Statistics for everyone: ch.31 two survival curves

easier R than SPSS with Rcmdr : Contents

ch.31 two survival curves

Calculating the samples size in survival analysis has a lot to consider. Let’s run it ourselves and think about it.

The ‘accrual duration’ is the time from the treatment of the first patient to the time of treatment of the last patient. The time period from the time of treatment of the last patient to the end of the study is a ‘follow-up durtion’. The time it takes for the last study to be finalized, i.e. the last time the data is collected, is the ‘total duration’.

The third entry, ‘survival ratio at n year in each group’, may be a bit misleading. The results show that this is the time at which the survival rate was measured (Follow-up duration for P1, P2). In other words, when we followed it for 2 years, it was estimated that 0.3 and 0.4 were survival.

If you graph this, it would ideally look like this. Suppose you survive for about 0.3 and 0.4 for two years. This curve is an exponential curve.

Note that compared to the left, the right side only had a longer total follow-up time, which resulted in a slight reduction in the number of samples. In other words, if you hold the time longer, the number of samples will decrease because more events will occur; that is, if you increase it from 5 to 8 years, the number of samples will decrease.

If I think 8 years is too long, so let’s try to finish the study for 5 years. During the 4 years of the Accrual duration(= Enrollment duration), 284 people must be drafted into a group. Therefore, you should recruit about 6 subjects in one group and 12 subjects in two groups per month. If you find it difficult to collect this number as usual, you should not start this study. You will need to collaborate with other centers. If the original multi-center study was expected to be able to recruit as many as 20 subjects a month, the duration of the study could be further reduced. I think we can recruit 240 subjects in 1 year, 480 in 2 years, and 600 subjects in 3 years.

Still, when it comes to recruiting for 1 year, it’s a staggering number.

In survival analysis, the formula is a little more complex than the others.

Since there are 608 subjects required for a 3-year period, a minimum of 3-year drafting period is required. Subjects who are tracked and lost also need a little more leeway, taking into account. In order to make the overall preparation in this way, the calculation of the number of samples is a necessary process.

In particular, the calculation of sample size in the survival data is complex, and different settings can be made depending on the number of subjects that can be collected each month, according to the enrollment duration, or the tracking period, so even the same formula can lead to multiple application formulas.

easier R than SPSS with Rcmdr : Contents

=================================================