Stage-dependent approach to stomach cancer

Accurate staging by unambiguously defined, comparable criteria is essential for a phase-adjusted therapeutic concept to cope with gastric carcinoma. Decision-making on gastrectomy or subtotal gastric resection has to depend primarily on tumour localisation and prognosis by histological classification according to Laurén. Between January 1, 1980, and May 1, 1988, operations for gastric carcinoma were performed on 203 patients at the Surgical Department of Charité, with gastric resection being applied to 144 of them. Postoperative lethality amounted to nine per cent. Indications were established in 66 cases for gastrectomy and in 78 patients for subtotal gastric resection. Two-year survival rates were 52 per cent for the intestinal cell type and 35 per cent for the diffuse type. Prognosis can be improved by radical resection and extensive lymph node removal at the first two lymph node stages. Resection of other infiltrated organ regions may quite often prove necessary for oncological radicality. Splenectomy will be chosen for stomach carcinomas localised in the upper and medium thirds or in tumour stages III and IV.     

Miniresection of small peripheral carcinomas of the lung

A total of 499 miniresective procedures were done for lung carcinoma over a 29-year period to July 1986. The operations were carried out in three groups: patients in whom a previous resection had been done, patients with stage III carcinoma, and those with stage I or II disease. Overall, perioperative mortality was 2.4 per cent. Cumulative survival in 296 patients with stage I or II carcinoma was 52 per cent at 5 years, 31 per cent at 10 years, and 11 per cent at 15 years.     

Clinical experience with continuous epidural infusion of bupivacaine at 6 ml per hour in obstetrics

Maintenance of continuous epidural analgesiafor obstetrics by slow infusion is a feasible alternative to maintenance by intermittent injection. The method described by Davies and Fettes using 0.25 per cent bupivacaine through a 6 ml/hr capillary flow device (Intraflo®), with the fast-flush cut-off, is simple and convenient. However, failure of the fast-flush valve, resulting in a high flow rate, has been reported. It is therefore imperative that, prior to connecting the pressurized system to the epidural catheter, correct function of the device be verified by observing the flow rate in the microdrip chamber. We have used this method for the past year and found the equipment reliable, but in order to achieve more satisfactory analgesia the concentration of the infusion was increased initially to 0.375per cent, and then reduced to 0.30 per cent. The records of the first 187 patients were reviewed retrospectively. Group I(n = 99) received an infusion of 0.25 per cent bupivacaine, Group II (n = 49) received 0.30 per cent and Group II (n = 39) received 0.375 per cent. Significantly more patients in Group I (30 per cent) required supplementary top-ups (in addition to the infusion) for the first stage of labour than in either Group II (10 per cent), or Group III (13 per cent). Top-ups for delivery were given to 55 per cent of patients in Group I, 64 per cent in Group II and 48 per cent in Group III. The incidence of motor block was significantly higher in Group III (21 per cent) than in either Group 1(1 per cent) or Group II (2 per cent). For continuous epidural infusion at 6 ml/hr, 0.30 per cent bupivacaine combines optimum analgesia with minimal side-effects.     

Reduced length of stay by implementation of a clinical pathway for bariatric surgery in an academic health care center.

Bariatric surgery is being performed in increasing numbers in an era when reimbursements are being reduced. Academic health centers bear the responsibility for training surgeons to perform these operations yet must keep costs to a minimum and retain high quality. The UCLA Bariatric Surgery Program developed a clinical pathway for the pre- and postoperative management for gastric bypass patients to achieve these goals. Medical records for 182 consecutive gastric bypass patients were retrospectively reviewed before implementation of the pathway (Group I) during the fiscal year of 1998/1999. Data on average length of stay, average intensive care unit length of stay, average standard variable cost, percentage readmission rate, and percentage return to the operating room were collected. This information was compared with the data collected prospectively from 182 patients after implementation of the pathway in July of 1999 (Group II) during the fiscal year of 1999/2000. Hospital cost per admission was reduced by 40 per cent in Group II compared with Group I (P < 0.02). The average length of stay was reduced from 4.05 days in Group I to 3.17 days in Group II (P < 0.033). Overall readmission rate was decreased from 4.2 per cent in Group I to 3.2 per cent in Group II (P < 0.05). There were no differences in morbidities between both groups. The pathway reduced costs by reducing the hospital length of stay, intensive care unit utilization, and readmission rates. Quality was maintained as evidenced by a similar pattern of postoperative morbidities yet readmission rates were reduced. Our results indicate that implementation of a clinical pathway for bariatric surgery reduces cost and improves quality of care in an academic institution.     

Co-carcinogenic effects of several Korean foods on gastric cancer induced by N-methyl-N'-nitro-N-nitrosoguanidine in rats

In study I, 48 ACI and Fisher inbred rats were given MNNG 100 micrograms/ml, with or without 1 per cent or 3 per cent red pepper diet; in study II, 164 Sprague-Dawley rats given MNNG 100 micrograms/ml, with or without 5 per cent or 10 per cent NaCl; in study III, 181 Wistar rats given MNNG 83 micrograms/ml with or without maejoo 10 gm per cent/diet; in study IV, 78 Wistar rats given MNNG 83 micrograms/ml with or without ginseng extract 150 micrograms/ml; in study V, 120 Wistar rats given MNNG 83 micrograms/ml with or without retinyl palmitate 150,000 IU/kg. Except for study II (28 weeks), all rats were fed the diets for 37 weeks and were examined at 38 weeks or 40 weeks. In study I, tumor incidence in rats fed a red pepper diet and MNNG solution were 57 per cent (ACI rats, 1 per cent red pepper) and 63 per cent (Fisher rats, 1 per cent or 3 per cent red pepper) which were higher than control group (44 per cent, 43 per cent); in study II, gastric cancer, 61.9 per cent (10 per cent NaCl-MNNG), 27.3 per cent (control); in study III, gastric cancer, 14.8 per cent (maejoo-MNNG), 24 per cent (control); in study IV, malignant tumor of gastroduodenum, 3.4 per cent (ginseng-MNNG), 32.1 per cent (control); in study V, forestomach papilloma, 10.7 per cent (retinoid-MNNG), 29.4 per cent (control), and cancer in duodenum and small intestine, 50.0 per cent (retinoid-MNNG), 17.6 per cent (control). Thus, gastric carcinogenesis was enhanced by red pepper and a high salt diet, was inhibited by a maejoo and ginseng diet and was not effected by vitamin A.     

Studies of fentanyl-supplemented anaesthesia: Effect of naloxone on the circulation and respiration

Ninety-nine unselected patients were given a standardized general anaesthetic with fentanyl 1.5 jag · kg^-1 every 30 minutes and were randomly divided into three equal Groups: Group I patients received naloxone 0.1 mg, Group II naloxone 0.2 mg, and Group III naloxone 0.4 mg, at the end of operation and after the reversal of neuromuscular blockade. After administration of naloxone systolic blood pressure increased by 4, 8 and 7 per cent and mean arterial blood pressure increased by 3, 8 and 8 per cent in Groups I, II and III respectively; heart rate increased by 4, 11 and 8 per cent and rate-pressure product increased by 7, 18 and 15 per cent in Groups I, II and III respectively. Tidal volume increased by 97, 101 and 95 per cent and minute volume increased by 122, 164 and 143 per cent in Groups I, II and III respectively after naloxone. Forty-nine percent of patients had a tidal volume of less than 5 ml · kg^-1 ora minute volume of less than 50 ml · kg^-1 before administration of naloxone; after naloxone three patients in Group I (naloxone 0.1 mg) had a tidal volume of less than 5 ml · kg^-1 and no patient had a respiratory minute volume of less than 50 ml · kg^-1. It is concluded that under the conditions of this study naloxone 0.1 mg is adequate to reverse the respiratory depressant effect of fentanyl in the majority of cases.     

Surgical treatment of perforated duodenal ulcer: a prospective trial between simple closure and definitive surgery

Seventy-eight patients with perforated duodenal ulcer were prospectively studied between 1977 and 1982. Patients were alternately allocated to receive simple closure (Group I, 33 patients) and definitive surgery (Group II, 32 patients). High-risk patients and those whose conditions dictated a definitive operation were excluded. All patients in Group II had a truncal vagotomy and drainage except one who had a proximal gastric vagotomy. There was no death in Group I or Group II; the complication rate and postoperative course were similar. Twenty-seven patients in Group I and 26 patients in Group II were available for follow-up 12 to 80 months after operation, mean 39 months. Good/excellent results were achieved in 30 per cent of Group I compared with 81 per cent of Group II (P less than 0.01). Eighty-five per cent of Group I patients developed recurrent ulcer symptoms and 33 per cent had already had a second definitive operation. Two patients (8 per cent) in Group II were reoperated upon for recurrent ulcer due to an incomplete vagotomy. In a population of patients where long-term follow-up and medical treatment for duodenal ulcer is unsatisfactory, truncal vagotomy with drainage should be the treatment of choice for perforation. Simple closure should be reserved for high-risk patients or when the surgeon is inexperienced.     

Predictors of operative morbidity and mortality in gastric cancer surgery

BACKGROUND: The aim of this study was to identify factors that predict morbidity and mortality in gastric cancer surgery. METHODS: Data on 719 consecutive patients who underwent operations for gastric cancer at Seoul National University Hospital between January and December 2002 were reviewed. RESULTS: Overall morbidity and mortality rates were 17.4 per cent (125 patients) and 0.6 per cent (four patients) respectively, and the rates of surgical and non-surgical complications were 14.7 per cent (106 patients) and 3.3 per cent (24 patients). Morbidity rates were higher in patients aged over 50 years (odds ratio (OR) 1.04 (95 per cent confidence interval (c.i.) 1.02 to 1.06)), when the gastric tumour was resected with another organ (36 per cent for combined resection versus 15.4 per cent for gastrectomy only; OR 3.25 (95 per cent c.i. 1.76 to 6.03)) and when gastrojejunostomy was used for reconstruction after subtotal gastrectomy (17.0 per cent for Billroth II versus 9.5 per cent for Billroth I; OR 2.00 (95 per cent c.i. 1.05 to 3.79)). Only three patients (2.8 per cent) with a surgical complication underwent reoperation, two for adhesive obstruction and one for intra-abdominal bleeding. CONCLUSION: Age, combined resection and Billroth II reconstruction after radical subtotal gastrectomy were independently associated with the development of complications after gastric cancer surgery.     

Co-carcinogenic effects of several Korean foods on gastric cancer induced by N-methyl-N′-nitro-N-nitrosoguanidine in rats

In study I, 48 ACI and Fisher inbred rats were given MNNG 100 μg/ml, with or without 1 per cent or 3 per cent red pepper diet; in study II, 164 Sprague-Dawley rats given MNNG 100 μg/ml, with or without 5 per cent or 10 per cent NaCl; in study III, 181 Wistar rats given MNNG 83 μg/ml with or without maejoo 10 gm per cent/diet; in study IV, 78 Wistar rats given MNNG 83 μg/ml with or without ginseng extract 150 μg/ml; in study V, 120 Wistar rats given MNNG 83 μg/ml with or without retinyl palmitate 150,000 IU/kg. Except for study II (28 weeks), all rats were fed the diets for 37 weeks and were examined at 38 weeks or 40 weeks. In study I, tumor incidence in rats fed a red pepper diet and MNNG solution were 57 per cent (ACI rats, 1 per cent red pepper) and 63 per cent (Fisher rats, 1 per cent or 3 per cent red pepper) which were higher than control group (44 per cent, 43 per cent); in study II, gastric cancer, 61.9 per cent (10 per cent NaCl-MNNG), 27.3 per cent (control); in study III, gastric cancer, 14.8 per cent (maejoo-MNNG), 24 per cent (control); in study IV, malignant tumor of gastroduodenum, 3.4 per cent (ginseng-MNNG), 32.1 per cent (control); in study V, forestomach papilloma, 10.7 per cent (retinoid-MNNG), 29.4 per cent (control), and cancer in duodenum and small intestine, 50.0 per cent (retinoid-MNNG), 17.6 per cent (control). Thus, gastric carcinogenesis was enhanced by red pepper and a high salt diet, was inhibited by a maejoo and ginseng diet and was not effected by vitamin A. Presented at the 85th Congress of the Japan Surgical Society, April 1985, in Sendai, Japan     

An analysis of filtration and volume of radionucleotide in sentinel lymph node biopsy in breast cancer patients

Controversy has occurred regarding whether to filter or not to filter the radionucleotide and what the optimal volume is when performing sentinel lymph node biopsies. To try and resolve this question we retrospectively looked at sentinel-node biopsies for breast cancer performed at our institution over an 18-month period. One hundred seven patients underwent sentinel node biopsy. Ninety-four patients had an axillary-node dissection. Twelve patients did not have a nodal dissection based on National Surgical Adjuvant Breast and Bowel Project protocol, and one patient refused dissection. Patients fell into three groups: Group I, filtered 8 cm3; Group II, unfiltered 8 cm3; and Group III, unfiltered 16 cm3 (NSABP protocol). Sentinel nodes were identified in 96.3 per cent of Group I, 84.9 per cent of Group II, and 96.3 per cent of Group III. These groups were not statistically different. With the addition of blue dye the ability to identify the sentinel lymph node was for Group I 96.3 per cent, for Group II 96.2 per cent, and for Group III 100 per cent. The false negative biopsy result was 0 per cent for all groups. The patients receiving 16-cm3 volume complained about a greater level of discomfort compared with the lower-volume patients. In conclusion neither the volume nor the filtration process affected the surgeon's ability to find the sentinel lymph node or the false negative rate. The higher injection volumes resulted in more pain. The costs and radiation exposure of the filtration process are not warranted.     

Oral transmucosal fentanyl citrate for premedication in paediatric outpatients

Two doses (10-15 micrograms.kg-1, Group I, and 15-20 micrograms.kg-1, Group II) of oral transmucosal fentanyl citrate (OTFC) plus a placebo (Group III) were evaluated for premedication in 105 healthy children, aged 2 to 13 yr, undergoing short (less than 1 hr) operations in the hospital short-stay unit. The study was randomized and double-blinded and 91 of the 105 children also received droperidol, 25 micrograms.kg-1 IV, after induction of anaesthesia with halothane and N2O in oxygen. Both doses of OTFC produced significantly greater sedation (first present at 20 min) and anxiolysis (first present in Group I at 40 min) than the placebo. Recovery times were similar in the three groups and analgesic requirements in the recovery room were significantly lower in Group I than Group III. Both OTFC groups took longer to tolerate oral fluids in the postoperative discharge unit than the placebo group and this caused patients in Group I to have a delayed discharge from the hospital compared to Group III. Preoperative pruritus occurred significantly more frequently in Groups I and II (58 and 76 per cent, respectively) than Group III (23 per cent). Although the incidences of nausea and vomiting tended to be slightly higher in the OTFC groups in the preoperative holding and postoperative discharge units, the differences among the groups were not statistically significant. Likewise droperidol did not reduce the incidence of postoperative nausea or vomiting. The data indicate that OTFC may be a safe and effective premedicant in paediatric patients having short operations but delays discharge from the hospital (by 30-50 min) by delaying the time patients tolerate fluids early after operation.     

Electrical pacing of the Roux limb resolves delayed gastric emptying

This study compares antegrade gastric and Roux-limb electrical pacing in the evaluation and treatment of delayed gastric emptying following vagotomy, antrectomy and Roux-Y enterostomy. Twenty-four male Sprague-Dawley rats (250 g) underwent midline laparotomy, truncal vagotomy, antrectomy, and Roux-Y jejunostomy. Pacemaker leads were implanted 1 cm apart in both the gastric fundus and proximal Roux limb. Pacing was accomplished using a 0.5 mA, 50 msec, 0.33 Hz signal and monitored by an oscilloscope. Animals were fasted for 2 hr and then gavaged with 1.0 cc of 99mTc-labeled egg white. At 1 hr rats were anesthetized. The stomach, Roux limb, small intestine, and colon were doubly ligated and excised without disturbing their contents. The total number of counts per minute per rat was determined in a gamma radiation counter, and percentage gastric emptying (GE) was evaluated. Group I controls (n = 8) retained 76 +/- 15.8% for a GE of 24%. Group II gastric paced rats (n = 8) retained 64.5 +/- 19.2% (GE 35.5%) and Group III Roux-limb paced rats (n = 8) retained 46.8% +/- 13.2 (GE 53.2%) (P less than 0.005 III vs I, P less than 0.05 III vs II). The amount of radioactive meal distal to the Roux limb was also evaluated. Group I had 15.7 +/- 16.1%, Group II (gastric paced) 20.5 +/- 19.0%, and Group III (Roux-limb paced) 37.2 +/- 11.9% (P less than 0.005 III vs I, P less than 0.05 III vs II). These data imply that Roux-en-Y limb dysmotility may contribute to delayed gastric emptying following vagotomy, antrectomy, and Roux-Y enterostomy.(ABSTRACT TRUNCATED AT 250 WORDS)     

Spinal narcotics for postoperative analgesia in total joint arthroplasty. A prospective study.

Sixty patients who were scheduled to have an elective total hip or knee arthroplasty were randomly assigned to one of three groups of twenty patients each before operation with spinal anesthesia. A double-blind technique was used throughout the study. The patients in Group I (control group) received hyperbaric 1 per cent tetracaine with epinephrine as the subarachnoid spinal anesthetic; the patients in Group II (morphine group), hyperbaric 1 per cent tetracaine with epinephrine and a single subarachnoid dose of Duramorph (morphine sulphate), 0.5 milligram; and those in Group III (Dilaudid group), hyperbaric 1 per cent tetracaine with epinephrine and a single subarachnoid dose of Dilaudid (hydromorphone hydrochloride), 0.002 milligram per kilogram of body weight. During the first twenty-four hours after the operation, the patients in Group II and Group III had significantly less pain compared with those in Group I. This was shown by the use of a visual linear-analog pain scale (p less than 0.05), the patients' ratings of the quality of relief of pain (p less than 0.02), and comparative measurements of the pain-altering medications that were used (p less than 0.05). The patients in Group II and Group III did not have any more complications or side effects than those in Group I. There was no significant difference in the quality and duration of analgesia between Group II and Group III.     

Changing pattern of histological type, location, stage and outcome of surgical treatment of gastric carcinoma

BACKGROUND: There are indications that some features of gastric carcinoma are changing, with a possible impact on prognosis. The aim of this study was to examine any changes in type, location, stage, resection rate, postoperative mortality rate or prognosis for patients with gastric carcinoma in a well defined population. METHODS: During 1974-1991, 1161 new cases of gastric adenocarcinoma were diagnosed in Osterg?tland County, Sweden. Tumour location, Laurén histological type, tumour node metastasis (TNM) stage, radicality of tumour resection and postoperative complications were recorded after histological re-evaluation of tissue specimens and examination of all patient records. Dates of death were obtained from the Swedish Central Bureau of Statistics. Time trends were studied by comparing the intervals 1974-1982 (period 1) and 1983-1991 (period 2). RESULTS: The proportion of diffuse type of adenocarcinoma increased (from 27 to 35 per cent), while that of mixed type decreased (from 16 to 9 per cent) and that of intestinal type was unchanged. The proportion of tumours located in the proximal two-thirds of the stomach increased (from 32 to 42 per cent) and the proportion of patients with tumours in TNM stage IV decreased (from 32 to 25 per cent). Overall tumour resection rates were unchanged, although the proportion of radical total gastrectomies increased (from 36 to 50 per cent). Excluding tumours of the cardia or gastric remnant after previous ulcer surgery, the 5-year relative survival rate after radical resection increased from 25 to 36 per cent and the postoperative mortality rate decreased for both radical (from 11 to 4 per cent) and palliative (from 18 to 6 per cent) resection. CONCLUSION: The patterns of tumour histology, location and stage of gastric carcinoma have changed in the authors' region. These changes were paralleled by a significant improvement in survival and postoperative mortality rates.     

Hepatic resection in the treatment of hepatocellular carcinoma: report of 60 cases

During the last 3 years and 9 months, hepatic resections were carried out on 60 patients with hepatocellular carcinoma (HCC). The resections were radical in 44 and palliative in 16 cases. Simultaneous operations were performed for oesophageal varices in 11 (9 prophylactic and 2 therapeutic) and for cholelithiasis in 9 patients. Fifty-nine patients had underlying hepatic disease; 52 had cirrhosis and 7 chronic hepatitis. Operative mortality rate within 1 month was 10 per cent and overall hospital mortality rate was 16.7 per cent. Excluding operative and hospital deaths, 76.8 per cent of patients who had radical excision and 18.8 per cent with palliative resections are alive. The result indicates that HCCs are frequently resectable even in the presence of liver cirrhosis provided that they are discovered at a relatively early stage.     

Distal pancreatectomy with and without splenectomy. A comparative study

Twenty-one patients undergoing distal pancreatectomy from January 1980 through April 1987 were studied retrospectively. Group I (n = 10) had distal pancreatectomy with splenectomy, and Group II (n = 11) had distal pancreatectomy with splenic preservation. The groups were comparable in mean age and extent of pancreatic resection. Operative time in Group I patients who did not require additional major procedures was 3.74 +/- 1.01 hours, compared with 2.86 +/- 1.68 hours for similar Group II patients. The overall complication rate in Group I was 40 per cent, including two pancreatic fistulas, one subphrenic abscess, and one gastric fistula. In Group II the overall complication rate was 36.4 per cent, with one pancreatic fistula, one subphrenic abscess, and one colonic fistula. Splenic infarction occurred in only one patient (Group II), in whom both splenic vessels were ligated. No patient developed insulin-dependent diabetes. There were no deaths. The mean hospitalization time was 18.8 days in Group I, and 17.5 days in Group II. Splenectomy should not be considered a routine part of distal pancreatic resection. Splenic preservation can be achieved in selected cases with no increase in complication rate, operative time, or length of postoperative hospitalization.     

Sleeve resection of the bronchus and the pulmonary artery for pulmonary lesions

The tissue preserving resections for non-small-cell bronchial carcinoma can be grouped into three main categories: I. bronchoplastic procedures, II. angioplastic procedures, and III. concomitant broncho- and angioplastic procedures, and into the subgroups, standard and extended sleeve resection. The indications are; elderly patients, impaired respiratory reserve, limited tumour growth, and palliative surgery. The analysis of 229 cases yielded follow-up data in 192. The estimated 5 years survival rate was 34 per cent, 19 per cent and 14 per cent in categories I, II and III, respectively. The decrease in survival was due to a greater tumour burden. The operative mortality rate was 8.9 per cent in category I and 17 per cent in category III, such being comparable with standard or extended pneumonectomy, respectively. Surgical techniques and postoperative complications are discussed. Presented at the 82nd Annual Congress of the Japan Surgical Society, 1982, Chiba, Japan     

Gastro-Gastric Fistulas and Marginal Ulcers in Gastric Bypass Procedures for Weight Reduction

Background: Gastro-gastric fistulas and marginal ulcers are frequent and serious complications of gastric compartmentalization procedures for obesity. Methods: The authors analyzed 810 patients after 911 operations for gastro-gastric fistulas and marginal ulcers over an 8-year period. All patients underwent a form of gastric bypass, in which a pouch is constructed along the lesser curvature of the stomach. The outlet of the pouch was restricted with a prosthetic band. In the first 189 patients (Group I), the pouch and stomach were stapled in continuity or partially divided. In the next 222 patients (Group II), segments were stapled and separated by transection. In the remaining 492 cases (Group III), in addition to transection of the stomach, a limb of jejunum was interposed between the pouch and excluded stomach. Stapled anastomoses were done in Group I and II patients and a portion of Group III patients. The remaining patients underwent hand-sewn anastomosis. Results: Gastro-gastric fistulas occurred in 49% of the patients in Group I, 2.6% of those in Group II, and 0% of those in Group III. In stapled anastomosis, the incidence of marginal ulceration in Groups I, II, and III were 8.5%, 5.4%, and 5.1%, respectively. In a subset of Group III patients, in whom a two-layer, hand-sewn anastomosis was done, the incidence was 1.6% when the outer layer was not absorbable and 0% when both layers were absorbable. Conclusions: Gastro-gastric fistulas and marginal ulcerations are likely the result of breakdown of the mucosa resulting from migrating staples and other foreign material. Lack of integrity of the gastric lining facilitates the action of the gastric digestive process. Transection of gastric segments with interposition of jejunum prevents gastro-gastric fistula formation. An intact serosa appears to block the digestion of bowel wall by gastric enzymes. Our early data suggest that the use of absorbable sutures at the gastrojejunostomy significantly decreases the incidence of marginal ulceration.     

Effect of surgical experience on the results of resection for oesophageal carcinoma

From 1957 to 1976 oesophageal resection for carcinoma was performed in 1119 patients reported to the West Midlands Cancer Registry. The operations were performed on 581 patients by 127 surgeons who averaged three or less resections per annum (the 'occasional' group). These were compared with 538 patients (the 'frequent' group) whose resections were performed by four surgeons who averaged six or more resections per annum. Operative mortality was 39.4 per cent in the 'occasional' group and 21.6 per cent in the 'frequent' group (P less than 0.001). The age adjusted 5-year survival was 11.1 and 15.2 per cent respectively (P less than 0.05) but when the operative deaths were excluded there was no significant difference. We suggest that oesophageal resection for carcinoma should be performed only where there is an acceptably low operative mortality rate.