Survival and disease-free curves including patients with incomplete follow-up: a practical consideration

抄録

We studied the influence of follow-up levels on survival and disease-free curves in 100patients with tongue carcinoma. According to the follow-up methods two levels were defined. The Level 1 data were defined as those obtained under condition that no systematic follow-up was performed. In this level the follow-up data were acquired by patient charts only. The Level 2 data were defined as those obtained after making meticulous follow-up. In this level the follow-up data were acquired by direct contact with the patients or family physicians, or by copies of death certificates, etc. As a result of follow-up, complete followup data were obtained for all but one patient, who was missed after a 15-year period of no evidence of disease. For aforementioned two levels' data, analyses of survival and diseasefree rates were performed with the Kaplan-Meier method. Consequently, we could get following results. First the more complete follow-up data were, the lower the survival rates. Second the disease-free rates were not changed with completing follow-up. Using the generalized Wilcoxon test and logrank test, their trend were proved statistically. This study indicates that data with incomplete follow-up may be used only for interim analysis, but not for survival analysis.