In addition, such parameters also include hazard rate, mortality rate (such as 5-year mortality rate) and survival rate. These parameters can be converted to each other, such as the experimental group and the control group under the assumption that the survival data meet the exponential distribution.
The estimation of parameters in sample size calculation is mainly the estimation of intervention effect. The commonly used clinical intervention effect parameters are mOS and T-year mortality (or survival rate), because these two groups of parameters are the most intuitive and easy to understand.
With two groups of mOS or T-year mortality (survival rate), all other parameters, such as HR, can be calculated.
Extended data
The mOS of the experimental group is relatively difficult to estimate. Because since it is an experimental group, there are usually few studies on this treatment scheme, and there are even fewer similar studies that can be directly referenced. If not, you can usually use the following indirect methods:
1) Refer to the previous test data. In the process of drug research, there are generally data of small-scale trials in the early stage, which can be used for reference. If the initial test result is OS, you can refer to it directly. If it is ORR, DCR, PFS and other data, we can compare the data of these indicators in the preliminary trials of other drugs. If the performance is better, then there is reason to believe that OS will be better than drug OS.
It should be noted that the sample size of the pre-test is generally small, and the correlation between ORR, DCR, PFS and OS is not necessarily strong, so the results of the pre-test cannot be fully believed.
2) Reference non-controlled studies. In clinical research, clinicians often summarize and publish the treatment of patients who have received this kind of treatment in our department. These studies generally have no control group, the sample size is not large enough, and even many of them are retrospective studies. For this kind of research, to find and compare the results as much as possible, we can use the method of meta-analysis to make a meta-analysis of the combined values for reference.
3) Refer to the effect of other intervention measures. The intervention mechanism to be studied can be compared with other intervention mechanisms, and then inferred from the large-scale experimental results of other interventions. For example, a new anticancer drug that has been marketed abroad has a patient mOS of 1 1 month, but the intervention measures to be evaluated are unlikely to be superior to the new drug in mechanism, so it is relatively reasonable to estimate that its mOS is less than 1 1 month.
4) It was pushed back by HR. Sometimes researchers will find that some studies are only for HR reference, so the experimental group mOS can be deduced according to the above formula.
5) The confidence and available resources of researchers. All the estimates can only give a rough range, and it is impossible to get accurate results. However, with the increase of reference channels, the range of estimation will gradually narrow. The determination of the final estimate is inevitably influenced by subjective speculation (based on clinical experience) and available resources, which is the risk of clinical trials and cannot be avoided.