Prediction of hospital stay after total gastrectomy.

The following factors were tested in the Cox proportional hazards model to determine their relationship to postoperative hospital stay in 174 patients who underwent total gastrectomy: age, sex, tumour stage, tumour free resection margins, anastomotic leakage, size of the EEA-stapler, preoperative hospital stay, location of the anastomosis and additional resection. Only anastomotic leakage and age independently affected postoperative hospital stay. Leakage, confirmed radiologically, developed in 20 patients (11.5%) of whom four died postoperatively. Another six patients who did not develop leakage also died, giving a postoperative mortality of 5.7%. The median age of the patients was 68 years (range 30-87), and the median postoperative hospital stay was 15 days (range 3-192). The operations were performed by 30 different surgeons of whom four performed more than ten, 9 performed 4-10 and 17 performed 1-3 total gastrectomies. There was a significant difference in postoperative mortality- and anastomotic leakage-rate between the surgeons who operated more than ten and less than four (p = 0.0122, p = 0.0453). However, when comparing the surgeons who operate more than 10 and 4-10 there was no difference in mortality or anastomotic leakage (p = 0.3493, p = 0.8867). When comparing the surgeons who operated on 4-10 with less than 4 there was no difference in postoperative mortality, but the anastomotic leakage rate was of borderline significance (p = 0.3916, p = 0.0582). Between the surgeons with the greatest experience there were significant differences as to leakage rate (p = 0.0003), time of operation (p < 0.001) but not in the number of thoracoabdominal incisions (p = 0.4939) or postoperative mortality (p = 0.8394). This study has shown conclusively that anastomotic leakage is the most important factor for prediction of postoperative hospital stay after total gastrectomy.