Surgical resection for elderly patients is accompanied with significant perioperative mortality and frequent postoperative complications. [4,5] In addition, the life expectancy of this population is limited, in comparison with that of young patients. Therefore, the indications of surgical treatment should be discussed comprehensively based on various patient- and tumor-related factors. In this study, 20 patients (27%) died from diseases other than gastric cancer during the follow-up period of five years after gastrectomy. This was an indispensable result, and OS, including deaths from diseases other than gastric cancer, should be considered equally important as DSS in elderly patients. Surgical indications should be determined by considering not only DSS but also OS, especially for elderly patients. The results of the current study clearly showed that NRC mortality is useful for predicting OS but not DSS.
Previous reports have identified sex, [18,19] ECOG-PS,  surgical procedure, [4,7] PNI, [7,20] and postoperative complications, [7,19] as predictors of OS among elderly patients with gastric cancer after gastrectomy. In our study, NRC-mortality, in addition to ECOG-PS and surgical procedure, was found to be the independent predictors of OS. When concerning surgical procedures, total gastrectomy may sometimes result in malnutrition  and aspiration pneumonia, particularly in elderly patients.  Given these insights, it may be better to avoid total gastrectomy for elderly patients with gastric cancer. On the other hand, postoperative complications have been recognized as reliable prognostic factors after curative resection of various cancers, including gastric cancer. [7,17,23-25] Based on our recent analysis that demonstrated a possible correlation between the adverse prognostic effect and immune status of patients,  the immunocompromised state of elderly patients might have affected our current findings.
The NCD Risk Calculator was originally a tool for the prediction surgical morbidity and short-term surgical mortality.  The NRC-mortality was calculated based on logistic analysis from a large-scale NCD database, in which surgery-related deaths were registered from various causes, including deaths secondary to complications, worsening of comorbidities, and sudden deaths of unknown cause. In this study, we found that the NRC-mortality predicted not only the short-term mortality rate but also the long-term outcomes of elderly patients with gastric cancer after gastrectomy. In this population, the perioperative mortality risk might correlate with the risks of death secondary to worsening of comorbidities and of sudden deaths in both the perioperative and late phases after gastrectomy and, consequently, might be related with OS, as shown by our analysis. Furthermore, no elderly patients with NRC-mortality >4.1 survived for 5 years postoperatively, which may indicate that the higher the NRC-mortality, the lower the expected survival.
POSSUM (Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity)  and E-PASS (Estimation of Physiologic Ability and Surgical Stress)  are well-known prognostic scoring systems that are based on statistical analysis of actual surgical data. However, both scoring systems provide a comprehensive prognostic assessment only after surgery and may not be suitable for accurately predicting the surgical risk prior to surgery. Moreover, the E-PASS scoring system shows that postoperative complications mainly depend on the surgical stress score, which is the actual surgical invasive outcome score.  Given this, the NRC-mortality rate, compared with POSSUM and E-PASS, may be more ideal tool for perioperative risk assessment of the short- and long-term outcomes. Risk assessment based on the POSSUM scoring system is known to deviate from the actual results of a limited population, such as low-risk cases and elderly patients. [29,30] Moreover, assessments using these scoring systems have been reported to not reflect the prognosis of elderly patients. 
There were some limitations in this study. First, it was small-scale and was done at a single institution. Further multi-institutional analyses of large number of patients should be conducted to confirm the current results. Second, the cohort included only a small number of patients who underwent laparoscopic gastrectomy, which is less invasive and may affect the long-term outcome in elderly patients. Concerning laparoscopic and robotic procedures for the elderly, huge numbers of patients’ clinicopathological information would be collected in the NCD system, which shall elucidate the clinical utility of the NRC-mortality in minimally invasive surgeries in the future.
In conclusion, the NRC-mortality, as well as ECOG-PS and surgical procedure, might be clinical useful for predicting not only surgical mortality but also OS after gastrectomy in elderly patients with gastric cancer. As much as possible, gastrointestinal surgeons should avoid performing total gastrectomy on elderly patients with gastric cancer and consider treatment without gastrectomy, particularly in cases with poor PS. The NCD Risk Calculator may help in the proper surgical risk assessment of such patients.