Vaxxed vs. unvaxxed study - Page 2

[Dar]

This study also didn't differentiate between kids who were fully vaxed according to the (German) schedule and kids who had received one inoculation against one disease... anyone who had been vaccinated at all was lumped together. That could also have made a difference - it might be interesting to exclude all kids who were partially vaxed and/or vaxed on a delayed schedule and then run the same statistical tests...

[Imakcerka]

Originally Posted by Bokonon 

I'm just identifying a flaw with this study, which you posted.  You have critiqued studies others have posted as well.

Are you just following her around now?  I'd honestly like to know more about this study and similar studies. 

[Bokonon]

Originally Posted by Imakcerka 

Are you just following her around now?  I'd honestly like to know more about this study and similar studies. 

No.  I read through the link and found serious flaws.

[Rrrrrachel]

Yup, the small sizes do mean big intervals and yes it is possible the sizes are just too small to detect an effect (that's called the power of a test, in general bigger sample = more power). These results are far from conclusive, just one more thing to weigh.

[AbbyGrant]

     *

Edited by AbbyGrant - 6/28/12 at 9:20pm

[crayfishgirl]

Originally Posted by Imakcerka 

Are you just following her around now? 

 I am...but not in a creepy way.      These threads have been really interesting and informative.

[prosciencemum]

It's an interesting study, because at the very least it puts an upper limit on the differences between vaccinated and unvaccinated children. That the unvaxed had more VPD was detectable because it was such a large effect. Other effects were not seen - so you can say from that "oh well they need a bigger sample" and that is true, but you can also take from that that there cannot be a huge difference between the two groups - because if there were it would have been found. 

Take Figure 3 for example - that looks at the rate of "atopic disorders" ("Allergic rhinoconjunctivitis", excema and asthama) between vaccinated and unvaccinated children. The biggest possible difference seems to be in the age range 11-17, where 1560 of the vaccinated children in the sample had these compared to 8 unvaccinated. Relative to the full sample size that was 30+/-1% of the vaccinated children, and 20+/-12% of the unvaccinated. If vaccinated children were twice as likely as unvaccinated to get these things - then you'd expect 40+/-1% of the vaccinated children to have them - and that would be detectable compared to 20% of unvaccinated. So that tells me that vaccinated children cannot be twice as likely to get these things. Do you see what I mean?  

I think that's a great illustration of what differences you can find in small samples. I work in the physical sciences, but this is similar maths to the kinds of studies I do there. :) 

[Rrrrrachel]

That's a good way to put it. It doesn't rule the possibility out, but it gives us a sense of at least how big (or small) the effect might be.

[prosciencemum]

Thanks. And thanks for posting the article. Very interesting. 

[Dar]

Originally Posted by prosciencemum 


Take Figure 3 for example - that looks at the rate of "atopic disorders" ("Allergic rhinoconjunctivitis
", excema and asthama) between vaccinated and unvaccinated children. The biggest possible difference seems to be in the age range 11-17, where 1560 of the vaccinated children in the sample had these compared to 8 unvaccinated. Relative to the full sample size that was 30+/-1% of the vaccinated children, and 20+/-12% of the unvaccinated. If vaccinated children were twice as likely as unvaccinated to get these things - then you'd expect 40+/-1% of the vaccinated children to have them - and that would be detectable compared to 20% of unvaccinated. So that tells me that vaccinated children cannot be twice as likely to get these things. Do you see what I mean?  

Sure they can be. It's not even that unlikely. The confidence intervals mean that 95% of the time - 19 times out of 20 - the actual, true incidence of those illnesses is somewhere between 29 and 31% of the vaxed kids and between 8 and 32% of the unvaxed kids. The real incidence, though, is only one of those numbers. So, possibly, the true incidence could be 10% for the unvaxed kids - with an n of 8, that might only mean that one extra kid got the disease (or his parents misremembered). That's not much of a stretch, and that would mean that unvaxed kids were only a third as likely to get the disease.

Or, because one in twenty times the actual number will be outside the confidence interval, maybe the actual incidence is 35% of unvaxed kids... or maybe parents who don't vax are less likely to honestly answer these questions, or to remember their children's disease history. It's important, I think, to look critically at all studies - lots of stuff out there, even in peer-reviewed journals, is just not well done - and to understand exactly what the numbers mean.

[prosciencemum]

OK maybe a factor of two is pushing it, but if there was a serious increase in the rate of these things in vaccinated children it's likely it would have been noticed. That was my point. 

It is true that 95% of the measurements will be in the confidence interval, but it's also true that the distribution is not flat, so it's more likely the true value is close to the estimate based on the sample measurement. For example its very unlikely that all three rates for unvaccinated kids were high while all three for vaccinated were low. And that's what would be needed to hide a big difference in the rates in the sense that proves vaccines are correlated to these diseases (which are all long term in this example, so I doubt parents "forgetting" could be a major issue).

I hope this will be repeated with a larger sample size. It is very important to keep looking for these effects.  

[Dar]

Originally Posted by prosciencemum 

OK maybe a factor of two is pushing it, but if there was a serious increase in the rate of these things in vaccinated children it's likely it would have been noticed. That was my point. 

It is true that 95% of the measurements will be in the confidence interval, but it's also true that the distribution is not flat, so it's more likely the true value is close to the estimate based on the sample measurement. For example its very unlikely that all three rates for unvaccinated kids were high while all three for vaccinated were low. And that's what would be needed to hide a big difference in the rates in the sense that proves vaccines are correlated to these diseases (which are all long term in this example, so I doubt parents "forgetting" could be a major issue).

I hope this will be repeated with a larger sample size. It is very important to keep looking for these effects.  

I think you have to quantify "likely", maybe.... which is what confidence intervals try to do. It would have been interesting to see the stats for all children at once, not subdivided into three groups, I think. As it is, when one child's (or one family's) answers could change the results by perhaps ten percentage points, I think likely is too strong a word.

The vaccine data was collected over the long-term, but the info about disease history was based on a retrospective, one-time survey sent to parents, so it's very subject to parental forgetfulness, or perhaps confabulation - if non-vaxing parents believe that not vaxing reduced their children's chance of getting certain diseases, then their memories will be more likely to support that. Or is there a bias in survey response - are non-vaxers with ill children less trusting of the government and therefore less likely to return surveys?

Also, it seems likely to me that many of the non-vaxed children would be from the same families - if parents don't vaccinate one child, why would they vaccinate subsequent ones? So they're not truly independent.

I think some qualitative research would have been helpful here, as well as a larger sample of unvaxed kids and a distinction between fully vaxed kids and partially vaxed kids. Research with people is really different from most physical science stuff - there are a lot more variables that can't be controlled, ethically, and it's hard to understand the whole picture without understanding the people.

[Rrrrrachel]

You mean not subdivided by age?  I totally disagree, I think that would completely muddle the issue.  However, I'll bet there's enough information in that study that you could backwards engineer those numbers.

[HouseofPeace]

there are a few running larger scale vax vs. nonvax studies being done.  there is a group out side of chicago which has treated non-vaxing people since the 70's, and they probably have the longest running data sample, but i can't remember the name of the group now.  i'll ask my friend...

that said, i'm contributing my data based on our medical records.  and have no hesitation to admit to anything.

i would also say that non-vaxxing parents come in a variety of groups, as do vaxxing, and some are more likely to attempt to expose their children to childhood diseases and NOT consider VPD a negative side effect of not vaxxing.  so if that is the only negative result of not vaxxing, it's not going to convince that group that vaccines are desirable.  

our family is in the Tuft's learning hospital system, so all our data is being collected by the most brilliant medical minds in New England.  and i take my non-vaxxing kids to all well-baby visits and am open and trust my dr who is also supportive of my choices.  i feel this is important to our family, and because i do love studies and research and want to help the modern medical world have a clear view of both the complimentary medical realm and the allopathic realm.  

[Rrrrrachel]

Thank you for participating in the research, Cassiani!  I'm glad you have a doctor you have such a good relationship with.  I think you're probably right that if the only negative effect of not vaccinating is more VPD many families won't have a problem with that (although I think some families that don't vaccinate don't think the vaccines actually work so don't consider themselves more likely to get sick, too), but I don't think the point is to convince non-vaccinating families to vaccinate or they'll get sick.

[AbbyGrant]

*

Edited by AbbyGrant - 6/28/12 at 9:20pm

[Rrrrrachel]

The problem with the homefirst data is that there's a lot of bias inherent in looking at data from just one practice like that.  Especially one that caters to a particular clientele. 

[HouseofPeace]

it's hard to not have bias- 

Most standard Dr's don't want unvaxxing kids in their practice.  Most practices in my area will not treat non-vaxxing families and they are forced to go homeopathic/osteopathic/chiropractic for their standard care practitioner.  This bias is definitely hurting the collection of data on non-vaxxing people by 'standard medicine'.  They even rate the effectiveness of different practices by % of vaccination instead of by effectiveness of treatment/healing.  it's easier and clearer data.  My doc happens to be young, curious, and uninvested in proving anything, and that has drawn a small group of questioning parents to him even though he is unique even in his practice.  That said, my friend whose son had some horrible vaccine reactions refuses to see him b/c she doesn't like him, he's too laid back and not alternative enough.  Many non-vaxxing/alternative schedule parents I know have walked away from most standard medicine in our area due to the negative treatment, judgement, threats of legal actions, and just disrespect given to them due to asking questions.

If modern medicine wants honest data collected by their practitioners, their practitioners need to be taught to be impartial at the least towards differing approaches to medicine.  Otherwise their bias affects the entire process as well.  

And yes- the group i was thinking of was Homefirst.  I've had many friends participate with that group and have only positive experiences.  Though not all of them were non-vaxxing.

[Rrrrrachel]

Sorry, not that kind of bias. It's going to have sampling bias because the individuals involved will tend to have certain things in common vs being a more broad cross section of the broader population.

[Rrrrrachel]

Re the other kind of bias, I'm not sure doctors should try and be unbiased towards alternative medicine which is unproven. I use various alternative medicine even though I know there's not solid science behind it because I feel like it works for me. I wouldn't want my doctor dispensing advise based on that, though.