Interesting and will take a look. I am also interested in the point about the Japanese and their outlook on the three disorders you mentioned. Recent Supreme Court of Canada case (Keays v. Honda) in last five years that dealt with Honda in Alliston Ontario and its termination of employment of Mr. Keays who alleged he had chronic fatigue syndrome. The trial judge found that Honda basically didn't believe the disorder existed and he awarded huge punitive damages and bad faith damages (for a Canadian court they were huge). Overturned by the SCC but did give a peak into the beliefs you mentioned.
You're welcome. You might want to check out chtp. They are pursuing fda approval of a low bp drug used in japan for 15 yrs. Plenty of safety history coupled with robust ph 3 data under an spa. Also, interim results from ph 2 study of a drug that beats holy grail of arthritis drugs, and without the nasty side effects (of methotrexate).
Beyond that, the low bp drug appears to work on 3 disorders that japan hasnt bothered pursuing because they think ADHD, fibromyalgia, and chronic fatigue syndrom are made up disorders. Fibro is in a ph 2 due to be reported next qtr. Small adhd study recently reported 47% improvement in symptoms. Cfs study is also small ph 2 and due next qtr.
It's worth a peak.
Thanks. Now that you've pointed it out, it makes sense that the number of events, not the number of patients enrolled, would determine a trial's statistical power. I see that my assumptions about how they design a trial were also wrong, and I appreciate being enlightened.
As I recall, I enjoyed my one stats course and am beginning to wish I'd taken a few more.
While designing the trial, using a power analysis, the co came up with 407 patients. Then the co said, let's finish this trial in 3 years. So to have 407 events in 3 years, the co needed to enroll 800 patients. They could have enrolled 600 and finish in 4 years with the same/similar statistical power. So for our purposes, 407 is the number we care about.
Also, in this case, since OGX-011 showed an early effect (see the P2 KM graph) and due to that HR decreased with time, it's better to finish the trial early by enrolling more rather than waiting more.
That's why I'm also hopeful about the interim.
At interim, right.
The other general point I should have mentioned for the benefit of statistics neophytes like myself is that the greater the actual treatment effect -- i.e., the effect in the general population -- the lower the likelihood of a sampling error. So if 011 actually increases median survival by 5 months compared to placebo, the chance that you'll find a statistically significant benefit in the 407-patient sample is greater than if it actually increases median survival by only 2 months.
At least I think that's right.
Thanks Summer, I think I get it. I remember this much from my statistics course: no matter what something (in this a case a drug) does in the population as a whole, there is always a chance that it won't be reflected in a sample of the population. Of course, the larger the sample (in this case, 800 patients, which is a pretty good size) the smaller the chance that that will happen.