This Is MS Multiple Sclerosis Community: Knowledge & Support

I confess, after my final exam I promptly forgot all my statistical analysis. Will someone give us the statistics on the figures from the Buffalo study? I am sure they are found somewhere in one or more of the press releases, but I was having a hard finding links to the study. Also, moderators, would you start a sticky of test results and press releases, all in one place?

What is the statistical significance on 55% CDMS AND CIS vs. 24% placebo with 500 patients divided 50/50, and the same for 80% CDMS vs 24% placebo with 250 placebo and however many CDMS patients were tested.

For comparison, Betaserone vs placebo was 0.00001 significant for annual relapse rate, 0.094 for relapse free, and not significant or progression. Even with all its side effects, with these statistics it became standard of care.

Yes...I remember telling my husband the day I started reading into Zamboni's research how even a small difference can be statistically significant and meaningful, and that what Zamboni was finding was huge. Now with the Buffalo results as released in the press release the spread was not as huge but still extremely significant and meaningful.

I think the Buffalo replication establishes this as real science and gives us all a better platfom from which to advocate for ourselves as patients.

_________________
"However, the truth in science ultimately emerges, although sometimes it takes a very long time," Arthur Silverstein, Autoimmunity: A History of the Early Struggle for Recognition

I used to work as an actuarial analyst for an insurance company.

I am very aware of the fact that with percentages, one can make them say almost anything one wants. The Buffalo study could have just as easily mentioned the fact that over 41% of the total group has CCSVI. This would downplay the relevance of CCSVI as MS.

Until the actual report comes out in April and the participants can study the numbers for themselves, the new release says that there is a correlation and I as a former analyst have to say "Okay so there is a correlation, but how much of a correlation?" What about CIS volunteers, CDMS volunteers, those in the questionable group, and the normal volunteers?
Right now all we can do is speculate at what the report is going to show us or work on our meditative skills and wait for the report.
I for one am going to try on work on my patience. I have been waiting more than 20 years for research to show what MS really is. I guess I can wait a few more months.

_________________
Dx 1988.
RRMS with permanent cognitive impairment
Permanent disability as of 4-2007

This was a serious question, not a challange to the data. If there is a correlation, it is statistically significant to some level of convidence. Depending on what factors you look at, you can make the data say dilfferent things, but for any given correlation wilth any given number of subjects there is one concrete level of confidence.

Here the question being tested was whether people with MS had a higher rate of CCSVI than those without. There are two groups, CDMS and
CIS. I was hoping this group had training in statistics and could give me that number.

I don't believe it was listed in the press release. I understand what you're asking. With it not in the press release, we won't get it until full results in April.

_________________
"However, the truth in science ultimately emerges, although sometimes it takes a very long time," Arthur Silverstein, Autoimmunity: A History of the Early Struggle for Recognition

I did not mean to ignore your post. I read somewhere in an article someone said the correlation was about 2 to 1 having CCSVI to MS, and another place I thought I read about a 4 to 1 corrrelation.

However. I believe the question you are asking given the data we know from the article is "What is the chance of having MS given you have CCSVI? " Given the limited data I know of I just did a quick calculation and arrived at this statement. If you have CCSVI, the probability of
having MS is 76.3%. So yes there is a correlation. I will look at my stat books and arrive at the variation of this data. (To think I used to do these types of calculations in my sleep and I loved it.) Oh how I love cog fog. I believe the variation of this data is 0.0586. (I looked at my stat books after my kids went to bed). I am still looking at my stat books it has been over 3 years.

As we start asking the question of having CIS and CCSVI, the percentage would get even higher and so on.

Once the actual data is available, hopefully your questions about CCSVI and MS will be answered or the data will be available to more fully answer your questions.

This study reminds me of the work I used to do and how with every answer, you ask another question that you requires more data (or more money). This is also why they want to keep this going.

_________________
Dx 1988.
RRMS with permanent cognitive impairment
Permanent disability as of 4-2007

From a brief look at the data. I would be a thrilled researcher looking at this data trying to find the flaws in the data or the questions that I am asking.
I am a cogntively disabled actuarial student that looked at numbers all day for almost 20 years and I always hestitated when the numbers looked to good to be true.
April is not that long to wait for the pros to look at this data and parsel out what looks wrong and what looks right.

CCSVI and MS are correlated but what does that mean?

Let's not jump to conclusions based off of a few incomplete studies that will give us the results in a few months.

Patience is a virtue. I plan to work on my virtue.

_________________
Dx 1988.
RRMS with permanent cognitive impairment
Permanent disability as of 4-2007

Msgigisji wrote: However. I believe the question you are asking given the data we know from the article is "What is the chance of having MS given you have CCSVI? "

Actually not. I am trying to find out what level of confidence is there in the difference in frequency actually observed. My understanding of this is that is a mathmatical expression of how often the difference observed would occur by chance given the number of observations made.

In other words, would this difference occur merely by chance? Again, my understanding of this is a bit rusty, but where the level of confidence is n=.05 which is often the lower limit, it would occur by chance 5 times in 100. If n=0.001 it would occur by chance only 1 in 1000..

That may not be quite right, but the "n" number is what I was wondering aboutl [/quote]

yes...the .001 or .0000001 or whatever it is, for statistical significance...no one knows it yet because it was not in the press release...

_________________
"However, the truth in science ultimately emerges, although sometimes it takes a very long time," Arthur Silverstein, Autoimmunity: A History of the Early Struggle for Recognition

The books I have do not have the necessary information I would need to answer that confidence interval question, but I will check will some ex-collegues of mine. I should be able to come up with an answer sometime next week. The sample size of 500 is still somewhat small to get a really good confidence interval.

_________________
Dx 1988.
RRMS with permanent cognitive impairment
Permanent disability as of 4-2007

Don't you have to know the exact n value to do that analysis? I thought how many were in which group was difficult to tell based on the various media reports, including the Buffalo press release.

Fogdweller:

From your posts, it sounds like you are really knowledgeable in the area of statistics in clinical trials. Unfortunately, I'm not sure the question you asked will be answered by the study.

You point out correctly that the percentages measured by the study group, and the statistical significance are two different things. My guess is the BNAC would not be releasing these numbers unless the p-value is less than .05 ( and they will probably give the p-value in their final study results). Of course they will probably give no specifics on how this was calculated, or on confidence interval limits, which are sometimes more meaningful.

I went through this after Zamboni released his first treatment results at last year's CX symposium. I ran the numbers every way I could - since he didn't provide specifics on the study population, I tried various combinations for the 2x2 table and logarithmic regression. No matter what I tried, I could not arrive at the same confidence interval limits.

Hello, I've been reading the CCSVI posts and following the CCSVI research since I learned of it about 4 months ago, but this is my first post.

I wrote to the author of a media article about the BNAC study that included a statistic that was not in the recent BNAC press release.

The article said that in the BNAC study, the rate of CCSVI found in people with RRMS was 80% while the rate of CCSVI found in people with CIS was 38%.

This seemed very significant to me!

I wrote asking the author whether this statistic is accurate and what was her source for the information?

She wrote back as follows:



I read her response as a "yes, the statistic is accurate".

Here is a link to the article that includes the statistic:

http://www.google.com/hostednews/canadi ... lWEAgfVkvw

and the quote from it:

<<The researchers also found fewer cases of CCSVI in patients who had experienced a single MS attack, called clinically isolated syndrome, compared to those with more advanced symptoms of the disease -38 per cent versus about 80 per cent.>>

I'm not a statistician but it seems like those two percentages might help with your analysis.

Mary Ann

DX 6-09 RRMS

Thank you Mary Ann. It is nice to hear a new voice in this forum who is obviously checking the sources !! We can speculate all we want but maybe our new goal will be April before details are reported.

ozarkcanoer

I did not speculate on anything. The original question, as I understand it, was whether the findings of the study were statistically significant. "Statistical significance" has a specific meaning in regards to experimental data. Just knowing the 80% CDMS number, 38% CIS number, or whatever says nothing about their statistical significance.