This is the first study to show that CGA and ECOG-PS are predictive of postoperative delirium in elderly patients undergoing gastrectomy for gastric cancer. Delirium was found to be the most common complication after gastrectomy among elderly gastric cancer patients, occurring in 12.2% of patients. The group with delirium was significantly more likely to have ECOG-PS ≥ 2, diabetes mellitus, coronary heart disease, or cerebrovascular disorder. In this group, CGA showed frailty in IADL, MMSE, VI, and GDS-15 items.

In the multivariate analysis, MMSE-frailty and ECOG-PS ≥ 2 were independent risk factors for delirium. Postoperative delirium was more strongly associated with the mental factor of MMSE than with the physical factor of each CGA item or the physical component of ECOG-PS, consistent with our previous reports.14,15 The finding that age, operative procedure, and ASA-PS are not predictors of postoperative delirium also is interesting and consistent with previous reports.14,25 For the prediction of postoperative delirium, use of an MMSE score of ≤ 23 or an ECOG-PS ≥ 2 as a cutoff resulted in a sensitivity of 80.7% and a specificity of 74.1%. These results suggest that postoperative delirium in elderly patients with gastric cancer might be more easily predicted by a combination of MMSE and ECOG-PS.

Delirium is a clinical diagnosis often unrecognized and overlooked.9 The main diagnostic features include acute onset and variable symptom course, inattention, impaired consciousness, and cognitive impairment.26 The CAM algorithm is the method most widely used for identification. It is said to have a sensitivity of 94% and specificity of 89%. It is highly reliable.23 The CAM algorithm was used for diagnosis in this study.

The primary choice for managing delirium symptoms involves non-pharmacologic approaches such as tapering or discontinuation of psychotropic medications; addressing acute medical issues such as infection, dehydration, abnormal glucose metabolism, and malnutrition; and adjusting circadian rhythms. The use of antipsychotic medications is recommended only for patients with severe agitation that might lead to treatment discontinuation.9,27 In this study, a multidisciplinary team consisting of geriatricians, psychiatrists, and other specialists intervened immediately after the diagnosis of postoperative delirium. Of 31 patients, 7 required haloperidol or another antipsychotic medication.

Consistent with previous reports, the group with postoperative delirium in this study had a significantly longer hospital stay (P < 0.001).28 Therefore, avoiding delirium is not only a medical safety issue, but might also be a way to reduce medical costs due to hospitalization and to avoid ADL declines due to prolonged hospitalization.29 In addition, proper preoperative assessment of the risk for delirium might help surgeons explain the risks to patients and families and help families better understand the recovery process and potential outcomes.

In this study, the most common adverse event due to postoperative delirium was self-extraction of tubes implanted in the body. Because the nasogastric tube is important as a drain and source of information about postoperative bleeding, and because intragastric decompression can reduce stress on the anastomosis, self-removal of the nasogastric tube might have a negative impact on postoperative management. Furthermore, in one case of ileus in this study, tube removal resulted in gastrointestinal perforation and reoperation followed by death due to aspiration pneumonia. Such adverse events cannot be ignored in terms of medical safety.

It has been reported that postoperative delirium is preventable in 30–40% of cases.9 If patients at high risk for postoperative delirium are identified early, adverse events might be prevented via more comprehensive intervention by a multidisciplinary team that includes psychiatrists, geriatricians, nurses, and rehabilitation specialists.30

Regarding the prediction of postoperative delirium, several studies have investigated the usefulness of CGA for assessing the risk of postoperative complications in elderly patients. Yamamoto et al.14 reported that MMSE and GDS-15 are important for predicting preoperative delirium in patients with esophageal cancer. In addition to MMSE and GDS-15, Arita et al.15 reported that preoperative grip strength measurement is useful for predicting delirium in colorectal cancer. Thus, assessing the risk of delirium in advance and intervening might help reduce delirium. Indeed, it has been reported that in hip arthroplasty, a multidisciplinary preoperative CGA evaluation and intervention by a geriatrician can reduce the risk of postoperative delirium.31,32 Moreover, it also has been reported that preoperative administration of antipsychotic drugs reduces the incidence of delirium and has a positive effect on the severity and duration of delirium.33,34 Therefore, appropriate identification of patients at high risk for postoperative delirium might allow for proactive therapeutic interventions and prevent delirium from occurring.

Although CGA is important for preoperative risk assessment for the elderly, in reality, not all hospitals have geriatricians on staff, and not all facilities perform CGA evaluation due to the time and cost involved. In fact, the CGA test in this study required approximately 30 min. Due to scheduling conflicts, 93 patients were unable to undergo CGA evaluation. Thus, if implemented as a routine procedure in all hospitals, it could be problematic in terms of time and human resources.

On the other hand, ECOG-PS is a very standard preoperative physical assessment tool, and many hospitals perform it routinely. The MMSE also is more convenient, requiring only approximately 10 min of evaluation by a non-expert geriatrician. Therefore, it is feasible to perform these two tests in a busy daily practice. They might be realistically feasible tools for predicting delirium.

This study had several limitations. First, because this was a retrospective study conducted at a single institution, it had some potential for bias. In the future, a prospective validation study will be conducted with a multidisciplinary team intervention for patients at high risk for postoperative delirium.

Second, several other methods besides CGA can be used to assess surgical risk for elderly patients with cancer, but we have not evaluated or compared them.35,36 In the future, it is necessary to compare evaluation tools and search for the best evaluation method.

In conclusion, this study suggests that mental assessment based on the MMSE among CGA items and physical assessment based on ECOG-PS might be important for predicting postoperative delirium in elderly patients with gastric cancer.