Rivulet Media Inc (fka RIVU)

[CPAP]

I recently completed two semesters of graduate bio-stats at a community health program in CT. It seems there's some truth in what everyone is saying...

When working with clinical trials and more specifically, randomized clinical trials (RCT's), we generally like to see sample sizes of n>30, because if we are dealing with interval data it can then be normalized and it becomes easier to run statistical analyses on. Additionally, this increases the power (or strength) of the findings. Another way studies gain statistical power is by blinding (single, double and triple - ie patient, physician and statistician, respectively).

Regardless of the test, we generally want to see alpha at either .05 or .01 (in rare cases, 0.10 is also acceptable, such as in AIDS testing), which means we correctly identify an observed group difference in 95 or 99 times out of 100 and the other 5 and 1 are observed merely "by chance," which is why sample size becomes so important.

That said, vital statistics indicate that the incidence and prevalence rates for aplastic anemia are incredibly low (<1 per 100,000 http://www.cancer.org/cancer/aplasticanemia/detailedguide/aplastic-anemia-key-statistics ), meaning it's challenging for researchers to assemble a large enough sample size with enough power to garner the weight needed to be recognized as being reliable.

Additionally, aplastic anemia has a high case-fatality rate with a recent publication from 2008 (Montane et al) indicating that a 2-year survival rate is less than 60%, making it difficult for companies to test their treatment because slightly more than half of the participants will live long enough to complete the study. Sadly, it would be poor science to give the treatment to all, because then you would lack a control group to compare your differences to (with the control being the currently accepted treatment for the disease). Simply put, we already have a very rare, very lethal disease...we then need to assemble representative samples with controls...all of whom have high case fatality rates. The sheer numbers make it challenging for the researchers to prove their findings. Not knocking these guys at all, just saying from a scientific/clinical research perspective, it's hard to prove their treatment works.

That said, I hope it (treatment) does work. I'm a blood banker and have only witnessed this disease once in three years and it ravaged a 12-year old girl that was incredibly active until she succumb to the disease last year.

Did these guys release their raw data or is it all proprietary at this point? I'd like to see some of their numbers and run them through SPSS and see what it looks like.

Thanks all, for the insightful input! BMSN