Consider an example from Machin and Campbell (2005) of a study comparing two forms of surgical
resection for patients with gastric cancer. From a prestudy survey, the baseline 5-year survival rate was expected to be 20% and an anticipated increase in survival in the experimental group expressed as a hazard ratio of 0.6667 (corresponding to a 5-year survival rate of approximately 34%) was desired to be detected with 90% power using a two-sided, 0.05 level, log-rank test under 1:1 randomization. To obtain the estimate of the sample size for this study, we use stpower logrank with survival proportion in the control group 0.2 supplied as an argument, the hratio(0.6667) option to request a hazard ratio of 0.6667, and the power(0.9) option to request 90% power.
Consider an example from Machin and Campbell (2005) of a study comparing two forms of surgicalresection for patients with gastric cancer. From a prestudy survey, the baseline 5-year survival rate was expected to be 20% and an anticipated increase in survival in the experimental group expressed as a hazard ratio of 0.6667 (corresponding to a 5-year survival rate of approximately 34%) was desired to be detected with 90% power using a two-sided, 0.05 level, log-rank test under 1:1 randomization. To obtain the estimate of the sample size for this study, we use stpower logrank with survival proportion in the control group 0.2 supplied as an argument, the hratio(0.6667) option to request a hazard ratio of 0.6667, and the power(0.9) option to request 90% power.
การแปล กรุณารอสักครู่..