Please tell me what you think of this study.

[Weak Kitten]

Unless you want to cite some specific examples re: the bolded portion, that has not been my experience. *snip*

Oh, "biased" probably isn't quite the right word. Possibly "distrustful" or even "aggravated". Any research that comes out that seems to even hint at the anti-vaxers and the heavy metal purge sellers having a point (not that I think they do) is met with rather aggressive skepticism. Papers are looked over with much greater diligence and the possible connections of the testers themselves are pursued. When someone who is unaware of this preexisting distrust brings a paper or study up they often seem shocked at the seeming hostility of the reaction.

I know this is just a forum and not a scientific community but people are people and I sometimes worry that similar behavioral patterns show up elsewhere. I'd hate for a piece of legitimate research to miss out on the attention it deserves because the bad research that came before tainted the well.

 

[Estellea]

Aggravated yes and yes to additional scrutiny. I think it's fair game to vet the authors (in any genre for that matter) as with various slants of autism research, there are some very dubious characters with a financial and personal gain. I haven't seen any legitimate autism research dismissed by non anti-vaxxers simply because it was written by a certain author or group or funded by a vested party. It's always the other way around though. The closest example that comes to mind for what you seem to be talking about would be the Burbacher et al. monkey/thiomersal study. But the criticisms were more about the abuse of the study by the mercury militia as opposed to the actual study, which was decent.

Given what anti-vaxxers have done in terms of hindering legitimate autism research to pursue bad hypotheses when their charges have been answered and they just keep shifting the goal-posts, it's no small wonder that these crap studies are met with hostility. I'm guilty I admit.

Este

 

[jdc324]

A lot of studies have been performed. Results all over the place depending upon the methodologies and biased investigators. I think the question is a good one; it just needs to be tested more rigorously.

Este

There's another small study here: http://www.ncbi.nlm.nih.gov/pubmed/21503799 The interesting bit for me is that they say they conducted a meta-analysis.

A meta-analysis including the present and previous similar studies excluded any association of autism with hair concentrations of mercury, cadmium, selenium, lithium and copper. A slight association was found for lead only, but it was very weak, as strictly dependent on the worst data from one study.

 

[marplots]

There's another small study here: http://www.ncbi.nlm.nih.gov/pubmed/21503799 The interesting bit for me is that they say they conducted a meta-analysis.

How many new studies does it take to overcome a meta-study? I guess I'm wondering about the self correcting nature of science. Seems like it's tough to turn the barge once it starts to fill up with studies. This would be troubling if the thing being studied had a time-varying element, so that the year-before-last's study wasn't examining the same phenomenon as this year's data.

 

[Weak Kitten]

Thanks Estellea, I'm glad to hear that my concerns are mainly unjustified. I guess when you listen to the loonies screaming about "scientists aren't listening" for long enough you start to worry that maybe there is something that is being missed.

It's sort of the Big Lie thing from WWII. You know, "people will believe a big lie sooner than a little one; and if you repeat it frequently enough people will sooner or later believe it." Even if you know that quote you can find yourself falling for the "big lie" as it slowly erodes your trust in those who are telling the truth.

 

[fuelair]

Thanks Estellea, I'm glad to hear that my concerns are mainly unjustified. I guess when you listen to the loonies screaming about "scientists aren't listening" for long enough you start to worry that maybe there is something that is being missed.

It's sort of the Big Lie thing from WWII. You know, "people will believe a big lie sooner than a little one; and if you repeat it frequently enough people will sooner or later believe it." Even if you know that quote you can find yourself falling for the "big lie" as it slowly erodes your trust in those who are telling the truth.

But, equally do not forget that sometimes - and way more often than it should be - scientists lie, report false information, make up data, exaggerate the importance of some data at the expense of other data, etc. Fortunately, due to the nature of science they often, though not always, get caught at it. Another reason I love Derek's blogs...............

 

[Acleron]

I only gave the paper a brief read and some things stood out to me. First, the participants were recruited by Autism Society of Greater Phoenix which is part of the umbrella organisation of Autism Society of America. Both are heavily invested in a vaccine causation and still cling to the belief that thimerosal causes autism. They also heavily promote "biomed" for "treating" autism which includes chelation based upon a challenge test which is not acceptable.

The authors also used some dubious labs for their metrics, DoctorsData and Vitamin Diagnostics which cater to so called 'autism doctors' and perform non-standardised tests which are then used to justify rather awful treatments. The authors also got pretty creative with their statistical analyses not only with some of their tests but also their own strata of statistically significant.

The results were kind of all over the place and they didn't control at all for medications and diets that the children were on. What they found for some metrics is not conclusive for causality but rather correlative which could very well be explained by neurobehaviours as many ASD children have sensory and texture issues that not only affect their dietary preferences but also effect inappropriate ingestion of non-food materials.

The study was funded in part by Autism Research Institute which also believes that autism is caused by vaccines and can be "cured". They support doctors who are not qualified in ASDs to treat ASD children. They have a lot of money behind them and there is no excuse for them to be funding such crap studies when they have actually funded some good ones.

Este

The statistics also appeared odd to me.

They were testing for a disclosed 29 biochemical markers, many metals and at least some amino acids in two-three blood compartments. It would be unusual not to find some correlations and significance. They mention cutting the significance when comparing groups by dividing by the number of groups, this seems wrong but it's too long since I did much. Oh for Linda to return.

 

[Deetee]

Here's the study.

The one which is cited from the OP is this one.
http://link.springer.com/article/10.1007/s12011-012-9551-1

How are they different?

How can the authors get away with this multiple publication of the same study?
Is this not tantamount to scientific fraud/fiddling, and should they not be reported?

 

[CapelDodger]

I know this is just a forum and not a scientific community but people are people and I sometimes worry that similar behavioral patterns show up elsewhere. I'd hate for a piece of legitimate research to miss out on the attention it deserves because the bad research that came before tainted the well.

When a subject is controversial beyond the scientific world you can be sure that any evidence presented, on either side, will not pass unremarked or unscrutinised. This may well attract confrontational personalities to the subject while discouraging others, but without data that remains purely speculative.

 

[Estellea]

The one which is cited from the OP is this one.
http://link.springer.com/article/10.1007/s12011-012-9551-1

How are they different?

How can the authors get away with this multiple publication of the same study?
Is this not tantamount to scientific fraud/fiddling, and should they not be reported?

DeeTee, they are the same group of investigators with the same cohort of children but using different tests. One was for nutritional/metabolic markers, the other for metals. You probably know that it isn't uncommon for investigators to generate multiple publications using different parameters on the same study groups. There is a term for it but it escapes me.

The study you link to is even worse in terms of over-reaching claims and 'statistical significance'.

Este

 

[Mojo]

They were testing for a disclosed 29 biochemical markers, many metals and at least some amino acids in two-three blood compartments. It would be unusual not to find some correlations and significance.

http://xkcd.com/882/

 

[jli]

The authors also used some dubious labs for their metrics, DoctorsData and Vitamin Diagnostics (edit to add: Know called Health Diagnostics)which cater to so called 'autism doctors' and perform non-standardised tests which are then used to justify rather awful treatments. The authors also got pretty creative with their statistical analyses not only with some of their tests but also their own strata of statistically significant.

Not only did the authors use these labs. Some of the authors are from these labs. Other authors are from
Integrative Developmental pediatrics and Southwest College of Naturopathic Medicine.

BTW. I didn't spot any attempts at blinding the samples.

 

[pakeha]

I only read the abstract and looked at a few of the references cited. The study seems to show that children with autism are less able to methylate and remove some toxic metals when compared to non-autistic children and this is correlated with the degree of autism seen. The implication is that autism has a metabolic component.

Note that the level of metals is predictive, even when cadmium (something found in lower amounts in autistic children) is considered. What I'd like to see is if there is an altered metabolic pathway that both clears cadmium better and doesn't clear the other metals listed as well.

Interestingly, it's the sort of answer I'd naively expect if autism has a genetic component which changes some important metabolic biology primarily and mentation secondarily.

As far as I can tell, the paper isn't identifying these metals as causative for autism, but markers or co-factors with an underlying, unidentified cause. But, I'd want to read the paper to do other than guess. Also, please understand I am not well read in autism generally.

I only gave the paper a brief read and some things stood out to me. First, the participants were recruited by Autism Society of Greater Phoenix which is part of the umbrella organisation of Autism Society of America. Both are heavily invested in a vaccine causation and still cling to the belief that thimerosal causes autism. They also heavily promote "biomed" for "treating" autism which includes chelation based upon a challenge test which is not acceptable.

The authors also used some dubious labs for their metrics, DoctorsData and Vitamin Diagnostics which cater to so called 'autism doctors' and perform non-standardised tests which are then used to justify rather awful treatments. The authors also got pretty creative with their statistical analyses not only with some of their tests but also their own strata of statistically significant.

The results were kind of all over the place and they didn't control at all for medications and diets that the children were on. What they found for some metrics is not conclusive for causality but rather correlative which could very well be explained by neurobehaviours as many ASD children have sensory and texture issues that not only affect their dietary preferences but also effect inappropriate ingestion of non-food materials.

The study was funded in part by Autism Research Institute which also believes that autism is caused by vaccines and can be "cured". They support doctors who are not qualified in ASDs to treat ASD children. They have a lot of money behind them and there is no excuse for them to be funding such crap studies when they have actually funded some good ones.

Este

This post stood out to me, but in general the thread is an example how good it gets when professionals look over a study.
Bookmarked.

 

[jdc324]

How many new studies does it take to overcome a meta-study? I guess I'm wondering about the self correcting nature of science. Seems like it's tough to turn the barge once it starts to fill up with studies. This would be troubling if the thing being studied had a time-varying element, so that the year-before-last's study wasn't examining the same phenomenon as this year's data.

I'm afraid I'm not knowledgeable enough to answer that question. I suppose it might depend on the size and quality of the new studies rather than just the number of new studies.

There were a couple of thoughts I had when reading your post though. One was whether a cumulative meta analysis could be used in this situation (or perhaps if there were new studies that were bigger and better, a new meta analysis could be conducted that used only those studies that met certain pre-defined criteria re size, methodology etc). The other was that you'd need to have a good basis for suspecting that the thing being studied had a time-varying element. Could you give an example of this being an issue? I'm trying to think of one, and can't off the top-of-my-head.

Cheers.

 

[marplots]

I'm afraid I'm not knowledgeable enough to answer that question. I suppose it might depend on the size and quality of the new studies rather than just the number of new studies.

There were a couple of thoughts I had when reading your post though. One was whether a cumulative meta analysis could be used in this situation (or perhaps if there were new studies that were bigger and better, a new meta analysis could be conducted that used only those studies that met certain pre-defined criteria re size, methodology etc). The other was that you'd need to have a good basis for suspecting that the thing being studied had a time-varying element. Could you give an example of this being an issue? I'm trying to think of one, and can't off the top-of-my-head.

Cheers.

Well, I can make something up.

Suppose autism is influenced by environmental lead, both in mothers making babies and then later, in babies themselves.

We quit using leaded gasoline in the US on January 1st, 1996 (EPA: http://yosemite.epa.gov/R10/airpage.nsf/webpage/Leaded+Gas+Phaseout). I'll also have to assume that lead levels related to roads falls sometime later on.

In any case, moms would have been exposed and children born during this period would have been exposed, but the levels gradually fall off. Studies done in the 1990's (and before and a bit after) would have this environmental issue that would gradually fall away.

Other things that might change are behaviors (percent of smokers falling off, less exposure to second hand smoke as a result of changes in the law).

Another, bigger change can come when we change instrumentation or diagnostic criteria, but that should be more visible, while other time dependent factors might be hidden.

 

[Estellea]

Not only did the authors use these labs. Some of the authors are from these labs. Other authors are from
Integrative Developmental pediatrics and Southwest College of Naturopathic Medicine.

BTW. I didn't spot any attempts at blinding the samples.

I didn't look closely at the authors' affiliations although I did see "support" from the Naturopathic College and no mention of blinding. Good grief, talk about the fox guarding the henhouse.

And thanks for your kind words pakeha.

Este

 

[jdc324]

Well, I can make something up.

Suppose autism is influenced by environmental lead, both in mothers making babies and then later, in babies themselves.

We quit using leaded gasoline in the US on January 1st, 1996 (EPA: http://yosemite.epa.gov/R10/airpage.nsf/webpage/Leaded+Gas+Phaseout). I'll also have to assume that lead levels related to roads falls sometime later on.

In any case, moms would have been exposed and children born during this period would have been exposed, but the levels gradually fall off. Studies done in the 1990's (and before and a bit after) would have this environmental issue that would gradually fall away.

Other things that might change are behaviors (percent of smokers falling off, less exposure to second hand smoke as a result of changes in the law).

Another, bigger change can come when we change instrumentation or diagnostic criteria, but that should be more visible, while other time dependent factors might be hidden.

If that were the case I suppose you'd expect to see a reduction in incidence of autism following the removal of leaded petrol / lead-based paint from general use (albeit with some time lag). Although depending on the association between lead and autism it might not be spotted if there were major changes in diagnostic criteria around the time that lead levels dropped below a certain level (i.e. if lead exposure was one risk factor of many, the continued existence of other risk factors + change in diagnostic criteria might mask it). I guess if a risk factor makes no visible difference then it's maybe not such an important risk factor.

I'm probably talking utter nonsense here but I guess I'm in the right place to do that (i.e. it's perhaps more easily spotted here than elsewhere).

 

[elipse]

The authors also used some dubious labs for their metrics, DoctorsData and Vitamin Diagnostics which cater to so called 'autism doctors' and perform non-standardised tests which are then used to justify rather awful treatments. The authors also got pretty creative with their statistical analyses not only with some of their tests but also their own strata of statistically significant.



Este

Isn't DoctorsData the place that's suing Stephen Barrett from Quackwatch? That doesn't inspire much confidence.

 

[Estellea]

If that were the case I suppose you'd expect to see a reduction in incidence of autism following the removal of leaded petrol / lead-based paint from general use (albeit with some time lag).

Exactly, just like the argument for thimerosal that the anti-vaxxers refuse to let go.

Although depending on the association between lead and autism it might not be spotted if there were major changes in diagnostic criteria around the time that lead levels dropped below a certain level (i.e. if lead exposure was one risk factor of many, the continued existence of other risk factors + change in diagnostic criteria might mask it). I guess if a risk factor makes no visible difference then it's maybe not such an important risk factor.

Perhaps but doubtful; you would have to have different risk factors emerging at the same time that leaded petrol was phased out but more importantly and lends to the unlikelyhood that leaded petrol was ever a significant contributor to ASDs, where was the dramatic increase in ASD cases when leaded petrol was at its height of use? I see what you are saying though with diagnostic criteria and probably not much of a risk factor to begin with even if it was contributing to ASD prevalence.

I'm probably talking utter nonsense here but I guess I'm in the right place to do that (i.e. it's perhaps more easily spotted here than elsewhere).

Not nonsense at all; it's a logical train of thought.

Isn't DoctorsData the place that's suing Stephen Barrett from Quackwatch? That doesn't inspire much confidence.

Yes, the very one. They are bottom-feeders of the bogus testing genre.

Este

 

[Dancing David]

I'm afraid I'm not knowledgeable enough to answer that question. I suppose it might depend on the size and quality of the new studies rather than just the number of new studies.

There were a couple of thoughts I had when reading your post though. One was whether a cumulative meta analysis could be used in this situation (or perhaps if there were new studies that were bigger and better, a new meta analysis could be conducted that used only those studies that met certain pre-defined criteria re size, methodology etc). The other was that you'd need to have a good basis for suspecting that the thing being studied had a time-varying element. Could you give an example of this being an issue? I'm trying to think of one, and can't off the top-of-my-head.

Cheers.

Meta analysis is never a good methodology, it depends upon the methodology of the studies under it, often they have different controls and protocols. Good meta analysis has large sample groups and demographic controls and rule outs.

Now it is good way of looking at what studies might show.