand paraspinal lung cancer. The lung neoplasmeroded into the esophagus with a contained leak,causing fever and leukocytosis. A covered esopha-geal stent was placed, which covered the necroticarea with no evident contrast extravasation onsubsequent barium swallow. The patient wasplaced ultimately in hospice care and died subse-quently of pneumonia.

The idea of nonoperative management ofesophageal perforation is not new. Althoughsome reports suggest that nonoperative manage-ment in highly selected groups can achieve a goodoutcome, the overall literature on the conservativemanagement of esophageal perforation is sparse.Cameron et al3 published their experience withnonoperative management of 8 patients in thelate 1970s; this report is still considered to bethe benchmark in the selection of patients fornonoperative management. It should be notedthat 5 out of 8 patients had an anastomotic leakrather than esophageal perforation. In the currentera, an anastomotic leak is considered a separateclinical entity from esophageal perforation with

Table VI. Comparison of outcomes in patientswith contained leak without respiratorycompromise in operative versus nonoperativegroups

Operative (31) Nonoperative (22) P value

Duration ofstay (d)

15 7.5 .001

Morbidity 71% 31.8% .006Mortality 3.2% (1/31) 4.5% (1/22)

an associated different management algorithm.Our series does not include patients withanastomotic leaks.

Mengoli and Klassen4 documented nonopera-tive management of thoracic esophageal perfora-tion in 18 patients with 6% mortality in 1 of theearliest series. Subsequently, Lyons et al5 com-pared the outcome of 18 patients with esophagealperforation treated operatively with 11 patientsmanaged nonoperatively. Mortality was 38% inthe operative group compared with 9% in the non-operative group. Recently, Vogel et al6 treated 32patients with nonoperative management with a100% survival rate. Four patients had perforationin the cervical esophagus and 28 had thoracicesophageal perforation. Three patients with tho-racic perforation required operative interventionsubsequently with no mortality. The 2 deaths outof the series of 47 patients were in the operativegroup. They favor nonoperative management inthe majority of patients with an emphasis onradiological drainage.

Fig 3. Overall survival of entire group (n = 119).

Fig 4. Survival stratified by type of perforation.

SurgeryOctober 2009

754 Abbas et al

Bufkin et al2 from Emory University publishedtheir experience with the management ofesophageal perforation in 66 patients. Twelve pa-tients with contained perforation and no signs ofsepsis were treated nonoperatively with no mortal-ity. The mortality rate in the operative group wasapproximately 25%. Similarly, Port et al7 treated 6patients with contained thoracic esophageal perfo-ration and no signs of sepsis conservatively with nomortality. Amir et al8 published their experiencewith the nonoperative management of esophageal

Fig 5. Survival stratified by approach.

perforation in 21 patients with an overall 95% sur-vival rate; however, 9 patients (43%) required oper-ative intervention subsequently owing to clinicaldeterioration. In contrast, Michel et al9 noted agreater mortality rate in their nonoperative groupas compared with other series. They examined 72patients with esophageal perforation over 21 years.Nineteen patients with contained leak were treatednonoperatively with 16% mortality, which was com-parable with the 17% mortality in the operativegroup.

Some recent reports suggest successful use ofcovered esophageal stents in the management ofesophageal perforation. Kiev et al10 managed 14thoracic esophageal perforations with endoscopicinsertion of a covered stent to occlude the perfora-tion. Esophageal leak healed in all patients with nomortality.

Primary repair of esophageal perforation,irrespective of delay in diagnosis, remains ourpreferred operative approach. The mortality ratewas 3% in primary repair group compared with18% in patients with esophageal perforation re-paired over a drain. Recent literature favors pri-mary repair as the best surgical approach withconsistently low mortality ranging from 3% to13%.11-13 Esophagectomy is preferred in the set-ting of obstructive pathology, including cancer,with an associated 11% mortality in our series.Orringer and Sterling14 reported an overall 88%survival in 24 patients with esophagectomy foresophageal perforation in the setting ofobstruction or cancer.

SurgeryVolume 146, Number 4

Abbas et al 755

The current study is limited by its retrospectivenature, introducing the possibility of selectionbias. A multi-institutional prospective randomizedstudy is necessary to better define the selectioncriteria for operative versus nonoperative manage-ment and to establish the best operative approachfor the management of esophageal perforation. Aperforation severity score using clinical variablessimilar to those presented herein can be used tovalidate its effectiveness.

Recent advances in imaging technology andpercutaneous interventions have changed the ther-apeutic armamentarium available to surgeons in themanagement of patients with esophageal perfora-tion. An increasing body of literature challenges usto reappraise the traditional operative paradigm,historically employed under these circumstances. Itis becoming increasingly clear that select subgroupsof patients (eg, contained perforation withoutsigns of respiratory compromise or sepsis) may bemanaged in a nonoperative fashion with outcomescomparable with (and in certain instances superiorto) traditional surgical approaches (Table VI). Pa-tients presenting with increased perforation severity(eg, uncontained perforations beyond 24 hoursfrom the time of injury) more often require aggres-sive operative intervention.

In conclusion, esophageal perforation con-tinues to be a challenging problem with substantialmorbidity and mortality. An aggressive, systematicapproach is necessary to improve the outcomes.Patients with contained leak without respiratorycompromise and other signs of sepsis, irrespectiveof etiology of perforation and delay in presenta-tion, may be managed nonoperatively with veryclose observation. Operative intervention may be-come necessary with clinical deterioration. Primaryrepair seems to have better outcomes comparedwith other operative approaches, and remains thepreferred approach in the majority of cases.

REFERENCES

1. Skinner DB, Lttle AG, DeMeester TR. Management ofesophageal perforation. Am J Surg 1988;139:760-4.

2. Bufkin BL, Miller JI Jr, Mansour KA. Esophageal perforation:emphasis on management. Ann Thorac Surg 1996;61:1447-51.

3. Cameron JL, Kieffer RF, Thomas R, et al. Selectivenonoperative management of contained intrathoracicEsophageal disruptions. Ann Thorac Surg 1978;27:404-8.

4. Mengoli LR, Klassen KP. Conservative management ofesophageal perforation. Arch Surg 1965;91:238.

5. Lyons WS, Seremetis MG, deGuzman VC, et al. Rupturesand perforations of the esophagus: the case for conservativesupportive management. Ann Thorac Surg 1978;25:346.

6. Vogel Sb, Rout WR, Martin TD, et al. Esophagealperforation in adults: aggressive, conservative treatment

lowers morbidity and mortality. Ann Surg 2005;241:1016-23.

7. Port JL, Kent MS, Korst RJ, et al. Thoracic esophagealperforations: a decade of experience. Ann Thorac Surg2003;75:1071-4.

8. Amir AI, van Dullemen H, Plukker JT. Selective approach inthe treatment of esophageal perforations. Scand J Gastroen-terol 2004;39:418-22.

9. Michel L, Grillo H, Malt R. Operative and nonoperativemanagement of esophageal perforations. Ann ThoracSurg 1982;39:203-10.

10. Kiev J, Amendola M, Bouhaidar D, et al. A managementalgorithm for esophageal perforation. Am J Surg 2007;194:103-6.

11. Wright C. Primary repair for delayed recognition of esoph-ageal perforation. In Difficult Decision in Thoracic Surgery:An Evidence-Based Approach. New York: Springer; 2008:298--304.

12. Richardson JD. Management of esophageal perforations:the value of aggressive surgical treatment. Am J Surg2005;190:161-5.

13. Wu JT, Mattox KL, Wall MJ Jr. Esophageal perforations: newperspectives and treatment paradigms. J Trauma 2007;63:1173-84.

14. Orringer MB, Stirling MC. Esophagectomy for esophagealdisruption. Ann Thorac Surg 1990;49:35-43.

DISCUSSION

Dr Michael F. Reed (Cincinnati, OH): There has beena steady increase in the opposition to the concept thatevery perforation after 24 hours must be managed non-operatively. And thank goodness we are finally stompingthat out. You made those key points that and I’m veryhappy that you made them for primary repair. Perfora-tions are now much more frequently instrumentation;I think the point of what you call nonoperative repairis critical. But this is a spectrum. Is placing a G tubeand placing a J tube, really nonoperative? Probablynot. But appropriate. And now with retrievable stents,it changes the management as well.

One thing you did not mention, though, that I’minterested in in particular, other than just do you alwayscover the repair in open procedures, but what is the rulefor minimally invasive surgery, in something that isperhaps contained without extravasation but there’smediastinal air, the patient is septic, clearly mediastinitishas set in? Is there a role for that? Or, what if theperforation is high in the neck but you have contami-nated the mediastinum, repair is not appropriate in theneck, is there role for minimally invasive management?Finally, do you see the role of stenting increasing?

Dr Ghulam Abbas (Pittsburgh, PA): The first ques-tion, the minimally invasive approach, we are doingmore and more of minimally invasive approach for re-pair. And the thoracic part like you do a minimally inva-sive esophagectomy and for events from a neckanastomosis to the chest anastomosis, we mobilize theesophagus. In a similar fashion, we mobilize the esopha-gus and endoscopically find where the perforation is anduse a J tube to do a myotomy superior to the perforationand then either you can staple it in the GI area or you

SurgeryOctober 2009

756 Abbas et al

can use the endostitch to do a repair, 1 layer or 2 layer,and then we routinely like to take like a Grillo patchfrom the pleura and patch the repair. So definitelythere’s a role for minimally invasive procedures, depend-ing what time of the night you are doing it and who ishelping, because obviously minimally invasive definitelyneeds somebody who can assist you, can use the 30�professionally.

As far as the stenting, definitely, we are seeing moreand more esophageal stents placed for occlusion ofesophageal perforation. If you look at the recent liter-ature, there are more reports out of---esophageal stent,anastomotic leak, and for esophageal perforation. Thebest reviewed is the Polyflex stent, although it does havea higher migration rate.

Dr Jeff N. Peters (Rochester, NY): That was a verynice review of a very large series of patients. I want youto focus for just a moment on a single question on theseptic, noncontained perforation, and I want to chal-lenge your conclusion that those patients all require op-erations. I think it showed up on 3 of your slides. As younicely showed, 2 papers, the most recent papers, provokeour thought process to suggest that maybe we ought notto operate on those patients, that noncontained septicpatient. And the real question here is not whether we op-erate on them or not but whether we help to close thehole in the esophagus. And there’s an emergent think-ing that that’s not necessary, that the primary aspect isjust to control the sepsis. So were your groups cleanwith regard to the noncontained nonseptic or the septicnoncontained? Or did you have somebody in the septicgroup that you managed nonoperatively? Shed a littleinsight.

Dr Ghulam Abbas (Pittsburgh, PA): These are verygood questions, and I totally agree with you. As you prob-ably know, we have a group of 12 thoracic surgeons, so

everybody has a different bias how to manage the esoph-ageal---but I personally totally agree with you. I don’t be-lieve that the esophageal hole has to be closed. I thinkthe key point is drainage of collection, adequate drain-age. The concept that if the patient is septic you haveto go and close the hole, I think that’s not true. Andthat’s not true on the basis of the literature. I think thekey point is that all the collection has to be drained ade-quately. And in this case, I think you have a point thatthat patient with a free perforation and is septic maybe better treated with putting in an esophageal stentand subcutaneous drain at the same time. It is not non-operatively, but it’s minimally invasive. And I think thosepatients may do better, especially if the patient is an olderpatient.

Dr Donald N. Reed, Jr. (Fort Wayne, IN): I had aquestion about your endoluminal stents when youalluded to that on the conservatively managed slide, ifI recall. Do you have any patients that were septic inwhom you elected to put a stent in with just drainagethat didn’t go to surgery? In other words, they weren’tin the aggressive group, but they were set.

Dr Ghulam Abbas (Pittsburgh, PA): There were 2 ofthese patients. But there will be more in the comingyears, I think, because more of us are realizing thefact that Dr Peters just brought out, that with the stent,probably drainage is what you need. And I think wewill be doing more and more stent placements, evenin septic patients. Especially myself, I’m convinced ifI have an older patient in the middle of the nightwho looks septic, I probably will do an esophagealstent and chest tube, or maybe percutaneous drainageand maybe go back later, I would do that, and justmake that sure that he has either a G tube, a long na-sogastric tube coming out of the pharynx going intothe jejunum or jejunostomy.