RESUMO
INTRODUCTION: Surgical risk calculators can estimate risk probabilities for postoperative outcomes utilizing patient-specific risk factors. They provide meaningful information for obtaining informed consent. The aim of the present paper was to evaluate the predictive value of the surgical risk calculators by the American College of Surgeons in German patients undergoing total pancreatectomy. METHODS: Data for patients who underwent total pancreatectomy between 2014 and 2018 were acquired from the Study, Documentation, and Quality Center of the German Society for General and Visceral Surgery. Risk factors were entered manually into the surgical risk calculators and calculated risks were compared with actual outcomes. RESULTS: Of the 408 patients analysed, predicted risk was higher in patients with complications except for the prediction of re-admission (P = 0.127), delayed gastric emptying (P = 0.243), and thrombosis (P = 0.256). In contrast, classification of patients into below, above, or average risk by the surgical risk calculators only produced meaningful results for discharge to nursing facility (P < 0.001), renal failure (P = 0.003), pneumonia (P = 0.001), serious complications, and overall morbidity (both P < 0.001). Assessment of discrimination and calibration showed poor results (scaled Brier scores 8.46 per cent or less). CONCLUSION: Overall surgical risk calculator performance was poor. This finding promotes the development of a specific surgical risk calculator applicable to the German healthcare system.