Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[GavJ]

Find a reputable cite for the dangers of vaccines

The point of the thread is that this does not exist.

Not enough people have been physically run in trials to know.  Thus, your preferred alternative plan you're suggesting (which would be my preferred plan too!) stops right there.

Medical professionals don't know any better, either. Nothing about being a medical professional allows you to magically know information that doesn't exist.

There are no "mountains" of evidence. There's hardly any at all. A tiny fraction of what would mathematically be needed.  I can't cite any additional evidence for my position because my position is pointing out the lack of sufficient citable evidence...

There's no such thing as evidence of lack of evidence, other than merely the other guy not being able to produce it. (and I've been doing the relevant research for YEARS and have never found it nor had the other guy produce it. The necessary studies have simply not been done.)

[Neonivek]

In other words is an unknown intangible risk that can only be found through heavy inference beat out the measurable quantifiable benefits.

[Putnam]

Wait, since when has 2,000 not been enough? Everything I've ever learned about statistics tells me that proper sample sizes are not actually proportional to the total population in question.

[Skyrunner]

Yep. Generally, it takes more and more samples to increase confidence in results, but by 400 or so people you already have a very good estimate, off by only 5% or so.

Let us say of the 2000 sample size, around 1%, or 20 people had reactions.
Second calculator here . Input 2000 sample size and 7.2 billion population size. along with 95% confidence. That's a confidence interval of .44, or there's a 95% chance that .78~1.22% of people will have a reaction. .44 is a preeetty accurate interval.

Heck, even in the worst accuracy (50%) it's only an interval of 4. This is definitely more than enough, mathematically, to suggest that there are no adverse affects that are hidden now but would magically appear if we just test more people. One does not simply claim it's not enough without using statistics. :V

[chaoticag]

You know, I had to look up one thing for this thread. Turns out, if you want to know the cost risk benefits of vaccinating through the MMR vaccine, then you'd need to look at some place where at least measles are common place. Say, Italy. Italy reported around 5000 cases of measles, 95% of all cases were among those that improperly or didn't vaccinate at all. That comes to roughly a 19 times at risk compared to those who do vaccinate. Of those 5500 cases, around 1800, or 23% of cases led to hospitalizations and 20% developed complications, with 135 developing pneumonia, and an immunocompromised adult dying.

So to put it bluntly, you're 19 times more at risk of developing measles if you do not immunize, at least, with the data at hand, and there's a good chance you may end up in a hospital if that does occur. I do not believe GravJ's has taken this into account, since what I skimmed over seems to be related to the cost death analysis through deaths caused, but a hospitalization is a serious financial burden to a family or individual.

And something of a post script on here, it looks like the 80% of measles cases in Europe have been unvaccinated, so it's worth re-iterating that the data I looked over was due to Italy, which reported the rates of measles of those who are unvaccinated or improperly vaccinated, rather than just unvaccinated only. Anyway, the data I used.

http://ecdc.europa.eu/en/publications/publications/measles-rubella-monitoring-may-2013.pdf

[Neonivek]

I think GavJ has.

He is talking on a "Well America doesn't have issues with Measles thus immunization is less important"

[chaoticag]

Isn't something like that less important? Plenty of people head overseas for vacation, so you can't really confine your analysis to just America. Not to mention, these things happen in outbreaks, rather than over time, and so it's more location based than just saying america, a country with practically 50 separate countries in it.

[GavJ]

Wait, since when has 2,000 not been enough? Everything I've ever learned about statistics tells me that proper sample sizes are not actually proportional to the total population in question.

It doesn't need to be proportional to the POPULATION.  It does need to be proportional to the OTHER STATISTICS of the opposite risks.
Population could be anything. That's not why I'm saying you need millions of subjects. The U.S. could have 7 people in it or 70 trillion people, and you would still need the same millions of research participants.
The thing that is constraining it is not how many people live here -- what's constraining is the rarity of measles vaccine benefits at a given level of vaccination coverage.



There is no magical number of people that's just sufficient for anything. It totally depends what you're asking. Some areas of science literally use ONE participant (like low level retinal level response tasks where nearly everybody has identical biology). I've seen papers published with one participant.

Other things need 30 or 2,000 or 300,000,000 participants, depending on what you want to demonstrate, and how tiny the effects are.

Clinical trials for a heart surgery technique where you're 50% likely to die without it only need like 10 subjects to show sufficient safety.
Clinical trials for a seizure drug where you're 5% likely to die without it might need, I dunno, a couple hundred subjects.
Clinical trials for verifying the safety of a higher than 90% vaccination rate for a vaccine for a disease where you're 1/60,000,000 likely to die from it at 90% coverage, you need 300,000,000 subjects.

It is the size of the effects youre dealing with that determines necessary subject populations. Going from 90% vaccination to 100% has tiny tiny tiny even potential benefits, so you need to run HUGE numbers of additional subjects to verify that it's safe to move from 90% to 100% vaccination.

Going from 70% to 90% (as you'll see further down in this post) has larger potential benefits, so you still need to run more subjects to know if it's safe (than what we have now), but you would need to run fewer than you do to verify that 90->100% is safe.

He is talking on a "Well America doesn't have issues with Measles thus immunization is less important"

Sigh, no. No no no.

Immunization is hugely important, and like I've said many times now, it is almost guaranteed that some vaccination is necessary to save the optimum number of lives.

But "vaccination" is not an ALL or NOTHING thing.  There is some optimum rate, and it MIGHT NOT be 0% or 100%.  It might be somewhere in between. In fact, it almost definitely is in between.

And whatever that optimal level is, it is going to be the optimal level everywhere in the entire world that has similar healthcare facilities as us. I.e. All of Europe, etc. And hitting that optimal level will be equally important everywhere (including Italy). Any higher than the optimal level = unnecessary deaths. Any LOWER = unnecessary deaths.




Italy doesn't really tell us that much, but it is another datapoint. In Campania (one of the provinces), the measles vaccine coverage rate is about 70%.  They had about 5 million people at the time of the data I have available, and they had ~3-4 deaths from measles a year. This is roughly 20,000 cases of measles, and about a 1/5,000-7,000 death rate (by the way, looks like I was right about modern medicine making it much less deadly than before, that's 1/4 the death rate of the 1960s)

So 1/1,250,000 people a year die of measles, or at that rate, over a lifetime, 1/16,000.

In order to show that MORE THAN 70% vaccination would be good for them, you'd have to show that the vaccine has a less than a 1/16,000 death rate. Which we don't have enough data to know.

It is mathematically impossible with clinical trials that only run 2,000 subjects (or 5,000 or 10,000 or whatever we have total). Simple as that. To prove this rate of death or lower, you'd have to run about 75,000 subjects statistically.

However, notice that the amount of data needed to prove that (if it's true) is way the hell lower than the amount needed to prove that a 90% rate is helpful. This has nothing to do with Campania having a lower population. It has everything to do with the higher potential and known benefits of increased vaccination from a 70% level. Thus, vaccine safety doesn't have to be AS stingently tested to justify vaccinating more than 70%. Though still more stringently than what we have tested so far.

As you go lower and observe higher death rates, the ability to gather enough data to say for sure is easier. Give me a place in Europe or somewhere else with good healthcare facilities with a rate lower than 70% and we might learn more. Does anybody know of one?

If I can get more datapoints, I can tell you exactly how much vaccination rate our current level of data verifies as lower than optimal. My gut instinct from these two datapoints is maybe 50%, but can't verify without actual more examples.

[Sergarr]

An optimum rate of vaccination is not determined only by the amount of deaths.

You'd also need to consider that you need a rate of vaccination higher than some value that guarantees virus-extermination at some point at the future. Or else we're looking at unlimited amount of deaths.

Also, have you considered the fact that the very issue of low mortality of diseases only came to be because the viruses you are talking about have been almost exterminated? Before vaccination, they were very widespread.

[Neonivek]

The thing about vaccination is its social benefits are compounded not linear.

70% vaccination rate is not only 10% better thank 80% for example.

[scrdest]

Actually, no. How can it possibly be constant everywhere?

If you calculate the optimum as a risk/reward function, it will vary HARD as risk/reward would be based on things like risk of infection - which would depend on population density, local climate, mobility, so on, so forth. So, the global optimum is the highest value of all local optima, and using that would mean you're over-vaccinating some individual populations.

Hell, you *would* over-vaccinate no matter what, since the optimum would be determined statistically and undershooting would mean you're not actually resolving the problem once and for all, so you'd have to overshoot by, like, 5 sigmas.

Also, funny thing that you demand larger human clinical testing. These, you know, involve actually vaccinating people and seeing what happens. So, to determine the rate for vaccination lethality, you first have to ACTUALLY KILL SOME PEOPLE (if the side-effects are untreatable) or at least harm them.

[GavJ]

The thing about vaccination is its social benefits are compounded not linear

I never said they were linear.

70% vaccination rate is not only 10% better than 80% for example

No, 80% is DEFINITELY less than 10% better than 70%, because the risk per vaccine doesn't change, and yet the social benefit lowers with every additional % poin (i.e., herd immunity contributes less and less additional benefit with each new vaccinated person). Therefore, 80% vaccination rate is guaranteed <10% better than 70% vaccination rate. How much less I don't know.

population density, local climate, mobility, son on, so forth

Yes I agree. This is not what the person I was responding to was talking about. I was pointing out that population size and current rate of vaccination -- those two things do not affect optimal rate.

Other things do matter, but I don't have space for 4 paragraphs of disclaimers and complexities for every sentence. I was not intending to discount climate etc. I was just responding to the point at hand.

you'd have to overshoot by, like, 5 sigmas

Uhhh wat.

Random, irresponsible number is random.

The amount you should overshoot by (I agree on that) should only be the amount within which you expect to see variance in your population awa from whatever you recommend to them. So if you think 0-10% of your population might not do what you tell them, then you should suggest [optimal rate + 12%] or what have you.  Remember, we are trying to minimize vaccine complications, there's no reason to wantonly overshoot. Keep in mind also that you can always just adjust the target if the population is acting squirrely.

So, to determine the rate for vaccination lethality, you first have to ACTUALLY KILL SOME PEOPLE

No, you don't know that. Since it is UNKNOWN whether vaccines or the disease's effect on unvaccinated people at this level of vaccination = more deadly, it is also UNKNOWN whether the clinical trial will kill or save more people than currently.

You can be sure that people will die in your trial, but you have no way of knowing yet whether more will die with or without the trial being run.

Therefore, it's perfectly ethical.  It would only be unethical if you already knew but ran it anyway. If that were the case, though, you wouldn't need to run it, because you already knew...

You also need to consider that you need a rate of vaccination higher than some value that guarantees virus-extermination

Please read the thread. This has been asked and answered at least half a dozen times.

Also have you considered the fact that the very issue of low mortailty of diseases came to be because the viruses you are talking about have been almost exterminated?

Scratch the previous response. Please read the very first post first! Then read the rest of the thread.

[scrdest]

The five sigmas was a deliberate exaggeration.

If I got that right, according to you, testing vaccines, even if you know they are lethal and there is no immediate threat of infection, is ethical, because you do not know if testing would have a positive or negative effect on the death rate.

It seems to me that this position is incongruent with the belief that, in absence of exact mortality rates for vaccines, you shouldn't just go all-out. After all you don't know if it will kill more people than it would otherwise, making that policy perfectly ethical.

[GavJ]

testing vaccines, even if you know they are lethal and there is no immediate threat of infection

Everything is lethal. Household gerbils I'm sure have been lethal at some point.

Testing vaccines is ethical (and in fact it is the whole reason to run them!) when the RATE of lethality is unknown currently relative to the RATE of marginal protection from death. that's when you don't know whether the trials will hurt or harm in and of themselves.

Which is currently the case for both the U.S. and Italy, our two datapoints considered so far.

It seems to me that this position is incongruent with the belief that, in absence of exact mortality rates for vaccines, you shouldn't just go all-out. After all you don't know if it will kill more people than it would otherwise, making that policy perfectly ethical.

Ethical is not the same thing as "not stupid"

In the absence of any information either way, you should simply not have a policy (in other words, the government should publish no opinion above and beyond whatever rate of vaccination their current level of data proves is beneficial. Which appears to be something lower than 70% from the current two data points, unless Italy's geography implies way more or less disease, or something). Not dramatically, aggressively push for one of the two policies you don't know is better...

[Skyrunner]

GavJ, why did you ignore me? 1% death rate can be determined easily with only 2000 samples. Math!