Studies demonstrating HPV vaccine is both safe and effective - Page 9

[kathymuggle]

Originally Posted by Chicharronita 

Yes but slightly higher than the saline group. I think we need more placebos in our life. 

You have to look at the sample size…..

Gardasil was 15000 and saline was 500.  

Did anyone see how long they measured death rates for?  I looked but missed it.  It might allow us to compare Gardasil to the general population, in the same age bracket.  

[Chicharronita]

Originally Posted by kathymuggle 

You have to look at the sample size…..

Gardasil was 15000 and saline was 500.  

I know. I was just making a joke.

[kathymuggle]

Originally Posted by Rrrrrachel 

You should do some reading about the idea of "statistical significance"

I know you are not talking to me, but I did!

I was reading about p value because twice this week I have run across studies where there have been slight differences in the test cases versus the control cases.  The difference are often dismissed due to "p value" - which is commonly around 0.05  (5%).  So if the difference in figures is less than 5%, the change is not noteworthy or could be chalked up to coincidence?  

I find that a little troubling - how to parse out small difference in stats wrt p value?

For example - a 4% difference when it comes to something as serious as asthma is huge!

Am I missing something?  

Edited by kathymuggle - 11/20/12 at 12:10pm

[kathymuggle]

Originally Posted by Chicharronita 
 

Glad to hear about the joke.  I can be a little literal about things 

You are right though - life could use more placebos!

[Dakotacakes]

Originally Posted by kathymuggle 

I know you are not talking to me, but I did!

I was reading about p value because twice this week I have run across studies where there have been slight differences in the test cases versus the control cases.  The difference are often dismissed due to "p value" - which is commonly around 0.05  (5%).  So if the difference in figures is less than 5%, the change is not noteworthy or could be chalked up to coincidence?  

I find that a little troubling - how to parse out small difference in stats wrt p value?

For example - a 4% difference when it comes to something as serious as asthma is huge!

Am I missing something?  

You can still find differences of under 5% that is the strength of the correlation or the effect size. The P value ishows the confidence that you are that the results are not due to statistical error. So you can have two groups showing only a .5%, or a .0001% difference between the two group and have a p value of .000 which would mean that you can be confident that there is a .5% or .00001% difference between the two groups. Simply put, you can be confident . However if the p value is higher than .025 ( or sometimes researchers will go up to .05 depending on sample sizes and methodology) you cannot be confident in the results and thus there is no difference between the two group. I have also seen very large differences between groups but the p value is also very large, and thus there is no difference (because there is no statistical significanc) HTH.

[Rrrrrachel]

It doesn't really have anything to do wih the absolute difference. You could have values hat differ by 50% and still have a p value less than .01.

Pardon the lengthy explanation.

Studies work using samples. We draw a sample and find some information about it and use that to draw a conclusion about a larger group (called the population). For example, if I'm trying to determine the average age of all posters at mdc, rather than finding all their ages I might select ten users randomly and use the average of their ages as an estimate for the age of all posters. Obviously my estimate would vary based on exactly which ten posters I chose, and if I draw samples and average ages over and over I would get a lot of different estimates.

Because of some mathemacal principles I won't go into, though, those estimates will behave in a predictable way. Most of them will be pretty close to the TRUE average of all posters ages, and a few of them will be more remote. They behave according to a mathematical pattern called the normal distribution.

Because they're so predictable, I can compare averages from two samples, say ten random posters from mdc and ten random posters from baby center. What the p value tells me is what's the probability, IF the two population averages are really the same, that my estimates are this different just by chance? You can also think of it as the probability the two values differ that much by random chance alone.

So if p value is low (and what exactly is low depends on a lot, typically .05 is the highest cutoff, .01 is another common one) that means its vey unlikely that the difference is due to random chance. If it was .05, for example, that would mean if the population averages were really the same I would only get estimates this remote 5% of the time. If he p value is low, we conclude its highly unlikely the difference is due to random chance and that the population averages must really be different, or that the difference is "statistically significant."

Hth

[Chicharronita]

Originally Posted by kathymuggle 

You are right though - life could use more placebos!

Especially the ones with sugar! :-)

[kathymuggle]

That does help.Thanks.

I am still trying to wrap my brain around a few things. 

So...it sounds like p value is the confidence they have that the results are correct - and that a low p value is fairly confident, while a high one means they are not confident.

I wonder what values factor into setting the p value.....(neither of you need to answere (obviously:) it might be beyond the scope of this conversation)

Would either of you say that studies where the rates are similar between a test and a control group have a high chance of their results being found  "statistically insignificant" due to p value ...or is this trend I am noticing likely coincidence (I can only cite two studies where I have seen this, after all)

How often does p value tend to render statistics "statistically insignificant?"  (no stat necessary - just your observations are fine).  Is it often?

Edited by kathymuggle - 11/20/12 at 5:09pm

[Rrrrrachel]

I'm not sure I understand all of your questions. As a statistician, I would not call a pvalue a confidence level, but a scientist probably would . It is, most simply and accurately, the probability the observed difference is from chance alone.

To oversimplify, p value depends on two things: size of the sample and the amount of variation in the population. Imagine trying to find the average age of a particular town. If you used a ten person sample your sample estimates would vary widely. It would be difficult to really be confident your results varied due to more than random chance. If you pick bigger and bigger samples your individual results would vary less and less.

Imagine trying to estimate the average age of all fifth graders in the us, though. Because the variation is so low, you could get a pretty reliable estimate with just a ten student sample.

Setting the cut off p value for significance (below .05, below .01, etc) is a matter of judgement. It is one way results of a study can be manipulated.

There's no real answer to how often a difference is labeled statistically insignificant. It depends. If you're comparing two things that are really the same, the difference should be found statistically insignificant 1-p% of the time and statistically significant (even though its ny not) p% of the time. This is one problem with studies that run lots and lots of tests. You would expect a certain percentage of the tests I show a significant difference, even though there really isn't one.

If thy really are different, the math is much more complicated.l and the exact numbers depen on a lot of factors. The difference would be found statistically significant the vast majority of the time, though.

[Rrrrrachel]

Whether or not similar results are likely to be found statistically significant depends.  Very small differences can be statistically significant if the populations have low variation.  Large differences can be insignificant, if overall populations are highly variable. 

[erigeron]

To me a p-value tells me how useful the results are. A p-value measures the likelihood that the statistical difference between two groups is due to random chance. So if p is, say, 0.5, that means there is a 50% chance that the differences in the results we are seeing are due to random chance. Say I am running a clinical trial to test a new cholesterol drug, and I'm testing it against a drug that is currently on the market. Say cholesterol levels went down an average of 40 points in the group getting the old drug and 45 points in the group getting the new drug. That sounds like a point in the new drug's favor, but if the p value is more than 0.05 (which is the level that's usually used in my field), that means there may really be no difference between the drugs. A p value of <0.05 is another way of saying that there is a 95% chance the differences between groups are NOT due to random chance, or a 5% or less chance that the differences ARE due to random chance (because 0.05 is the same as 5%). 

So if I read a study and some difference is reported, I look at the p-value, and if it's greater than 0.05 I know that reported difference between groups could just be due to chance (i.e. there isn't actually a difference, it just looks like it), so I don't get too worked up about it. 

[HappyHappyMommy]

Hi All, There's been some reports about posts in this thread that addressed a member posting, not a topic. Please edit any posts you've made that address another member and revise them so that they stick to the topic. Thank you!
 

[Chicharronita]

Originally Posted by chickabiddy 

Here's some actual data to read: http://www.fda.gov/downloads/biologicsbloodvaccines/vaccines/approvedproducts/ucm111263.pdf

Under Warnings and Precautions it says:

Quote:

5.1 Syncope 

Because vaccinees may develop syncope, sometimes resulting in falling with injury, observation for 15 minutes after administration is recommended. Syncope, sometimes associated with tonic-clonic [jerking] movements and other seizure-like activity, has  been reported following vaccination with GARDASIL. 

 

When syncope is associated with tonic-clonic movements, the activity is usually transient [...]

So what do they do if the "activity" isn't transient?  

[Rrrrrachel]

There are some serious adverse events with vaccines. Sometimes they're permanent. These are extremely rare, and far less common than serious adverse events that occur as a result of actually contracting a vpd. No one claims vaccination is risk free.

[Chicharronita]

I know, but what can doctors do for you if the problem becomes permanent?

[Taximom5]

Originally Posted by Rrrrrachel 

There are some serious adverse events with vaccines. Sometimes they're permanent. These are extremely rare, and far less common than serious adverse events that occur as a result of actually contracting a vpd. 

We don't know this, because of the flaws and limitations in the reporting systems, and because doctors are not trained to recognize any serious adverse reactions except for immediate ones, like allergic reaction.

If you have a systemic or autoimmune reaction that takes more than a few hours to appear, or if you have a seizure a week later, most doctors will not realize that the vaccine was the trigger.  If they don't realize it, they don't report it.

I posted before how complications from a disease are far, far more likely to be recognized, understood--and reported--than complications from a vaccine.

Those of us who have suffered such reactions, or whose children have suffered them, are getting very frustrated with those who keep insisting that such reactions are extremely rare, because we've had no trouble finding people--THOUSANDS of people--who have had similar reactions, and who, like us, were told that such reactions hardly ever happen.

Funny how those complications from VPDs have been misreported in a way to make you fear the flu (OMG, 36 million people have copmlications from the flu!), while safety/efficacy of the flu shot has also been misreported in a way to keep you from learning that the flu shot is about as effective a s a placebo and at the same time, has a high rate of reported adverse reactions...

Obviously, if someone chooses not to look, they won't see.

[Rrrrrachel]

Vaers is not our only method of tracking the rate of adverse events. There are other monitoring systems that don't rely on reporting by doctors, as well as experimental evidence.

[Rrrrrachel]

Vaers is not our only method of tracking the rate of adverse events. There are other monitoring systems that don't rely on reporting by doctors, as well as experimental evidence.

[Chicharronita]

Originally Posted by Taximom5 

Those of us who have suffered such reactions, or whose children have suffered them, are getting very frustrated with those who keep insisting that such reactions are extremely rare, because we've had no trouble finding people--THOUSANDS of people--who have had similar reactions, and who, like us, were told that such reactions hardly ever happen.

Especially now thanks to the Internet. I know a few in my city because of a local online parenting group I belong to.

And of course Facebook: vaccine awareness groups are popping up all the time, and lots of people are posting their experiences.

The  "My child's vaccine reaction" page was just created on the 16th, and already has over 2,600 likes and numerous personal stories posted. Here's the description:

Quote:

I invite parents whose child has suffered a vaccine-related reaction, injury, disability or death to post their story with a photo or photos on this Facebook page to help inform other parents about the risks of vaccinating. Thank you so much for sharing your and your child's story!

[prosciencemum]

Originally Posted by Chicharronita 

Especially now thanks to the Internet. 

The internet is very good at connecting what feel like large groups, but in reality are very rare. 

I don't know anyone personally who has had a vaccine reaction, except soreness at the vaccine site and mild fever following the injection (which demonstrates that the immune system is working as intended in response to the vaccine).