ECOG PS and palliative chemotherapy have been reported as independent prognostic factors for OS  . Specifically, patients with a good ECOG PS are recommended to undergo palliative chemotherapy , and a meta-analysis of randomized controlled trials of advanced gastric cancer reported that chemotherapy extended OS by 6.7 months relative to BSC . In this study, we accordingly classified patients into clusters and determined that patients in Cluster 3 were most likely to have an ECOG PS of 0–2 and to have received palliative chemotherapy compared to the other clusters. We noted that metastases of the peritoneum, lymph nodes, bone, lung, ovary, and brain have also been identified as potential prognostic factors for OS     . In this study, only the likelihood of peritoneal metastasis differed, with a higher incidence in Cluster 2 relative to the other clusters. These results might be attributable to the use of data collected at the first consultation which did not account for newly developed metastases.
Lauren’s histological classification classifies gastric cancers into two histological subtypes, intestinal and diffuse, as this variable has been reported to predict survival and responses to chemotherapy  . In this study, however, we did not observe a significant association of either type with a long OS. We noted that the histological diagnoses in our cases were generally made from biopsy rather than surgical samples. Therefore, the diagnoses might not have reflected the true nature of the disease. In contrast to histology, age is not generally considered a prognostic factor in patients with advanced gastric cancer, and several studies reported that palliative chemotherapy is equally tolerable and effective in older and younger patients  . Consistent with those observations, we did not observe an association of age with a long OS.
QOL has become increasingly important as the number of newly diagnosed patients with cancer continues to increase. Over time, improvements in the management of certain chemotherapy-associated toxicities have led to a shift from issues of physical QOL to issues of psychosocial QOL . The most widely used measures of cancer-specific health-related QOL are the European Organization for Research and Treatment Quality of Life Questionnaire, version 3.0 (EORTC-QLQ-C30), McGill QOL questionnaire, and the EuroQol-5D (EQ-5D) . These questionnaires have been used to investigate the trajectory of QOL among advanced-stage cancer patients in several studies  . However, the physical and/or mental condition of the patient may make it difficult to administer a questionnaire, particularly if the patient has a poor ECOG PS and has reached the end-of-life stage. Moreover, many studies of the changes in QOL experienced by patients with incurable cancer patients are limited to specific clinical course phases, such as limited cycles of chemotherapy or phase III chemotherapy trials  .
Hospitalization has negative effects on the QOL. However, this option is needed in many situations. Planned hospitalization is necessary for the administration of palliative chemotherapy, radiation therapy, and surgery. In contrast, unplanned hospitalization is generally needed to treat chemotherapy-related toxicities and cancer-related symptoms and is especially undesirable for patients  . Patients with end-stage disease exhibit a significant loss of QOL during hospitalization . Notably, one previous study suggested the importance of an optimal discharge-planning system and early referral to palliative care to prevent hospital readmission .
The Japanese healthcare system features a unique combination of characteristics that have led to the overuse of tests and drugs, as well as relatively longer hospital stays than those in other countries . One study reported that 27% of older patients in Japan experienced ≥90 days of hospitalization during their last year of life . For patients with incurable malignancies, decision making regarding treatment is complex; therefore, oncologists need to assist the patients and their families.  In decision making, not only OS but the trajectory of QOL throughout the clinical course might be helpful.
In this study, a long OS was shown to correlate strongly with hospitalization. However, the trajectory of hospitalization tended to accumulate rapidly due to unplanned hospitalization during the last phase of the clinical course.
The major strength of this study was the classification of patients with incurable gastric cancer into three groups by OS/hospitalization as well as our analysis of various characteristics of these groups, including planned and unplanned hospitalization.
This study has several limitations. We used planned/unplanned hospitalization as a predictor of QOL; however, other factors are also associated with psychosocial QOL. In this study, predicting the length of OS/hospitalization was challenging. In the future, predicting OS and the trajectory of QOL throughout the clinical course is essential.