Can the MMR or other vac ever cause autism?

[StarlightMcKenzie]

www.ncbi.nlm.nih.gov/m/pubmed/25114790/

Excluding the 117 from the racial analysis is what Hooker did to get his headline figure. But even he realised that he had to make some show of trying to control for other factors so he also analysed the black boys who were a normal weight at birth. By looking at the birth certificate population. And guess what? He found the effect was reduced.
I wonder what had happened if he had controlled for other factors too?

If Hooker was so confused by the decision to analyse race data from the birth certificate population only, in order to control for a factor that was only available from birth certificate data, then why did he do the exact same thing in his study, and then describe this analysis as a strength of the study?

Yes, sorry, I should have specified that they were missing race information for 21 case children. I didn't misread the table - it is pretty easy to read!

I didn't mention control children because they didn't match controls and cases according to race. They matched them according to age and gender. They didn't need to ditch controls due to lack of information about race.

In the total sample, case and control children were matched appropriately on age and gender, with a preponderance of boys in both groups (Table 2). The racial distributions were also fairly similar, although a larger proportion of control (10%) than case (6%) children were classified as “other” race and both groups had an appreciable number for which race information was missing.

BeachComber, having looked at this further, i see that the same set of symptoms is called x before 36 months, and y afterwards.so semantics really.

Having read the thread this morning while wrangling dc, and the paper this evening while cooking dinner... can anyone clarify why the 2nd paper measures results to 31 months, but reports results as before/after 36 months? I haven't time to trace it all through.

For the reasons already described, especially the issue with sample sizes not being quoted, and also the major confounders that have not been adjusted for, the paper in question is absolutely useless as a source of evidence. Nothing can detract from this, so there is no point arguing about it....

Which is why the bloody paper has been removed by the journal.

I'm pretty sure that it is considered dodgy, if when analysing a bunch of people based on their race, that you simply bin some of your records which are missing race information, and keep others.

Even using race information is dodgy. FGS.

You guys all lost me a long time ago as to who is arguing what, but could I add a few factors re habits and culture and things like that in the USA that may be relevant to the conclusions you are making. I can't say that they are definitely relevant as I am totally lost about who is saying what.

Anyway....here in Los Angeles there is an incentive (it's long and complicated and completely inane) with regard to getting into the better state schools to identify your child as white vs. African American. Most mixed race couples that I know had not figured this out when their child was born, but had by the time they were the age to send them to school. So it's perfectly possible that someone might show a child as black on their birth certificate and white on a form filled out at age 3. In addition, the rules for vaccination requirements have changed over time and are enforced differently in different school districts. Also, the programs and/or ability to have autism diagnosed at an early age vary enormously from one area to the next and vary greatly depending on your health insurance which isn't necessarily linked to your race or economic status, but might be depending on where you live. All of this is to say that there are a lot of factors to adjust for and without an enormous sample size, I question any accuracy claimed in adjustments.

Did I mention that it's stupid in human terms to analyse data by race?
It's only useful to identify largely isolated and inbred groups, such as Ashkenazi Jews. But most ethnic groups are genetically mixed. It may be possible to identify genetic groups within a population (population structure), but that requires the study of genetic markers. Proper genetic markers, not very subjective phenotypes like skin colour, nose shape or hair type. As I mentioned, there can be greater differences between African genetic groups than between European groups.
In that sense both studies are flawed.
Treating data according to race is only useful in socio economic terms, as the (perceived as) Black population in the US (and not only) is still disadvantaged.
So, I'd be very very weary of suggesting any group specific effect (which would be genetic) of giving a vaccine (or any other intervention),

Then there is the timing. Without time information regarding diagnosis and vaccine (and reasons for late diagnoses) any association results only point to association (not cause-effect!). Cause-effects require crucial time information. (among other things)
Even assuming that this was not a statistical fluke (which could well be - science is full of these as we need to operate within margins of error).
Finally, the association found, given the time periods mentioned (age group and vaccine timing late, as well as ethnic group) strongly suggest that it's not cause-effect. Instead the reverse, diagnosis causing vaccine, is more likely. Not discounting that going for a vaccine could lead to a dianosis, because health professionals might flag it, while slightly less educated parents or less health alert parents might not have spotted certain signs.
I wouldn't have lost sleep over 10 years if I had discounted that (tiny and very flawed) result.

It is possible that vaccines could have an effect on some (very rare) susceptible individuals. This should be properly investigated, using proper genetic data in well designed genetic epidemiology studies.
I will point out, though, that for a vaccine to have an adverse effect, it's more likely that it would be due to the virus or bacteria content (I.e. the agent of the disease). As such, it's highly likely that an individual affected by the vaccine would also have an adverse serious reaction to the disease.
These are diseases that can kill or have serious effects. Any consequences of the vaccine are only likely to be smaller than the disease itself.

Are you noblegiraffe? Pretty sure it's dodgy? But that is exactly what Destefano et al did. Except they didn't bin 'some' of their records. They binned nearly half.

BTW, all of you complaining that Hooker doesn't reveal his data - has it occurred to you that it isn't his data. It is the CDC's and Hooker obtained it under a FOIA request. The data is protected and classified. Hooker was able to access it, but I'm not sure that he would be allowed to publish it. Just a thought for all of you jumping to conclusions about his competence and or honesty.

Human data must be treated carefully regarding identities.

Hooker would only have to show the sample sizes. Which everyone does. It's very strange that he doesn't.

No, Beachcomber, Destefano didn't bin some of the records which were missing race information but keep others according to whether they were autistic, which is what you were proposing.

And re 'binning records' being dreadful

"However, CDC researchers did not include any children that did not have a valid State of Georgia birth certificate – reducing the study’s sample size by 41% . Dr. Hooker explains that by introducing this discretionary criteria into the analysis, the cohort size was sharply reduced, eliminating what would have been a higher statistical finding. This minimized the strong MMR-autism link in African American boys. When asked if there could be any scientific basis for excluding children born outside of Georgia, Dr. Hooker responded, “I know of none, and none has been provided by the authors of the DeStefano study.”"

focusautisminc.org/focus-autism-releases-findings-on-2003-cdc-autism-study-higher-autism-rate-among-african-american-boys-receiving-mmr-shot-earlier-than-36-months/

From Hooker’s study: Through investigating the cohort demographic data, it was found that there was a higher proportion of low birth weight African Americans compared to the entire cohort, specifically within the portion of the cohort that possessed a Georgia state birth certificate. The number of individuals with a birth weight under 2500 g for African Americans was 11.9% as compared to the total cohort at 8.66%. A final analysis was completed on African American children in the “birth certificate” cohort, excluding low birth weight individuals (Table 4).

Did Hooker “can’t think of any scientific reason for excluding children born outside of Georgia blah blah cover-up blah fraud blah” just exclude children born outside of Georgia from his analysis? Yes, yes he did.

Further to this: A strength of the current study is.....Also the current study controlled for a possible association between low birth weight individuals and autism in the final analysis on the African American cohort, all children of birth weight less than 2500 g were eliminated from the cohort.

So it is a strength of his study to exclude children born outside of Georgia in his analysis, but completely inexplicable for DeStefano to do the same thing? And must be to hide results and commit fraud and run a racist experiment to give black boys autism?

I don't know how you could read his quote to the press and the quotes from his study and not conclude that he is a liar with agenda and not to be trusted.

here in Los Angeles there is an incentive (it's long and complicated and completely inane) with regard to getting into the better state schools to identify your child as white vs. African American

The stuff about not analysing on race based on physical features is interesting.

Can I just say that it's disgraceful that a country actually includes "race" in a birth certificate?

noblegiraffe...yes it's rather a black (no pun intended) humor that the african american and hispanic families spend a lot of time trying to find the white relative in their past. Or my friend who is caucasian but looks hispanic desperately trying to prove her white lineage to the school office when registering her daughter in middle school. Or my mixed race friends who didn't figure out the trick until their middle child, so have 2 caucasian children and 1 african american one even though they are all 3 full siblings. And then there is the white south african (and the algerian) who doesn't understand why he's not african-american.

Noblegiraffe, I'm beginning to wonder if you are really discussing this in good faith.

Earlier in the thread we talked about protocol. When you do a study you have a protocol. You design a study based on the hypothesis you want to test, you provide a protocol, you apply for ethical clearance if necessary, you submit requests for data, etc. All this is normal standard stuff. Basic rules.

Your protocol is decided before you gather and analyse data. You aren't allowed to gather data, analyse it and then go back and change your protocol to then give your data analysis more or less significance according to the results that you want to highlight or bury. That is not allowed and I'm sure you can see why.

But this is what Thompson said they did. He says they had a protocol, they gathered and analysed data and then they changed their protocol because the data threw up something 'problematic'. And the change they made (according to Thompson) was to bin the non birth certificate group in order to dilute findings that had statistical significance by reducing numbers.

I regret that my coauthors and I omitted statistically significant information in our 2004 article published in the journal Pediatrics. The omitted data suggested that African American males who received the MMR vaccine before age 36 months were at increased risk for autism. Decisions were made regarding which findings to report after the data were collected, and I believe that the final study protocol was not followed.

We have no reason to think that Hooker didn't follow his protocol. There is no reason for him to not run a separate analysis on the 'birth certificate cohort' - as long as it was in his protocol.

The problem isn't having two cohorts - one with birth certs and one without. The problem is not running different analysis for these two groups. The problem is breaking your protocol in order to allow you to publish a conclusion that would not have been supported by the data should you have stuck to the protocol and the study design.

What do you think of these documents?

letter from Thompson to Gerberding

email from Thompson to Wharton

I mean I guess they may be fake but we cannot assume so. Especially considering Thompson's statement.

Why is he talking about 'problematic results' and 'sensitive results' and 'sensitive legal issues'. Why is he so concerned that he is thinking of withdrawing his name from the draft manuscript?

I can see nothing 'sensitive' or 'problematic' in the DeStefano paper. Quite the opposite.

Lweji, Hooker states the following about the data;

Cohort data were obtained directly as a “restricted access data set” from the Centers for Disease Control and Prevention (CDC) via a Data Use Agreement. Data were deidentified by the CDC in accordance with Family Education Rights and Privacy Act (FERPA) and the Health Insurance Portability and Accountability Act (HIPAA) prior to receipt by the study authors. Use of the CDC specifically for the study described herein was approved by the Simpson University Institutional Review Board, in accordance with U.S. Federal regulations.

I don't think we can assume that the decision to not publish the actual numbers came from Hooker. And IMO assuming without actual having the information that it was his decision and then saying that this makes him dishonest/a liar/incompetent suggests that one is not being entirely open minded nor honest oneself.

I don't know what a 'restricted access data set' is or how the Date Use Agreement works. I suspect that no-one on this thread does. Considering that the data was classified and it took a FOIA request and help from congress members for Hooker to get it, I wouldn't be entirely surprised if he was not allowed to publish it or if the journal wasn't or didn't want to. Who knows? I think assuming the lack of published numbers makes Hooker a bad 'un is a bizarre way to look at all this and suggests a strong bias frankly. The data should have been available to the public in the first place - Hooker shouldn't have needed a FOIA request to access it.

Of course you're allowed to change your protocol if you hit a problem. (Like not having enough subjects in a group for your data to make sense).

You might have to clear it with ethics again, but you can control for whatever you want, as long as you describe what you've done in the write-up.

As long as you report the results accurately, you can write whatever crap you want in the conclusion, because it's basically your opinion. The idea is that you give your results, and other people can look at them and form their own opinion.

Which is why people are getting twitchy about the paper which omits the number of subjects from its results section. Not telling something that's so vital to the analysis is really odd.

Also, I've just read the full text of the original Pediatrics article, and they do describe how they controlled for the factors mentioned. No hidden data. No conspiracy. Sorry.

"I don't think we can assume that the decision to not publish the actual numbers came from Hooker."

Sorry, missed this - in order to run the statistical analysis that he ran, he had to have the numbers in each group, it's really really vital because the difference you need to see between your groups to claim 'statistical significance' depends on how many samples you have.

That's why we are harping on about something that I can see appears a bit irrelevant if you haven't ever had to go through the whole tedious process.

Yes, Boulevard. What I'm saying you can't do is change your protocol because your data throws up a result that you don't want to be transparent about.

Thompson didn't say that they needed to change their protocol because they hit a problem such as not having enough data. He says that they had the data and that they omitted it on purpose.

Also, I've just read the full text of the original Pediatrics article, and they do describe how they controlled for the factors mentioned. No hidden data.

Can you post a quote from the paper where they explain that they had to drop the non birth certificate group for analysis on race? I've read the paper several times and I can't find anything like that.

Lweji you keep saying that it is stupid to analyse by race. If you were looking at how people make vitamin D or how vitamin D deficiency comes about and affects people then the skin colour of your subjects would be entirely relevant. Or if you were examining skin cancer incidence in different populations.

There will obviously be different legislation covering ethics in the US as oppose to the UK but here our ethics approval concerned the source of patients we could recruit and the information we were allowed to ask them (this was a study on a different neurological condition rather than autism). Ethics approval had to be re-sought if we wanted to include patients from a different source or if extra information was asked. We did not have to re-seek ethics approval whenever we had a change to how we analysed the data e.g. for instance if we wanted to compare one bit of the epidemiological data to the frequency of a particular gene allele in the population or sub-population. I guess it depends on how the ethics is worded but most I imagine will be written quite vaguely to allow this as it is a very time-consuming process to get ethics here.

Personally I would be unhappy if this level of information was available about my ds's without my consent. I appreciate names were not give but if you know school, age and diagnosis and the children are in mainstream then they are identifiable. Certainly, I would not want it in the public domain unless identifiers were well hidden.

And I understand the importance of sample sizes.

The point I am making is that I don't think we can jump to the conclusion that Hooker has decided not to publish sample sizes because he is an incompetent liar. Of course he had access to the data in order to do his analysis.

I'm just suggesting that rather than jump to the conclusion that he is hiding something, we should consider that the CDC may have released the data to him as a result of his FOIA request, but not given him the authority to publish it.

As I say, I don't know. I don't think anyone else does either. I prefer to keep an open mind. If it turns out that it is entirely down to Hooker that the sample sizes are not in his paper then I agree that that is remiss.

scroll a bit for

Upon request for more a more specific response to Dr. Hooker’s accusations, I received the following detailed reply from Belsie González, Senior Public Affairs Specialist at the CDC:

“As this topic was so sensitive and complex, the CDC study published in Pediatrics in February 2004 underwent clearance at CDC, the usual process of internal review for scientific accuracy that all CDC papers undergo. In addition, before submission to the journal, the manuscript was reviewed by five experts outside of CDC and an independent CDC statistician (see acknowledgements section of the paper for specific names). Finally, all reputable journals undergo peer-review of all submitted papers before final publication.

The 2004 CDC study was designed as a case-control study. This means, children with autism (cases) were specifically identified, and children without autism (controls) were identified to be similar to the children with autism in other respects. When data are collected in a specific way for a specific type of statistical analysis (a case-control study in this instance), using those data in a different type of analysis can produce confusing results. Because the methods in Dr. Hooker’s reanalysis were not described in detail, it is hard to speculate why his results differed from CDC’s. We would be interested to learn more about how he analyzed the data to reach the conclusions made in his study.”

In short, not only did experts outside of the CDC verify the original study, but Dr. Hooker’s new paper likely reaches the findings it does because he uses the wrong kind of analysis for the data set.

and yes, Hooker is certainly incompetent - it is not clear why though, when he had Thompson to consult..

Lweji you keep saying that it is stupid to analyse by race. If you were looking at how people make vitamin D or how vitamin D deficiency comes about and affects people then the skin colour of your subjects would be entirely relevant. Or if you were examining skin cancer incidence in different populations.

But that is not by "race" but about one specific phenotype that is skin colour.
I know a few "white" people with darker skins that people that could be classed as "black" and there is quite an overlap between the two. There are also other groups of people that have dark skin.

To study autism (again, a disease that is not likely to be linked to skin colour, but has so far been shown to have a fairly high genetic component - quick google so from Wikipedia: "Early studies of twins estimated the heritability of autism to be more than 90%; in other words, that 90% of the differences between autistic and non-autistic individuals is due to genetic effects") "race" does not go much into it and any study looking at populations or specific groups should be backed up with a decent genetic characterization of the individuals.

As for publishing, I can't see why Hooker wouldn't be able to actually say what his population samples were if he managed to say what the p values were. It's a basic publishable result.