ABSTRACT
OBJECTIVE: To analyze treatment outcomes and approaches to predicting the postoperative morbidity and mortality in patients with perforated ulcers and cancer. MATERIAL AND METHODS: A non-randomized trial included 194 patients. The first group enrolled 45 (23%) patients with perforated ulcers and concomitant cancer who underwent at the oncology center; the second group included 149 (77%) patients with perforated ulcers and no cancer who underwent surgery in general surgical hospitals. Organ-sparing procedures prevailed (40 (88.9%) and 138 (92.6%) cases, respectively). Resections were performed in 5 (11.1%) and 11 (7.4%) patients respectively. Analyzing the factors affecting treatment outcomes, we studied crude (COR) and adjusted (AOR) odds ratios. ROC-analysis was used to assess diagnostic significance of the models for prognosis of morbidity and mortality. RESULTS: Length of hospital-stay was 10 (range 9-14) and 8 (range 7-9) days respectively. Postoperative complications (Clavien-Dindo grading system) occurred in 18 (40%) in 37 (24.8%) patients, respectively. According to multivariate analysis, predictors of complications in patients of the first group were treatment with NSAIDs/glucocorticoids and Charlson-Deyo index >3. Sensitivity of this model was 82.4%, specificity - 75.0%. Postoperative mortality was 15.6% (n=7) and 7.4% (n=11) respectively. According to multivariate analysis, predictors of mortality were age over 65 years and more than 5 chemotherapeutic courses. Sensitivity of the model was 85.7%, specificity - 97.4%. CONCLUSION: The stratified approach makes it possible to improve prediction of postoperative morbidity and mortality in patients with perforated ulcers.
