Management of spontaneous rupture of the oesophagus

Abstract

Abstract Aims Spontaneous rupture of the oesophagus (SRO) is a rare and often fatal event. The aim of this study was to evaluate the presentation, management and outcome of SRO in a single unit. Methods Data were collected on all patients presenting with SRO over a 5-year period with respect to presenting features, diagnostic investigations and subsequent management. Statistical analysis was by Student's t test, χ2 and Fisher's exact tests. Results Fourteen patients were identified, 12 men and two women with a median age of 64 (range 18–78) years; eight were tertiary referrals. Thirteen of 14 patients presented with chest or upper abdominal pain following vomiting or retching and 13 had an abnormal initial chest radiograph; only one presented with Mackler's triad of pain, vomiting and surgical emphysema. The median delay to diagnosis was 21 (range 1–84) h; this delay did not significantly affect outcome (P = 0·16). An endoscopic assessment and contrast swallow were performed in all patients. Nine of ten patients with a demonstrable leak and full-thickness tear were managed surgically and the four patients with no leak were managed conservatively (P = 0·005); surgical management consisted of thoracotomy, lavage, repair of the perforation and a feeding jejunostomy. Seven patients had a repair over a T tube and two had a primary repair. All conservatively managed patients had contained, controlled or intramural perforations and two also required a feeding jejunostomy. Patients requiring surgery had a longer hospital stay (mean(s.d.) 57·9(34·8) versus 22·2(30·7) days; P = 0·081) and a significantly longer intensive care unit stay (P = 0·044). The overall mortality rate from SRO was 14 per cent (two patients); no deaths occurred in the conservatively managed group. Conclusions SRO continues to be diagnosed late despite a classical history and/or abnormal chest radiograph. Endoscopic assessment of perforations is safe and in combination with a contrast swallow can confidently predict patients with contained or controlled rupture in whom non-operative management is successful.