Advanced detection of recent changing trends in gastric cancer survival: up-to-date comparison by period analysis

OBJECTIVE

To establish a comprehensive cancer treatment and prevention policy, data collection should be performed in a timely manner, and survival analysis needs to reflect changes in treatment strategy. Therefore, we introduced the concept of period analysis for gastric cancer, the most prevalent cancer in Korea. We estimated 5- and 10-year survival trend of gastric cancer, based on data from the Yonsei Cancer Center Tumor Registry between 1990 and 2004.

METHODS

We compared the differences in survival between cohort, complete and period analyses for two different periods, 1995-99 and 2000-04.

RESULTS

A total of 11 724 cases were included. The median age of cancer diagnosis gradually increased over time, and more patients were diagnosed with Stage I disease in recent years. In the basic comparison of three estimated analytic methods (cohort, complete and period), period analysis (45.8%) was most similar to the actual 5-year observed survival rate (48.5%), when compared with cohort (43.6%) and complete (44.8%) analyses. When we compared survival between different 10-year periods (1990-99 and 1995-2004), period analysis demonstrated a greater difference than complete analysis (9.0 versus 3.9%). Subgroup analysis indicated that the survival improvement was determined by period analysis, and it was more pronounced for the age group <74 years and in Stages III-IV patients.

CONCLUSIONS

We observed that period analysis demonstrates the most similar results to the actual observed survival and is, therefore, a useful method to derive precise cancer survival in gastric cancer. This information is useful to understand survival differences that are influenced by changing treatment strategy.

[Usefulness of comorbidity indices in operative gastric cancer cases]

OBJECTIVES

The purpose of the current study was to evaluate the usefulness of the following four comorbidity indices in gastric cancer patients who underwent surgery: Charlson Comorbidity Index (CCI), Cumulative Illness rating scale (CIRS), Index of Co-existent Disease (ICED), and Kaplan-Feinstein Scale (KFS).

METHODS

The study subjects were 614 adults who underwent surgery for gastric cancer at K hospital between 2005 and 2007. We examined the test-retest and inter-rater reliability of 4 comorbidity indices for 50 patients. Reliability was evaluated with Spearman rho coefficients for CCI and CIRS, while Kappa values were used for the ICED and KFS indices. Logistic regression was used to determine how these comorbidity indices affected unplanned readmission and death. Multiple regression was used for determining if the comorbidity indices affected length of stay and hospital costs.

RESULTS

The test-retest reliability of CCI and CIRS was substantial (Spearman rho=0.746 and 0.775, respectively), while for ICED and KFS was moderate (Kappa=0.476 and 0.504, respectively). The inter-rater reliability of the CCI, CIRS, and ICED was moderate (Spearman rho=0.580 and 0.668, and Kappa=0.433, respectively), but for KFS was fair (Kappa=0.383). According to the results from logistic regression, unplanned readmissions and deaths were not significantly different between the comorbidity index scores. But, according to the results from multiple linear regression, the CIRS group showed a significantly increased length of hospital stay (p<0.01). Additionally, CCI showed a significant association with increased hospital costs (p<0.01).

CONCLUSIONS

This study suggests that the CCI index may be useful in the estimation of comorbidities associated with hospital costs, while the CIRS index may be useful where estimatation of comorbiditie associated with the length of hospital stay are concerned.