Medical instrumentation	Percentage risk of iatrogenic oesophageal disruption
Dilatation	0.5
Dilatation for achalasia	2
Endoscopic thermal therapy	1–2
Treatment of variceal bleeding	1–6
Endoscopic laser therapy	1–5
Photodynamic therapy	5
Stent placement	5–25

As a result, in spontaneous perforation, the diagnostic error is high, with only 5% of cases diagnosed at presentation. This leads to diagnostic delay of greater than 12 hours in the majority of cases.⁷ It may be that less than 35% of cases are correctly diagnosed pre-mortem⁸ (Box 19.1). As time passes, the critical condition of the patient further obscures relevant clinical features and the pursuit of incorrect investigations makes the diagnosis even more elusive.

Depending on the aetiology and amount of contamination, pain may be severe, constant, retrosternal or epigastric, distressing, exacerbated by movement and poorly relieved by narcotics or relatively mild. Dysphagia and odynophagia are common. Patients can be tachypnoeic and may sit up to splint their diaphragm. Abdominal pain or tenderness are not uncommon and can lead to a negative laparotomy.⁶ Similarly, subcutaneous emphysema takes time to develop; mediastinal emphysema precedes this and may be visible on a plain chest radiograph. With time the negative intrathoracic pressure draws air, food and fluids into the mediastinum and pleural cavities and a chemical pleuromediastinitis develops. A low-grade pyrexia ensues, and a sympathetic nervous system response develops with pallor, sweating, peripheral circulatory shutdown, tachycardia, tachypnoea and overt haemodynamic shock, which worsens as the systemic inflammatory response gives way to sepsis. Within 24–48 hours cardiopulmonary embarrassment and collapse develop as a consequence of overwhelming bacterial mediastinitis and septic shock. The combination of chest pain and shock may inappropriately, but all too commonly lead to a cardiological referral. Survival is dependent on the evacuation of the contamination, from the mediastinal and pleural cavities at the earliest possible opportunity.⁹ Systemic effects are less common when the cervical oesophagus is damaged, with neck pain, torticollis, dysphonia, cervical dysphagia, hoarseness and subcutaneous emphysema predominating.

Penetrating oesophageal trauma manifests in the same pattern but a high index of suspicion based on the likely tract of the insult is essential for diagnosis. Any deep penetrating transcervical or transmediastinal injury, especially gunshot derived, should be deemed suspect for oesophageal trauma. In contrast, except in the most violent of circumstances, blunt trauma rarely causes oesophageal injury but in high impact events a high index of suspicion should be exercised and injury actively excluded.

Investigations

Plain radiography

The typical findings on plain chest radiography are subtle – dependent on the site and the time interval following the insult. These are documented in Box 19.2 and Fig. 19.1. A plain abdominal radiograph may help to exclude a perforated intra-abdominal viscus.⁷

Box 19.2 Typical chest radiograph findings in spontaneous perforation of the oesophagus

• Pleural effusion

• Pneumomediastinum

• Subcutaneous emphysema

• Hydropneumothorax

• Pneumothorax

• Collapse/consolidation

Figure 19.1 (a,b) Typical chest radiograph findings of intrapleural oesophageal perforation.

Contrast radiography

Oral water-soluble contrast radiography ascertains the site, the degree of containment and the degree of drainage of the perforation (Fig. 19.2). Aqueous agents are rapidly absorbed, do not exacerbate inflammation and have minimal tissue effects. However, false-negative results in 27–66% and the limited applicability to a collapsed, unwell patient have downgraded their usefulness.

Figure 19.2 Contrast swallow demonstrating free extravasation of contrast media after oesophageal perforation during balloon dilatation of achalasia.

Upper gastrointestinal endoscopy

Endoscopic assessment excludes the diagnosis if normal, influences management if underlying pathology is discovered and facilitates the placement of a nasojejunal tube to allow enteral feeding. Risks are minimised using modern, flexible videoscopes together with fluoroscopic guidance, but should only be performed by a highly experienced endoscopist conversant with the consequences of their actions (Figs 19.3 and 19.4). Endoscopy can be performed in the sickest of patients, if necessary ‘on table’, when other injuries or instability of the patient preclude radiological assessment.

Figure 19.3 Endoscopic appearances of full-thickness spontaneous oesophageal perforation.

Figure 19.4 Endoscopic appearances of iatrogenic perforation. (a) Food bolus with false iatrogenic lumen alongside. (b) Appearance after food bolus removed. (c) Contained mediastinal cavity. (d) Six weeks later following conservative management a small pit remains.

In a retrospective review of 55 trauma patients, Horwitz et al. demonstrated 100% sensitivity and 92.4% specificity for upper gastrointestinal endoscopy in confirming oesophageal perforation and although injuries were infrequent (prevalence 3.6%), no injuries were missed and the examination was safe.¹⁰ In a similar study of 31 patients (24 of whom were intubated at the time of the examination), video endoscopy had a sensitivity of 100% and a specificity of 96% with no associated morbidity.¹¹ Video endoscopy has also been used to examine the oesophagogastric anastomosis post-oesophagectomy without additional morbidity.^10–12

Computed tomography (CT)

CT is increasingly useful in patients stable enough to undergo scanning. It is especially helpful in cases of multi-trauma and in critically ill patients with an atypical presentation, but the radiology department remains a dangerous place for an unstable patient. In combination with complex interventional radiology, CT has also revolutionised the management of intrathoracic collections. It plays a significant role post-therapy, be that assessing the patient postoperatively or assessing the adequacy of non-operative management.

In an intubated patient, the sensitivity of CT for spontaneous perforation is increased by placing a nasogastric tube just past the cricopharyngeus to run in a small amount of contrast media ¹³ (Fig. 19.5).

Figure 19.5 CT appearances of spontaneous oesophageal perforation. (a) Left pleural hydropneumothorax. (b) Left basal intercostal chest drain in the same patient as in (a).

Other investigations

Thoracocentesis of frank gastric contents is diagnostic – a pH of less than 6.0, a high amylase or microscopic squamous cells in the fluid can also confirm oesophageal perforation in difficult cases. Swallowed or injected oral/nasogastric dyes, such as methylene blue, may be diagnostically useful if a communicating drain is in situ; however, dye staining can be troublesome in the operative field if surgery is subsequently required.

Management

The rarity and severe consequences of inappropriate treatment have limited the ability to evaluate management options. As a result, published observa- tional case series often span many years, many centres, many surgeons and many techniques. Survival is dependent on controlling mediastinal and pleural contamination so surgery remains mandatory when gross contamination is present and is the mainstay for spontaneous perforation. However, non-operative treatment has become standard for iatrogenic trauma where contamination is more limited and delay in diagnosis is uncommon. Patients require a multidisciplinary approach with input from intensive care, radiology, physiotherapy and rehabilitation services. Hospitals lacking these specialist facilities or the versatile surgical cover necessary to deal with the oesophagus by abdominal or left or right thoracic operative approaches should transfer the patients at the earliest opportunity after stabilisation.

All patients with an oesophageal perforation are critically ill. The immediate priorities are the establishment of a secure and adequate airway, stabilisation of cardiovascular status and relief of pain, often using opiate-based analgesia. Regular reassessment is obligatory as an initially stable patient can rapidly decompensate. An early anaesthetic review is recommended. Box 19.3 documents the initial resuscitation.

Box 19.3 Initial resuscitation in spontaneous oesophageal perforation

• Control of airway and administration of supplementary oxygen

• Early anaesthetic and critical care involvement and support

• Large-bore intravenous access and intravenous fluid resuscitation

• Central venous access and arterial line monitoring with or without inotropic support

• Urethral catheterisation and close monitoring of fluid balance

• Broad-spectrum antibiotic and antifungal agents

• Intravenous proton-pump inhibitors

• Strictly nil by mouth

• Large-bore intercostal chest drainage – possibly bilaterally

• Nasogastric tube (only to be placed under endoscopic vision or radiological guidance)

• Enteral access – nasojejunal tube/formal feeding jejunostomy

• Multidisciplinary approach with low threshold for aggressive/operative intervention

Non-operative management

Non-operative management, endoscopic and minimally invasive operative management have all been shown to be safe and feasible in carefully selected patients who have either been diagnosed with minimal contamination and no mediastinitis or with a contained perforation. It may also be considered in those with a delayed diagnosis who have demonstrated tolerance.⁹

Non-operative treatment comprises observation in intensive care or ward-based high-dependency units with patients kept nil by mouth and fed enterally, if necessary via a feeding jejunostomy. A nasogastric tube should be placed under endoscopic and/or radiological assistance past the perforation to decompress the stomach and to limit refluxate. Contrast radiology, endoscopy and CT are used to monitor the status of the perforation and collections should be drained. The timing of investigations is best guided by the clinical condition of the patient but weekly serial contrast or CT studies are not unreasonable. All patients should be given broad-spectrum intravenous antibiotics, antifungal and antisecretory agents. Non-operative treatment is not ‘conservative’; patients require intensive observation and a low threshold for intervention, with 20% of patients requiring aggressive surgical salvage.

Iatrogenic cervical perforations are usually contained and thus managed non-operatively with percutaneous drainage of collections where necessary. Any resulting oesophagocutaneous fistulas heal rapidly in the absence of distal obstruction. Occasionally, operative prevertebral lavage, primary closure and drainage using a left lateral incision anterior to the sternocleidomastoid are required, and are well tolerated by even critically ill patients.

Criteria have been developed to aid the selection of suitable patients for non-operative management. These are detailed in Box 19.4. Case series applying these criteria demonstrate a mortality rate between zero and 16%, but numbers are small and results are skewed by both selection and publication bias.

Adjuncts to non-operative management

An endoluminal approach can be used to support patients undergoing non-operative management and can replicate some of the principles of open surgery with less associated trauma. This is pertinent in patients where the benefits of surgical exploration are outweighed by the risk and the ultimate outcome (advanced cancer) or in patients in whom the defect is small, clean and easily dealt with at the time of injury. All endoscopic approaches are technically difficult and should not be attempted by inexperienced operators unable to deal with the consequences of their actions.

Closure: clips and sealants: Endoclips are well established in closing small, clean defects after endoscopic mucosal resection or submucosal dissection for early cancer.^14,¹⁵ In the absence of significant contamination, small iatrogenic perforations may be closed immediately using endoclips in addition to supportive non-operative treatment. However, endoclipping ‘en face’ in the oesophagus is extremely challenging and should only be attempted by highly skilled endoscopists. It is also debatable whether this significantly alters the clinical course over a simple non-operative approach.² There is at least one case report of clipping a spontaneous oesophageal perforation but this cannot be recommended in the face of gross contamination.¹⁶

Diversion: stents: Self-expanding stents have been used to seal oesophageal perforations, chronic fistulas and even postoperative anastomotic leaks.17–19 Stents were not designed for use in a normal oesophagus and migration rates approach 30%, and concerns have been raised in terms of extending the defect through pressure necrosis and through the trauma of their subsequent removal.^14,²⁰ Publication bias means that failure and the consequences of failure remain unknown. There is considerable variation in the timing of stent placement and number of stents used. It is evident that the majority of cases also involve aggressive non-operative management.^21–24 It is therefore difficult to attribute successful outcomes to the stent placement alone. For example, one ‘successful’ report documents a patient who had five stents placed over an 8-month period before eventually proceeding to oesophagectomy at a tertiary referral centre.²¹ The one prospective stenting study lists 10 patients with a Boerhaave perforation.²⁵ Stent migration was high (11 out of 33) and there was a 50% complication rate (bleeding/stent fracture/impacted stent) if stents were not removed before 6 weeks.