In the present study, the comparison of the incidence rates for the postoperative delirium between the combined-administration (with suvorexant and ramelteon) group and the no-hypnotic group as well as the results of the logistic regression analyses in the delirium group and the no-delirium group suggested the efficacy of the combined administration with suvorexant and ramelteon in preventing the postoperative delirium for the cancer patients. These results indicated a similar tendency with the results of the previous study by Hatta et al. investigating the efficacy of the combined administration with suvorexant and ramelteon for the patients with various diseases including heart failure and brain stroke [11]. One of the similarities with the previous studies was the dates of incidence. Days 1 and 2 in the present study for most of the patients who had the postoperative delirium in the combined-administration group and the no-hypnotic group both coincided with the results reported by Booka et al. [12]. Since postoperative delirium is supposed to occur from 10 min to 7 days [13, 14] after surgery, this study examines the cumulative incidence of delirium over a 7-day period. Likewise, a significant difference was also observed in the analysis limited to the period of frequent occurrence of postoperative delirium up to 3 days after the surgeries, suggesting the efficacy of the combination of suvorexant and ramelteon.

According to the report by Booka et al. [12], the incidence rate for the postoperative delirium was 2.4% (1/41) when cancer patients received either the sole remedy with ramelteon or the combined administration with suvorexant and ramelteon. Among these cases, the incidence rate for the postoperative delirium was 0% (0/19) when the patients received the combined administration with suvorexant and ramelteon. The multivariate analyses in the report indicated that the use of ramelteon was the preventive factor for the postoperative delirium, regardless of using suvorexant in combination, in the surgical treatment in the esophageal as well as the head and neck regions (OR: 0.06, CI: 0.0066–0.55). On the other hand, in the present study, the incidence rate for the postoperative delirium could be lowered by using the combined administration with suvorexant and ramelteon. In addition, in the report by Booka et al., the patients who didn’t take either suvorexant or ramelteon unexceptionally received benzodiazepine hypnotics, while those who didn’t receive the combined administration with suvorexant and ramelteon in the present study used no hypnotics at all, which should be noted as a distinctive difference.

Honda et al. reported in their study that the elderly patients and the male ones indicated the risk factors affecting the delirium incidence [3]. Although the multivariate logistic analyses in the present study didn’t indicate any significant differences between the elderly/male patients and the younger/female ones, the results of the univariate logistic analyses reflected a similar trend with the previous report by Honda et al. The reasons why there were not any significant differences between the elderly/male patients and the younger/female ones in the present study include the fact that the number of the study subjects was rather small, and as a result, the statistical power might not have been enough for substantiating the trend. Although the significance of the regression analysis with the present number of factors was maintained at P = 0.002, the small number of cases also affected the number of factors for which multivariate logistics analysis could be performed.　Meanwhile, another previous study reported that the incidence rate for the postoperative delirium was higher in female patients than male ones [15]. Therefore, future studies should include study subjects in bigger numbers than the present study in order to gain more precise and extended results to prove how gender and the delirium incidence rate are associated.

There are some limitations in this study. Firstly, the limitations of the present study can be the bias in the patient background factors. In the combined-administration group, as the numbers of the patients aged 75 or older and those with the history of mental disorder were significantly big and the number of male patients was significantly small, age and gender may have affected the delirium incidence rates in each group. In one of the previous studies, Kazmierski et al. reported that the delirium incidence rate after cardiac procedures among the depression patients was significantly high [16]. Among the patient background factors in the present study, the history of mental disorder turned out to be a significantly superior one for the combined-administration group. Secondly, since all the applicable patients were extracted from the electronic medical records by surveying all the drugs prescribed by the attending physician of the Division of Hepato-Biliary-Pancreatic Surgery, and the drugs prescribed by the physicians at other divisions excluding anesthesiology, as well as the drugs which the patients had brought in from other hospitals, some of the cases included unknown intentions for prescribing suvorexant and ramelteon. As the combined remedy with suvorexant and ramelteon was administered for the purpose of improving insomnia in the first place, it is possible to speculate that delirium tended to be induced more often in the combined-administration group than the no-hypnotic group, because insomnia is one of the mental disorders inducing delirium. In addition, it is undeniable that the prescription for suvorexant and ramelteon may have been biased, as the physician who prescribed them could have thought that the patients had high risk for delirium. However, it was controlled well in the present study. If some of the patients in the combined-administration group had similar background factors as the no-hypnotic group, which included the cases where the patients were not elderly and/or didn’t have the history of mental disorder for example, the incidence of delirium in the combined-administration group is well controlled, and the difference in incidence between the two groups may be greater. Thirdly, there were some other factors affecting the delirium incidence rates than those covered in the present study. They include lights in the hospital rooms, nutrition status, and sleep hours for instance. These are the factors which the nursing skills of the ward staff, who are responsible for providing non-drug therapy, matter in most cases and may have possibly affected the delirium incidence rates in the present study [17]. Also noteworthy was that, while longer operation time [18], longer duration of anesthesia [19], and worse postoperative pain control [15, 20] affect the occurrence of postoperative delirium, some of the cases in the present study included no clear descriptions for operation time or endpoints (Numerical Rating Scale, Visual Analogue Scale, etc.). Therefore, comparisons using statistical analysis could not be made. However, the electronic medical records of the patients did not contain any reports on abnormally long operation time, prolonged duration of anesthesia, or increased pains, which suggests that their effects on the surgeries and postoperative management must have been small. Finally, as the investigator (S.I.) in this study was also involved in the analysis itself, this study was not blinded, and the influence of rater bias cannot be denied for concluding that suvorexant-ramelteon is effective in preventing postoperative delirium. It is suggested that if a prospective, blinded study is conducted in the future, in which this bias can be eliminated, the difference between the two groups may be smaller than the present study.