Statistics of the Clinical Trials, page-20

The two lost patients? From the STIG poster in November 2014 – the data in the overall survival in second remission Cvac group doesn’t add up. 10 patients started. 1 died. At 24 months only 7 are at risk. Two were lost. This is discussed in one of the investor call transcripts (SEC filing ages ago).

Now the company will claim it is censoring these observations as per protocol. Recording them as neither alive or dead. But when you are lost to follow up in an oncology it generally isn’t good - in other words it is informative of outcome.

So properly what you should do is say heres the estimate worst case / best case. When the sample is large and the censoring is small in both groups it won’t matter. When the sample is small and both cases are in the treatment group it can make all the difference in the world.

Bilbo ... I’m not sure what that all of that meant but a hazard ratio is not a correlation. So your formula is the wrong one if you are trying to covert from a CI to a test statistic. But I'm not sure quite why you would be wanting to do that in the first place?